[AArch64] Fix name of macros called in the vdup_lane Neon intrinsics
[official-gcc.git] / gcc / ChangeLog
blob34d18b8fd41b895a4d75cd832811a0cd1b993bee
1 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
3         * config/aarch64/arm_none.h
4         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
6 2013-08-12  Nick Clifton  <nickc@redhat.com>
8         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
10 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
12         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
13         expand for QI/HImode operand to produce more effictive code for
14         unsigned char(short) --> float(double) conversion.
16 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
18         * doc/invoke.texi: Mention that -ftls-model does not force the final
19         model.
21 2013-08-12  Marek Polacek  <polacek@redhat.com>
22             Marc Glisse  <marc.glisse@inria.fr>
24         PR tree-optimization/57980
25         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
26         when creating -1 constant.
28 2013-08-10  Jan Hubicka  <jh@suse.cz>
30         Workaround binutils PR14342.
31         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
32         (gimple_init_edge_profiler): Likewise.
33         (gimple_gen_ic_func_profiler): Likewise.
35 2013-08-09  Jan Hubicka  <jh@suse.cz>
37         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
39 2013-08-09  Xinliang David Li  <davidxl@google.com>
41         * config/i386/stringop.def: New file.
42         * config/i386/stringop.opt: New file.
43         * config/i386/i386-opts.h: Include stringopt.def.
44         * config/i386/i386.opt: Include stringopt.opt.
45         * config/i386/i386.c (ix86_option_override_internal):
46         Override default size based stringop inline strategies
47         with options.
48         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
49         New function.
51 2013-08-09  Jan Hubicka  <jh@suse.cz>
53          * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
55 2013-08-09  Jan Hubicka  <jh@suse.cz>
57         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
58         CGRAPH_FREQ_MAX.
59         (dump_cgraph_node): Dump profile-id.
60         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
61         and common_target_probability.
62         * lto-cgraph.c (lto_output_edge): Stream common targets.
63         (lto_output_node): Stream profile ids.
64         (input_node): Stream profile ids.
65         (input_edge): Stream common targets.
66         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
67         * ipa.c: Include value-prof.h
68         (ipa_profile_generate_summary): Turn indirect call statement histograms
69         into common targets.
70         (ipa_profile): Turn common targets into speculative edges.
72 2013-08-09  Jan Hubicka  <jh@suse.cz>
74         * cgraph.h (cgraph_node): Add profile_id.
75         * value-prof.c (cgraph_node_map): Turn into pointer_map.
76         (init_node_map): Rewrite to handle hashes increas of incremental
77         IDs.
78         (del_node_map): Update.
79         (find_func_by_funcdef_no): Replace by ...
80         (find_func_by_profile_id): ... this one.
81         (gimple_ic_transform): Do not remove useful histograms when
82         speculation is not done; dump info when indirect call removal
83         can happen at LTO.
84         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
85         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
86         (__gcov_indirect_call_profiler_v2): .. this one.
87         * profile.h (init_node_map): Update.
88         * coverage.c (coverage_compute_profile_id): New function.
89         * coverage.h (coverage_compute_profile_id): Declare.
90         * tree-profile.c (init_ic_make_global_vars): Make
91         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
92         (gimple_init_edge_profiler): Update prototype of
93         __gcov_indirect_call_profiler.
94         (gimple_gen_ic_func_profiler): Simplify.
95         (tree_profiling): Use init_node_map
97 2013-08-09  Jan Hubicka  <jh@suse.cz>
99         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only non-speculative
100         refs.
101         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
102         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
103         (cgraph_set_call_stmt): Likewise.
104         (cgraph_create_edge_1): Fix release checking compilatoin;
105         clear lto_stmt_uid.
106         (cgraph_free_edge): Free indirect info.
107         (cgraph_turn_edge_to_speculative): New function.
108         (cgraph_speculative_call_info): New function.
109         (cgraph_make_edge_direct): Return direct edge; handle speculation.
110         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative
111         edges.
112         (dump_cgraph_node): Dump speculation.
113         (verify_edge_count_and_frequency): Accept speculative edges.
114         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
115         (verify_cgraph_node): Handle speculation.
116         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
117         (cgraph_set_call_stmt): Update prototype.
118         (cgraph_make_edge_direct): Update prototype.
119         (cgraph_speculative_call_info): Declare.
120         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
121         to change; update call of ipa_find_references.
122         * ipa-ref.c (ipa_record_reference): Fix return value; clear
123         lto_stmt_uid and speculative flags.
124         (ipa_dump_references): Dump speculation.
125         (ipa_clone_references): Clone speculative flag.
126         (ipa_clone_referring): Likewise.
127         (ipa_clone_ref): New function.
128         (ipa_find_reference): Look into lto_stmt_uids
129         (ipa_clear_stmts_in_references): Do not clear speculative calls.
130         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
131         (ipa_find_reference): Update declaration.
132         (ipa_clone_ref): Declare.
133         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
134         stream speculative flag.
135         (lto_output_ref): Stream statements uids and speculation.
136         (input_ref): Likewise.
137         (input_edge): Stream speuclation.
138         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
139         (cgraph_set_call_stmt_including_clones): Handle speculation.
140         * ipa-inline.c (heap_edge_removal_hook): New function.
141         (inline_small_functions): Register it.
142         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
143         also initialize refs.
144         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
145         edge to change.
146         (try_make_edge_direct_simple_call): Likewise.
147         (try_make_edge_direct_simple_call): Likewise.
148         (update_indirect_edges_after_inlining): Likewise.
149         (remove_described_reference): Look proper lto_stmt_uid.
150         (propagate_controlled_uses): Likewise.
151         (propagate_controlled_uses): Liekwise.
152         * tree-inline.c (copy_bb): Copy speculative edges.
153         (redirect_all_calls): New function.
154         (copy_cfg_body): Do redirection after loop info
155         is updated.
156         (delete_unreachable_blocks_update_callgraph): Updadte
157         speculation.
159 2013-08-09  Jan Hubicka  <jh@suse.cz>
161         * lto-streamer-out.c (output_function): Renumber PHIs.
162         * lto-streamer-in.c (input_function): Likewise.
164 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
166         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
167         (get_lane_unsigned): Likewise.
168         (dup_lane_scalar): Likewise.
169         (get_lane): enable for VALL.
170         * config/aarch64/aarch64-simd.md
171         (aarch64_dup_lane_scalar<mode>): Remove.
172         (aarch64_get_lane_signed<mode>): Likewise.
173         (aarch64_get_lane_unsigned<mode>): Likewise.
174         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
175         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
176         (aarch64_get_lane<mode>): Enable for all vector modes.
177         (aarch64_get_lanedi): Remove misleading constraints.
178         * config/aarch64/arm_neon.h
179         (__aarch64_vget_lane_any): Define.
180         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
181         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
182         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
183         * config/aarch64/iterators.md (VDQQH): New.
184         (VDQQHS): Likewise.
185         (vwcore): Likewise.
187 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
189         * configure.ac: Add GAS check for LEON instructions on SPARC.
190         * configure: Regenerate.
191         * config.in: Likewise.
192         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
193         sparc*-*-* block.
194         * config/sparc/sparc.opt (LEON, LEON3): New masks.
195         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
196         for LEON or LEON3.
197         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
198         (AS_LEON_FLAG): New macro.
199         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
200         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
201         Deal with LEON and LEON3 for the memory model.
202         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
203         (atomic_compare_and_swap<mode>_1): Likewise.
204         (*atomic_compare_and_swap<mode>_1): Likewise.
206 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
208         * config/arm/neon.md (vcond): Fix floating-point vector
209         comparisons against 0.
211 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
213         * lra-constraints.c (emit_spill_move): Remove assert.
214         (process_alt_operands): Add more debugging
215         output.  Increase reject for spilling into memory.  Decrease
216         reject for reloading scratch.
217         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
219 2013-08-08  Steve Ellcey  <sellcey@mips.com>
221         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
222         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
223         micromips incompatible.  Add nan2008.
224         (MULTILIB_DIRNAMES): Add nan2008.
225         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
226         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
227         and micromips incompatible.  Add nan2008.
228         (MULTILIB_DIRNAMES): Add nan2008.
229         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
231 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
233         PR rtl-optimization/58079
234         * combine.c (combine_simplify_rtx): Avoid using SUBST if
235         simplify_comparison has widened a comparison with an integer.
237 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
239         * config/arm/neon.md (movmisalign<mode>): Disable when we
240         don't allow unaligned accesses.
241         (*movmisalign<mode>_neon_store): Likewise.
242         (*movmisalign<mode>_neon_load): Likewise.
243         (*movmisalign<mode>_neon_store): Likewise.
244         (*movmisalign<mode>_neon_load): Likewise.
246 2013-08-08  Jan Hubicka  <jh@suse.cz>
248         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
249         (make_pass_rebuild_cgraph_edges): Also clear references.
250         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
251         * ipa-inline-transform.c (inline_transform): Remove all references
252         after inlining.
253         * cgraphunit.c (expand_function): Remove all references after expansion.
254         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
255         (ipa_find_reference): Rewrite to iterator.
256         (remove_stmt_references): Likewise.
257         (ipa_clear_stmts_in_references): New function.
258         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
259         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
260         clear references.
261         * ipa-split.c (split_function): Remove references in split function.
263 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
265         PR target/57431
266         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
267         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
269 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
271         PR target/56979
272         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
273         suggested mode for the assignment isn't compatible with the
274         registers required.
276 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
278         PR target/58065
279         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
281 2013-08-07  Xinliang David Li  <davidxl@google.com>
283         * config/i386/i386.opt: New option -mtune-ctrl=.
284         * config/i386/x86-tune.def: New file.
285         * config/i386/i386.h: include x86-tune.def.
286         * config/i386/i386.c (ix86_option_override_internal):
287         Parsing -mtune-ctrl= option and set tune features.
289 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
291         PR other/12081
292         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
293         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
294         to gen_2arg_fn_t.
296 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
298         * rtl.h (update_alignments): Declare.
299         * final.c (grow_label_align): New function extracted from...
300         (shorten_branches): ...here.  Call it.
301         (update_alignments): New function.
302         * reorg.c (sibling_labels): New variable.
303         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
304         the new label along with it onto the sibling_labels vector.
305         (fill_simple_delay_slots): Adjust call to get_label_before.
306         (fill_slots_from_thread): Likewise.
307         (relax_delay_slots): Likewise.
308         (make_return_insns): Likewise.
309         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
311 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
313         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
314         document its semantics.
315         (diagnostic_report_diagnostic): Adjust accordingly.
317 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
319         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
320         (sparc_option_override): ...and port to new C++ pass API.
321         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
323 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
325         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
327 2013-08-06  Caroline Tice  <cmtice@google.com>
329         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
330         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
331         * tree-pass.h: Add pass_vtable_verify.
332         * varasm.c (assemble_variable): Add code to properly set the comdat
333         section and name for the .vtable_map_vars section.
334         (assemble_vtyv_preinit_initializer): New function.
335         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
336         LINK_ONCE flag.
337         * output.h: Add function decl for assemble_vtv_preinit_initializer.
338         * vtable-verify.c: New file.
339         * vtable-verify.h: New file.
340         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
341         initialiation levels.
342         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
343         * passes.def: Insert pass_vtable_verify.
344         * aclocal.m4: Reorder includes.
345         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
346         -fvtv-counts options.
347         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
348         as appropriate, if -fvtable-verify=... is used.
349         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
350         -fvtable-verify=... is used.
351         * Makefile.in (OBJS):  Add vtable-verify.o to list.
352         (vtable-verify.o): Add new build rule.
353         (GTFILES): Add vtable-verify.c to list.
354         * common.opt (fvtable-verify=): New flag.
355         (vtv_priority): Values for fvtable-verify= flag.
356         (fvtv-counts): New flag.
357         (fvtv-debug): New flag.
358         * tree.h (save_vtable_map_decl): New extern function decl.
360 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
362         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
363         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
364         (pass_data_rl78_devirt): ...new pass_data instance and...
365         (make_pass_rl78_devirt): ...new function.
366         (rl78_asm_file_start): Port pass registration to new C++ API.
368 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
370         * coretypes.h (rtl_opt_pass): Add.
371         (gcc::context): Add.
372         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
373         (epiphany_init): Port to new C++ pass API.
374         (epiphany_optimize_mode_switching): Likewise.
375         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
376         (pass_manager::get_pass_mode_switching): New.
377         (pass_manager::get_pass_peephole2): New.
378         * mode-switching.c (pass_mode_switching): Add clone method.
379         * recog.c (pass_peephole2): Add clone method.
380         (pass_split_all_insns): Add clone method.
382 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
384         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
385         (mips_option_override): ...here, porting to new C++ API for passes.
387 2013-08-06  Jan Hubicka  <jh@suse.cz>
389         * cgraph.c (cgraph_get_body): New function based on lto.c
390         implementation.
391         * cgraph.h (cgraph_get_body): Declare.
392         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
393         LTO paths.
394         * cgraphunit.c (expand_function): Get body prior expanding.
395         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
396         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
397         really need.
398         * passes.c (do_per_function_toporder): Get body.
399         * tree-inline.c (expand_call_inline): Get body prior inlining it.
400         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
402 2013-08-06  Martin Jambor  <mjambor@suse.cz>
404         PR fortran/57987
405         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
406         is not re-finalized.  Rename second parameter to no_collect.
408 2013-08-06  Martin Jambor  <mjambor@suse.cz>
410         PR middle-end/58041
411         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
412         MEM_REF has proper alignment information.
414 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
416         PR other/12081
417         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
418         class insn_gen_fn.
419         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
420         rtx (*) (rtx, ...) with insn_gen_fn.
421         * genoutput.c (output_insn_data): Cast gen_? function pointers to
422         insn_gen_fn::stored_funcptr.  Add initializer braces.
424 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
426         Rewrite how instances of passes are cloned to remove assumptions
427         about their sizes (thus allowing pass subclasses to have
428         additional data fields, albeit non-GC-managed ones at this point).
430         * passes.c (make_pass_instance): Now that passes have clone
431         methods, rewrite this function to eliminate XNEW and memcpy
432         calls that used hardcoded sizes.  Since this function no longer
433         creates pass instances, rename it to...
434         (add_pass_instance): ...this.  Document the old way that passes were
435         numbered and flagged, and rework this function to continue using it.
436         (next_pass_1): Add an initial_pass argument for use by
437         add_pass_instance.
438         (position_pass): When adding multiple instances of a pass, use
439         the pass's clone method, rather than relying on the XNEW/memcpy
440         within the former make_pass_instance (now add_pass_instance).
441         (pass_manager::pass_manager): When invoking next_pass_1, also supply
442         the initial instance of the current pass within the pass manager.
444 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
446         This is the automated part of the conversion of passes from C
447         structs to C++ classes.
449         Patch autogenerated by refactor_passes.py from
450         https://github.com/davidmalcolm/gcc-refactoring-scripts
451         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
453         * asan.c (pass_asan): Convert from a global struct to a subclass of
454         gimple_opt_pass along with...
455         (pass_data_asan): ...new pass_data instance and...
456         (make_pass_asan): ...new function.
457         (pass_asan_O0): Convert from a global struct to a subclass of
458         gimple_opt_pass along with...
459         (pass_data_asan_O0): ...new pass_data instance and...
460         (make_pass_asan_O0): ...new function.
461         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
462         subclass of rtl_opt_pass along with...
463         (pass_data_inc_dec): ...new pass_data instance and...
464         (make_pass_inc_dec): ...new function.
465         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
466         a subclass of rtl_opt_pass along with...
467         (pass_data_reorder_blocks): ...new pass_data instance and...
468         (make_pass_reorder_blocks): ...new function.
469         (pass_duplicate_computed_gotos): Convert from a global struct to a
470         subclass of rtl_opt_pass along with...
471         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
472         (make_pass_duplicate_computed_gotos): ...new function.
473         (pass_partition_blocks): Convert from a global struct to a subclass of
474         rtl_opt_pass along with...
475         (pass_data_partition_blocks): ...new pass_data instance and...
476         (make_pass_partition_blocks): ...new function.
477         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
478         struct to a subclass of rtl_opt_pass along with...
479         (pass_data_branch_target_load_optimize1): ...new pass_data instance
480         and...
481         (make_pass_branch_target_load_optimize1): ...new function.
482         (pass_branch_target_load_optimize2): Convert from a global struct to a
483         subclass of rtl_opt_pass along with...
484         (pass_data_branch_target_load_optimize2): ...new pass_data instance
485         and...
486         (make_pass_branch_target_load_optimize2): ...new function.
487         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
488         of rtl_opt_pass along with...
489         (pass_data_jump): ...new pass_data instance and...
490         (make_pass_jump): ...new function.
491         (pass_jump2): Convert from a global struct to a subclass of
492         rtl_opt_pass along with...
493         (pass_data_jump2): ...new pass_data instance and...
494         (make_pass_jump2): ...new function.
495         * cfgexpand.c (pass_expand): Convert from a global struct to a
496         subclass of rtl_opt_pass along with...
497         (pass_data_expand): ...new pass_data instance and...
498         (make_pass_expand): ...new function.
499         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
500         of rtl_opt_pass along with...
501         (pass_data_free_cfg): ...new pass_data instance and...
502         (make_pass_free_cfg): ...new function.
503         (pass_into_cfg_layout_mode): Convert from a global struct to a
504         subclass of rtl_opt_pass along with...
505         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
506         (make_pass_into_cfg_layout_mode): ...new function.
507         (pass_outof_cfg_layout_mode): Convert from a global struct to a
508         subclass of rtl_opt_pass along with...
509         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
510         (make_pass_outof_cfg_layout_mode): ...new function.
511         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
512         struct to a subclass of gimple_opt_pass along with...
513         (pass_data_build_cgraph_edges): ...new pass_data instance and...
514         (make_pass_build_cgraph_edges): ...new function.
515         (pass_rebuild_cgraph_edges): Convert from a global struct to a
516         subclass of gimple_opt_pass along with...
517         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
518         (make_pass_rebuild_cgraph_edges): ...new function.
519         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
520         subclass of gimple_opt_pass along with...
521         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
522         and...
523         (make_pass_remove_cgraph_callee_edges): ...new function.
524         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
525         struct to a subclass of rtl_opt_pass along with...
526         (pass_data_stack_adjustments): ...new pass_data instance and...
527         (make_pass_stack_adjustments): ...new function.
528         * combine.c (pass_combine): Convert from a global struct to a subclass
529         of rtl_opt_pass along with...
530         (pass_data_combine): ...new pass_data instance and...
531         (make_pass_combine): ...new function.
532         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
533         global struct to a subclass of rtl_opt_pass along with...
534         (pass_data_compare_elim_after_reload): ...new pass_data instance
535         and...
536         (make_pass_compare_elim_after_reload): ...new function.
537         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
538         of rtl_opt_pass along with...
539         (pass_data_rtl_cprop): ...new pass_data instance and...
540         (make_pass_rtl_cprop): ...new function.
541         * cse.c (pass_cse): Convert from a global struct to a subclass of
542         rtl_opt_pass along with...
543         (pass_data_cse): ...new pass_data instance and...
544         (make_pass_cse): ...new function.
545         (pass_cse2): Convert from a global struct to a subclass of
546         rtl_opt_pass along with...
547         (pass_data_cse2): ...new pass_data instance and...
548         (make_pass_cse2): ...new function.
549         (pass_cse_after_global_opts): Convert from a global struct to a
550         subclass of rtl_opt_pass along with...
551         (pass_data_cse_after_global_opts): ...new pass_data instance and...
552         (make_pass_cse_after_global_opts): ...new function.
553         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
554         of rtl_opt_pass along with...
555         (pass_data_ud_rtl_dce): ...new pass_data instance and...
556         (make_pass_ud_rtl_dce): ...new function.
557         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
558         rtl_opt_pass along with...
559         (pass_data_fast_rtl_dce): ...new pass_data instance and...
560         (make_pass_fast_rtl_dce): ...new function.
561         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
562         a subclass of rtl_opt_pass along with...
563         (pass_data_df_initialize_opt): ...new pass_data instance and...
564         (make_pass_df_initialize_opt): ...new function.
565         (pass_df_initialize_no_opt): Convert from a global struct to a
566         subclass of rtl_opt_pass along with...
567         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
568         (make_pass_df_initialize_no_opt): ...new function.
569         (pass_df_finish): Convert from a global struct to a subclass of
570         rtl_opt_pass along with...
571         (pass_data_df_finish): ...new pass_data instance and...
572         (make_pass_df_finish): ...new function.
573         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
574         rtl_opt_pass along with...
575         (pass_data_rtl_dse1): ...new pass_data instance and...
576         (make_pass_rtl_dse1): ...new function.
577         (pass_rtl_dse2): Convert from a global struct to a subclass of
578         rtl_opt_pass along with...
579         (pass_data_rtl_dse2): ...new pass_data instance and...
580         (make_pass_rtl_dse2): ...new function.
581         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
582         subclass of rtl_opt_pass along with...
583         (pass_data_dwarf2_frame): ...new pass_data instance and...
584         (make_pass_dwarf2_frame): ...new function.
585         * except.c (pass_set_nothrow_function_flags): Convert from a global
586         struct to a subclass of rtl_opt_pass along with...
587         (pass_data_set_nothrow_function_flags): ...new pass_data instance
588         and...
589         (make_pass_set_nothrow_function_flags): ...new function.
590         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
591         subclass of rtl_opt_pass along with...
592         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
593         and...
594         (make_pass_convert_to_eh_region_ranges): ...new function.
595         * final.c (pass_compute_alignments): Convert from a global struct to a
596         subclass of rtl_opt_pass along with...
597         (pass_data_compute_alignments): ...new pass_data instance and...
598         (make_pass_compute_alignments): ...new function.
599         (pass_final): Convert from a global struct to a subclass of
600         rtl_opt_pass along with...
601         (pass_data_final): ...new pass_data instance and...
602         (make_pass_final): ...new function.
603         (pass_shorten_branches): Convert from a global struct to a subclass of
604         rtl_opt_pass along with...
605         (pass_data_shorten_branches): ...new pass_data instance and...
606         (make_pass_shorten_branches): ...new function.
607         (pass_clean_state): Convert from a global struct to a subclass of
608         rtl_opt_pass along with...
609         (pass_data_clean_state): ...new pass_data instance and...
610         (make_pass_clean_state): ...new function.
611         * function.c (pass_instantiate_virtual_regs): Convert from a global
612         struct to a subclass of rtl_opt_pass along with...
613         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
614         (make_pass_instantiate_virtual_regs): ...new function.
615         (pass_leaf_regs): Convert from a global struct to a subclass of
616         rtl_opt_pass along with...
617         (pass_data_leaf_regs): ...new pass_data instance and...
618         (make_pass_leaf_regs): ...new function.
619         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
620         subclass of rtl_opt_pass along with...
621         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
622         and...
623         (make_pass_thread_prologue_and_epilogue): ...new function.
624         (pass_match_asm_constraints): Convert from a global struct to a
625         subclass of rtl_opt_pass along with...
626         (pass_data_match_asm_constraints): ...new pass_data instance and...
627         (make_pass_match_asm_constraints): ...new function.
628         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
629         subclass of rtl_opt_pass along with...
630         (pass_data_rtl_fwprop): ...new pass_data instance and...
631         (make_pass_rtl_fwprop): ...new function.
632         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
633         rtl_opt_pass along with...
634         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
635         (make_pass_rtl_fwprop_addr): ...new function.
636         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
637         rtl_opt_pass along with...
638         (pass_data_rtl_pre): ...new pass_data instance and...
639         (make_pass_rtl_pre): ...new function.
640         (pass_rtl_hoist): Convert from a global struct to a subclass of
641         rtl_opt_pass along with...
642         (pass_data_rtl_hoist): ...new pass_data instance and...
643         (make_pass_rtl_hoist): ...new function.
644         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
645         subclass of gimple_opt_pass along with...
646         (pass_data_lower_cf): ...new pass_data instance and...
647         (make_pass_lower_cf): ...new function.
648         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
649         from a global struct to a subclass of gimple_opt_pass along with...
650         (pass_data_strength_reduction): ...new pass_data instance and...
651         (make_pass_strength_reduction): ...new function.
652         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
653         of rtl_opt_pass along with...
654         (pass_data_rtl_ifcvt): ...new pass_data instance and...
655         (make_pass_rtl_ifcvt): ...new function.
656         (pass_if_after_combine): Convert from a global struct to a subclass of
657         rtl_opt_pass along with...
658         (pass_data_if_after_combine): ...new pass_data instance and...
659         (make_pass_if_after_combine): ...new function.
660         (pass_if_after_reload): Convert from a global struct to a subclass of
661         rtl_opt_pass along with...
662         (pass_data_if_after_reload): ...new pass_data instance and...
663         (make_pass_if_after_reload): ...new function.
664         * init-regs.c (pass_initialize_regs): Convert from a global struct to
665         a subclass of rtl_opt_pass along with...
666         (pass_data_initialize_regs): ...new pass_data instance and...
667         (make_pass_initialize_regs): ...new function.
668         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
669         of ipa_opt_pass_d along with...
670         (pass_data_ipa_cp): ...new pass_data instance and...
671         (make_pass_ipa_cp): ...new function.
672         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
673         global struct to a subclass of gimple_opt_pass along with...
674         (pass_data_inline_parameters): ...new pass_data instance and...
675         (make_pass_inline_parameters): ...new function.
676         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
677         subclass of gimple_opt_pass along with...
678         (pass_data_early_inline): ...new pass_data instance and...
679         (make_pass_early_inline): ...new function.
680         (pass_ipa_inline): Convert from a global struct to a subclass of
681         ipa_opt_pass_d along with...
682         (pass_data_ipa_inline): ...new pass_data instance and...
683         (make_pass_ipa_inline): ...new function.
684         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
685         struct to a subclass of gimple_opt_pass along with...
686         (pass_data_local_pure_const): ...new pass_data instance and...
687         (make_pass_local_pure_const): ...new function.
688         (pass_ipa_pure_const): Convert from a global struct to a subclass of
689         ipa_opt_pass_d along with...
690         (pass_data_ipa_pure_const): ...new pass_data instance and...
691         (make_pass_ipa_pure_const): ...new function.
692         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
693         to a subclass of ipa_opt_pass_d along with...
694         (pass_data_ipa_reference): ...new pass_data instance and...
695         (make_pass_ipa_reference): ...new function.
696         * ipa-split.c (pass_split_functions): Convert from a global struct to
697         a subclass of gimple_opt_pass along with...
698         (pass_data_split_functions): ...new pass_data instance and...
699         (make_pass_split_functions): ...new function.
700         (pass_feedback_split_functions): Convert from a global struct to a
701         subclass of gimple_opt_pass along with...
702         (pass_data_feedback_split_functions): ...new pass_data instance and...
703         (make_pass_feedback_split_functions): ...new function.
704         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
705         global struct to a subclass of simple_ipa_opt_pass along with...
706         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
707         instance and...
708         (make_pass_ipa_function_and_variable_visibility): ...new function.
709         (pass_ipa_free_inline_summary): Convert from a global struct to a
710         subclass of simple_ipa_opt_pass along with...
711         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
712         (make_pass_ipa_free_inline_summary): ...new function.
713         (pass_ipa_whole_program_visibility): Convert from a global struct to a
714         subclass of ipa_opt_pass_d along with...
715         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
716         and...
717         (make_pass_ipa_whole_program_visibility): ...new function.
718         (pass_ipa_profile): Convert from a global struct to a subclass of
719         ipa_opt_pass_d along with...
720         (pass_data_ipa_profile): ...new pass_data instance and...
721         (make_pass_ipa_profile): ...new function.
722         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
723         ipa_opt_pass_d along with...
724         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
725         (make_pass_ipa_cdtor_merge): ...new function.
726         * ira.c (pass_ira): Convert from a global struct to a subclass of
727         rtl_opt_pass along with...
728         (pass_data_ira): ...new pass_data instance and...
729         (make_pass_ira): ...new function.
730         (pass_reload): Convert from a global struct to a subclass of
731         rtl_opt_pass along with...
732         (pass_data_reload): ...new pass_data instance and...
733         (make_pass_reload): ...new function.
734         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
735         subclass of rtl_opt_pass along with...
736         (pass_data_cleanup_barriers): ...new pass_data instance and...
737         (make_pass_cleanup_barriers): ...new function.
738         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
739         of rtl_opt_pass along with...
740         (pass_data_loop2): ...new pass_data instance and...
741         (make_pass_loop2): ...new function.
742         (pass_rtl_loop_init): Convert from a global struct to a subclass of
743         rtl_opt_pass along with...
744         (pass_data_rtl_loop_init): ...new pass_data instance and...
745         (make_pass_rtl_loop_init): ...new function.
746         (pass_rtl_loop_done): Convert from a global struct to a subclass of
747         rtl_opt_pass along with...
748         (pass_data_rtl_loop_done): ...new pass_data instance and...
749         (make_pass_rtl_loop_done): ...new function.
750         (pass_rtl_move_loop_invariants): Convert from a global struct to a
751         subclass of rtl_opt_pass along with...
752         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
753         (make_pass_rtl_move_loop_invariants): ...new function.
754         (pass_rtl_unswitch): Convert from a global struct to a subclass of
755         rtl_opt_pass along with...
756         (pass_data_rtl_unswitch): ...new pass_data instance and...
757         (make_pass_rtl_unswitch): ...new function.
758         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
759         subclass of rtl_opt_pass along with...
760         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
761         and...
762         (make_pass_rtl_unroll_and_peel_loops): ...new function.
763         (pass_rtl_doloop): Convert from a global struct to a subclass of
764         rtl_opt_pass along with...
765         (pass_data_rtl_doloop): ...new pass_data instance and...
766         (make_pass_rtl_doloop): ...new function.
767         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
768         a subclass of rtl_opt_pass along with...
769         (pass_data_lower_subreg): ...new pass_data instance and...
770         (make_pass_lower_subreg): ...new function.
771         (pass_lower_subreg2): Convert from a global struct to a subclass of
772         rtl_opt_pass along with...
773         (pass_data_lower_subreg2): ...new pass_data instance and...
774         (make_pass_lower_subreg2): ...new function.
775         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
776         struct to a subclass of ipa_opt_pass_d along with...
777         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
778         (make_pass_ipa_lto_gimple_out): ...new function.
779         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
780         of ipa_opt_pass_d along with...
781         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
782         (make_pass_ipa_lto_finish_out): ...new function.
783         * mode-switching.c (pass_mode_switching): Convert from a global struct
784         to a subclass of rtl_opt_pass along with...
785         (pass_data_mode_switching): ...new pass_data instance and...
786         (make_pass_mode_switching): ...new function.
787         * modulo-sched.c (pass_sms): Convert from a global struct to a
788         subclass of rtl_opt_pass along with...
789         (pass_data_sms): ...new pass_data instance and...
790         (make_pass_sms): ...new function.
791         * omp-low.c (pass_expand_omp): Convert from a global struct to a
792         subclass of gimple_opt_pass along with...
793         (pass_data_expand_omp): ...new pass_data instance and...
794         (make_pass_expand_omp): ...new function.
795         (pass_lower_omp): Convert from a global struct to a subclass of
796         gimple_opt_pass along with...
797         (pass_data_lower_omp): ...new pass_data instance and...
798         (make_pass_lower_omp): ...new function.
799         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
800         of gimple_opt_pass along with...
801         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
802         (make_pass_diagnose_omp_blocks): ...new function.
803         * passes.c (pass_early_local_passes): Convert from a global struct to
804         a subclass of simple_ipa_opt_pass along with...
805         (pass_data_early_local_passes): ...new pass_data instance and...
806         (make_pass_early_local_passes): ...new function.
807         (pass_all_early_optimizations): Convert from a global struct to a
808         subclass of gimple_opt_pass along with...
809         (pass_data_all_early_optimizations): ...new pass_data instance and...
810         (make_pass_all_early_optimizations): ...new function.
811         (pass_all_optimizations): Convert from a global struct to a subclass
812         of gimple_opt_pass along with...
813         (pass_data_all_optimizations): ...new pass_data instance and...
814         (make_pass_all_optimizations): ...new function.
815         (pass_all_optimizations_g): Convert from a global struct to a subclass
816         of gimple_opt_pass along with...
817         (pass_data_all_optimizations_g): ...new pass_data instance and...
818         (make_pass_all_optimizations_g): ...new function.
819         (pass_rest_of_compilation): Convert from a global struct to a subclass
820         of rtl_opt_pass along with...
821         (pass_data_rest_of_compilation): ...new pass_data instance and...
822         (make_pass_rest_of_compilation): ...new function.
823         (pass_postreload): Convert from a global struct to a subclass of
824         rtl_opt_pass along with...
825         (pass_data_postreload): ...new pass_data instance and...
826         (make_pass_postreload): ...new function.
827         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
828         subclass of rtl_opt_pass along with...
829         (pass_data_gcse2): ...new pass_data instance and...
830         (make_pass_gcse2): ...new function.
831         * postreload.c (pass_postreload_cse): Convert from a global struct to
832         a subclass of rtl_opt_pass along with...
833         (pass_data_postreload_cse): ...new pass_data instance and...
834         (make_pass_postreload_cse): ...new function.
835         * predict.c (pass_profile): Convert from a global struct to a subclass
836         of gimple_opt_pass along with...
837         (pass_data_profile): ...new pass_data instance and...
838         (make_pass_profile): ...new function.
839         (pass_strip_predict_hints): Convert from a global struct to a subclass
840         of gimple_opt_pass along with...
841         (pass_data_strip_predict_hints): ...new pass_data instance and...
842         (make_pass_strip_predict_hints): ...new function.
843         * recog.c (pass_peephole2): Convert from a global struct to a subclass
844         of rtl_opt_pass along with...
845         (pass_data_peephole2): ...new pass_data instance and...
846         (make_pass_peephole2): ...new function.
847         (pass_split_all_insns): Convert from a global struct to a subclass of
848         rtl_opt_pass along with...
849         (pass_data_split_all_insns): ...new pass_data instance and...
850         (make_pass_split_all_insns): ...new function.
851         (pass_split_after_reload): Convert from a global struct to a subclass
852         of rtl_opt_pass along with...
853         (pass_data_split_after_reload): ...new pass_data instance and...
854         (make_pass_split_after_reload): ...new function.
855         (pass_split_before_regstack): Convert from a global struct to a
856         subclass of rtl_opt_pass along with...
857         (pass_data_split_before_regstack): ...new pass_data instance and...
858         (make_pass_split_before_regstack): ...new function.
859         (pass_split_before_sched2): Convert from a global struct to a subclass
860         of rtl_opt_pass along with...
861         (pass_data_split_before_sched2): ...new pass_data instance and...
862         (make_pass_split_before_sched2): ...new function.
863         (pass_split_for_shorten_branches): Convert from a global struct to a
864         subclass of rtl_opt_pass along with...
865         (pass_data_split_for_shorten_branches): ...new pass_data instance
866         and...
867         (make_pass_split_for_shorten_branches): ...new function.
868         * ree.c (pass_ree): Convert from a global struct to a subclass of
869         rtl_opt_pass along with...
870         (pass_data_ree): ...new pass_data instance and...
871         (make_pass_ree): ...new function.
872         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
873         subclass of rtl_opt_pass along with...
874         (pass_data_stack_regs): ...new pass_data instance and...
875         (make_pass_stack_regs): ...new function.
876         (pass_stack_regs_run): Convert from a global struct to a subclass of
877         rtl_opt_pass along with...
878         (pass_data_stack_regs_run): ...new pass_data instance and...
879         (make_pass_stack_regs_run): ...new function.
880         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
881         subclass of rtl_opt_pass along with...
882         (pass_data_cprop_hardreg): ...new pass_data instance and...
883         (make_pass_cprop_hardreg): ...new function.
884         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
885         subclass of rtl_opt_pass along with...
886         (pass_data_reginfo_init): ...new pass_data instance and...
887         (make_pass_reginfo_init): ...new function.
888         * regmove.c (pass_regmove): Convert from a global struct to a subclass
889         of rtl_opt_pass along with...
890         (pass_data_regmove): ...new pass_data instance and...
891         (make_pass_regmove): ...new function.
892         * regrename.c (pass_regrename): Convert from a global struct to a
893         subclass of rtl_opt_pass along with...
894         (pass_data_regrename): ...new pass_data instance and...
895         (make_pass_regrename): ...new function.
896         * reorg.c (pass_delay_slots): Convert from a global struct to a
897         subclass of rtl_opt_pass along with...
898         (pass_data_delay_slots): ...new pass_data instance and...
899         (make_pass_delay_slots): ...new function.
900         (pass_machine_reorg): Convert from a global struct to a subclass of
901         rtl_opt_pass along with...
902         (pass_data_machine_reorg): ...new pass_data instance and...
903         (make_pass_machine_reorg): ...new function.
904         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
905         of rtl_opt_pass along with...
906         (pass_data_sched): ...new pass_data instance and...
907         (make_pass_sched): ...new function.
908         (pass_sched2): Convert from a global struct to a subclass of
909         rtl_opt_pass along with...
910         (pass_data_sched2): ...new pass_data instance and...
911         (make_pass_sched2): ...new function.
912         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
913         to a subclass of rtl_opt_pass along with...
914         (pass_data_stack_ptr_mod): ...new pass_data instance and...
915         (make_pass_stack_ptr_mod): ...new function.
916         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
917         to a subclass of rtl_opt_pass along with...
918         (pass_data_rtl_store_motion): ...new pass_data instance and...
919         (make_pass_rtl_store_motion): ...new function.
920         * tracer.c (pass_tracer): Convert from a global struct to a subclass
921         of gimple_opt_pass along with...
922         (pass_data_tracer): ...new pass_data instance and...
923         (make_pass_tracer): ...new function.
924         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
925         to a subclass of gimple_opt_pass along with...
926         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
927         (make_pass_diagnose_tm_blocks): ...new function.
928         (pass_lower_tm): Convert from a global struct to a subclass of
929         gimple_opt_pass along with...
930         (pass_data_lower_tm): ...new pass_data instance and...
931         (make_pass_lower_tm): ...new function.
932         (pass_tm_init): Convert from a global struct to a subclass of
933         gimple_opt_pass along with...
934         (pass_data_tm_init): ...new pass_data instance and...
935         (make_pass_tm_init): ...new function.
936         (pass_tm_mark): Convert from a global struct to a subclass of
937         gimple_opt_pass along with...
938         (pass_data_tm_mark): ...new pass_data instance and...
939         (make_pass_tm_mark): ...new function.
940         (pass_tm_edges): Convert from a global struct to a subclass of
941         gimple_opt_pass along with...
942         (pass_data_tm_edges): ...new pass_data instance and...
943         (make_pass_tm_edges): ...new function.
944         (pass_tm_memopt): Convert from a global struct to a subclass of
945         gimple_opt_pass along with...
946         (pass_data_tm_memopt): ...new pass_data instance and...
947         (make_pass_tm_memopt): ...new function.
948         (pass_ipa_tm): Convert from a global struct to a subclass of
949         simple_ipa_opt_pass along with...
950         (pass_data_ipa_tm): ...new pass_data instance and...
951         (make_pass_ipa_tm): ...new function.
952         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
953         subclass of gimple_opt_pass along with...
954         (pass_data_call_cdce): ...new pass_data instance and...
955         (make_pass_call_cdce): ...new function.
956         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
957         subclass of gimple_opt_pass along with...
958         (pass_data_build_cfg): ...new pass_data instance and...
959         (make_pass_build_cfg): ...new function.
960         (pass_split_crit_edges): Convert from a global struct to a subclass of
961         gimple_opt_pass along with...
962         (pass_data_split_crit_edges): ...new pass_data instance and...
963         (make_pass_split_crit_edges): ...new function.
964         (pass_warn_function_return): Convert from a global struct to a
965         subclass of gimple_opt_pass along with...
966         (pass_data_warn_function_return): ...new pass_data instance and...
967         (make_pass_warn_function_return): ...new function.
968         (pass_warn_function_noreturn): Convert from a global struct to a
969         subclass of gimple_opt_pass along with...
970         (pass_data_warn_function_noreturn): ...new pass_data instance and...
971         (make_pass_warn_function_noreturn): ...new function.
972         (pass_warn_unused_result): Convert from a global struct to a subclass
973         of gimple_opt_pass along with...
974         (pass_data_warn_unused_result): ...new pass_data instance and...
975         (make_pass_warn_unused_result): ...new function.
976         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
977         a subclass of gimple_opt_pass along with...
978         (pass_data_merge_phi): ...new pass_data instance and...
979         (make_pass_merge_phi): ...new function.
980         * tree-complex.c (pass_lower_complex): Convert from a global struct to
981         a subclass of gimple_opt_pass along with...
982         (pass_data_lower_complex): ...new pass_data instance and...
983         (make_pass_lower_complex): ...new function.
984         (pass_lower_complex_O0): Convert from a global struct to a subclass of
985         gimple_opt_pass along with...
986         (pass_data_lower_complex_O0): ...new pass_data instance and...
987         (make_pass_lower_complex_O0): ...new function.
988         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
989         subclass of gimple_opt_pass along with...
990         (pass_data_lower_eh): ...new pass_data instance and...
991         (make_pass_lower_eh): ...new function.
992         (pass_refactor_eh): Convert from a global struct to a subclass of
993         gimple_opt_pass along with...
994         (pass_data_refactor_eh): ...new pass_data instance and...
995         (make_pass_refactor_eh): ...new function.
996         (pass_lower_resx): Convert from a global struct to a subclass of
997         gimple_opt_pass along with...
998         (pass_data_lower_resx): ...new pass_data instance and...
999         (make_pass_lower_resx): ...new function.
1000         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1001         of gimple_opt_pass along with...
1002         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1003         (make_pass_lower_eh_dispatch): ...new function.
1004         (pass_cleanup_eh): Convert from a global struct to a subclass of
1005         gimple_opt_pass along with...
1006         (pass_data_cleanup_eh): ...new pass_data instance and...
1007         (make_pass_cleanup_eh): ...new function.
1008         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1009         to a subclass of simple_ipa_opt_pass along with...
1010         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1011         (make_pass_ipa_lower_emutls): ...new function.
1012         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1013         a subclass of gimple_opt_pass along with...
1014         (pass_data_if_conversion): ...new pass_data instance and...
1015         (make_pass_if_conversion): ...new function.
1016         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1017         subclass of gimple_opt_pass along with...
1018         (pass_data_build_ssa): ...new pass_data instance and...
1019         (make_pass_build_ssa): ...new function.
1020         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1021         global struct to a subclass of gimple_opt_pass along with...
1022         (pass_data_loop_distribution): ...new pass_data instance and...
1023         (make_pass_loop_distribution): ...new function.
1024         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1025         subclass of gimple_opt_pass along with...
1026         (pass_data_mudflap_1): ...new pass_data instance and...
1027         (make_pass_mudflap_1): ...new function.
1028         (pass_mudflap_2): Convert from a global struct to a subclass of
1029         gimple_opt_pass along with...
1030         (pass_data_mudflap_2): ...new pass_data instance and...
1031         (make_pass_mudflap_2): ...new function.
1032         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1033         subclass of gimple_opt_pass along with...
1034         (pass_data_mudflap_1): ...new pass_data instance and...
1035         (make_pass_mudflap_1): ...new function.
1036         (pass_mudflap_2): Convert from a global struct to a subclass of
1037         gimple_opt_pass along with...
1038         (pass_data_mudflap_2): ...new pass_data instance and...
1039         (make_pass_mudflap_2): ...new function.
1040         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1041         gimple_opt_pass along with...
1042         (pass_data_nrv): ...new pass_data instance and...
1043         (make_pass_nrv): ...new function.
1044         (pass_return_slot): Convert from a global struct to a subclass of
1045         gimple_opt_pass along with...
1046         (pass_data_return_slot): ...new pass_data instance and...
1047         (make_pass_return_slot): ...new function.
1048         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1049         to a subclass of gimple_opt_pass along with...
1050         (pass_data_object_sizes): ...new pass_data instance and...
1051         (make_pass_object_sizes): ...new function.
1052         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1053         global struct to a subclass of gimple_opt_pass along with...
1054         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1055         and...
1056         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1057         (pass_fixup_cfg): Convert from a global struct to a subclass of
1058         gimple_opt_pass along with...
1059         (pass_data_fixup_cfg): ...new pass_data instance and...
1060         (make_pass_fixup_cfg): ...new function.
1061         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1062         (make_pass_mudflap_1): ...new function.
1063         (pass_mudflap_2): Replace declaration with that of...
1064         (make_pass_mudflap_2): ...new function.
1065         (pass_asan): Replace declaration with that of...
1066         (make_pass_asan): ...new function.
1067         (pass_asan_O0): Replace declaration with that of...
1068         (make_pass_asan_O0): ...new function.
1069         (pass_tsan): Replace declaration with that of...
1070         (make_pass_tsan): ...new function.
1071         (pass_tsan_O0): Replace declaration with that of...
1072         (make_pass_tsan_O0): ...new function.
1073         (pass_lower_cf): Replace declaration with that of...
1074         (make_pass_lower_cf): ...new function.
1075         (pass_refactor_eh): Replace declaration with that of...
1076         (make_pass_refactor_eh): ...new function.
1077         (pass_lower_eh): Replace declaration with that of...
1078         (make_pass_lower_eh): ...new function.
1079         (pass_lower_eh_dispatch): Replace declaration with that of...
1080         (make_pass_lower_eh_dispatch): ...new function.
1081         (pass_lower_resx): Replace declaration with that of...
1082         (make_pass_lower_resx): ...new function.
1083         (pass_build_cfg): Replace declaration with that of...
1084         (make_pass_build_cfg): ...new function.
1085         (pass_early_tree_profile): Replace declaration with that of...
1086         (make_pass_early_tree_profile): ...new function.
1087         (pass_cleanup_eh): Replace declaration with that of...
1088         (make_pass_cleanup_eh): ...new function.
1089         (pass_sra): Replace declaration with that of...
1090         (make_pass_sra): ...new function.
1091         (pass_sra_early): Replace declaration with that of...
1092         (make_pass_sra_early): ...new function.
1093         (pass_early_ipa_sra): Replace declaration with that of...
1094         (make_pass_early_ipa_sra): ...new function.
1095         (pass_tail_recursion): Replace declaration with that of...
1096         (make_pass_tail_recursion): ...new function.
1097         (pass_tail_calls): Replace declaration with that of...
1098         (make_pass_tail_calls): ...new function.
1099         (pass_tree_loop): Replace declaration with that of...
1100         (make_pass_tree_loop): ...new function.
1101         (pass_tree_loop_init): Replace declaration with that of...
1102         (make_pass_tree_loop_init): ...new function.
1103         (pass_lim): Replace declaration with that of...
1104         (make_pass_lim): ...new function.
1105         (pass_tree_unswitch): Replace declaration with that of...
1106         (make_pass_tree_unswitch): ...new function.
1107         (pass_predcom): Replace declaration with that of...
1108         (make_pass_predcom): ...new function.
1109         (pass_iv_canon): Replace declaration with that of...
1110         (make_pass_iv_canon): ...new function.
1111         (pass_scev_cprop): Replace declaration with that of...
1112         (make_pass_scev_cprop): ...new function.
1113         (pass_empty_loop): Replace declaration with that of...
1114         (make_pass_empty_loop): ...new function.
1115         (pass_record_bounds): Replace declaration with that of...
1116         (make_pass_record_bounds): ...new function.
1117         (pass_graphite): Replace declaration with that of...
1118         (make_pass_graphite): ...new function.
1119         (pass_graphite_transforms): Replace declaration with that of...
1120         (make_pass_graphite_transforms): ...new function.
1121         (pass_if_conversion): Replace declaration with that of...
1122         (make_pass_if_conversion): ...new function.
1123         (pass_loop_distribution): Replace declaration with that of...
1124         (make_pass_loop_distribution): ...new function.
1125         (pass_vectorize): Replace declaration with that of...
1126         (make_pass_vectorize): ...new function.
1127         (pass_slp_vectorize): Replace declaration with that of...
1128         (make_pass_slp_vectorize): ...new function.
1129         (pass_complete_unroll): Replace declaration with that of...
1130         (make_pass_complete_unroll): ...new function.
1131         (pass_complete_unrolli): Replace declaration with that of...
1132         (make_pass_complete_unrolli): ...new function.
1133         (pass_parallelize_loops): Replace declaration with that of...
1134         (make_pass_parallelize_loops): ...new function.
1135         (pass_loop_prefetch): Replace declaration with that of...
1136         (make_pass_loop_prefetch): ...new function.
1137         (pass_iv_optimize): Replace declaration with that of...
1138         (make_pass_iv_optimize): ...new function.
1139         (pass_tree_loop_done): Replace declaration with that of...
1140         (make_pass_tree_loop_done): ...new function.
1141         (pass_ch): Replace declaration with that of...
1142         (make_pass_ch): ...new function.
1143         (pass_ccp): Replace declaration with that of...
1144         (make_pass_ccp): ...new function.
1145         (pass_phi_only_cprop): Replace declaration with that of...
1146         (make_pass_phi_only_cprop): ...new function.
1147         (pass_build_ssa): Replace declaration with that of...
1148         (make_pass_build_ssa): ...new function.
1149         (pass_build_alias): Replace declaration with that of...
1150         (make_pass_build_alias): ...new function.
1151         (pass_build_ealias): Replace declaration with that of...
1152         (make_pass_build_ealias): ...new function.
1153         (pass_dominator): Replace declaration with that of...
1154         (make_pass_dominator): ...new function.
1155         (pass_dce): Replace declaration with that of...
1156         (make_pass_dce): ...new function.
1157         (pass_dce_loop): Replace declaration with that of...
1158         (make_pass_dce_loop): ...new function.
1159         (pass_cd_dce): Replace declaration with that of...
1160         (make_pass_cd_dce): ...new function.
1161         (pass_call_cdce): Replace declaration with that of...
1162         (make_pass_call_cdce): ...new function.
1163         (pass_merge_phi): Replace declaration with that of...
1164         (make_pass_merge_phi): ...new function.
1165         (pass_split_crit_edges): Replace declaration with that of...
1166         (make_pass_split_crit_edges): ...new function.
1167         (pass_pre): Replace declaration with that of...
1168         (make_pass_pre): ...new function.
1169         (pass_profile): Replace declaration with that of...
1170         (make_pass_profile): ...new function.
1171         (pass_strip_predict_hints): Replace declaration with that of...
1172         (make_pass_strip_predict_hints): ...new function.
1173         (pass_lower_complex_O0): Replace declaration with that of...
1174         (make_pass_lower_complex_O0): ...new function.
1175         (pass_lower_complex): Replace declaration with that of...
1176         (make_pass_lower_complex): ...new function.
1177         (pass_lower_vector): Replace declaration with that of...
1178         (make_pass_lower_vector): ...new function.
1179         (pass_lower_vector_ssa): Replace declaration with that of...
1180         (make_pass_lower_vector_ssa): ...new function.
1181         (pass_lower_omp): Replace declaration with that of...
1182         (make_pass_lower_omp): ...new function.
1183         (pass_diagnose_omp_blocks): Replace declaration with that of...
1184         (make_pass_diagnose_omp_blocks): ...new function.
1185         (pass_expand_omp): Replace declaration with that of...
1186         (make_pass_expand_omp): ...new function.
1187         (pass_expand_omp_ssa): Replace declaration with that of...
1188         (make_pass_expand_omp_ssa): ...new function.
1189         (pass_object_sizes): Replace declaration with that of...
1190         (make_pass_object_sizes): ...new function.
1191         (pass_strlen): Replace declaration with that of...
1192         (make_pass_strlen): ...new function.
1193         (pass_fold_builtins): Replace declaration with that of...
1194         (make_pass_fold_builtins): ...new function.
1195         (pass_stdarg): Replace declaration with that of...
1196         (make_pass_stdarg): ...new function.
1197         (pass_early_warn_uninitialized): Replace declaration with that of...
1198         (make_pass_early_warn_uninitialized): ...new function.
1199         (pass_late_warn_uninitialized): Replace declaration with that of...
1200         (make_pass_late_warn_uninitialized): ...new function.
1201         (pass_cse_reciprocals): Replace declaration with that of...
1202         (make_pass_cse_reciprocals): ...new function.
1203         (pass_cse_sincos): Replace declaration with that of...
1204         (make_pass_cse_sincos): ...new function.
1205         (pass_optimize_bswap): Replace declaration with that of...
1206         (make_pass_optimize_bswap): ...new function.
1207         (pass_optimize_widening_mul): Replace declaration with that of...
1208         (make_pass_optimize_widening_mul): ...new function.
1209         (pass_warn_function_return): Replace declaration with that of...
1210         (make_pass_warn_function_return): ...new function.
1211         (pass_warn_function_noreturn): Replace declaration with that of...
1212         (make_pass_warn_function_noreturn): ...new function.
1213         (pass_cselim): Replace declaration with that of...
1214         (make_pass_cselim): ...new function.
1215         (pass_phiopt): Replace declaration with that of...
1216         (make_pass_phiopt): ...new function.
1217         (pass_forwprop): Replace declaration with that of...
1218         (make_pass_forwprop): ...new function.
1219         (pass_phiprop): Replace declaration with that of...
1220         (make_pass_phiprop): ...new function.
1221         (pass_tree_ifcombine): Replace declaration with that of...
1222         (make_pass_tree_ifcombine): ...new function.
1223         (pass_dse): Replace declaration with that of...
1224         (make_pass_dse): ...new function.
1225         (pass_nrv): Replace declaration with that of...
1226         (make_pass_nrv): ...new function.
1227         (pass_rename_ssa_copies): Replace declaration with that of...
1228         (make_pass_rename_ssa_copies): ...new function.
1229         (pass_sink_code): Replace declaration with that of...
1230         (make_pass_sink_code): ...new function.
1231         (pass_fre): Replace declaration with that of...
1232         (make_pass_fre): ...new function.
1233         (pass_check_data_deps): Replace declaration with that of...
1234         (make_pass_check_data_deps): ...new function.
1235         (pass_copy_prop): Replace declaration with that of...
1236         (make_pass_copy_prop): ...new function.
1237         (pass_vrp): Replace declaration with that of...
1238         (make_pass_vrp): ...new function.
1239         (pass_uncprop): Replace declaration with that of...
1240         (make_pass_uncprop): ...new function.
1241         (pass_return_slot): Replace declaration with that of...
1242         (make_pass_return_slot): ...new function.
1243         (pass_reassoc): Replace declaration with that of...
1244         (make_pass_reassoc): ...new function.
1245         (pass_rebuild_cgraph_edges): Replace declaration with that of...
1246         (make_pass_rebuild_cgraph_edges): ...new function.
1247         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1248         (make_pass_remove_cgraph_callee_edges): ...new function.
1249         (pass_build_cgraph_edges): Replace declaration with that of...
1250         (make_pass_build_cgraph_edges): ...new function.
1251         (pass_local_pure_const): Replace declaration with that of...
1252         (make_pass_local_pure_const): ...new function.
1253         (pass_tracer): Replace declaration with that of...
1254         (make_pass_tracer): ...new function.
1255         (pass_warn_unused_result): Replace declaration with that of...
1256         (make_pass_warn_unused_result): ...new function.
1257         (pass_diagnose_tm_blocks): Replace declaration with that of...
1258         (make_pass_diagnose_tm_blocks): ...new function.
1259         (pass_lower_tm): Replace declaration with that of...
1260         (make_pass_lower_tm): ...new function.
1261         (pass_tm_init): Replace declaration with that of...
1262         (make_pass_tm_init): ...new function.
1263         (pass_tm_mark): Replace declaration with that of...
1264         (make_pass_tm_mark): ...new function.
1265         (pass_tm_memopt): Replace declaration with that of...
1266         (make_pass_tm_memopt): ...new function.
1267         (pass_tm_edges): Replace declaration with that of...
1268         (make_pass_tm_edges): ...new function.
1269         (pass_split_functions): Replace declaration with that of...
1270         (make_pass_split_functions): ...new function.
1271         (pass_feedback_split_functions): Replace declaration with that of...
1272         (make_pass_feedback_split_functions): ...new function.
1273         (pass_strength_reduction): Replace declaration with that of...
1274         (make_pass_strength_reduction): ...new function.
1275         (pass_ipa_lower_emutls): Replace declaration with that of...
1276         (make_pass_ipa_lower_emutls): ...new function.
1277         (pass_ipa_function_and_variable_visibility): Replace declaration with
1278         that of...
1279         (make_pass_ipa_function_and_variable_visibility): ...new function.
1280         (pass_ipa_tree_profile): Replace declaration with that of...
1281         (make_pass_ipa_tree_profile): ...new function.
1282         (pass_early_local_passes): Replace declaration with that of...
1283         (make_pass_early_local_passes): ...new function.
1284         (pass_ipa_whole_program_visibility): Replace declaration with that
1285         of...
1286         (make_pass_ipa_whole_program_visibility): ...new function.
1287         (pass_ipa_lto_gimple_out): Replace declaration with that of...
1288         (make_pass_ipa_lto_gimple_out): ...new function.
1289         (pass_ipa_increase_alignment): Replace declaration with that of...
1290         (make_pass_ipa_increase_alignment): ...new function.
1291         (pass_ipa_inline): Replace declaration with that of...
1292         (make_pass_ipa_inline): ...new function.
1293         (pass_ipa_free_lang_data): Replace declaration with that of...
1294         (make_pass_ipa_free_lang_data): ...new function.
1295         (pass_ipa_free_inline_summary): Replace declaration with that of...
1296         (make_pass_ipa_free_inline_summary): ...new function.
1297         (pass_ipa_cp): Replace declaration with that of...
1298         (make_pass_ipa_cp): ...new function.
1299         (pass_ipa_reference): Replace declaration with that of...
1300         (make_pass_ipa_reference): ...new function.
1301         (pass_ipa_pure_const): Replace declaration with that of...
1302         (make_pass_ipa_pure_const): ...new function.
1303         (pass_ipa_pta): Replace declaration with that of...
1304         (make_pass_ipa_pta): ...new function.
1305         (pass_ipa_lto_finish_out): Replace declaration with that of...
1306         (make_pass_ipa_lto_finish_out): ...new function.
1307         (pass_ipa_tm): Replace declaration with that of...
1308         (make_pass_ipa_tm): ...new function.
1309         (pass_ipa_profile): Replace declaration with that of...
1310         (make_pass_ipa_profile): ...new function.
1311         (pass_ipa_cdtor_merge): Replace declaration with that of...
1312         (make_pass_ipa_cdtor_merge): ...new function.
1313         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1314         of...
1315         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1316         (pass_init_datastructures): Replace declaration with that of...
1317         (make_pass_init_datastructures): ...new function.
1318         (pass_fixup_cfg): Replace declaration with that of...
1319         (make_pass_fixup_cfg): ...new function.
1320         (pass_expand): Replace declaration with that of...
1321         (make_pass_expand): ...new function.
1322         (pass_instantiate_virtual_regs): Replace declaration with that of...
1323         (make_pass_instantiate_virtual_regs): ...new function.
1324         (pass_rtl_fwprop): Replace declaration with that of...
1325         (make_pass_rtl_fwprop): ...new function.
1326         (pass_rtl_fwprop_addr): Replace declaration with that of...
1327         (make_pass_rtl_fwprop_addr): ...new function.
1328         (pass_jump): Replace declaration with that of...
1329         (make_pass_jump): ...new function.
1330         (pass_jump2): Replace declaration with that of...
1331         (make_pass_jump2): ...new function.
1332         (pass_lower_subreg): Replace declaration with that of...
1333         (make_pass_lower_subreg): ...new function.
1334         (pass_cse): Replace declaration with that of...
1335         (make_pass_cse): ...new function.
1336         (pass_fast_rtl_dce): Replace declaration with that of...
1337         (make_pass_fast_rtl_dce): ...new function.
1338         (pass_ud_rtl_dce): Replace declaration with that of...
1339         (make_pass_ud_rtl_dce): ...new function.
1340         (pass_rtl_dce): Replace declaration with that of...
1341         (make_pass_rtl_dce): ...new function.
1342         (pass_rtl_dse1): Replace declaration with that of...
1343         (make_pass_rtl_dse1): ...new function.
1344         (pass_rtl_dse2): Replace declaration with that of...
1345         (make_pass_rtl_dse2): ...new function.
1346         (pass_rtl_dse3): Replace declaration with that of...
1347         (make_pass_rtl_dse3): ...new function.
1348         (pass_rtl_cprop): Replace declaration with that of...
1349         (make_pass_rtl_cprop): ...new function.
1350         (pass_rtl_pre): Replace declaration with that of...
1351         (make_pass_rtl_pre): ...new function.
1352         (pass_rtl_hoist): Replace declaration with that of...
1353         (make_pass_rtl_hoist): ...new function.
1354         (pass_rtl_store_motion): Replace declaration with that of...
1355         (make_pass_rtl_store_motion): ...new function.
1356         (pass_cse_after_global_opts): Replace declaration with that of...
1357         (make_pass_cse_after_global_opts): ...new function.
1358         (pass_rtl_ifcvt): Replace declaration with that of...
1359         (make_pass_rtl_ifcvt): ...new function.
1360         (pass_into_cfg_layout_mode): Replace declaration with that of...
1361         (make_pass_into_cfg_layout_mode): ...new function.
1362         (pass_outof_cfg_layout_mode): Replace declaration with that of...
1363         (make_pass_outof_cfg_layout_mode): ...new function.
1364         (pass_loop2): Replace declaration with that of...
1365         (make_pass_loop2): ...new function.
1366         (pass_rtl_loop_init): Replace declaration with that of...
1367         (make_pass_rtl_loop_init): ...new function.
1368         (pass_rtl_move_loop_invariants): Replace declaration with that of...
1369         (make_pass_rtl_move_loop_invariants): ...new function.
1370         (pass_rtl_unswitch): Replace declaration with that of...
1371         (make_pass_rtl_unswitch): ...new function.
1372         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1373         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1374         (pass_rtl_doloop): Replace declaration with that of...
1375         (make_pass_rtl_doloop): ...new function.
1376         (pass_rtl_loop_done): Replace declaration with that of...
1377         (make_pass_rtl_loop_done): ...new function.
1378         (pass_web): Replace declaration with that of...
1379         (make_pass_web): ...new function.
1380         (pass_cse2): Replace declaration with that of...
1381         (make_pass_cse2): ...new function.
1382         (pass_df_initialize_opt): Replace declaration with that of...
1383         (make_pass_df_initialize_opt): ...new function.
1384         (pass_df_initialize_no_opt): Replace declaration with that of...
1385         (make_pass_df_initialize_no_opt): ...new function.
1386         (pass_reginfo_init): Replace declaration with that of...
1387         (make_pass_reginfo_init): ...new function.
1388         (pass_inc_dec): Replace declaration with that of...
1389         (make_pass_inc_dec): ...new function.
1390         (pass_stack_ptr_mod): Replace declaration with that of...
1391         (make_pass_stack_ptr_mod): ...new function.
1392         (pass_initialize_regs): Replace declaration with that of...
1393         (make_pass_initialize_regs): ...new function.
1394         (pass_combine): Replace declaration with that of...
1395         (make_pass_combine): ...new function.
1396         (pass_if_after_combine): Replace declaration with that of...
1397         (make_pass_if_after_combine): ...new function.
1398         (pass_ree): Replace declaration with that of...
1399         (make_pass_ree): ...new function.
1400         (pass_partition_blocks): Replace declaration with that of...
1401         (make_pass_partition_blocks): ...new function.
1402         (pass_match_asm_constraints): Replace declaration with that of...
1403         (make_pass_match_asm_constraints): ...new function.
1404         (pass_regmove): Replace declaration with that of...
1405         (make_pass_regmove): ...new function.
1406         (pass_split_all_insns): Replace declaration with that of...
1407         (make_pass_split_all_insns): ...new function.
1408         (pass_fast_rtl_byte_dce): Replace declaration with that of...
1409         (make_pass_fast_rtl_byte_dce): ...new function.
1410         (pass_lower_subreg2): Replace declaration with that of...
1411         (make_pass_lower_subreg2): ...new function.
1412         (pass_mode_switching): Replace declaration with that of...
1413         (make_pass_mode_switching): ...new function.
1414         (pass_sms): Replace declaration with that of...
1415         (make_pass_sms): ...new function.
1416         (pass_sched): Replace declaration with that of...
1417         (make_pass_sched): ...new function.
1418         (pass_ira): Replace declaration with that of...
1419         (make_pass_ira): ...new function.
1420         (pass_reload): Replace declaration with that of...
1421         (make_pass_reload): ...new function.
1422         (pass_clean_state): Replace declaration with that of...
1423         (make_pass_clean_state): ...new function.
1424         (pass_branch_prob): Replace declaration with that of...
1425         (make_pass_branch_prob): ...new function.
1426         (pass_value_profile_transformations): Replace declaration with that
1427         of...
1428         (make_pass_value_profile_transformations): ...new function.
1429         (pass_postreload_cse): Replace declaration with that of...
1430         (make_pass_postreload_cse): ...new function.
1431         (pass_gcse2): Replace declaration with that of...
1432         (make_pass_gcse2): ...new function.
1433         (pass_split_after_reload): Replace declaration with that of...
1434         (make_pass_split_after_reload): ...new function.
1435         (pass_branch_target_load_optimize1): Replace declaration with that
1436         of...
1437         (make_pass_branch_target_load_optimize1): ...new function.
1438         (pass_thread_prologue_and_epilogue): Replace declaration with that
1439         of...
1440         (make_pass_thread_prologue_and_epilogue): ...new function.
1441         (pass_stack_adjustments): Replace declaration with that of...
1442         (make_pass_stack_adjustments): ...new function.
1443         (pass_peephole2): Replace declaration with that of...
1444         (make_pass_peephole2): ...new function.
1445         (pass_if_after_reload): Replace declaration with that of...
1446         (make_pass_if_after_reload): ...new function.
1447         (pass_regrename): Replace declaration with that of...
1448         (make_pass_regrename): ...new function.
1449         (pass_cprop_hardreg): Replace declaration with that of...
1450         (make_pass_cprop_hardreg): ...new function.
1451         (pass_reorder_blocks): Replace declaration with that of...
1452         (make_pass_reorder_blocks): ...new function.
1453         (pass_branch_target_load_optimize2): Replace declaration with that
1454         of...
1455         (make_pass_branch_target_load_optimize2): ...new function.
1456         (pass_leaf_regs): Replace declaration with that of...
1457         (make_pass_leaf_regs): ...new function.
1458         (pass_split_before_sched2): Replace declaration with that of...
1459         (make_pass_split_before_sched2): ...new function.
1460         (pass_compare_elim_after_reload): Replace declaration with that of...
1461         (make_pass_compare_elim_after_reload): ...new function.
1462         (pass_sched2): Replace declaration with that of...
1463         (make_pass_sched2): ...new function.
1464         (pass_stack_regs): Replace declaration with that of...
1465         (make_pass_stack_regs): ...new function.
1466         (pass_stack_regs_run): Replace declaration with that of...
1467         (make_pass_stack_regs_run): ...new function.
1468         (pass_df_finish): Replace declaration with that of...
1469         (make_pass_df_finish): ...new function.
1470         (pass_compute_alignments): Replace declaration with that of...
1471         (make_pass_compute_alignments): ...new function.
1472         (pass_duplicate_computed_gotos): Replace declaration with that of...
1473         (make_pass_duplicate_computed_gotos): ...new function.
1474         (pass_variable_tracking): Replace declaration with that of...
1475         (make_pass_variable_tracking): ...new function.
1476         (pass_free_cfg): Replace declaration with that of...
1477         (make_pass_free_cfg): ...new function.
1478         (pass_machine_reorg): Replace declaration with that of...
1479         (make_pass_machine_reorg): ...new function.
1480         (pass_cleanup_barriers): Replace declaration with that of...
1481         (make_pass_cleanup_barriers): ...new function.
1482         (pass_delay_slots): Replace declaration with that of...
1483         (make_pass_delay_slots): ...new function.
1484         (pass_split_for_shorten_branches): Replace declaration with that of...
1485         (make_pass_split_for_shorten_branches): ...new function.
1486         (pass_split_before_regstack): Replace declaration with that of...
1487         (make_pass_split_before_regstack): ...new function.
1488         (pass_convert_to_eh_region_ranges): Replace declaration with that
1489         of...
1490         (make_pass_convert_to_eh_region_ranges): ...new function.
1491         (pass_shorten_branches): Replace declaration with that of...
1492         (make_pass_shorten_branches): ...new function.
1493         (pass_set_nothrow_function_flags): Replace declaration with that of...
1494         (make_pass_set_nothrow_function_flags): ...new function.
1495         (pass_dwarf2_frame): Replace declaration with that of...
1496         (make_pass_dwarf2_frame): ...new function.
1497         (pass_final): Replace declaration with that of...
1498         (make_pass_final): ...new function.
1499         (pass_rtl_seqabstr): Replace declaration with that of...
1500         (make_pass_rtl_seqabstr): ...new function.
1501         (pass_release_ssa_names): Replace declaration with that of...
1502         (make_pass_release_ssa_names): ...new function.
1503         (pass_early_inline): Replace declaration with that of...
1504         (make_pass_early_inline): ...new function.
1505         (pass_inline_parameters): Replace declaration with that of...
1506         (make_pass_inline_parameters): ...new function.
1507         (pass_update_address_taken): Replace declaration with that of...
1508         (make_pass_update_address_taken): ...new function.
1509         (pass_convert_switch): Replace declaration with that of...
1510         (make_pass_convert_switch): ...new function.
1511         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1512         to a subclass of simple_ipa_opt_pass along with...
1513         (pass_data_ipa_tree_profile): ...new pass_data instance and...
1514         (make_pass_ipa_tree_profile): ...new function.
1515         * tree-sra.c (pass_sra_early): Convert from a global struct to a
1516         subclass of gimple_opt_pass along with...
1517         (pass_data_sra_early): ...new pass_data instance and...
1518         (make_pass_sra_early): ...new function.
1519         (pass_sra): Convert from a global struct to a subclass of
1520         gimple_opt_pass along with...
1521         (pass_data_sra): ...new pass_data instance and...
1522         (make_pass_sra): ...new function.
1523         (pass_early_ipa_sra): Convert from a global struct to a subclass of
1524         gimple_opt_pass along with...
1525         (pass_data_early_ipa_sra): ...new pass_data instance and...
1526         (make_pass_early_ipa_sra): ...new function.
1527         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1528         subclass of gimple_opt_pass along with...
1529         (pass_data_ccp): ...new pass_data instance and...
1530         (make_pass_ccp): ...new function.
1531         (pass_fold_builtins): Convert from a global struct to a subclass of
1532         gimple_opt_pass along with...
1533         (pass_data_fold_builtins): ...new pass_data instance and...
1534         (make_pass_fold_builtins): ...new function.
1535         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1536         subclass of gimple_opt_pass along with...
1537         (pass_data_copy_prop): ...new pass_data instance and...
1538         (make_pass_copy_prop): ...new function.
1539         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1540         global struct to a subclass of gimple_opt_pass along with...
1541         (pass_data_rename_ssa_copies): ...new pass_data instance and...
1542         (make_pass_rename_ssa_copies): ...new function.
1543         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1544         subclass of gimple_opt_pass along with...
1545         (pass_data_dce): ...new pass_data instance and...
1546         (make_pass_dce): ...new function.
1547         (pass_dce_loop): Convert from a global struct to a subclass of
1548         gimple_opt_pass along with...
1549         (pass_data_dce_loop): ...new pass_data instance and...
1550         (make_pass_dce_loop): ...new function.
1551         (pass_cd_dce): Convert from a global struct to a subclass of
1552         gimple_opt_pass along with...
1553         (pass_data_cd_dce): ...new pass_data instance and...
1554         (make_pass_cd_dce): ...new function.
1555         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1556         subclass of gimple_opt_pass along with...
1557         (pass_data_dominator): ...new pass_data instance and...
1558         (make_pass_dominator): ...new function.
1559         (pass_phi_only_cprop): Convert from a global struct to a subclass of
1560         gimple_opt_pass along with...
1561         (pass_data_phi_only_cprop): ...new pass_data instance and...
1562         (make_pass_phi_only_cprop): ...new function.
1563         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1564         subclass of gimple_opt_pass along with...
1565         (pass_data_dse): ...new pass_data instance and...
1566         (make_pass_dse): ...new function.
1567         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1568         a subclass of gimple_opt_pass along with...
1569         (pass_data_forwprop): ...new pass_data instance and...
1570         (make_pass_forwprop): ...new function.
1571         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1572         struct to a subclass of gimple_opt_pass along with...
1573         (pass_data_tree_ifcombine): ...new pass_data instance and...
1574         (make_pass_tree_ifcombine): ...new function.
1575         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1576         subclass of gimple_opt_pass along with...
1577         (pass_data_ch): ...new pass_data instance and...
1578         (make_pass_ch): ...new function.
1579         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1580         subclass of gimple_opt_pass along with...
1581         (pass_data_tree_loop): ...new pass_data instance and...
1582         (make_pass_tree_loop): ...new function.
1583         (pass_tree_loop_init): Convert from a global struct to a subclass of
1584         gimple_opt_pass along with...
1585         (pass_data_tree_loop_init): ...new pass_data instance and...
1586         (make_pass_tree_loop_init): ...new function.
1587         (pass_lim): Convert from a global struct to a subclass of
1588         gimple_opt_pass along with...
1589         (pass_data_lim): ...new pass_data instance and...
1590         (make_pass_lim): ...new function.
1591         (pass_tree_unswitch): Convert from a global struct to a subclass of
1592         gimple_opt_pass along with...
1593         (pass_data_tree_unswitch): ...new pass_data instance and...
1594         (make_pass_tree_unswitch): ...new function.
1595         (pass_predcom): Convert from a global struct to a subclass of
1596         gimple_opt_pass along with...
1597         (pass_data_predcom): ...new pass_data instance and...
1598         (make_pass_predcom): ...new function.
1599         (pass_vectorize): Convert from a global struct to a subclass of
1600         gimple_opt_pass along with...
1601         (pass_data_vectorize): ...new pass_data instance and...
1602         (make_pass_vectorize): ...new function.
1603         (pass_graphite): Convert from a global struct to a subclass of
1604         gimple_opt_pass along with...
1605         (pass_data_graphite): ...new pass_data instance and...
1606         (make_pass_graphite): ...new function.
1607         (pass_graphite_transforms): Convert from a global struct to a subclass
1608         of gimple_opt_pass along with...
1609         (pass_data_graphite_transforms): ...new pass_data instance and...
1610         (make_pass_graphite_transforms): ...new function.
1611         (pass_check_data_deps): Convert from a global struct to a subclass of
1612         gimple_opt_pass along with...
1613         (pass_data_check_data_deps): ...new pass_data instance and...
1614         (make_pass_check_data_deps): ...new function.
1615         (pass_iv_canon): Convert from a global struct to a subclass of
1616         gimple_opt_pass along with...
1617         (pass_data_iv_canon): ...new pass_data instance and...
1618         (make_pass_iv_canon): ...new function.
1619         (pass_scev_cprop): Convert from a global struct to a subclass of
1620         gimple_opt_pass along with...
1621         (pass_data_scev_cprop): ...new pass_data instance and...
1622         (make_pass_scev_cprop): ...new function.
1623         (pass_record_bounds): Convert from a global struct to a subclass of
1624         gimple_opt_pass along with...
1625         (pass_data_record_bounds): ...new pass_data instance and...
1626         (make_pass_record_bounds): ...new function.
1627         (pass_complete_unroll): Convert from a global struct to a subclass of
1628         gimple_opt_pass along with...
1629         (pass_data_complete_unroll): ...new pass_data instance and...
1630         (make_pass_complete_unroll): ...new function.
1631         (pass_complete_unrolli): Convert from a global struct to a subclass of
1632         gimple_opt_pass along with...
1633         (pass_data_complete_unrolli): ...new pass_data instance and...
1634         (make_pass_complete_unrolli): ...new function.
1635         (pass_parallelize_loops): Convert from a global struct to a subclass
1636         of gimple_opt_pass along with...
1637         (pass_data_parallelize_loops): ...new pass_data instance and...
1638         (make_pass_parallelize_loops): ...new function.
1639         (pass_loop_prefetch): Convert from a global struct to a subclass of
1640         gimple_opt_pass along with...
1641         (pass_data_loop_prefetch): ...new pass_data instance and...
1642         (make_pass_loop_prefetch): ...new function.
1643         (pass_iv_optimize): Convert from a global struct to a subclass of
1644         gimple_opt_pass along with...
1645         (pass_data_iv_optimize): ...new pass_data instance and...
1646         (make_pass_iv_optimize): ...new function.
1647         (pass_tree_loop_done): Convert from a global struct to a subclass of
1648         gimple_opt_pass along with...
1649         (pass_data_tree_loop_done): ...new pass_data instance and...
1650         (make_pass_tree_loop_done): ...new function.
1651         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
1652         struct to a subclass of gimple_opt_pass along with...
1653         (pass_data_cse_reciprocals): ...new pass_data instance and...
1654         (make_pass_cse_reciprocals): ...new function.
1655         (pass_cse_sincos): Convert from a global struct to a subclass of
1656         gimple_opt_pass along with...
1657         (pass_data_cse_sincos): ...new pass_data instance and...
1658         (make_pass_cse_sincos): ...new function.
1659         (pass_optimize_bswap): Convert from a global struct to a subclass of
1660         gimple_opt_pass along with...
1661         (pass_data_optimize_bswap): ...new pass_data instance and...
1662         (make_pass_optimize_bswap): ...new function.
1663         (pass_optimize_widening_mul): Convert from a global struct to a
1664         subclass of gimple_opt_pass along with...
1665         (pass_data_optimize_widening_mul): ...new pass_data instance and...
1666         (make_pass_optimize_widening_mul): ...new function.
1667         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
1668         subclass of gimple_opt_pass along with...
1669         (pass_data_phiopt): ...new pass_data instance and...
1670         (make_pass_phiopt): ...new function.
1671         (pass_cselim): Convert from a global struct to a subclass of
1672         gimple_opt_pass along with...
1673         (pass_data_cselim): ...new pass_data instance and...
1674         (make_pass_cselim): ...new function.
1675         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
1676         subclass of gimple_opt_pass along with...
1677         (pass_data_phiprop): ...new pass_data instance and...
1678         (make_pass_phiprop): ...new function.
1679         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
1680         subclass of gimple_opt_pass along with...
1681         (pass_data_pre): ...new pass_data instance and...
1682         (make_pass_pre): ...new function.
1683         (pass_fre): Convert from a global struct to a subclass of
1684         gimple_opt_pass along with...
1685         (pass_data_fre): ...new pass_data instance and...
1686         (make_pass_fre): ...new function.
1687         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
1688         subclass of gimple_opt_pass along with...
1689         (pass_data_reassoc): ...new pass_data instance and...
1690         (make_pass_reassoc): ...new function.
1691         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
1692         subclass of gimple_opt_pass along with...
1693         (pass_data_sink_code): ...new pass_data instance and...
1694         (make_pass_sink_code): ...new function.
1695         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
1696         subclass of gimple_opt_pass along with...
1697         (pass_data_strlen): ...new pass_data instance and...
1698         (make_pass_strlen): ...new function.
1699         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
1700         struct to a subclass of gimple_opt_pass along with...
1701         (pass_data_build_alias): ...new pass_data instance and...
1702         (make_pass_build_alias): ...new function.
1703         (pass_build_ealias): Convert from a global struct to a subclass of
1704         gimple_opt_pass along with...
1705         (pass_data_build_ealias): ...new pass_data instance and...
1706         (make_pass_build_ealias): ...new function.
1707         (pass_ipa_pta): Convert from a global struct to a subclass of
1708         simple_ipa_opt_pass along with...
1709         (pass_data_ipa_pta): ...new pass_data instance and...
1710         (make_pass_ipa_pta): ...new function.
1711         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
1712         subclass of gimple_opt_pass along with...
1713         (pass_data_uncprop): ...new pass_data instance and...
1714         (make_pass_uncprop): ...new function.
1715         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
1716         global struct to a subclass of gimple_opt_pass along with...
1717         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
1718         (make_pass_late_warn_uninitialized): ...new function.
1719         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
1720         to a subclass of gimple_opt_pass along with...
1721         (pass_data_init_datastructures): ...new pass_data instance and...
1722         (make_pass_init_datastructures): ...new function.
1723         (pass_early_warn_uninitialized): Convert from a global struct to a
1724         subclass of gimple_opt_pass along with...
1725         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
1726         (make_pass_early_warn_uninitialized): ...new function.
1727         (pass_update_address_taken): Convert from a global struct to a
1728         subclass of gimple_opt_pass along with...
1729         (pass_data_update_address_taken): ...new pass_data instance and...
1730         (make_pass_update_address_taken): ...new function.
1731         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
1732         struct to a subclass of gimple_opt_pass along with...
1733         (pass_data_release_ssa_names): ...new pass_data instance and...
1734         (make_pass_release_ssa_names): ...new function.
1735         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
1736         subclass of gimple_opt_pass along with...
1737         (pass_data_stdarg): ...new pass_data instance and...
1738         (make_pass_stdarg): ...new function.
1739         * tree-switch-conversion.c (pass_convert_switch): Convert from a
1740         global struct to a subclass of gimple_opt_pass along with...
1741         (pass_data_convert_switch): ...new pass_data instance and...
1742         (make_pass_convert_switch): ...new function.
1743         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
1744         to a subclass of gimple_opt_pass along with...
1745         (pass_data_tail_recursion): ...new pass_data instance and...
1746         (make_pass_tail_recursion): ...new function.
1747         (pass_tail_calls): Convert from a global struct to a subclass of
1748         gimple_opt_pass along with...
1749         (pass_data_tail_calls): ...new pass_data instance and...
1750         (make_pass_tail_calls): ...new function.
1751         * tree-vect-generic.c (pass_lower_vector): Convert from a global
1752         struct to a subclass of gimple_opt_pass along with...
1753         (pass_data_lower_vector): ...new pass_data instance and...
1754         (make_pass_lower_vector): ...new function.
1755         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
1756         gimple_opt_pass along with...
1757         (pass_data_lower_vector_ssa): ...new pass_data instance and...
1758         (make_pass_lower_vector_ssa): ...new function.
1759         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
1760         to a subclass of gimple_opt_pass along with...
1761         (pass_data_slp_vectorize): ...new pass_data instance and...
1762         (make_pass_slp_vectorize): ...new function.
1763         (pass_ipa_increase_alignment): Convert from a global struct to a
1764         subclass of simple_ipa_opt_pass along with...
1765         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
1766         (make_pass_ipa_increase_alignment): ...new function.
1767         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
1768         gimple_opt_pass along with...
1769         (pass_data_vrp): ...new pass_data instance and...
1770         (make_pass_vrp): ...new function.
1771         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
1772         subclass of simple_ipa_opt_pass along with...
1773         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
1774         (make_pass_ipa_free_lang_data): ...new function.
1775         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
1776         gimple_opt_pass along with...
1777         (pass_data_tsan): ...new pass_data instance and...
1778         (make_pass_tsan): ...new function.
1779         (pass_tsan_O0): Convert from a global struct to a subclass of
1780         gimple_opt_pass along with...
1781         (pass_data_tsan_O0): ...new pass_data instance and...
1782         (make_pass_tsan_O0): ...new function.
1783         * var-tracking.c (pass_variable_tracking): Convert from a global
1784         struct to a subclass of rtl_opt_pass along with...
1785         (pass_data_variable_tracking): ...new pass_data instance and...
1786         (make_pass_variable_tracking): ...new function.
1787         * web.c (pass_web): Convert from a global struct to a subclass of
1788         rtl_opt_pass along with...
1789         (pass_data_web): ...new pass_data instance and...
1790         (make_pass_web): ...new function.
1791         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
1792         declaration with that of...
1793         (make_pass_mode_switch_use): ...new function.
1794         (pass_resolve_sw_modes): Replace declaration with that of...
1795         (make_pass_resolve_sw_modes): ...new function.
1796         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
1797         from a global struct to a subclass of rtl_opt_pass along with...
1798         (pass_data_mode_switch_use): ...new pass_data instance and...
1799         (make_pass_mode_switch_use): ...new function.
1800         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
1801         from a global struct to a subclass of rtl_opt_pass along with...
1802         (pass_data_resolve_sw_modes): ...new pass_data instance and...
1803         (make_pass_resolve_sw_modes): ...new function.
1804         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
1805         struct to a subclass of rtl_opt_pass along with...
1806         (pass_data_insert_vzeroupper): ...new pass_data instance and...
1807         (make_pass_insert_vzeroupper): ...new function.
1808         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
1809         global struct to a subclass of rtl_opt_pass along with...
1810         (pass_data_work_around_errata): ...new pass_data instance and...
1811         (make_pass_work_around_errata): ...new function.
1812         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
1813         struct to a subclass of rtl_opt_pass along with...
1814         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
1815         (make_pass_mips_machine_reorg2): ...new function.
1817 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1819         * passes.c (pass_manager::operator new): New.
1821 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1823         Handwritten part of conversion of passes to C++ classes.
1825         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
1826         (toplev.o): Add dep on PASS_MANAGER_H.
1827         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
1828         of early local pases to reflect this moving from a global to a
1829         member of gcc::pass_manager.
1830         (cgraph_add_new_function): Likewise.
1831         * lto-cgraph.c (lto_output_node): Update for conversion of
1832         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
1833         * passes.c (opt_pass::clone): New.
1834         (opt_pass::gate): New.
1835         (opt_pass::execute): New.
1836         (opt_pass::opt_pass): New.
1837         (pass_manager::execute_early_local_passes): New.
1838         (pass_manager::execute_pass_mode_switching): new.
1839         (finish_optimization_passes): Convert to...
1840         (pass_manager::finish_optimization_passes): ...this.
1841         (finish_optimization_passes): Update for conversion of passes to
1842         C++ classes.
1843         (register_dump_files_1): Use has_gate since we cannot portably
1844         check a vtable entry against NULL.
1845         (dump_one_pass): Likewise.
1846         (ipa_write_summaries_2): Likewise.
1847         (ipa_write_optimization_summaries_1): Likewise.
1848         (ipa_read_summaries_1): Likewise.
1849         (ipa_read_optimization_summaries_1): Likewise.
1850         (execute_ipa_stmt_fixups): Likewise.
1851         (pass_manager::pass_manager): Rewrite pass-creation, invoking
1852         pass-creation functions rather than wiring up globals, and
1853         storing the results in fields of pass_manager generated using
1854         pass-instances.def.
1855         (pass_manager::dump_profile_report): Update for conversion of
1856         passes to C++ classes.
1857         (pass_manager::execute_ipa_summary_passes): Likewise.
1858         (execute_one_ipa_transform_pass): Likewise.
1859         (execute_one_pass): Use has_gate and has_execute since we cannot
1860         portably check a vtable entry against NULL.
1861         * pass_manager.h (pass_manager::finish_optimization_passes): New.
1862         (pass_manager): Use pass-instances.def to add fields for the
1863         various pass instances.
1864         * toplev.c (finalize): Update for move of
1865         finish_optimization_passes to a method of gcc::pass_manager.
1866         * toplev.h (finish_optimization_passes): Move to method of class
1867         pass_manager.
1868         * tree-pass.h (struct pass_data): New.
1869         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
1870         (opt_pass::gate): Convert to virtual function.
1871         (opt_pass::~opt_pass): New.
1872         (opt_pass::clone): New.
1873         (opt_pass::execute): Convert to virtual function.
1874         (opt_pass::opt_pass): New.
1875         (opt_pass::ctxt_): new.
1876         (gimple_opt_pass): Convert to subclass of opt_pass.
1877         (gimple_opt_pass::gimple_opt_pass): New.
1878         (rtl_opt_pass): Convert to subclass of opt_pass.
1879         (rtl_opt_pass::rtl_opt_pass): New.
1880         (ipa_opt_pass_d): Convert to subclass of opt_pass.
1881         (ipa_opt_pass_d::ipa_opt_pass_d): New.
1882         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
1883         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
1884         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
1885         invocation of pass_mode_switching to reflect this moving from a
1886         global to a member of gcc::pass_manager.
1887         (ix86_option_override): Rework how pass_insert_vzeroupper is
1888         added to the pass_manager to reflect autogenerated changes.
1889         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
1891 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
1893         PR rtl-optimization/57708
1894         * recog.c (peep2_find_free_register): Validate all regs in a
1895         multi-reg mode.
1897 2013-08-05  Jan Hubicka  <jh@suse.cz>
1899         PR lto/57602
1900         * cgraph.c (verify_cgraph_node): Accept local flags from other
1901         partitions.
1902         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
1903         (function_and_variable_visibility): Likewise.
1904         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
1906 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1908         * graph.c (init_graph_slim_pretty_print): Remove.
1909         (print_graph_cfg): Do not call it.  Use local pretty printer.
1910         (start_graph_dump): Likewise.
1912 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1914         * gimple-pretty-print.c (buffer): Remove.
1915         (initialized): Likewise.
1916         (maybe_init_pretty_print): Likewise.
1917         (print_gimple_stmt): Do not call it.  Use non-static local
1918         pretty_printer variable.
1919         (print_gimple_expr): Likewise.
1920         (print_gimple_seq): Likewise.
1921         (gimple_dump_bb): Likewise.
1923 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1925         * asan.c (asan_pp): Remove.
1926         (asan_pp_initialized): Likewise.
1927         (asan_pp_initialize): Likewise.
1928         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
1929         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
1930         (asan_add_global): Likewise.
1932 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1934         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
1935         * diagnostic.h (diagnostic_flush_buffer): Adjust.
1936         * pretty-print.c (pp_formatted_text_data): Likewise.
1937         (pp_indent): Rename from pp_base_indent.
1938         (pp_format): Rename from pp_base_format.
1939         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
1940         (pp_format_verbatim): Rename from pp_base_format_verbatim.
1941         (pp_flush): Rename from pp_base_flush.
1942         (pp_set_line_maximum_length): Rename from
1943         pp_base_set_line_maximum_length.
1944         (pp_clear_output_area): Rename from pp_base_clear_output_area.
1945         (pp_set_prefix): Rename from pp_base_set_prefix.
1946         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
1947         (pp_emit_prefix): Rename from pp_base_emit_prefix.
1948         (pp_append_text): Rename from pp_base_append_text.
1949         (pp_formatted_text): Rename from pp_base_formatted_text.
1950         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
1951         (pp_remaining_character_count_for_line): Rename from
1952         pp_base_remaining_character_count_for_line.
1953         (pp_newline): Rename from pp_base_newline.
1954         (pp_character): Rename from pp_base_character.
1955         (pp_string): Rename from pp_base_string.
1956         (pp_maybe_space): Rename from pp_base_maybe_space.
1957         * asan.c (asan_pp_string): Adjust.
1958         (asan_emit_stack_protection): Likewise.
1959         (asan_add_global): Likewise.
1960         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
1961         * tree-mudflap.c (mf_varname_tree): Likewise.
1962         * tree-pretty-print.c (pp_tree_identifier): Rename from
1963         pp_base_tree_identifier.
1964         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
1965         Declare as function.
1967 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1969         * pretty-print.h (pp_bar_bar): New.
1970         (pp_ampersand_ampersand): Likewise.
1971         (pp_less_equal): Likewise.
1972         (pp_greater_equal): Likewise.
1973         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
1974         printer functions instead of pp_string or operators and punctuators.
1975         (dump_gimple_call): Likewise.
1976         (dump_gimple_omp_for): Likewise.
1977         (dump_gimple_transaction): Likewise.
1978         (dump_gimple_phi): Likewise.
1979         (pp_gimple_stmt_1): Likewise.
1980         * sched-vis.c (print_insn): Likewise.
1981         * tree-mudflap.c (mf_varname_tree): Likewise.
1982         * tree-pretty-print.c (dump_block_node): Likewise.
1983         (dump_generic_node): Likewise.
1985 2013-08-02  Jan Hubicka  <jh@suse.cz>
1987         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
1988         boundaries.
1989         * lto-streamer-out.c (tree_is_indexable): Results decls and
1990         parm decls are not indexable.
1991         (DFS_write_tree_body): Do not follow args and results.
1992         (hash_tree): Likewise.
1993         (output_functions): Rearrange so struct function is needed
1994         only when real body is output; be able to also ouptut abstract
1995         functions; output DECL_ARGUMENTS and DECL_RESULT.
1996         (lto_output): When not in WPA, ale store abstract functions.
1997         (write_symbol): Do not care about RESULT_DECL.
1998         (output_symbol_p): Handle correctly sbtract decls.
1999         * lto-streamer-in.c (input_function): Rearrange so struct
2000         function can be NULL at entry; allow streaming of
2001         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2002         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2003         sanity check during LTO.
2004         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2005         RESULT_DECl and DECL_ARGUMENTS.
2006         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2007         Likewise.
2009 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2011         * pretty-print.h (pp_underscore): New.
2012         (pp_comma): Tidy.
2013         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2014         printer functions instead of pp_character.
2015         (dump_binary_rhs): Likewise.
2016         (dump_ternary_rhs): Likewise.
2017         (dump_gimple_call_args): Likewise.
2018         (pp_points_to_solution): Likewise.
2019         (dump_gimple_call): Likewise.
2020         (dump_gimple_switch): Likewise.
2021         (dump_gimple_cond): Likewise.
2022         (dump_gimple_bind): Likewise.
2023         (dump_gimple_try): Likewise.
2024         (dump_gimple_omp_for): Likewise.
2025         (dump_gimple_omp_continue): Likewise.
2026         (dump_gimple_omp_single): Likewise.
2027         (dump_gimple_omp_sections): Likewise.
2028         (dump_gimple_omp_block): Likewise.
2029         (dump_gimple_omp_critical): Likewise.
2030         (dump_gimple_transaction): Likewise.
2031         (dump_gimple_asm): Likewise.
2032         (dump_gimple_phi): Likewise.
2033         (dump_gimple_omp_parallel): Likewise.
2034         (dump_gimple_omp_task): Likewise.
2035         (dump_gimple_omp_atomic_load): Likewise.
2036         (dump_gimple_omp_atomic_store): Likewise.
2037         (dump_gimple_mem_ops): Likewise.
2038         (pp_gimple_stmt_1): Likewise.
2039         (pp_cfg_jump): Likewise.
2040         (dump_implicit_edges): Likewise.
2041         (gimple_dump_bb_for_graph): Likewise.
2042         * graph.c (draw_cfg_node): Likewise.
2043         * langhooks.c (lhd_print_error_function): Likewise.
2044         * sched-vis.c (print_exp): Likewise.
2045         (print_value): Likewise.
2046         (print_pattern): Likewise.
2047         (print_insn): Likewise.
2048         (rtl_dump_bb_for_graph): Likewise.
2049         * tree-pretty-print.c (dump_function_declaration): Likewise.
2050         (dump_array_domain): Likewise.
2051         (dump_omp_clause): Likewise.
2052         (dump_location): Likewise.
2053         (dump_generic_node): Likewise.
2054         (print_struct_decl): Likewise.
2055         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2057 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2059         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2060         candidate table when replacing a candidate statement.
2061         (replace_rhs_if_not_dup): Likewise.
2062         (replace_one_candidate): Likewise.
2064 2013-08-02  Jan Hubicka  <jh@suse.cz>
2065             Martin Liska  <marxin.liska@gmail.com>
2067         * cgraphunit.c (add_new_function): Fix logic when adding from
2068         late IPA pass.
2069         (assemble_thunk): Rename to ...
2070         (expand_thunk); .. this one; export; get it working with
2071         general functions; make produced gimple valid.
2072         * cgraph.h (expand_thunk): Declare.
2074 2013-08-02  Jan Hubicka  <jh@suse.cz>
2076         * ipa-cp.c (gather_context_independent_values): Use
2077         ipa_get_param_move_cost.
2078         (get_replacement_map): Remove PARAM; move parameter folding
2079         into tree-inline.c
2080         (create_specialized_node): Update.
2081         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2082         assert that we have gimple body; update move_cost.
2083         (count_formal_params): Assert that we have gimple body.
2084         (ipa_dump_param): New function.
2085         (ipa_alloc_node_params): Break out from ...
2086         (ipa_initialize_node_params): ... here.
2087         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2088         (ipa_write_node_info): Stream move costs.
2089         (ipa_read_node_info): Read move costs.
2090         (ipa_update_after_lto_read): Do not recompute node params.
2091         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2092         (ipa_get_param): Check we are not in WPA.
2093         (ipa_get_param_move_cost): New.
2094         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2095         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2096         parm numbers to be present.
2098 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2100         PR rtl-optimization/58048
2101         * lra-constraints.c (process_alt_operands): Don't check asm
2102         operand on register.
2104 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2106         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2107         the implied StoreLoad barrier for atomic operations if before.
2109 2013-08-02  Jan Hubicka  <jh@suse.cz>
2110             Martin Liska  <marxin.liska@gmail.com>
2112         * cgraph.c (cgraph_function_body_availability): Do not check
2113         cgraph flags.
2114         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2115         symtab_node_availability): Declare.
2116         * ipa.c (can_replace_by_local_alias): New.
2117         (function_and_variable_visibility): Use it.
2118         * symtab.c (symtab_for_node_and_aliases,
2119         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2121 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2123         PR rtl-optimization/57963
2124         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2125         (lra_constraints): Use them.
2127 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2129         * config/arm/types.md (define_attr "type"): Add "load_acq"
2130         and "store_rel".
2131         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2132         changes.
2133         (cortex_a53_store1): Likewise.
2135 2013-08-01  Jan Hubicka  <jh@suse.cz>
2137         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2138         partitions are not needed.
2140 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
2142         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2143         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2144         MAYBE_NON_Q_CLASS_P where appropriate.
2146 2013-08-01  Jan Hubicka  <jh@suse.cz>
2148         * cgraph.h (release_function_body): Declare.
2149         * tree.c (free_lang_data_in_decl): Free, parameters and return values
2150         of unused delcarations.
2152 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2154         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2155         RTL form when subtracting a constant.
2157 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2159         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2160         Generate canonical plus rtx with negated immediate instead of minus
2161         where appropriate.
2162         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2164 2013-08-01  Jan Hubicka  <jh@suse.cz>
2166         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2167         (cgraph_release_function_body): Likewise.
2168         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2169         * cgraph.h (cgrpah_node): Rename abstract_and_needed
2170         to used_as_abstract_origin.
2171         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2172         symbols used as abstract origins.
2173         * cgraphunit.c (analyze_functions): Update.
2174         * ipa.c (symtab_remove_unreachable_nodes): Recompute
2175         used_as_abstract_origin.
2176         * tree-inline.c (tree_function_versioning): Update
2177         used_as_abstract_origin; be ready for DECL_RESULT and
2178         DECL_ARGUMENTS to be NULL.
2180         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2181         for abstract functions.
2182         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2183         real symbols.
2185 2013-08-01  Jan Hubicka  <jh@suse.cz>
2187         * profile.c (compute_value_histograms): Fix thinko.
2189 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2191         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
2192         aarch-common-protos.h to extra_headers.
2193         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2194         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2195         * config/aarch64/t-aarch64 (aarch-common.o): Define.
2197 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2199         * config/aarch64/aarch64.md (define_attr "type"): Delete.
2200         Include "../arm/types.md".  Define "type" attribute for all patterns.
2201         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2202         attribute changes.
2204 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
2206         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2207         to support power8 load fusion.
2208         (fusion_gpr_mem_load): Likewise.
2210         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2212         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2213         declarations for power8 load fusion.
2214         (emit_fusion_gpr_load): Likewise.
2216         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2217         tuning for power8, turn on fusion mode by default.  Turn on sign
2218         extending fusion mode if normal fusion mode is on, and we are at
2219         -O2 or -O3.
2220         (fusion_gpr_load_p): New function, return true if we can fuse an
2221         addis instruction with a dependent load to a GPR.
2222         (emit_fusion_gpr_load): Emit the instructions for power8 load
2223         fusion to GPRs.
2225         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2226         (VSX load fusion peepholes): New peepholes to fuse together an
2227         addi instruction with a VSX load instruction.
2229         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2230         peepholes to fuse an addis instruction with a load to a GPR base
2231         register.  If we are supporting sign extending fusions, convert
2232         sign extending loads to zero extending loads and add an explicit
2233         sign extension.
2235 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2237         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
2238         aarch-common-protos.h to extra_headers.
2239         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2240         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2241         (arm_early_store_addr_dep): Likewise.
2242         (arm_no_early_alu_shift_dep): Likewise.
2243         (arm_no_early_alu_shift_value_dep): Likewise.
2244         (arm_no_early_mul_dep): Likewise.
2245         (arm_no_early_store_addr_dep): Likewise.
2246         (arm_mac_accumulator_is_mul_result): Likewise.
2247         (arm_mac_accumulator_is_result): Likewise.
2248         * config/arm/aarch-common.c: ... here.  New file.
2249         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2250         here to ...
2251         (arm_early_store_addr_dep): Likewise.
2252         (arm_no_early_alu_shift_dep): Likewise.
2253         (arm_no_early_alu_shift_value_dep): Likewise.
2254         (arm_no_early_mul_dep): Likewise.
2255         (arm_no_early_store_addr_dep): Likewise.
2256         (arm_mac_accumulator_is_mul_result): Likewise.
2257         (arm_mac_accumulator_is_result): Likewise.
2258         * config/arm/aarch-common-protos.h: ... here.  New file.
2259         * config/arm/t-arm (aarch-common.o): Define.
2261 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2263         * config/arm/arm.md: Include new file "types.md".
2264         (define_attr "type"): Move from here to ...
2265         (define_attr "mul32"): Likewise.
2266         (define_attr "mul64"): Likewise.
2267         * config/arm/types.md: ... here.  New file.
2269 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2271         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2272         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2274 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2276         * gen-pass-instances.awk: Fix offset of substr().
2278 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
2280         * Makefile.in (pass-instances.def): New.
2281         (passes.o): Replace dependency on passes.def with one on
2282         pass-instances.def
2284         * gen-pass-instances.awk: New.
2286         * passes.c (pass_manager::pass_manager): Use pass-instances.def
2287         rather than passes.def, updating local definition of NEXT_PASS
2288         macro to add an extra NUM parameter (currently unused).
2290 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
2292         * Makefile.in (PASS_MANAGER_H): New.
2293         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2294         (passes.o): Likewise.
2295         (statistics.o): Likewise.
2296         (cgraphunit.o): Likewise.
2297         (context.o): Depend on PASS_MANAGER_H.
2299         * pass_manager.h: New.
2301         * cgraphunit.c (cgraph_add_new_function): Update for moves
2302         of globals to fields of pass_manager.
2303         (analyze_function): Likewise.
2304         (expand_function): Likewise.
2305         (ipa_passes): Likewise.
2306         (compile): Likewise.
2308         * context.c (context::context): New.
2309         * context.h  (context::context): New.
2310         (context::get_passes): New.
2311         (context::passes_): New.
2313         * lto-cgraph.c (input_node): Update for moves of globals to
2314         fields of pass_manager.
2316         * passes.c (all_passes): Remove, in favor of a field of the
2317         same name within the new class pass_manager.
2318         (all_small_ipa_passes): Likewise.
2319         (all_lowering_passes): Likewise.
2320         (all_regular_ipa_passes): Likewise.
2321         (all_late_ipa_passes): Likewise.
2322         (all_lto_gen_passes): Likewise.
2323         (passes_by_id): Likewise.
2324         (passes_by_id_size): Likewise.
2325         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2326         the new class pass_manager.
2327         (set_pass_for_id): Convert to...
2328         (pass_manager::set_pass_for_id): ...method.
2329         (get_pass_for_id): Convert to...
2330         (pass_manager::get_pass_for_id): ...method.
2331         (register_one_dump_file): Move body of implementation into...
2332         (pass_manager::register_one_dump_file): ...here.
2333         (register_dump_files_1): Convert to...
2334         (pass_manager::register_dump_files_1): ...method.
2335         (register_dump_files): Convert to...
2336         (pass_manager::register_dump_files): ...method.
2337         (create_pass_tab): Update for moves of globals to fields of
2338         pass_manager.
2339         (dump_passes): Move body of implementation into...
2340         (pass_manager::dump_passes): ...here.
2341         (register_pass): Move body of implementation into...
2342         (pass_manager::register_pass): ...here.
2343         (init_optimization_passes): Convert into...
2344         (pass_manager::pass_manager): ...constructor for new
2345         pass_manager class, and initialize the pass_lists array.
2346         (check_profile_consistency): Update for moves of globals to
2347         fields of pass_manager.
2348         (dump_profile_report): Move body of implementation into...
2349         (pass_manager::dump_profile_report): ...here.
2350         (ipa_write_summaries_1): Update for moves of pass lists from
2351         being globals to fields of pass_manager.
2352         (ipa_write_optimization_summaries): Likewise.
2353         (ipa_read_summaries):  Likewise.
2354         (ipa_read_optimization_summaries): Likewise.
2355         (execute_all_ipa_stmt_fixups): Likewise.
2357         * statistics.c (statistics_fini): Update for moves of globals to
2358         fields of pass_manager.
2360         * toplev.c (general_init): Replace call to
2361         init_optimization_passes with construction of the pass_manager
2362         instance.
2364         * tree-pass.h (all_passes): Remove, in favor of a field of the
2365         same name within the new class pass_manager.
2366         (all_small_ipa_passes): Likewise.
2367         (all_lowering_passes): Likewise.
2368         (all_regular_ipa_passes): Likewise.
2369         (all_lto_gen_passes): Likewise.
2370         (all_late_ipa_passes): Likewise.
2371         (passes_by_id): Likewise.
2372         (passes_by_id_size): Likewise.
2373         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2374         the new class pass_manager.
2375         (get_pass_for_id): Remove.
2377 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2379         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2380         configs.
2382 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2384         * arm.md (mulhi3): New expand pattern.
2386 2013-07-30  Jan Hubicka  <jh@suse.cz>
2387             Martin Liska  <marxin.liska@gmail.com>
2389         * profile.c (compute_value_histograms): Do not ICE when
2390         there is mismatch only on some counters.
2392 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2394         PR rtl-optimization/57637
2395         * function.c (move_insn_for_shrink_wrap): Also check the
2396         GEN set of the LIVE problem for the liveness analysis
2397         if it exists, otherwise give up.
2399 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2401         PR tree-optimization/57993
2402         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2403         replaced statement in the candidate table.
2404         (phi_add_costs): Return infinite cost when the hidden basis does
2405         not dominate all phis on which the candidate is dependent.
2406         (replace_one_candidate): Record replaced statement in the
2407         candidate table.
2409 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
2411         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2412         (ashlv2si3): New expander.
2413         (*ashlv2si3_i): New define_insn_and_split.
2414         * predicates.md (float_operation): Allow patterns with three
2415         basic sub-patterns.
2417         PR rtl-optimization/58021
2418         * mode-switching.c (create_pre_exit): Always split off preceding
2419         insns if we are not at the basic block head.
2421 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
2423         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2424         (UCLIBC_DYNAMIC_LINKER): New macro.
2425         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2426         `-mnan=2008'.
2427         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2428         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
2429         `-mnan=2008'.
2430         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2431         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2432         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2433         for SF and DF modes.  Use ieee_quad_format for TF mode.
2434         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2435         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2436         (mips_option_override): Handle `-mnan=legacy'.
2437         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2438         `-mabs=2008' and `-mnan=2008'.
2439         (OPTION_DEFAULT_SPECS): Add "nan" default.
2440         (ASM_SPEC): Handle `-mnan='.
2441         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2442         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2443         comment accordingly.
2444         (neg<mode>2): Likewise.
2445         * config/mips/mips.opt (mabs, mnan): New options.
2446         * doc/install.texi (Configuration): Document `--with-nan=' option.
2447         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2448         `-mnan=' options.
2449         (MIPS Options): Document them.
2450         * config.gcc <mips*-*-*>: Handle `--with-nan='.
2451         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2452         * configure: Regenerate.
2453         * config.in: Regenerate.
2455 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2457         * config/i386/i386.md (float post-reload splitters): Do not check
2458         for subregs of SSE registers.
2460 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2461             H.J. Lu  <hongjiu.lu@intel.com>
2463         PR target/57954
2464         PR target/57988
2465         * config/i386/i386.md (post-reload splitter
2466         to avoid partial SSE reg dependency stalls): New pattern.
2468 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2470         * config/s390/s390.md ("movcc"): Swap load and store instructions.
2472 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
2474         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2475         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2477 2013-07-26  Cary Coutant  <ccoutant@google.com>
2479         * dwarf2out.c (die_checksum_ordered): Don't include template
2480         instantiations in signature.
2481         (is_template_parameter): New function.
2482         (is_template_instantiation): New function.
2483         (generate_skeleton_bottom_up): Don't include template instantiations
2484         in type unit DIE.
2485         (generate_skeleton): Likewise.
2486         (break_out_comdat_types): Move recursive call to break out nested
2487         types earlier.
2488         (prune_unused_types_mark_generic_parms_dies): Call
2489         is_template_parameter.
2491 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
2493         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2494         uses vector registers.
2495         * config/aarch64/iterators.md: Add attributes rtn and vas.
2497 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2498             Richard Earnshaw  <richard.earnshaw@arm.com>
2500         * combine.c (simplify_comparison): Re-canonicalize operands
2501         where appropriate.
2502         * config/arm/arm.md (movcond_addsi): New splitter.
2504 2013-07-25  Sterling Augustine  <saugustine@google.com>
2506         * dwarf2out.c (size_of_pubnames): Move code to...
2507         (include_pubname_in_output): ...here.  New.
2508         (want_pubnames): Rearrange.
2509         (output_pubnames): Call include_pubname_in_output.  Move assertion.
2511 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2513         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2515 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2517         PR target/38836
2518         * doc/extend.texi: Remove obsolete builtins. Fix
2519         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2521 2013-07-25  Jan Hubicka  <jh@suse.cz>
2523         * cgraph.c (release_function_body): Break out from ...
2524         (cgraph_release_function_body): ... this one; also release DECL_RESULT
2525         and DECL_ARGUMENTS.
2526         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2527         old_tree in the map.
2528         (create_specialized_node): Update.
2529         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2530         into index.
2531         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2532         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2533         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2534         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2535         DECL_RESULT.
2537 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2539         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2540         addsi3_carryin_alt2_<optab>): Correct output template.
2542 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2544         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2545         Adjust for arm_restrict_it.
2546         Remove trailing whitespace.
2548 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
2550         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2551         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2553         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2555 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
2557         PR rtl-optimization/57960
2558         * lra-constraints.c (process_alt_operands): Use the right mode
2559         when checking strict_low.
2561 2013-07-25  Jan Hubicka  <jh@suse.cz>
2563         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2564         * cgraph.c (cgraph_remove_node): Do not release function body
2565         when in cgraph streaming.
2566         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2567         in other partitions are not considered reachable; fix handling of
2568         clones.
2570 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2572         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2574 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2576         PR target/19599
2577         PR target/57731
2578         PR target/57837
2579         * config/arm/arm.md ("*sibcall_insn): Replace use of
2580         Ss with US. Adjust output for v5 and v4t.
2581         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2583         * config/arm/constraints.md ("Ss"): Rename to US.
2585 2013-07-25  Terry Guo  <terry.guo@arm.com>
2587         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2588         shift_add/shift_sub0/shift_sub1 RTXs.
2590 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
2591             Anton Blanchard <anton@au1.ibm.com>
2593         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2594         (altivec_vpks<VI_char>ss): Likewise.
2595         (altivec_vpks<VI_char>us): Likewise.
2596         (altivec_vpku<VI_char>us): Likewise.
2597         (altivec_vpku<VI_char>um): Likewise.
2599 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
2601         Introduce context class.
2603         * Makefile.in (CONTEXT_H): New.
2604         (OBJS): Add context.o.
2605         (toplev.o): Add CONTEXT_H to dependencies.
2606         (context.o): New.
2608         * toplev.c (general_init): Create the singleton gcc::context instance.
2610         * context.c: New.
2612         * context.h: New.
2614 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2616         PR rtl-optimization/57968
2617         * mode-switching.c (create_pre_exit): Allow instructions that
2618         don't set a return register to need a non-exit mode.
2620 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2621             Anton Blanchard <anton@au1.ibm.com>
2623         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
2624         operands to vperm for little endian.
2625         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
2626         of lvsl to create the control mask for a vperm for little endian.
2628 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2629             Anton Blanchard <anton@au1.ibm.com>
2631         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2632         two operands for little-endian.
2634 2013-07-23  Steve Ellcey  <sellcey@mips.com>
2636         * config/mips/mips.c (mips_case_values_threshold): New.
2637         (TARGET_CASE_VALUES_THRESHOLD): Define.
2639 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2640             Anton Blanchard <anton@au1.ibm.com>
2642         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
2643         selection of field for vector splat in little endian mode.
2645 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
2647         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
2648         expanders to rs6000.md.
2649         (ior<mode>3): Likewise.
2650         (and<mode>3): Likewise.
2651         (one_cmpl<mode>2): Likewise.
2652         (nor<mode>3): Likewise.
2653         (andc<mode>3): Likewise.
2654         (eqv<mode>3): Likewise.
2655         (nand<mode>3): Likewise.
2656         (orc<mode>3): Likewise.
2658         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
2659         declaration.
2661         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
2662         to split multi-word logical operations.
2663         (rs6000_split_logical_di): Likewise.
2664         (rs6000_split_logical): Likewise.
2666         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
2667         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
2668         and allow TImode operations in 32-bit.
2669         (vsx_and<mode>3_64bit): Likewise.
2670         (vsx_ior<mode>3_32bit): Likewise.
2671         (vsx_ior<mode>3_64bit): Likewise.
2672         (vsx_xor<mode>3_32bit): Likewise.
2673         (vsx_xor<mode>3_64bit): Likewise.
2674         (vsx_one_cmpl<mode>2_32bit): Likewise.
2675         (vsx_one_cmpl<mode>2_64bit): Likewise.
2676         (vsx_nor<mode>3_32bit): Likewise.
2677         (vsx_nor<mode>3_64bit): Likewise.
2678         (vsx_andc<mode>3_32bit): Likewise.
2679         (vsx_andc<mode>3_64bit): Likewise.
2680         (vsx_eqv<mode>3_32bit): Likewise.
2681         (vsx_eqv<mode>3_64bit): Likewise.
2682         (vsx_nand<mode>3_32bit): Likewise.
2683         (vsx_nand<mode>3_64bit): Likewise.
2684         (vsx_orc<mode>3_32bit): Likewise.
2685         (vsx_orc<mode>3_64bit): Likewise.
2687         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
2688         logical types in GPRs.
2690         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
2691         logical insns to rs6000.md, and allow TImode operations in
2692         32-bit.
2693         (altivec_ior<mode>3): Likewise.
2694         (altivec_xor<mode>3): Likewise.
2695         (altivec_one_cmpl<mode>2): Likewise.
2696         (altivec_nor<mode>3): Likewise.
2697         (altivec_andc<mode>3): Likewise.
2699         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
2700         attributes for moving the 128-bit logical operations into
2701         rs6000.md.
2702         (BOOL_REGS_OUTPUT): Likewise.
2703         (BOOL_REGS_OP1): Likewise.
2704         (BOOL_REGS_OP2): Likewise.
2705         (BOOL_REGS_UNARY): Likewise.
2706         (BOOL_REGS_AND_CR0): Likewise.
2707         (one_cmpl<mode>2): Add support for DI logical operations on
2708         32-bit, splitting the operations to 32-bit.
2709         (anddi3): Likewise.
2710         (iordi3): Likewise.
2711         (xordi3): Likewise.
2712         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
2713         changes to combine the 32/64-bit code, allow logical operations on
2714         TI mode in 32-bit, and to use similar match_operator patterns like
2715         scalar mode uses.  Combine the Altivec and VSX code for logical
2716         operations, and move it here.
2717         (ior<mode>3, 128-bit types): Likewise.
2718         (xor<mode>3, 128-bit types): Likewise.
2719         (one_cmpl<mode>3, 128-bit types): Likewise.
2720         (nor<mode>3, 128-bit types): Likewise.
2721         (andc<mode>3, 128-bit types): Likewise.
2722         (eqv<mode>3, 128-bit types): Likewise.
2723         (nand<mode>3, 128-bit types): Likewise.
2724         (orc<mode>3, 128-bit types): Likewise.
2725         (and<mode>3_internal): Likewise.
2726         (bool<mode>3_internal): Likewise.
2727         (boolc<mode>3_internal1): Likewise.
2728         (boolc<mode>3_internal2): Likewise.
2729         (boolcc<mode>3_internal1): Likewise.
2730         (boolcc<mode>3_internal2): Likewise.
2731         (eqv<mode>3_internal1): Likewise.
2732         (eqv<mode>3_internal2): Likewise.
2733         (one_cmpl1<mode>3_internal): Likewise.
2735 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2737         * config/microblaze/microblaze.c (microblaze_expand_prologue):
2738         Rename flag_stack_usage to flag_stack_usage_info.
2740 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2742         * config/microblaze/sync.md: New file.
2743         * config/microblaze/microblaze.md: Include sync.md
2744         * config/microblaze/microblaze.c: Add print_operand 'y'.
2745         * config/microblaze/constraints.md: Add memory_contraint
2746         'Q' which is a single register.
2748 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
2750         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
2752 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2754         * reload.c (find_reloads): Exit loop once we find this operand
2755         cannot be reloaded somehow for this alternative.
2757         * reload.c (find_reloads): Exit loop once we find a hard register.
2759         * rtlanal.c (computed_jump_p): Exit loop once we find label
2760         reference is used.
2762         * i386.c (ix86_pad_returns): Exit loop after setting replace.
2764         * cfgloopmanip.c (remove_path): Exit loop after setting
2765         irred_invalidated.
2767         * gensupport.c (subst_dup): Avoid loop if code is not
2768         MATCH_DUP nor MATCH_OP_DUP.
2770 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
2772         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
2774 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2776         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
2777         true for SP_REGNUM if mode == ptr_mode.
2778         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
2779         with value R0_REGNUM + 31.
2781 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2783         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
2784         pad pointer-typed argument downward.
2786 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2788         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
2789         and __ILP32__ when the ILP32 model is in use.
2791 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2793         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
2794         (aarch64_load_symref_appropriately): In the case of
2795         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
2796         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
2797         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
2798         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
2799         if 'mode' doesn't equal to 'ptr_mode'.
2800         (aarch64_output_mi_thunk): Add an assertion on the alignment of
2801         'vcall_offset'; change to call aarch64_emit_move differently depending
2802         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
2803         to calculate the upper bound of 'vcall_offset'.
2804         (aarch64_cannot_force_const_mem): Change to also return true if
2805         mode != ptr_mode.
2806         (aarch64_legitimize_reload_address): In the case of large
2807         displacements, add new local variable 'xmode' and an assertion
2808         based on it; change to use 'xmode' to generate the new rtx and
2809         reload.
2810         (aarch64_asm_trampoline_template): Change to generate the template
2811         differently depending on TARGET_ILP32 or not; change to use
2812         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
2813         (aarch64_trampoline_size): Removed.
2814         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
2815         and replace immediate literals with it.  Change to use 'ptr_mode'
2816         instead of 'DImode' and call convert_memory_address if the mode
2817         of 'fnaddr' doesn't equal to 'ptr_mode'.
2818         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
2819         to output symbol.
2820         (aarch64_elf_asm_destructor): Likewise.
2821         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
2822         on TARGET_ILP32 instead of aarch64_trampoline_size.
2823         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
2824         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
2825         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
2826         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
2827         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
2828         (storewb_pair<GPI:mode>_<P:mode>): ... this.
2829         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
2830         depending on the value of 'mode'.
2831         (add_losym_<mode>): New.
2832         (ldr_got_small_<mode>): New, based on ldr_got_small.
2833         (ldr_got_small): Remove.
2834         (ldr_got_small_sidi): New.
2835         * config/aarch64/iterators.md (P): New.
2836         (PTR): Change to 'ptr_mode' in the condition.
2838 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2840         * config.gcc (aarch64*-*-*): Support --with-abi.
2841         (aarch64*-*-elf): Support --with-multilib-list.
2842         (aarch64*-*-linux*): Likewise.
2843         (supported_defaults): Add abi to aarch64*-*-*.
2844         * configure.ac: Mention AArch64 for --with-multilib-list.
2845         * configure: Re-generated.
2846         * config/aarch64/biarchilp32.h: New file.
2847         * config/aarch64/biarchlp64.h: New file.
2848         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
2849         (ABI_SPEC): Ditto.
2850         (MULTILIB_DEFAULTS): Ditto.
2851         (DRIVER_SELF_SPECS): Ditto.
2852         (ASM_SPEC): Update to also substitute -mabi.
2853         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
2854         file whose name depends on -mabi= and -mbig-endian.
2855         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
2856         TARGET_ILP32.
2857         (POINTER_SIZE): New define.
2858         (POINTERS_EXTEND_UNSIGNED): Ditto.
2859         (enum aarch64_abi_type): New enumeration tag.
2860         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
2861         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
2862         (TARGET_ILP32): New define.
2863         * config/aarch64/aarch64.opt (mabi): New.
2864         (aarch64_abi): New.
2865         (ilp32, lp64): New values for -mabi.
2866         * config/aarch64/t-aarch64 (comma): New define.
2867         (MULTILIB_OPTIONS): Ditto.
2868         (MULTILIB_DIRNAMES): Ditto.
2869         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
2870         * doc/invoke.texi: Document -mabi for AArch64.
2872 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
2874         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
2876 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2877             Anton Blanchard <anton@au1.ibm.com>
2879         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
2880         endianness when selecting field to splat.
2882 2013-07-22  Eric Christopher  <echristo@gmail.com>
2884         * dwarf2out.c (die_odr_checksum): New function to use
2885         CHECKSUM_ macros and ULEB128 for DIE tag.
2886         (generate_type_signature): Use.
2888 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
2890         * config.gcc (sparc*-*-*): Accept leon3 processor.
2891         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
2892         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
2893         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
2894         * config/sparc/sparc.opt (enum processor_type): Add leon3.
2895         (mfix-ut699): Adjust comment.
2896         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
2897         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
2898         (CPP_CPU_SPEC): Likewise.
2899         (ASM_CPU_SPEC): Likewise.
2900         * config/sparc/sparc.c (leon3_cost): New constant.
2901         (sparc_option_override): Add leon3 support.
2902         (mem_ref): New function.
2903         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
2904         (sparc_do_work_around_errata): Look into the instruction in the delay
2905         slot and adjust accordingly.  Add fix for the data cache nullify issues
2906         of the UT699.  Change insertion position for the NOP.
2907         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
2908         (leon3_load): New reservation.
2909         (leon_store): Bump latency to 2.
2910         (grfpu): New automaton.
2911         (grfpu_alu): New unit.
2912         (grfpu_ds): Likewise.
2913         (leon_fp_alu): Adjust.
2914         (leon_fp_mult): Delete.
2915         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
2916         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
2917         * config/sparc/sparc.md (cpu): Add leon3.
2918         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
2919         (swapsi): Likewise.
2920         (atomic_test_and_set): Likewise.
2921         (ldstub): Likewise.
2923 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
2925         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
2926         default for R5900 targets.
2927         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
2928         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
2929         * config/mips/mips.c (mips_option_override): Report an error for
2930         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
2931         for -march=r5900 -mhard-float.
2933 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2935         * df-problems.c (can_move_insns_across): Exit loop once we
2936         find a non-fixed, non-global register.
2938         * ipa-pure-const.c (propagate_nothrow): Exit loop after
2939         setting can_throw.
2941         * omega.c (omega_eliminate_red): Break after setting red_found.
2942         (omega_problem_has_red_equations): Similarly after setting found.
2943         (omega_query_variable): Similarly after setting coupled.
2945 2013-07-22  Marek Polacek  <polacek@redhat.com>
2947         * gimplify.c: Don't include gimple.h twice.
2949 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2951         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
2952         instead of TARGET_THUMB1.
2953         (Pz): New constraint.
2954         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
2955         encodings.
2956         (compare_negsi_si): Likewise.
2957         (compare_addsi2_op0): Likewise.
2958         (compare_addsi2_op1): Likewise.
2959         (addsi3_carryin_<optab>): Likewise.
2960         (addsi3_carryin_alt2_<optab>): Likewise.
2961         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
2962         for arm_restrict_it.
2963         (subsi3_carryin): Likewise.
2964         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
2965         (minmax_arithsi): Disable for arm_restrict_it.
2966         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
2967         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
2968         (satsi_<SAT:code>_shift): Likewise.
2969         (arm_shiftsi3): Add alternative for 16-bit encoding.
2970         (arm32_movhf): Disable for arm_restrict_it.
2971         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
2972         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
2974 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
2976         * config/arm/arm.md (attribute "insn"): Delete.
2977         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
2978         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
2979         (not_shiftsi): Update for attribute change.
2980         (not_shiftsi_compare0): Likewise.
2981         (not_shiftsi_compare0_scratch): Likewise.
2982         (arm_one_cmplsi2): Likewise.
2983         (thumb1_one_cmplsi2): Likewise.
2984         (notsi_compare0): Likewise.
2985         (notsi_compare0_scratch): Likewise.
2986         (thumb1_movdi_insn): Likewise.
2987         (arm_movsi_insn): Likewise.
2988         (movhi_insn_arch4): Likewise.
2989         (movhi_bytes): Likewise.
2990         (arm_movqi_insn): Likewise.
2991         (thumb1_movqi_insn): Likewise.
2992         (arm32_movhf): Likewise.
2993         (thumb1_movhf): Likewise.
2994         (arm_movsf_soft_insn): Likewise.
2995         (thumb1_movsf_insn): Likewise.
2996         (thumb_movdf_insn): Likewise.
2997         (movsicc_insn): Likewise.
2998         (movsfcc_soft_insn): Likewise.
2999         (and_scc): Likewise.
3000         (cond_move): Likewise.
3001         (if_move_not): Likewise.
3002         (if_not_move): Likewise.
3003         (if_shift_move): Likewise.
3004         (if_move_shift): Likewise.
3005         (if_shift_shift): Likewise.
3006         (if_not_arith): Likewise.
3007         (if_arith_not): Likewise.
3008         (cond_move_not): Likewise.
3009         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3010         (neon_mov<mode>): Likewise.
3011         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3012         (thumb2_movsi_vfp): Likewise.
3013         (movsf_vfp): Likewise.
3014         (thumb2_movsf_vfp): Likewise.
3015         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3016         change.
3017         (cortexa7_older_only): Likewise.
3018         (cortexa7_younger): Likewise.
3019         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3020         (1020alu_shift_op): Likewise.
3021         (1020alu_shift_reg_op): Likewise.
3022         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3023         (alu_shift_op): Likewise.
3024         (alu_shift_reg_op): Likewise.
3025         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3026         (11_alu_shift_op): Likewise.
3027         (11_alu_shift_reg_op): Likewise.
3028         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3029         (9_alu_shift_reg_op): Likewise.
3030         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3031         change.
3032         (cortex_a15_alu_shift): Likewise.
3033         (cortex_a15_alu_shift_reg): Likewise.
3034         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3035         (cortex_a5_alu_shift): Likewise.
3036         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3037         change.
3038         (cortex_a53_alu_shift): Likewise.
3039         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3040         change.
3041         (cortex_a7_alu_reg): Likewise.
3042         (cortex_a7_alu_shift): Likewise.
3043         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3044         (cortex_a8_alu_shift): Likewise.
3045         (cortex_a8_alu_shift_reg): Likewise.
3046         (cortex_a8_mov): Likewise.
3047         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3048         (cortex_a9_dp_shift): Likewise.
3049         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3050         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3051         (cortex_r4_mov): Likewise.
3052         (cortex_r4_alu_shift): Likewise.
3053         (cortex_r4_alu_shift_reg): Likewise.
3054         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3055         (526_alu_shift_op): Likewise.
3056         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3057         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3058         (626te_alu_shift_op): Likewise.
3059         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3060         (726te_alu_op): Likewise.
3061         (726te_alu_shift_op): Likewise.
3062         (726te_alu_shift_reg_op): Likewise.
3063         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3064         (mp626_alu_shift_op): Likewise.
3065         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3066         (pj4_alu_e1_conds): Likewise.
3067         (pj4_alu): Likewise.
3068         (pj4_alu_conds): Likewise.
3069         (pj4_shift): Likewise.
3070         (pj4_shift_conds): Likewise.
3071         (pj4_alu_shift): Likewise.
3072         (pj4_alu_shift_conds): Likewise.
3074 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3076         * config/arm/predicates.md (shiftable_operator_strict_it):
3077         New predicate.
3078         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3079         Disable cond_exec version for arm_restrict_it.
3080         (thumb2_smaxsi3): Convert to generate cond_exec.
3081         (thumb2_sminsi3): Likewise.
3082         (thumb32_umaxsi3): Likewise.
3083         (thumb2_uminsi3): Likewise.
3084         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3085         (thumb2_neg_abssi2): Likewise.
3086         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3087         (thumb2_movsicc_insn): Adjust alternatives.
3088         (thumb2_mov_negscc): Disable for arm_restrict_it.
3089         (thumb2_mov_negscc_strict_it): New pattern.
3090         (thumb2_mov_notscc_strict_it): New pattern.
3091         (thumb2_mov_notscc): Disable for arm_restrict_it.
3092         (thumb2_ior_scc): Likewise.
3093         (thumb2_ior_scc_strict_it): New pattern.
3094         (thumb2_cond_move): Adjust for arm_restrict_it.
3095         (thumb2_cond_arith): Disable for arm_restrict_it.
3096         (thumb2_cond_arith_strict_it): New pattern.
3097         (thumb2_cond_sub): Adjust for arm_restrict_it.
3098         (thumb2_movcond): Likewise.
3099         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3100         (thumb2_zero_extendhisi2_v6): Likewise.
3101         (thumb2_zero_extendqisi2_v6): Likewise.
3102         (orsi_notsi_si): Likewise.
3103         (orsi_not_shiftsi_si): Likewise.
3105 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3107         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3108         instruction sequence is 1 byte shorter.
3110 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3112         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3113         it is not needed after split.
3115 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3117         PR target/51784
3118         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3119         second label for nonlocal goto receivers. Don't output pic base labels
3120         unless we're producing PIC; mark that action unreachable().
3121         (ix86_save_reg): If the function contains a nonlocal label, save the
3122         PIC base reg.
3123         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3124         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3125         (update_pic_label_number_if_needed): New.
3126         (machopic_output_function_base_name): Adjust for nonlocal receiver
3127         case.
3128         (machopic_should_output_picbase_label): New.
3129         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3130         (nonlocal_goto_receiver): New insn and split.
3132 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3134         * config/aarch64/aarch64-builtins.c
3135         (aarch64_fold_builtin): Fold abs in all modes.
3136         * config/aarch64/aarch64-simd-builtins.def
3137         (abs): Enable for all modes.
3138         * config/aarch64/arm_neon.h
3139         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3140         (vabs_f64): Add missing intrinsic.
3142 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
3144         * config/aarch64/arm_neon.h (vabs_s64): New function
3146 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
3148         PR target/57516
3149         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3150         * config/avr/avr.md (adjust_len): Add `round'.
3151         * config/avr/avr-protos.h (avr_out_round): New prototype.
3152         (avr_out_plus): Add `out_label' argument.
3153         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3154         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3155         Handle the case where `insn' is just a pattern.
3156         (avr_out_bitop): Handle the case where `insn' is just a pattern.
3157         (avr_out_round): New function.
3158         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3160 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
3162         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3163         Add check for flag_stack_usage to handle -fstack-usage support
3165 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
3167         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3168         interaction for new Power8 flags and VSX.
3170 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
3172         PR middle-end/57698
3173         * tree-inline.c (expand_call_inline): Emit errors during
3174         early_inlining only if optimization is not turned on.
3176 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3178         * passes.def: New.
3180         * passes.c (init_optimization_passes): Move the construction of
3181         the pass hierarchy into a new passes.def file.
3183         * Makefile.in (passes.o): Add dependency on passes.def.
3185 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3187         * passes.c (init_optimization_passes): Introduce macros for
3188         constructing the tree of passes (INSERT_PASSES_AFTER,
3189         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3190         TERMINATE_PASS_LIST).
3192 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
3193             Wei Mi  <wmi@google.com>
3195         PR rtl-optimization/57878
3196         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3197         top.
3198         (reload_pseudo_compare_func): Check nregs first for reload
3199         pseudos.
3201 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3203         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3205 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
3207         * read-rtl.c (validate_const_int): Once an invalid character is
3208         seen, quit the loop.
3210         * gengtype.c (write_roots): Similarly once we find the "deletable"
3211         or "if_marked" option.
3213 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3215         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3216         "xtab" and "sat".  Move value "clz" from here to ...
3217         (attriubte "type"): ... here.
3218         (satsi_<SAT:code>): Delete "insn" attribute.
3219         (satsi_<SAT:code>_shift): Likewise.
3220         (arm_zero_extendqisi2addsi): Likewise.
3221         (arm_extendqisi2addsi): Likewise.
3222         (clzsi2): Update for attribute changes.
3223         (rbitsi2): Likewise.
3224         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3225         attribute.
3226         (arm_usatsihi): Likewise.
3227         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3229 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3231         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3232         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
3233         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
3234         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
3235         in alphabetical order.
3236         (attribute "core_cycles"): Update for attribute changes.
3237         (arm_addsi3): Likewise.
3238         (addsi3_compare0): Likewise.
3239         (addsi3_compare0_scratch): Likewise.
3240         (addsi3_compare_op1): Likewise.
3241         (addsi3_compare_op2): Likewise.
3242         (compare_addsi2_op0): Likewise.
3243         (compare_addsi2_op1): Likewise.
3244         (addsi3_carryin_shift_<optab>): Likewise.
3245         (subsi3_carryin_shift): Likewise.
3246         (rsbsi3_carryin_shift): Likewise.
3247         (arm_subsi3_insn): Likewise.
3248         (subsi3_compare0): Likewise.
3249         (subsi3_compare): Likewise.
3250         (arm_andsi3_insn): Likewise.
3251         (thumb1_andsi3_insn): Likewise.
3252         (andsi3_compare0): Likewise.
3253         (andsi3_compare0_scratch): Likewise.
3254         (zeroextractsi_compare0_scratch
3255         (andsi_not_shiftsi_si): Likewise.
3256         (iorsi3_insn): Likewise.
3257         (iorsi3_compare0): Likewise.
3258         (iorsi3_compare0_scratch): Likewise.
3259         (arm_xorsi3): Likewise.
3260         (thumb1_xorsi3_insn): Likewise.
3261         (xorsi3_compare0): Likewise.
3262         (xorsi3_compare0_scratch): Likewise.
3263         (satsi_<SAT:code>_shift): Likewise.
3264         (rrx): Likewise.
3265         (arm_shiftsi3): Likewise.
3266         (shiftsi3_compare0): Likewise.
3267         (not_shiftsi): Likewise.
3268         (not_shiftsi_compare0): Likewise.
3269         (not_shiftsi_compare0_scratch): Likewise.
3270         (arm_one_cmplsi2): Likewise.
3271         (thumb_one_complsi2): Likewise.
3272         (notsi_compare0): Likewise.
3273         (notsi_compare0_scratch): Likewise.
3274         (thumb1_zero_extendhisi2): Likewise.
3275         (arm_zero_extendhisi2): Likewise.
3276         (arm_zero_extendhisi2_v6): Likewise.
3277         (arm_zero_extendhisi2addsi): Likewise.
3278         (thumb1_zero_extendqisi2): Likewise.
3279         (thumb1_zero_extendqisi2_v6): Likewise.
3280         (arm_zero_extendqisi2): Likewise.
3281         (arm_zero_extendqisi2_v6): Likewise.
3282         (arm_zero_extendqisi2addsi): Likewise.
3283         (thumb1_extendhisi2): Likewise.
3284         (arm_extendhisi2): Likewise.
3285         (arm_extendhisi2_v6): Likewise.
3286         (arm_extendqisi): Likewise.
3287         (arm_extendqisi_v6): Likewise.
3288         (arm_extendqisi2addsi): Likewise.
3289         (thumb1_extendqisi2): Likewise.
3290         (thumb1_movdi_insn): Likewise.
3291         (arm_movsi_insn): Likewise.
3292         (movsi_compare0): Likewise.
3293         (movhi_insn_arch4): Likewise.
3294         (movhi_bytes): Likewise.
3295         (arm_movqi_insn): Likewise.
3296         (thumb1_movqi_insn): Likewise.
3297         (arm32_movhf): Likewise.
3298         (thumb1_movhf): Likewise.
3299         (arm_movsf_soft_insn): Likewise.
3300         (thumb1_movsf_insn): Likewise.
3301         (movdf_soft_insn): Likewise.
3302         (thumb_movdf_insn): Likewise.
3303         (arm_cmpsi_insn): Likewise.
3304         (cmpsi_shiftsi): Likewise.
3305         (cmpsi_shiftsi_swp): Likewise.
3306         (arm_cmpsi_negshiftsi_si): Likewise.
3307         (movsicc_insn): Likewise.
3308         (movsfcc_soft_insn): Likewise.
3309         (arith_shiftsi): Likewise.
3310         (arith_shiftsi_compare0
3311         (arith_shiftsi_compare0_scratch
3312         (sub_shiftsi): Likewise.
3313         (sub_shiftsi_compare0
3314         (sub_shiftsi_compare0_scratch
3315         (and_scc): Likewise.
3316         (cond_move): Likewise.
3317         (if_plus_move): Likewise.
3318         (if_move_plus): Likewise.
3319         (if_move_not): Likewise.
3320         (if_not_move): Likewise.
3321         (if_shift_move): Likewise.
3322         (if_move_shift): Likewise.
3323         (if_shift_shift): Likewise.
3324         (if_not_arith): Likewise.
3325         (if_arith_not): Likewise.
3326         (cond_move_not): Likewise.
3327         (thumb1_ashlsi3): Set type attribute.
3328         (thumb1_ashrsi3): Likewise.
3329         (thumb1_lshrsi3): Likewise.
3330         (thumb1_rotrsi3): Likewise.
3331         (shiftsi3_compare0_scratch): Likewise.
3332         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3333         (neon_mov<mode>): Likewise.
3334         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3335         attribute changes.
3336         (thumb2_movsi_insn): Likewise.
3337         (thumb2_cmpsi_neg_shiftsi): Likewise.
3338         (thumb2_extendqisi_v6): Likewise.
3339         (thumb2_zero_extendhisi2_v6): Likewise.
3340         (thumb2_zero_extendqisi2_v6): Likewise.
3341         (thumb2_shiftsi3_short): Likewise.
3342         (thumb2_addsi3_compare0_scratch): Likewise.
3343         (orsi_not_shiftsi_si): Likewise.
3344         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3345         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3346         changes.
3347         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3348         (1020alu_shift_op): Likewise.
3349         (1020alu_shift_reg_op): Likewise.
3350         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3351         (alu_shift_op): Likewise.
3352         (alu_shift_reg_op): Likewise.
3353         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3354         (11_alu_shift_op): Likewise.
3355         (11_alu_shift_reg_op): Likewise.
3356         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3357         (9_alu_shift_reg_op): Likewise.
3358         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3359         changes.
3360         (cortex_a15_alu_shift): Likewise.
3361         (cortex_a15_alu_shift_reg): Likewise.
3362         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3363         changes.
3364         (cortex_a5_alu_shift): Likewise.
3365         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3366         changes.
3367         (cortex_a53_alu_shift): Likewise.
3368         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3369         changes.
3370         (cortex_a7_alu_reg): Likewise.
3371         (cortex_a7_alu_shift): Likewise.
3372         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3373         changes.
3374         (cortex_a8_alu_shift): Likewise.
3375         (cortex_a8_alu_shift_reg): Likewise.
3376         (cortex_a8_mov): Likewise.
3377         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3378         (cortex_a9_dp_shift): Likewise.
3379         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3380         changes.
3381         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3382         changes.
3383         (cortex_r4_mov): Likewise.
3384         (cortex_r4_alu_shift): Likewise.
3385         (cortex_r4_alu_shift_reg): Likewise.
3386         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3387         (526_alu_shift_op): Likewise.
3388         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3389         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3390         (626te_alu_shift_op): Likewise.
3391         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3392         (726te_alu_op): Likewise.
3393         (726te_alu_shift_op): Likewise.
3394         (726te_alu_shift_reg_op): Likewise.
3395         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3396         (mp626_alu_shift_op): Likewise.
3397         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3398         (pj4_alu_e1_conds): Likewise.
3399         (pj4_alu): Likewise.
3400         (pj4_alu_conds): Likewise.
3401         (pj4_shift): Likewise.
3402         (pj4_shift_conds): Likewise.
3403         (pj4_alu_shift): Likewise.
3404         (pj4_alu_shift_conds): Likewise.
3405         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3406         changes.
3407         (cortexa7_older_only): Likewise.
3408         (cortexa7_younger): Likewise.
3410 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3412         * ipa-pure-const.c (generate_summary): Rename to...
3413         (pure_const_generate_summary): ... this.
3415 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
3417         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3419 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
3421         PR target/57909
3422         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3423         usage in HI mode.
3425 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3427         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3428         enabled without -march=zEC12.
3429         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3430         flags to be set.
3432 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3434         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3435         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3436         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3437         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3438         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3439         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3440         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3441         ISA_HAS_FP4.
3442         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3443         and ISA_HAS_NMADD3_NMSUB3.
3444         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3445         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3446         (nmsub4<mode>, nmsub3<mode>): Likewise.
3447         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3449 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3451         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3452         TARGET_MIPS5400 checking.
3454 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
3455             Peter Bergner  <bergner@vnet.ibm.com>
3457         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3458         registers in the comment.
3459         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3460         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3461         rather than FIRST_PSEUDO_REGISTERS.
3463 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
3465         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3466         enable extra ISA flags with TARGET_HTM.
3468 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3470         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3471         Fix comment typos.
3473 2013-07-15  Cong Hou  <congh@google.com>
3475         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3476         in compare function for sorting.
3478 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
3480         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3481         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3482         * config/rs6000/rs6000.opt: Add -mhtm option.
3483         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3484         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3485         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3486         __HTM__ if the HTM instructions are available.
3487         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3488         htm_spr_reg_operand): New define_predicates.
3489         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3490         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3491         Include htm.md.
3492         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3493         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3494         HTM builtin functions.
3495         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3496         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3497         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3498         (rs6000_builtin_mask_calculate): Likewise.
3499         (rs6000_option_override_internal): Likewise.
3500         (bdesc_htm): Add new HTM builtin support.
3501         (htm_spr_num): New function.
3502         (htm_spr_regno): Likewise.
3503         (rs6000_htm_spr_icode): Likewise.
3504         (htm_expand_builtin): Likewise.
3505         (htm_init_builtins): Likewise.
3506         (rs6000_expand_builtin): Add support for HTM builtin functions.
3507         (rs6000_init_builtins): Likewise.
3508         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3509         option.
3510         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3511         (TARGET_HTM, MASK_HTM): Define macros.
3512         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3513         (FIXED_REGISTERS): Likewise.
3514         (CALL_USED_REGISTERS): Likewise.
3515         (CALL_REALLY_USED_REGISTERS): Likewise.
3516         (REG_ALLOC_ORDER): Likewise.
3517         (enum reg_class): Likewise.
3518         (REG_CLASS_NAMES): Likewise.
3519         (REG_CLASS_CONTENTS): Likewise.
3520         (REGISTER_NAMES): Likewise.
3521         (ADDITIONAL_REGISTER_NAMES): Likewise.
3522         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3523         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3524         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3525         * config/rs6000/htm.md: New file.
3526         * config/rs6000/htmintrin.h: New file.
3527         * config/rs6000/htmxlintrin.h: New file.
3529 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3531         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3532         Define SYMBOL_TINY_GOT, update comment.
3533         * config/aarch64/aarch64.c
3534         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3535         (aarch64_expand_mov_immediate): Likewise.
3536         (aarch64_print_operand): Likewise.
3537         (aarch64_classify_symbol): Likewise.
3538         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3539         (ldr_got_tiny): Define.
3541 2013-07-13  Tobias Grosser  <tobias@grosser.es>
3543         PR tree-optimization/54094
3544         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3545           scheduling dimension for the parallelism check from the polyhedral
3546           information in the AST.
3547         * graphite-dependences.c (carries_deps): Do not assume the schedule is
3548           in 2D + 1 form.
3550 2013-07-13  Jason Merrill  <jason@redhat.com>
3552         * print-tree.c (debug_vec_tree): Use debug_raw.
3553         (debug_raw (vec<tree, va_gc> &)): New.
3554         (debug_raw (vec<tree, va_gc> *)): New.
3555         * tree.h: Declare them.
3557 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
3559         * ifcvt.c (ifcvt_after_combine): New static variable.
3560         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3561         for size.
3562         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
3563         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3564         rest_of_handle_if_after_reload): Pass new argument for if_convert.
3566 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
3568         * config/mips/mips.c (mips_expand_call): Remove empty statement.
3570 2013-07-12  Michael Matz  <matz@suse.de>
3572         PR middle-end/55771
3573         * convert.c (convert_to_real): Reject non-float inner types.
3575 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
3577         * config/aarch64/aarch64-protos.h
3578         (aarch64_simd_immediate_valid_for_move): Remove.
3579         * config/aarch64/aarch64.c (simd_immediate_info): New member.
3580         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3581         cases.
3582         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3584 2013-07-11  Steve Ellcey  <sellcey@mips.com>
3586         * config/mips/mips.c (mips_conditional_register_usage): Do not
3587         use t[0-7] registers in MIPS16 mode when optimizing for size.
3589 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
3591         * config/i386/i386.c (dispatch_function_versions): Fix array
3592         indexing of function_version_info to match actual_versions.
3594 2013-07-11  Teresa Johnson  <tejohnson@google.com>
3596         * vec.h (struct va_gc): Move release out-of-line.
3597         (va_gc::release): Call ggc_free on released vec.
3599 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3601         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3602         Require GOT register as additional operand in UNSPEC.
3603         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3604         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3605         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3606         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3607         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3608         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3609         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3611 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
3613         PR target/57631
3614         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3615         name seen by assembler/linker rather if available.
3617 2013-07-11  Andreas Schwab  <schwab@suse.de>
3619         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
3621 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
3623         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
3625 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
3627         * doc/tm.texi.in: Move hook documentation to ....
3628         * target.def: ... here.
3630         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
3631         text on @hook line.
3632         * doc/tm.texi: Regenerate.
3634 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
3636         PR c++/57869
3637         * doc/invoke.texi: Document -Wconditionally-supported.
3639 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3641         PR target/57844
3642         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
3643         of my_fp.
3645 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3647         PR target/57506
3648         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
3649         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
3650         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
3651         Remove duplicate devices.
3652         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
3653         * config/avr/t-multilib: Regenerate.
3654         * config/avr/avr-tables.opt: Regenerate.
3655         * doc/avr-mmcu.texi: Regenerate.
3657 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3659         PR target/56987
3660         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
3662 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
3664         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
3665         the cost of MULT when optimizing for size.
3667 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3669         * config/cr16/cr16-protos.h: Don't include target.h.
3671 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
3673         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
3674         adjust register size for TDmode and TFmode for VSX registers.
3676 2013-07-08  Kai Tietz  <ktietz@redhat.com>
3678         PR target/56892
3679         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
3680         hook_bool_const_tree_true.
3682 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3684         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
3685         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
3686         * config/s390/s390.md: Define FPR*_REGNUM constants.
3687         Fix FPR2_REGNUM constant (18 -> 17).
3688         ("*trunc<BFP:mode><DFP_ALL:mode>2")
3689         ("*trunc<DFP_ALL:mode><BFP:mode>2")
3690         ("trunc<BFP:mode><DFP_ALL:mode>2")
3691         ("trunc<DFP_ALL:mode><BFP:mode>2")
3692         ("*extend<BFP:mode><DFP_ALL:mode>2")
3693         ("*extend<DFP_ALL:mode><BFP:mode>2")
3694         ("extend<BFP:mode><DFP_ALL:mode>2")
3695         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
3696         FPR4_REGNUM.
3698 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
3700         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
3702 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3704         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
3705         and cfun_fpr_bit_p to cfun_fpr_save_p.
3706         (s390_frame_area, s390_register_info, s390_frame_info)
3707         (s390_emit_prologue, s390_emit_epilogue)
3708         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
3709         register numbers.
3710         * config/s390/s390.h: Define *_REGNUM macros for floating point
3711         register numbers.
3713 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
3715         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
3717 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
3719         PR rtl-optimization/57786
3720         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
3721         and break out of the loop when it is set to false.
3723 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
3725         PR target/57819
3726         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
3727         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
3728         (const_int 63)) 0)).
3729         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
3730         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
3731         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
3733         PR rtl-optimization/57829
3734         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
3735         mask bits outside of mode are just sign-extension from mode to HWI.
3737 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
3739         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
3740         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
3741         adjust_address instead of change_address to keep info about alignment.
3742         (emit_strmov): Remove.
3743         (emit_memmov): New function.
3744         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
3745         (expand_movmem_epilogue): Likewise and return updated rtx for
3746         destination.
3747         (expand_constant_movmem_prologue): Likewise and return updated rtx for
3748         destination and source.
3749         (decide_alignment): Refactor, handle vector_loop.
3750         (ix86_expand_movmem): Likewise.
3751         (ix86_expand_setmem): Likewise.
3752         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
3754 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
3756         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
3757         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
3759 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3761         * config/i386/sse.md (sse_movlhps): Change alternative 3
3762         of operand 2 to "m".
3764 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3766         PR target/57807
3767         * config/i386/sse.md (iptr): New mode attribute.
3768         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
3769         (<sse>_vm<plusminus_insn><mode>3): Ditto.
3770         (<sse>_vmmul<mode>3): Ditto.
3771         (<sse>_vmdiv<mode>3): Ditto.
3772         (sse_vmrcpv4sf2): Ditto.
3773         (<sse>_vmsqrt<mode>2): Ditto.
3774         (sse_vmrsqrtv4sf2): Ditto.
3775         (<sse>_vm<code><mode>3): Ditto.
3776         (avx_vmcmp<mode>3): Ditto.
3777         (<sse>_vmmaskcmp<mode>3): Ditto.
3778         (<sse>_comi): Ditto.
3779         (<sse>_ucomi): Ditto.
3780         (*xop_vmfrcz_<mode>): Ditto.
3781         (*fmai_fmadd_<mode>): Ditto.
3782         (*fmai_fmsub_<mode>): Ditto.
3783         (*fmai_fnmadd_<mode>): Ditto.
3784         (*fmai_fnmsub_<mode>): Ditto.
3785         (*fma4i_vmfmadd_<mode>): Ditto.
3786         (*fma4i_vmfmsub_<mode>): Ditto.
3787         (*fma4i_vmfnmadd_<mode>): Ditto.
3788         (*fma4i_vmfnmsub_<mode>): Ditto.
3789         (*xop_vmfrcz_<mode>): Ditto.
3790         (sse_cvtps2pi): Ditto.
3791         (sse_cvttps2pi): Ditto.
3792         (sse_cvtss2si): Ditto.
3793         (sse_cvtss2si_2): Ditto.
3794         (sse_cvtss2siq_2): Ditto.
3795         (sse_cvttss2si): Ditto.
3796         (sse_cvttss2siq): Ditto.
3797         (sse_cvtsd2si): Ditto.
3798         (sse_cvtsd2si_2): Ditto.
3799         (sse_cvtsd2siq_2): Ditto.
3800         (sse_cvttsd2si): Ditto.
3801         (sse_cvttsd2siq): Ditto.
3802         (sse_cvtsd2ss): Ditto.
3803         (sse_cvtss2sd): Ditto.
3804         (avx2_pbroadcast<mode>): Ditto.
3805         (avx2_pbroadcast<mode>_1): Ditto.
3806         (*avx_vperm_broadcast_v4sf): Ditto.
3808         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
3809         (sse_movlhps): Ditto.
3810         (sse_storehps): Ditto.
3811         (sse_loadhps): Ditto.
3812         (sse_storelps): Ditto.
3813         (sse_loadlps): Ditto.
3814         (*vec_concatv4sf): Ditto.
3815         (*vec_interleave_highv2df): Ditto.
3816         (*vec_interleave_lowv2df): Ditto.
3817         (*vec_extractv2df_1_sse): Ditto.
3818         (*vec_extractv2df_0_sse): Ditto.
3819         (sse2_storelpd): Ditto.
3820         (sse2_loadlpd): Ditto.
3821         (sse2_movsd): Ditto.
3822         (*vec_concatv4si): Ditto.
3823         (vec_concatv2di): Ditto.
3825         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
3826         for Intel asm dialect.
3827         (mmx_punpcklwd): Ditto.
3828         (mmx_punpckldq): Ditto.
3830         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
3831         for intel assembler dialect.
3833 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
3835         PR target/29776
3836         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
3837         for BUILT_IN_C{LZ,LRSB}*.
3838         * tree.h (CASE_INT_FN): Add FN##IMAX case.
3839         * tree-vrp.c (extract_range_basic): Handle
3840         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
3841         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
3842         fall thru to code calling set_value*.
3843         * builtins.c (expand_builtin): Remove *IMAX cases.
3844         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
3845         if width is bigger than 2*HWI.
3847 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
3849         PR rtl-optimization/55342
3850         * lra-int.h (lra_subreg_reload_pseudos): New.
3851         * lra.c: Add undoing optional reloads to the block diagram.
3852         (lra_subreg_reload_pseudos): New.
3853         (lra_optional_reload_pseudos): Change comments.
3854         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
3855         lra_optional_reload_pseudos after undo transformations.
3856         * lra-assigns.c (pseudo_prefix_title): New.
3857         (lra_setup_reg_renumber): Use it.
3858         (spill_for): Ditto.  Check subreg reload pseudos too.
3859         (assign_by_spills): Consider subreg reload pseudos too.
3860         * lra-constraints.c (simplify_operand_subreg): Use
3861         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
3862         (curr_insn_transform): Recognize and do optional reloads.
3863         (undo_optional_reloads): New.
3864         (lra_undo_inheritance): Call undo_optional_reloads.
3866 2013-07-05  Thomas Quinot  <quinot@adacore.com>
3868         * tree-complex.c (expand_complex_operations_1): Fix typo.
3870 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
3872         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
3873         (tune_params): New member 'const vec_costs'.
3874         * config/aarch64/aarch64.c (generic_vector_cost): New.
3875         (generic_tunings): New member 'generic_vector_cost'.
3876         (aarch64_builtin_vectorization_cost): New.
3877         (aarch64_add_stmt_cost): New.
3878         (TARGET_VECTORIZE_ADD_STMT_COST): New.
3879         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
3881 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
3883         PR target/57777
3884         * config/i386/predicates.md (vsib_address_operand): Disallow
3885         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
3887 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
3889         PR middle-end/55030
3890         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
3891         expand_builtin_setjmp_receiver.
3892         (expand_label): Adjust, call expand_builtin_setjmp_receiver
3893         with NULL for the label parameter.
3894         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
3895         the frame-pointer.  Adjust comments.
3896         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
3897         only if LABEL is non-NULL.
3899 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
3901         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
3902         (ARM_ABI_AAPCS64): Ditto.
3903         (arm_abi): Ditto.
3904         (ARM_DEFAULT_ABI): Ditto.
3906 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
3908         * config/aarch64/aarch64-builtins.c
3909         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
3910         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
3911         (st1): Likewise.
3912         * config/aarch64/aarch64-simd.md
3913         (aarch64_ld1<VALL:mode>): New.
3914         (aarch64_st1<VALL:mode>): Likewise.
3915         * config/aarch64/arm_neon.h
3916         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
3918 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
3920         * config/i386/i386.c (gate_insert_vzeroupper): Check if
3921         target ISA is AVX.
3922         (ix86_option_override_internal):Turn on all -mavx target flags by
3923         default as they are dependent on AVX anyway.
3925 2013-07-02  Cary Coutant  <ccoutant@google.com>
3927         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
3928         deterministic hash.
3929         (loc_checksum_ordered): Likewise.
3930         (hash_loc_operands): Remove inline keyword.
3932 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
3934         PR tree-optimization/57741
3935         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
3936         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
3937         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
3938         Allow REAL_CST step_exprs if flag_associative_math.
3939         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
3941 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
3943         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
3945 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
3947         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
3949 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3951         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
3952         encoding.
3953         (iorsi3_insn): Likewise.
3954         (arm_xorsi3): Likewise.
3956 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
3958         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
3959         to ...
3960         (attribute "type"): ... here, and prefix with "wmmx_".
3961         (attribute "core_cycles"): Update for attribute changes.
3962         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
3963         (tbcstv4hi): Likewise.
3964         (tbcstv2si): Likewise.
3965         (iwmmxt_iordi3): Likewise.
3966         (iwmmxt_xordi3): Likewise.
3967         (iwmmxt_anddi3): Likewise.
3968         (iwmmxt_nanddi3): Likewise.
3969         (iwmmxt_arm_movdi): Likewise.
3970         (iwmmxt_movsi_insn): Likewise.
3971         (mov<mode>_internal): Likewise.
3972         (and<mode>3_iwmmxt): Likewise.
3973         (ior<mode>3_iwmmxt): Likewise.
3974         (xor<mode>3_iwmmxt): Likewise.
3975         (add<mode>3_iwmmxt): Likewise.
3976         (ssaddv8qi3): Likewise.
3977         (ssaddv4hi3): Likewise.
3978         (ssaddv2si3): Likewise.
3979         (usaddv8qi3): Likewise.
3980         (usaddv4hi3): Likewise.
3981         (usaddv2si3): Likewise.
3982         (sub<mode>3_iwmmxt): Likewise.
3983         (sssubv8qi3): Likewise.
3984         (sssubv4hi3): Likewise.
3985         (sssubv2si3): Likewise.
3986         (ussubv8qi3): Likewise.
3987         (ussubv4hi3): Likewise.
3988         (ussubv2si3): Likewise.
3989         (mulv4hi3_iwmmxt): Likewise.
3990         (smulv4hi3_highpart): Likewise.
3991         (umulv4hi3_highpart): Likewise.
3992         (iwmmxt_wmacs): Likewise.
3993         (iwmmxt_wmacsz): Likewise.
3994         (iwmmxt_wmacu): Likewise.
3995         (iwmmxt_wmacuz): Likewise.
3996         (iwmmxt_clrdi): Likewise.
3997         (iwmmxt_clrv8qi): Likewise.
3998         (iwmmxt_clr4hi): Likewise.
3999         (iwmmxt_clr2si): Likewise.
4000         (iwmmxt_uavgrndv8qi3): Likewise.
4001         (iwmmxt_uavgrndv4hi3): Likewise.
4002         (iwmmxt_uavgv8qi3): Likewise.
4003         (iwmmxt_uavgv4hi3): Likewise.
4004         (iwmmxt_tinsrb): Likewise.
4005         (iwmmxt_tinsrh): Likewise.
4006         (iwmmxt_tinsrw): Likewise.
4007         (iwmmxt_textrmub): Likewise.
4008         (iwmmxt_textrmsb): Likewise.
4009         (iwmmxt_textrmuh): Likewise.
4010         (iwmmxt_textrmsh): Likewise.
4011         (iwmmxt_textrmw): Likewise.
4012         (iwmxxt_wshufh): Likewise.
4013         (eqv8qi3): Likewise.
4014         (eqv4hi3): Likewise.
4015         (eqv2si3): Likewise.
4016         (gtuv8qi3): Likewise.
4017         (gtuv4hi3): Likewise.
4018         (gtuv2si3): Likewise.
4019         (gtv8qi3): Likewise.
4020         (gtv4hi3): Likewise.
4021         (gtv2si3): Likewise.
4022         (smax<mode>3_iwmmxt): Likewise.
4023         (umax<mode>3_iwmmxt): Likewise.
4024         (smin<mode>3_iwmmxt): Likewise.
4025         (umin<mode>3_iwmmxt): Likewise.
4026         (iwmmxt_wpackhss): Likewise.
4027         (iwmmxt_wpackwss): Likewise.
4028         (iwmmxt_wpackdss): Likewise.
4029         (iwmmxt_wpackhus): Likewise.
4030         (iwmmxt_wpackwus): Likewise.
4031         (iwmmxt_wpackdus): Likewise.
4032         (iwmmxt_wunpckihb): Likewise.
4033         (iwmmxt_wunpckihh): Likewise.
4034         (iwmmxt_wunpckihw): Likewise.
4035         (iwmmxt_wunpckilb): Likewise.
4036         (iwmmxt_wunpckilh): Likewise.
4037         (iwmmxt_wunpckilw): Likewise.
4038         (iwmmxt_wunpckehub): Likewise.
4039         (iwmmxt_wunpckehuh): Likewise.
4040         (iwmmxt_wunpckehuw): Likewise.
4041         (iwmmxt_wunpckehsb): Likewise.
4042         (iwmmxt_wunpckehsh): Likewise.
4043         (iwmmxt_wunpckehsw): Likewise.
4044         (iwmmxt_wunpckelub): Likewise.
4045         (iwmmxt_wunpckeluh): Likewise.
4046         (iwmmxt_wunpckeluw): Likewise.
4047         (iwmmxt_wunpckelsb): Likewise.
4048         (iwmmxt_wunpckelsh): Likewise.
4049         (iwmmxt_wunpckelsw): Likewise.
4050         (ror<mode>3): Likewise.
4051         (ashr<mode>3_iwmmxt): Likewise.
4052         (lshr<mode>3_iwmmxt): Likewise.
4053         (ashl<mode>3_iwmmxt): Likewise.
4054         (ror<mode>3_di): Likewise.
4055         (ashr<mode>3_di): Likewise.
4056         (lshr<mode>3_di): Likewise.
4057         (ashl<mode>3_di): Likewise.
4058         (iwmmxt_wmadds): Likewise.
4059         (iwmmxt_wmaddu): Likewise.
4060         (iwmmxt_tmia): Likewise.
4061         (iwmmxt_tmiaph): Likewise.
4062         (iwmmxt_tmiabb): Likewise.
4063         (iwmmxt_tmiatb): Likewise.
4064         (iwmmxt_tmiabt): Likewise.
4065         (iwmmxt_tmiatt): Likewise.
4066         (iwmmxt_tmovmskb): Likewise.
4067         (iwmmxt_tmovmskh): Likewise.
4068         (iwmmxt_tmovmskw): Likewise.
4069         (iwmmxt_waccb): Likewise.
4070         (iwmmxt_wacch): Likewise.
4071         (iwmmxt_waccw): Likewise.
4072         (iwmmxt_waligni): Likewise.
4073         (iwmmxt_walignr): Likewise.
4074         (iwmmxt_walignr0): Likewise.
4075         (iwmmxt_walignr1): Likewise.
4076         (iwmmxt_walignr2): Likewise.
4077         (iwmmxt_walignr3): Likewise.
4078         (iwmmxt_wsadb): Likewise.
4079         (iwmmxt_wsadh): Likewise.
4080         (iwmmxt_wsadbz): Likewise.
4081         (iwmmxt_wsadhz): Likewise.
4082         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4083         (iwmmxt_wabsdiffb): Likewise.
4084         (iwmmxt_wabsdiffh): Likewise.
4085         (iwmmxt_wabsdiffw): Likewise.
4086         (iwmmxt_waddsubhx): Likewise
4087         (iwmmxt_wsubaddhx): Likewise.
4088         (addc<mode>3): Likewise.
4089         (iwmmxt_avg4): Likewise.
4090         (iwmmxt_avg4r): Likewise.
4091         (iwmmxt_wmaddsx): Likewise.
4092         (iwmmxt_wmaddux): Likewise.
4093         (iwmmxt_wmaddsn): Likewise.
4094         (iwmmxt_wmaddun): Likewise.
4095         (iwmmxt_wmulwsm): Likewise.
4096         (iwmmxt_wmulwum): Likewise.
4097         (iwmmxt_wmulsmr): Likewise.
4098         (iwmmxt_wmulumr): Likewise.
4099         (iwmmxt_wmulwsmr): Likewise.
4100         (iwmmxt_wmulwumr): Likewise.
4101         (iwmmxt_wmulwl): Likewise.
4102         (iwmmxt_wqmulm): Likewise.
4103         (iwmmxt_wqmulwm): Likewise.
4104         (iwmmxt_wqmulmr): Likewise.
4105         (iwmmxt_wqmulwmr): Likewise.
4106         (iwmmxt_waddbhusm): Likewise.
4107         (iwmmxt_waddbhusl): Likewise.
4108         (iwmmxt_wqmiabb): Likewise.
4109         (iwmmxt_wqmiabt): Likewise.
4110         (iwmmxt_wqmiatb): Likewise.
4111         (iwmmxt_wqmiatt): Likewise.
4112         (iwmmxt_wqmiabbn): Likewise.
4113         (iwmmxt_wqmiabtn): Likewise.
4114         (iwmmxt_wqmiatbn): Likewise.
4115         (iwmmxt_wqmiattn): Likewise.
4116         (iwmmxt_wmiabb): Likewise.
4117         (iwmmxt_wmiabt): Likewise.
4118         (iwmmxt_wmiatb): Likewise.
4119         (iwmmxt_wmiatt): Likewise.
4120         (iwmmxt_wmiabbn): Likewise.
4121         (iwmmxt_wmiabtn): Likewise.
4122         (iwmmxt_wmiatbn): Likewise.
4123         (iwmmxt_wmiattn): Likewise.
4124         (iwmmxt_wmiawbb): Likewise.
4125         (iwmmxt_wmiawbt): Likewise.
4126         (iwmmxt_wmiawtb): Likewise.
4127         (iwmmxt_wmiawtt): Likewise.
4128         (iwmmxt_wmiawbbn): Likewise.
4129         (iwmmxt_wmiawbtn): Likewise.
4130         (iwmmxt_wmiawtbn): Likewise.
4131         (iwmmxt_wmiawttn): Likewise.
4132         (iwmmxt_wmerge): Likewise.
4133         (iwmmxt_tandc<mode>3): Likewise.
4134         (iwmmxt_torc<mode>3): Likewise.
4135         (iwmmxt_torvsc<mode>3): Likewise.
4136         (iwmmxt_textrc<mode>3): Likewise.
4137         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4138         (wmmxt_pack): Likewise.
4139         (wmmxt_mult_c1): Likewise.
4140         (wmmxt_mult_c2): Likewise.
4141         (wmmxt_alu_c1): Likewise.
4142         (wmmxt_alu_c2): Likewise.
4143         (wmmxt_alu_c3): Likewise.
4144         (wmmxt_transfer_c1): Likewise.
4145         (wmmxt_transfer_c2): Likewise.
4146         (wmmxt_transfer_c3): Likewise.
4147         (marvell_f_iwmmxt_wstr): Likewise.
4148         (marvell_f_iwmmxt_wldr): Likewise.
4150 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
4152         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4154 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4156         Revert:
4157         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4158         * lra-constraints.c (need_for_split_p): Check call used hard regs
4159         living through calls.
4161         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4162         call used regs for call insn.
4164 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
4166         PR target/57736
4167         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4168         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4170 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
4172         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4174 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4176         * lra-constraints.c (need_for_split_p): Check call used hard regs
4177         living through calls.
4179 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
4181         PR target/57744
4182         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4183         to tie with any other modes.  Eliminate Altivec vector mode tests,
4184         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
4185         code, to return 0 if testing MODE2 for a condition, if we've
4186         already tested MODE1 for the same condition.
4188 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4190         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4191         layout.
4193 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4195         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4196         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4198 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4200         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4201         Define.
4202         (aarch64_symbolic_constant_p): Remove.
4203         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4204         static.  Fix line length and white space.
4205         (aarch64_symbolic_constant_p): Remove.
4206         * config/aarch64/predicates.md (aarch64_valid_symref):
4207         Use aarch64_classify_symbol_expression.
4209 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4211         * config/arm/constraints.md (Ts): New constraint.
4212         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4213         16-bit encodings.
4214         (compare_scc): Use "Ts" constraint for operand 0.
4215         (ior_scc_scc): Likewise.
4216         (and_scc_scc): Likewise.
4217         (and_scc_scc_nodom): Likewise.
4218         (ior_scc_scc_cmp): Likewise for operand 7.
4219         (and_scc_scc_cmp): Likewise.
4220         * config/arm/thumb2.md (thumb2_movsi_insn):
4221         Add alternatives for 16-bit encodings.
4222         (thumb2_movhi_insn): Likewise.
4223         (thumb2_movsicc_insn): Likewise.
4224         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
4225         (thumb2_negscc): Use "Ts" constraint.
4226         Move mvn instruction outside cond_exec block.
4227         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4228         for 16-bit encodings.
4230 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4232         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4233         encoding.
4234         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4235         (mulsi3subsi): Likewise.
4236         (mulsidi3adddi): Likewise.
4237         (mulsidi3_v6): Likewise.
4238         (umulsidi3_v6): Likewise.
4239         (umulsidi3adddi_v6): Likewise.
4240         (smulsi3_highpart_v6): Likewise.
4241         (umulsi3_highpart_v6): Likewise.
4242         (mulhisi3tb): Likewise.
4243         (mulhisi3bt): Likewise.
4244         (mulhisi3tt): Likewise.
4245         (maddhisi4): Likewise.
4246         (maddhisi4tb): Likewise.
4247         (maddhisi4tt): Likewise.
4248         (maddhidi4): Likewise.
4249         (maddhidi4tb): Likewise.
4250         (maddhidi4tt): Likewise.
4251         (zeroextractsi_compare0_scratch): Likewise.
4252         (insv_zero): Likewise.
4253         (insv_t2): Likewise.
4254         (anddi_notzesidi_di): Likewise.
4255         (anddi_notsesidi_di): Likewise.
4256         (andsi_notsi_si): Likewise.
4257         (iordi_zesidi_di): Likewise.
4258         (xordi_zesidi_di): Likewise.
4259         (andsi_iorsi3_notsi): Likewise.
4260         (smax_0): Likewise.
4261         (smax_m1): Likewise.
4262         (smin_0): Likewise.
4263         (not_shiftsi): Likewise.
4264         (unaligned_loadsi): Likewise.
4265         (unaligned_loadhis): Likewise.
4266         (unaligned_loadhiu): Likewise.
4267         (unaligned_storesi): Likewise.
4268         (unaligned_storehi): Likewise.
4269         (extv_reg): Likewise.
4270         (extzv_t2): Likewise.
4271         (divsi3): Likewise.
4272         (udivsi3): Likewise.
4273         (arm_zero_extendhisi2addsi): Likewise.
4274         (arm_zero_extendqisi2addsi): Likewise.
4275         (compareqi_eq0): Likewise.
4276         (arm_extendhisi2_v6): Likewise.
4277         (arm_extendqisi2addsi): Likewise.
4278         (arm_movt): Likewise.
4279         (thumb2_ldrd): Likewise.
4280         (thumb2_ldrd_base): Likewise.
4281         (thumb2_ldrd_base_neg): Likewise.
4282         (thumb2_strd): Likewise.
4283         (thumb2_strd_base): Likewise.
4284         (thumb2_strd_base_neg): Likewise.
4285         (arm_negsi2): Add alternative for 16-bit encoding.
4286         (arm_one_cmplsi2): Likewise.
4288 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4290         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4291         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4292         (movdfcc): Likewise.
4293         * config/arm/vfp.md (*thumb2_movsf_vfp):
4294         Disable predication for arm_restrict_it.
4295         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4296         (*thumb2_movdfcc_vfp): Likewise.
4297         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4298         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4299         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4300         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4301         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4302         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4303         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4304         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4305         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4306         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4307         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4308         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4309         Disable predication for arm_restrict_it.
4311 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
4313         * config/i386/bmiintrin.h (_bextr_u32): New.
4314         (_bextr_u64): Ditto.
4316 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
4318         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4319         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4320         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4321         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4322         * config/mips/n32-elf.h: ...this new file.
4324 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
4326         PR target/57224
4327         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4328         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4330 2013-06-27  Catherine Moore <clm@codesourcery.com>
4332         * config/mips/mips-tables.opt: Regenerate.
4333         * config/mips/mips-cpus.def: Add m14ke and m14kec.
4334         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4335         * doc/invoke.texi: Add -m14kc.
4337 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4339         PR target/57623
4340         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4341         constraints of operand 1 and 2.
4343         PR target/57623
4344         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4345         to match RTL canonicalization.  Swap predicates and
4346         constraints of operand 1 and 2.
4348 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
4350         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4351         Process OP_INOUT regs for splitting too.
4353 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4355         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4356         decl before the loop, initialize to NULL.
4357         (vectorizable_load): Initialize ptr_incr to NULL.
4359 2013-06-27  Martin Jambor  <mjambor@suse.cz>
4361         PR lto/57208
4362         * ipa-ref.h (ipa_maybe_record_reference): Declare.
4363         * ipa-ref.c (ipa_maybe_record_reference): New function.
4364         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4365         * ipa-cp.c (create_specialized_node): Record potential references from
4366         aggvals.
4367         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4369 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4371         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4372         parameter 'mode' of type 'enum machine_mode mode'; change to pass
4373         'mode' to force_reg.
4374         (aarch64_add_offset): Update calls to aarch64_force_temporary.
4375         (aarch64_expand_mov_immediate): Likewise.
4377 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4379         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4380         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4382 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4384         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4385         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4386         (struct machine_function): Add tbegin_p.
4387         (s390_canonicalize_comparison): Fold CC mode compares to
4388         conditional jump if possible.
4389         (s390_emit_jump): Return the emitted jump.
4390         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4391         Handle CCRAWmode compares.
4392         (s390_option_override): Default to -mhtm if available.
4393         (s390_reg_clobbered_rtx): Handle floating point regs as well.
4394         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4395         FPRs instead of df_regs_ever_live_p.
4396         (s390_optimize_nonescaping_tx): New function.
4397         (s390_init_frame_layout): Extend clobbered_regs array to cover
4398         FPRs as well.
4399         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4400         (s390_expand_tbegin): New function.
4401         (enum s390_builtin): New enum definition.
4402         (code_for_builtin): New array definition.
4403         (s390_init_builtins): New function.
4404         (s390_expand_builtin): New function.
4405         (TARGET_INIT_BUILTINS): Define.
4406         (TARGET_EXPAND_BUILTIN): Define.
4407         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4408         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4409         (s390_alc_comparison): Likewise.
4410         * config/s390/s390-modes.def: Add CCRAWmode.
4411         * config/s390/s390.h (processor_flags): Add PF_TX.
4412         (TARGET_CPU_HTM): Define macro.
4413         (TARGET_HTM): Define macro.
4414         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4415         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4416         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4417         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4418         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4419         (TBEGIN_MASK, TBEGINC_MASK): New constants.
4420         ("*cc_to_int"): Move up.
4421         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4422         constants other than 0.
4423         ("*ccraw_to_int"): New insn and splitter definition.
4424         ("tbegin", "tbegin_nofloat", "tbegin_retry")
4425         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4426         ("tx_assist"): New expander.
4427         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4428         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4429         * config/s390/s390.opt: Add -mhtm option.
4430         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4431         * config/s390/htmxlintrin.h: New file.
4432         * config/s390/htmintrin.h: New file.
4433         * config/s390/s390intrin.h: New file.
4434         * doc/extend.texi: Document htm builtins.
4435         * config.gcc: Add the new header files to extra_headers.
4437 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
4439         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4440         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4442 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
4443             Pat Haugen <pthaugen@us.ibm.com>
4444             Peter Bergner <bergner@vnet.ibm.com>
4446         * config/rs6000/power8.md: New.
4447         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4448         setting for power8 entry.
4449         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4450         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4451         test for Power4/Power5 only.
4452         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4453         support.
4454         (force_new_group): Adjust comment.
4455         * config/rs6000/rs6000.md: Include power8.md.
4457 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
4459         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4460         * config/arm/arm-protos.h (arm_max_conditional_execute): New
4461         declaration.
4462         (tune_params): Update comment.
4463         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4464         (arm_max_conditional_execute): New function.
4465         (thumb2_final_prescan_insn): Use max_insn_skipped and
4466         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4468 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
4470         PR tree-optimization/57705
4471         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4472         SSA_NAME step, provided that it is not defined inside the loop.
4473         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4474         (get_initial_def_for_induction): Handle SSA_NAME IV step.
4476 2013-06-25  Martin Jambor  <mjambor@suse.cz>
4478         PR middle-end/57670
4479         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4480         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4481         calls in the dump.
4482         (ipa_note_param_call): Initialize member_ptr flag.
4483         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4484         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4485         (ipa_write_indirect_edge_info): Stream member_ptr flag.
4486         (ipa_read_indirect_edge_info): Likewise.
4488 2013-06-25  Richard Biener  <rguenther@suse.de>
4490         PR middle-end/56977
4491         * passes.c (init_optimization_passes): Move pass_fold_builtins
4492         and pass_dce earlier with -Og.
4494 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
4496         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4497         <BIT_FIELD_REF>: Remove trailing TAB.
4498         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4499         remove blank line.
4501 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4503         PR tree-optimization/57358
4504         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4505         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4506         (ipa_analyze_params_uses): Generate pessimistic info when true.
4508 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4510         PR tree-optimization/57539
4511         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4512         global.inlined_to of the new node to it.  All callers changed.
4513         * ipa-inline-transform.c (clone_inlined_nodes): New variable
4514         inlining_into, pass it to cgraph_clone_node.
4515         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4516         ipa_free_edge_args_substructures.
4517         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4518         rdesc linked list.  Do not assert rdesc edges have inlined caller.
4519         Assert we have found an rdesc in the rdesc list.
4521 2013-06-24  Richard Biener  <rguenther@suse.de>
4523         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4524         (pointer_set_lookup): Declare.
4525         (class pointer_map): New template class implementing a
4526         generic pointer to T map.
4527         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4528         pointer_map<T>::contains, pointer_map<T>::insert,
4529         pointer_map<T>::traverse): New functions.
4530         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4531         (pointer_set_lookup): New function.
4532         (pointer_set_contains): Use pointer_set_lookup.
4533         (pointer_set_insert): Likewise.
4534         (insert_aux): Remove.
4535         (struct pointer_map_t): Embed a pointer_set_t.
4536         (pointer_map_create): Adjust.
4537         (pointer_map_destroy): Likewise.
4538         (pointer_map_contains): Likewise.
4539         (pointer_map_insert): Likewise.
4540         (pointer_map_traverse): Likewise.
4541         * tree-streamer.h (struct streamer_tree_cache_d): Use a
4542         pointer_map<unsigned> instead of a pointer_map_t.
4543         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4544         (streamer_tree_cache_lookup): Likewise.
4545         (streamer_tree_cache_create): Likewise.
4546         (streamer_tree_cache_delete): Likewise.
4547         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4548         pointer_map<unsigned> instead of a pointer_map_t.
4549         (lto_init_tree_ref_encoder): Adjust.
4550         (lto_destroy_tree_ref_encoder): Likewise.
4551         * lto-section-out.c (lto_output_decl_index): Likewise.
4552         (lto_record_function_out_decl_state): Likewise.
4553         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4555 2013-06-24  Richard Biener  <rguenther@suse.de>
4557         PR tree-optimization/57488
4558         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4560 2013-06-24  Alan Modra  <amodra@gmail.com>
4562         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4563         (gen_easy_altivec_constant): Likewise.
4564         * config/rs6000/predicates.md (easy_vector_constant_add_self,
4565         easy_vector_constant_msb): Likewise.
4567 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
4569         PR target/57688
4570         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4571         add missing return true.
4573 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
4575         PR target/52483
4576         * config/sh/predicates.md (general_extend_operand): Invoke
4577         general_movsrc_operand for memory operands.
4578         (general_movsrc_operand): Allow reg+reg addressing, do not use
4579         general_operand for memory operands.
4581 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
4583         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
4584         when current target options does not apply.
4585         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4586         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4587         * config/i386/bmiintrin.h: Pass appropriate target
4588         attributes to header.
4589         * config/i386/mmintrin.h: Ditto.
4590         * config/i386/nmmintrin.h: Ditto.
4591         * config/i386/avx2intrin.h: Ditto.
4592         * config/i386/fxsrintrin.h: Ditto.
4593         * config/i386/tbmintrin.h: Ditto.
4594         * config/i386/xsaveintrin.h: Ditto.
4595         * config/i386/f16cintrin.h: Ditto.
4596         * config/i386/xtestintrin.h: Ditto.
4597         * config/i386/xsaveoptintrin.h: Ditto.
4598         * config/i386/bmi2intrin.h: Ditto.
4599         * config/i386/lzcntintrin.h: Ditto.
4600         * config/i386/smmintrin.h: Ditto.
4601         * config/i386/wmmintrin.h: Ditto.
4602         * config/i386/x86intrin.h: Remove all header include guards.
4603         * config/i386/prfchwintrin.h: Ditto.
4604         * config/i386/pmmintrin.h: Ditto.
4605         * config/i386/tmmintrin.h: Ditto.
4606         * config/i386/xmmintrin.h: Ditto.
4607         * config/i386/popcntintrin.h: Ditto.
4608         * config/i386/rdseedintrin.h: Ditto.
4609         * config/i386/ammintrin.h: Ditto.
4610         * config/i386/emmintrin.h: Ditto.
4611         * config/i386/immintrin.h: Remove all header include guards.
4612         * config/i386/fma4intrin.h: Ditto.
4613         * config/i386/lwpintrin.h: Ditto.
4614         * config/i386/xopintrin.h: Ditto.
4615         * config/i386/ia32intrin.h: Ditto.
4616         * config/i386/avxintrin.h: Ditto.
4617         * config/i386/rtmintrin.h: Ditto.
4618         * config/i386/fmaintrin.h: Ditto.
4619         * config/i386/mm3dnow.h: Ditto.
4621 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
4623         * common/config/i386/i386-common.c: Handle LZCNT.
4625 2013-06-22  Andi Kleen  <ak@linux.intel.com>
4627         * doc/extend.texi: Use __atomic_store_n instead of
4628         __atomic_store in HLE example.
4630 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
4632         * config/sh/sh.c: Remove <cstdlib> workaround.
4634 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4636         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
4638 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4640         * doc/extend.texi: Document that __atomic_clear and
4641           __atomic_test_and_set should only be used with bool.
4643 2013-06-20  Jan Hubicka  <jh@suse.cz>
4645         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
4646         types_same_for_odr.
4647         * tree.c (decls_same_for_odr): New function.
4648         (same_for_edr): New function.
4649         (types_same_for_odr): New function.
4650         (get_binfo_at_offset): Use it.
4651         * tree.h (types_same_for_odr): Declare.
4653 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
4654             Jason Merrill  <jason@redhat.com>
4656         * system.h: Include <cstdlib> as well as <stdlib.h>.
4658 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
4660         PR target/57655
4661         * config/i386/i386.c (construct_container): Report error if
4662         long double is used with disabled x87 float returns.
4664 2013-06-20  Jan Hubicka  <jh@suse.cz>
4666         * lto-cgraph.c (input_symtab): Do not set cgraph state.
4668 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
4670         PR rtl-optimization/57425
4671         PR rtl-optimization/57569
4672         * alias.c (write_dependence_p): Remove parameters mem_mode and
4673         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
4674         Changed all callers.
4675         (canon_anti_dependence): Get comments and semantics in sync.
4676         Add parameter mem_canonicalized.  Changed all callers.
4677         * rtl.h (canon_anti_dependence): Update prototype.
4679 2013-06-20  Richard Biener  <rguenther@suse.de>
4681         * data-streamer-in.c (streamer_read_uhwi): Optimize single
4682         byte case, inline streamer_read_uchar and defer section
4683         overrun check.
4685 2013-06-20  Richard Biener  <rguenther@suse.de>
4687         PR tree-optimization/57584
4688         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
4689         SSA names into the expanded expression that take part in
4690         abnormal coalescing.
4692 2013-06-19  Sharad Singhai  <singhai@google.com>
4694         * gcov.c (print_usage): Handle new option.
4695         (process_args): Ditto.
4696         (get_gcov_intermediate_filename): New function.
4697         (output_intermediate_file): New function.
4698         (output_gcov_file): New function
4699         (generate_results): Handle new option.
4700         (release_function): Relase demangled name.
4701         (read_graph_file): Handle demangled name.
4702         (output_lines): Ditto.
4703         * doc/gcov.texi: Document gcov intermediate format.
4705 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
4707         PR bootstrap/57604
4708         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
4709         (lra_emit_add): Use the functions.  Add comment about Y as an
4710         address segment.
4712 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
4714         PR driver/57652
4715         * collect2.c (collect_atexit): New.
4716         (collect_exit): Delete.
4717         (main): Register collect_atexit with atexit.
4718         (collect_wait): Change collect_exit to exit.
4719         (do_wait): Same.
4720         * collect2.h (collect_exit): Delete.
4721         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
4723 2013-06-19  Wei Mi  <wmi@google.com>
4725         PR rtl-optimization/57518
4726         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
4727         if regno is used in paradoxical subreg.
4728         (update_equiv_regs): Check pdx_subregs[regno] before
4729         set a reg to be equivalent with a mem.
4731 2013-06-19  Matthias Klose  <doko@ubuntu.com>
4733         PR driver/57651
4734         * file-find.h (find_a_file): Add a mode parameter.
4735         * file-find.c (find_a_file): Likewise.
4736         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
4737         with X_OK for the executables.
4738         * collect2.c (main): Call find_a_file with X_OK.
4740 2013-06-19  Steve Ellcey  <sellcey@mips.com>
4742         PR target/56942
4743         * config/mips/mips.md (casesi_internal_mips16_<mode>):
4744         Use NEXT_INSN instead of next_real_insn.
4746 2013-06-19  Jan Hubicka  <jh@suse.cz>
4748         * cgraph.h (const_value_known_p): Replace by ...
4749         (ctor_for_folding): .. this one.
4750         * cgraphunit.c (process_function_and_variable_attributes): Use it.
4751         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
4752         * expr.c (expand_expr_real_1): Likewise.
4753         (string_constant): Likewise.
4754         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
4755         * ipa.c (process_references): Likewise.
4756         (symtab_remove_unreachable_nodes): Likewise.
4757         * ipa-inline-analysis.c (param_change_prob): Likewise.
4758         * gimple-fold.c (canonicalize_constructor_val): Likewise.
4759         (get_base_constructor): Likwise.
4760         * varpool.c (varpool_remove_node): Likewise.
4761         (varpool_remove_initializer): LIkewise.
4762         (dump_varpool_node): LIkwise.
4763         (const_value_known_p): Rewrite to ...
4764         (ctor_for_folding): ... this one.
4766 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
4768         PR driver/57651
4769         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
4770         PERSONALITY in $PATH derived prefixes.
4772 2013-06-19  Jeff Law  <law@redhat.com>
4774         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
4775         in comment.
4777         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
4778         (simplify_bitwise_binary): Use it to simpify certain binary ops on
4779         booleans.
4781 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
4783         * config/arm/vfp.md: Move VFP instruction classification documentation
4784         to ...
4785         * config/arm/arm.md: ... here.  Update instruction classification
4786         documentation.
4788 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
4790         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
4791         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
4792         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
4794 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
4796         PR target/57609
4797         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4798         with NEXT_INSN.  Use tablejump_p to check for jump table data
4799         insns.
4801 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
4803         PR c++/56544
4804         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
4805         that now in C++ the value is correct per the C++ standards.
4807 2013-06-19  Richard Biener  <rguenther@suse.de>
4809         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
4810         for global context.
4812 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4814         Revert:
4815         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4817         PR target/57609
4818         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4819         with next_active_insn.
4821 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
4823         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
4824         functions are inlined during failures to flag an error.
4825         * tree-inline.c (expand_call_inline): Allow the error to be flagged
4826         in early inline pass.
4828 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
4830         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
4831         in comments.
4833 2013-06-18  Julian Brown  <julian@codesourcery.com>
4835         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
4836         Permit virtual register pre-reload if !strict.
4837         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
4838         change.
4839         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
4840         prototype.
4841         * config/arm/neon.md (movmisalign<mode>): Use
4842         neon_perm_struct_or_reg_operand instead of
4843         neon_struct_or_register_operand.
4844         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
4845         neon_permissive_struct_operand instead of neon_struct_operand.
4846         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
4847         neon_vector_mem_operand.
4848         * config/arm/predicates.md (neon_struct_operand): Adjust call to
4849         neon_vector_mem_operand.
4850         (neon_permissive_struct_operand): New.
4851         (neon_struct_or_register_operand): Rename to...
4852         (neon_perm_struct_or_reg_operand): This. Adjust call to
4853         neon_vector_mem_operand.
4855 2013-06-18  Richard Biener  <rguenther@suse.de>
4857         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
4858         * lto-streamer.h: Include pointer-set.h.
4859         (struct lto_decl_slot): Remove.
4860         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
4861         Remove next_index entry.
4862         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4863         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4864         (lto_init_tree_ref_encoder): Adjust.
4865         (lto_destroy_tree_ref_encoder): Likewise.
4866         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4867         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4868         (lto_output_decl_index): Adjust.
4869         (lto_new_out_decl_state): Likewise.
4870         (lto_record_function_out_decl_state): Likewise.
4871         * lto-streamer-out.c (copy_function): Likewise.
4873 2013-06-18  Richard Biener  <rguenther@suse.de>
4875         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
4876         * cgraphunit.c: Include cfgloop.h.
4877         (init_lowered_empty_function): Initialize the loop tree.
4878         (assemble_thunk): Insert new BBs into loops.
4880 2013-06-18  Richard Biener  <rguenther@suse.de>
4882         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
4883         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
4884         the map from cache entry to cache index optional.
4885         (streamer_tree_cache_replace_tree): Adjust accordingly.
4886         (streamer_tree_cache_append): Likewise.
4887         (streamer_tree_cache_delete): Likewise.
4888         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
4889         streamer cache map from cache entry to cache index.
4890         * lto-streamer-out.c (create_output_block): Adjust.
4892 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
4894         * config/arm/arm.md (attribute "insn"): Move multiplication and
4895         division attributes to...
4896         (attribute "type"): ... here.  Remove mult.
4897         (attribute "mul32"): New attribute.
4898         (attribute "mul64"): Add umaal.
4899         (*arm_mulsi3): Update attributes.
4900         (*arm_mulsi3_v6): Likewise.
4901         (*thumb_mulsi3): Likewise.
4902         (*thumb_mulsi3_v6): Likewise.
4903         (*mulsi3_compare0): Likewise.
4904         (*mulsi3_compare0_v6): Likewise.
4905         (*mulsi_compare0_scratch): Likewise.
4906         (*mulsi_compare0_scratch_v6): Likewise.
4907         (*mulsi3addsi): Likewise.
4908         (*mulsi3addsi_v6): Likewise.
4909         (*mulsi3addsi_compare0): Likewise.
4910         (*mulsi3addsi_compare0_v6): Likewise.
4911         (*mulsi3addsi_compare0_scratch): Likewise.
4912         (*mulsi3addsi_compare0_scratch_v6): Likewise.
4913         (*mulsi3subsi): Likewise.
4914         (*mulsidi3adddi): Likewise.
4915         (*mulsi3addsi_v6): Likewise.
4916         (*mulsidi3adddi_v6): Likewise.
4917         (*mulsidi3_nov6): Likewise.
4918         (*mulsidi3_v6): Likewise.
4919         (*umulsidi3_nov6): Likewise.
4920         (*umulsidi3_v6): Likewise.
4921         (*umulsidi3adddi): Likewise.
4922         (*umulsidi3adddi_v6): Likewise.
4923         (*smulsi3_highpart_nov6): Likewise.
4924         (*smulsi3_highpart_v6): Likewise.
4925         (*umulsi3_highpart_nov6): Likewise.
4926         (*umulsi3_highpart_v6): Likewise.
4927         (mulhisi3): Likewise.
4928         (*mulhisi3tb): Likewise.
4929         (*mulhisi3bt): Likewise.
4930         (*mulhisi3tt): Likewise.
4931         (maddhisi4): Likewise.
4932         (*maddhisi4tb): Likewise.
4933         (*maddhisi4tt): Likewise.
4934         (maddhidi4): Likewise.
4935         (*maddhidi4tb): Likewise.
4936         (*maddhidi4tt): Likewise.
4937         (divsi3): Likewise.
4938         (udivsi3): Likewise.
4939         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
4940         (thumb2_mulsi_short_compare0): Likewise.
4941         (thumb2_mulsi_short_compare0_scratch): Likewise.
4942         * config/arm/arm1020e.md (1020mult1): Update attribute change.
4943         (1020mult2): Likewise.
4944         (1020mult3): Likewise.
4945         (1020mult4): Likewise.
4946         (1020mult5): Likewise.
4947         (1020mult6): Likewise.
4948         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
4949         change.
4950         (cortex_a15_mult64): Likewise.
4951         (cortex_a15_sdiv): Likewise.
4952         (cortex_a15_udiv): Likewise.
4953         * config/arm/arm1026ejs.md (mult1): Update attribute change.
4954         (mult2): Likewise.
4955         (mult3): Likewise.
4956         (mult4): Likewise.
4957         (mult5): Likewise.
4958         (mult6): Likewise.
4959         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
4960         (pj4_ir_div): Likewise.
4961         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
4962         (11_mult2): Likewise.
4963         (11_mult3): Likewise.
4964         (11_mult4): Likewise.
4965         (11_mult5): Likewise.
4966         (11_mult6): Likewise.
4967         (11_mult7): Likewise.
4968         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
4969         (cortex_a8_mla): Likewise.
4970         (cortex_a8_mull): Likewise.
4971         (cortex_a8_smulwy): Likewise.
4972         (cortex_a8_smlald): Likewise.
4973         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
4974         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
4975         (cortex_r4_mul_3): Likewise.
4976         (cortex_r4_mla_4): Likewise.
4977         (cortex_r4_mla_3): Likewise.
4978         (cortex_r4_smlald): Likewise.
4979         (cortex_r4_mull): Likewise.
4980         (cortex_r4_sdiv): Likewise.
4981         (cortex_r4_udiv): Likewise.
4982         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
4983         (cortex_a7_idiv): Likewise.
4984         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
4985         (9_mult2): Likewise.
4986         (9_mult3): Likewise.
4987         (9_mult4): Likewise.
4988         (9_mult5): Likewise.
4989         (9_mult6): Likewise.
4990         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
4991         (cortex_a53_sdiv): Likewise.
4992         (cortex_a53_udiv): Likewise.
4993         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
4994         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
4995         (mp626_mult2): Likewise.
4996         (mp626_mult3): Likewise.
4997         (mp626_mult4): Likewise.
4998         * config/arm/fa526.md (526_mult1): Update attribute change.
4999         (526_mult2): Likewise.
5000         * config/arm/arm-generic.md (mult): Update attribute change.
5001         (mult_ldsched_strongarm): Likewise.
5002         (mult_ldsched): Likewise.
5003         (multi_cycle): Likewise.
5004         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5005         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5006         (606te_mult2): Likewise.
5007         (606te_mult3): Likewise.
5008         (606te_mult4): Likewise.
5009         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5010         (cortex_a9_mac16): Likewise.
5011         (cortex_a9_multiply): Likewise.
5012         (cortex_a9_mac): Likewise.
5013         (cortex_a9_multiply_long): Likewise.
5014         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5015         (626te_mult2): Likewise.
5016         (626te_mult3): Likewise.
5017         (626te_mult4): Likewise.
5019 2013-06-18  Richard Biener  <rguenther@suse.de>
5021         PR lto/57334
5022         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5024 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5026         PR target/57609
5027         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5028         with next_active_insn.
5030 2013-06-18  Alan Modra  <amodra@gmail.com>
5032         * config/rs6000/rs6000.h (enum data_align): New.
5033         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5034         (DATA_ABI_ALIGNMENT): Define.
5035         (CONSTANT_ALIGNMENT): Correct comment.
5036         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5037         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5039 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5041         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5042         ATTRIBUTE_UNUSED marking.
5044 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5046         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5047         alternative and update.
5048         (aarch64_dup_lanedi): Delete.
5049         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5050         * config/aarch64/aarch64-simd-builtins.def: Update.
5052 2013-06-17  Richard Biener  <rguenther@suse.de>
5054         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5055         (lto_input_scc): Declare.
5056         (lto_input_tree_1): Likewise.
5057         (struct lto_stats_d): Add num_tree_bodies_output and
5058         num_pickle_refs_output.
5059         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5060         (lto_read_tree_1): Split out from ...
5061         (lto_read_tree): ... this.
5062         (lto_input_scc): New function.
5063         (lto_input_tree_1): Split out from ...
5064         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5065         (lto_data_in_create): Create the streamer cache without hashes.
5066         * lto-streamer-out.c (create_output_block): Create the streamer
5067         cache with hashes when not doing WPA.
5068         (lto_write_tree_1): Split out from ...
5069         (lto_write_tree): ... this.
5070         (get_symbol_initial_value): New function.
5071         (lto_output_tree_1): Split out from ...
5072         (lto_output_tree): ... this.  Write trees as series of SCCs
5073         using a DFS walk via DFS_write_tree.
5074         (struct sccs, struct scc_entry): New types.
5075         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5076         (DFS_write_tree_body): New function.
5077         (DFS_write_tree): Likewise.
5078         (hash_tree): Likewise.
5079         (scc_entry_compare): Likewise.
5080         (hash_scc): Likewise.
5081         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5082         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5083         TREE_CHAIN as regular reference.
5084         (streamer_read_integer_cst): Remove.
5085         (streamer_get_pickled_tree): Adjust.
5086         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5087         of DECL_EXTERNALs in BLOCK_VARS for now.
5088         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5089         reference.
5090         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5091         Add hash value argument and record that if hashes are recorded
5092         in the cache.
5093         (streamer_tree_cache_insert_1): Adjust.
5094         (streamer_tree_cache_insert): Likewise.
5095         (streamer_tree_cache_insert_at): Rename to ...
5096         (streamer_tree_cache_replace_tree): ... this and adjust.
5097         (streamer_tree_cache_append): Adjust.
5098         (record_common_node): Likewise.
5099         (streamer_tree_cache_create): Add argument whether to
5100         record hash values together with trees.
5101         (streamer_tree_cache_delete): Adjust.
5102         * tree-streamer.h (struct streamer_tree_cache_d): Add
5103         vector of hashes.
5104         (streamer_read_integer_cst): Remove.
5105         (streamer_tree_cache_insert): Adjust.
5106         (streamer_tree_cache_append): Likewise.
5107         (streamer_tree_cache_insert_at): Rename to ...
5108         (streamer_tree_cache_replace_tree): ... this and adjust.
5109         (streamer_tree_cache_create): Add argument whether to record hashes.
5110         (streamer_tree_cache_get): Rename to ...
5111         (streamer_tree_cache_get_tree): ... this.
5112         (streamer_tree_cache_get_hash): New function.
5113         * tree.c (cache_integer_cst): New function.
5114         * tree.h (cache_integer_cst): Declare.
5115         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5116         * lto-symtab.c (lto_varpool_replace_node): Only release
5117         DECL_INITIAL of non-prevailing decls.
5118         * varpool.c (varpool_remove_initializer): Do not release
5119         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5121 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5123         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5124         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5125         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5126         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5127         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5128         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5129         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5130         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5131         instead of TARGET_64BIT.
5132         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5133         Require ISA_HAS_<D>DIV.
5135 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
5137         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5138         (mips*-*-linux*): Move default with_llsc setting to where other
5139         defaults are set.
5140         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5141         with_arch block.
5142         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5143         Likewise.  Remove default with_tune setting.  Move default float
5144         setting to its own block.  Handle with_llsc in the same block as above.
5146 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
5148         PR rtl-optimization/57425
5149         PR rtl-optimization/57569
5150         * alias.c (write_dependence_p): Add new parameters mem_mode,
5151         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
5152         Changed all callers.
5153         (canon_anti_dependence): New function.
5154         * cse.c (check_dependence): Use canon_anti_dependence.
5155         * cselib.c (cselib_invalidate_mem): Likewise.
5156         * rtl.h (canon_anti_dependence): Declare.
5158 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5160         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5161         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5162         ".set mips3" for 64-bit targets.
5164 2013-06-15  Dehao Chen  <dehao@google.com>
5166         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5167         * gimple-low.c (gimple_check_call_matching_types): Likewise.
5168         (gimple_check_call_args): Likewise.
5169         * value-prof.c (check_ic_target): Likewise.
5170         * ipa-inline.c (early_inliner): Likewise.
5171         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5172         * cgraph.c (cgraph_create_edge_1): Likewise.
5173         (cgraph_make_edge_direct): Likewise.
5175 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
5177         PR target/57615
5178         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5179         rs6000_output_move_128bit to handle emitting quad memory
5180         operations.  Set attribute length to 8 bytes.
5182 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
5184         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5185         New pattern.
5186         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5187         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5188         (aarch64_<su>mlsl<mode>): Likewise.
5190 2013-06-14  Mike Stump  <mikestump@comcast.net>
5192         * Makefile.in (TARGET_H): Add insn-codes.h.
5194 2013-06-14  Alan Modra  <amodra@gmail.com>
5196         PR middle-end/57134
5197         PR middle-end/57586
5198         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5199         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
5200         bitfield expansion when EXPAND_MEMORY.
5201         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5203 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
5205         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5206         test for clearing quad memory on 32-bit later.
5208 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5210         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5211         (fold_negate_expr): Likewise.
5212         (fold_real_zero_addition_p): Handle vectors.
5213         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5215 2013-06-14  Alan Modra  <amodra@gmail.com>
5217         * varasm.c (force_const_mem): Revert 2013-06-07 change.
5219 2013-06-13  Jan Hubicka  <jh@suse.cz>
5221         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5222         Local comdats are not externally visible.
5223         * symtab.c (dump_symtab_base): Dump externally visible.
5224         (verify_symtab_base): Verify back links in the symtab hash.
5226 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5228         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5229         CONVERT_EXPR as equal nodes.
5231 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5233         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5235 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5237         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5238         Generalize to complex and vector.
5239         * tree.c (build_all_ones_cst): New function.
5240         * tree.h (build_all_ones_cst): Declare it.
5242 2013-06-13  Alan Modra  <amodra@gmail.com>
5244         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5245         * config/rs6000/rs6000.md (signbittf2): New insn.
5246         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5247         (abstf2_internal, cmptf_internal2): Likewise.
5248         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5250 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5251             Pat Haugen <pthaugen@us.ibm.com>
5252             Peter Bergner <bergner@vnet.ibm.com>
5254         * config/rs6000/rs6000.c (emit_load_locked): Add support for
5255         power8 byte, half-word, and quad-word atomic instructions.
5256         (emit_store_conditional): Likewise.
5257         (rs6000_expand_atomic_compare_and_swap): Likewise.
5258         (rs6000_expand_atomic_op): Likewise.
5260         * config/rs6000/sync.md (larx): Add new modes for power8.
5261         (stcx): Likewise.
5262         (AINT): New mode iterator to include TImode as well as normal
5263         integer modes on power8.
5264         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5265         that VSX registers are not considered.  Use AINT mode iterator
5266         instead of INT1 to allow inclusion of quad word atomic operations
5267         on power8.
5268         (load_locked<mode>): Likewise.
5269         (store_conditional<mode>): Likewise.
5270         (atomic_compare_and_swap<mode>): Likewise.
5271         (atomic_exchange<mode>): Likewise.
5272         (atomic_nand<mode>): Likewise.
5273         (atomic_fetch_<fetchop_name><mode>): Likewise.
5274         (atomic_nand_fetch<mode>): Likewise.
5275         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5276         each type.
5277         (ATOMIC): On power8, add QImode, HImode modes.
5278         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5279         modes that promote to SImode.
5280         (load_lockedti): Convert TImode arguments to PTImode, so that we
5281         get a guaranteed even/odd register pair.
5282         (load_lockedpti): Likewise.
5283         (store_conditionalti): Likewise.
5284         (store_conditionalpti): Likewise.
5286         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5287         atomic load/store instructions.
5288         (HSI): Likewise.
5290 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
5292         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5293         loads.
5294         (insn_count): New attribute, with most cases extracted from...
5295         (length): ...here.  Redefine most cases in terms of insn_count.
5296         (single_insn): Delete.
5297         (can_delay): Use insn_count to check for single instructions.
5298         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5299         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5300         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5301         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5302         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5303         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5304         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5305         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5306         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5307         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5308         rather than "length".
5309         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
5310         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5311         Use "insn_count" rather than "length".
5312         * config/mips/mips-dsp.md
5313         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5314         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5315         length attributes.
5317 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
5319         PR tree-optimization/57361
5320         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5322 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
5324         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5325         to split.
5326         (aarch64_simd_combine<mode>): New instruction expansion.
5327         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5328         function prototype.
5329         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5330         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5332 2013-06-12  Jan Hubicka  <jh@suse.cz>
5334         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5335         decl has when in streaming stage.
5336         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5337         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5339 2013-06-12  Roland Stigge <stigge@antcom.de>
5341         PR target/57578
5342         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5344 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5346         PR tree-optimization/57537
5347         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5348         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5350 2013-06-12  Richard Biener  <rguenther@suse.de>
5352         * data-streamer.h (streamer_write_char_stream): CSE
5353         obs->current_pointer.
5354         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5355         streamer_write_char_stream manually and optimize the resulting loop.
5356         (streamer_write_hwi_stream): Likewise.
5358 2013-06-12  Jan Hubicka  <jh@suse.cz>
5360         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5361         * cgraph.h (varpool_create_empty_node): Declare.
5362         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5363         duplicated nodes.
5364         * symtab.c (symtab_unregister_node): Be lax about missin entries
5365         in node hash.
5366         (symtab_get_node): Update comment.
5367         * varpool.c (varpool_create_empty_node): Break out from ...
5368         (varpool_node_for_decl): ... here.
5369         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5371 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
5373         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5374         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5375         part.  Use straight-line flow at the end.
5376         <COMPONENT_REF>: Remove superfluous else.
5377         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5379 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5381         PR target/56564
5382         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5383         target hook even for !TREE_PUBLIC decls.  If no resolution info
5384         is available, return false for common and external decls.
5386 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
5388         * config/rl78/constraints.md (U): New constraint.
5389         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5390         valloc attribute.
5392 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5394         PR target/57589
5395         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5396         to allow returning address to AT_PLATFORM name.
5398 2013-06-11  Jan Hubicka  <jh@suse.cz>
5400         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5401         * cgraph.h (symtab_node_base): Add weakref flag.
5402         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5403         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5404         (output_weakrefs): Use weakref flag.
5405         * fold-const.c (simple_operand_p): Handle WEAK.
5406         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5407         * ipa.c (varpool_externally_visible_p): Drop weakref.
5408         (function_and_variable_visibility): Update comment; fix weakref
5409         sanity checks; do not clear DECL_WEAK on them.
5410         * lto-cgraph.c (lto_output_node): update.
5411         (lto_output_varpool_node): Update.
5412         (input_overwrite_node): Update.
5413         (input_node): Update.
5414         (input_varpool_node): Update.
5415         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5416         (lto_symtab_merge_symbols): Add sanity check.
5417         (lto_symtab_prevailing_decl): Do not special case weakrefs.
5418         * passes.c (rest_of_decl_compilation): Set static flag, too.
5419         * symtab.c (dump_symtab_base): Dump weakref.
5420         (verify_symtab_base): Sanity check weakrefs.
5421         (symtab_make_decl_local): Remove duplicated code.
5422         (symtab_alias_ultimate_target): Simplify.
5423         * varpool.c (varpool_create_variable_alias): Set weakref flag.
5425 2013-06-11  DJ Delorie  <dj@redhat.com>
5427         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5428         (rl78_unwind_word_mode): New.
5430 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5432         * final.c (debug_prefix_maps): Make static.
5434 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5436         * function.c (initial_trampoline): Remove stray copy.
5438 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
5440         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5442 2013-06-11  Martin Jambor  <mjambor@suse.cz>
5444         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5445         within bounds at the beginning of the function.
5447 2013-06-11  Alan Modra  <amodra@gmail.com>
5449         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
5450         reporting.
5451         (get_named_section): Don't NULL !DECL_P decl.
5453 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
5455         * doc/invoke.texi (core-avx2): Document.
5456         (slm): Likewise.
5457         (atom): Updated with MOVBE.
5459 2013-06-11  Richard Biener  <rguenther@suse.de>
5461         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5463 2013-06-11  Anton Blanchard  <anton@samba.org>
5465         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5466         correct shift value in little-endian mode.
5468 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
5470         PR target/56564
5471         * varasm.c (get_variable_align): Move #endif to the right place.
5473 2013-06-10  Cary Coutant  <ccoutant@google.com>
5475         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5476         for hash so that hash table traversal order is deterministic.
5478 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
5479             Pat Haugen <pthaugen@us.ibm.com>
5480             Peter Bergner <bergner@vnet.ibm.com>
5482         * config/rs6000/vector.md (GPR move splitter): Do not split moves
5483         of vectors in GPRS if they are direct moves or quad word load or
5484         store moves.
5486         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5487         declaration.
5488         (direct_move_p): Likewise.
5489         (quad_load_store_p): Likewise.
5491         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5492         classes into bins based on the physical register type.
5493         (reg_class_to_reg_type): Likewise.
5494         (IS_STD_REG_TYPE): Likewise.
5495         (IS_FP_VECT_REG_TYPE): Likewise.
5496         (reload_fpr_gpr): Arrays to determine what insn to use if we can
5497         use direct move instructions.
5498         (reload_gpr_vsx): Likewise.
5499         (reload_vsx_gpr): Likewise.
5500         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5501         information that is a simplification of register classes.  Also
5502         precalculate direct move reload helpers.
5503         (direct_move_p): New function to return true if the operation can
5504         be done as a direct move instruciton.
5505         (quad_load_store_p): New function to return true if the operation
5506         is a quad memory operation.
5507         (rs6000_legitimize_address): If quad memory, only allow register
5508         indirect for TImode addresses.
5509         (rs6000_legitimate_address_p): Likewise.
5510         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5511         (rs6000_reload_register_type): Likewise.
5512         (register_to_reg_type): Return register type.
5513         (rs6000_secondary_reload_simple_move): New helper function for
5514         secondary reload and secondary memory needed to identify anything
5515         that is a simple move, and does not need reloading.
5516         (rs6000_secondary_reload_direct_move): New helper function for
5517         secondary reload to identify cases that can be done with several
5518         instructions via the direct move instructions.
5519         (rs6000_secondary_reload_move): New helper function for secondary
5520         reload to identify moves between register types that can be done.
5521         (rs6000_secondary_reload): Add support for quad memory operations
5522         and for direct move.
5523         (rs6000_secondary_memory_needed): Likewise.
5524         (rs6000_debug_secondary_memory_needed): Change argument names.
5525         (rs6000_output_move_128bit): New function to return the move to
5526         use for 128-bit moves, including knowing about the various
5527         limitations of quad memory operations.
5529         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5530         memory operations.  call rs6000_output_move_128bit for the actual
5531         instruciton(s) to generate.
5532         (vsx_movti_64bit): Likewise.
5534         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5535         (UNSPEC_P8V_MTVSRWZ): Likewise.
5536         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5537         (UNSPEC_P8V_MTVSRD): Likewise.
5538         (UNSPEC_P8V_XXPERMDI): Likewise.
5539         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5540         (UNSPEC_FUSION_GPR): Likewise.
5541         (FMOVE128_GPR): New iterator for direct move.
5542         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5543         (f32_sv): Likewise.
5544         (f32_dm): Likewise.
5545         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5546         loads and direct move instructions.
5547         (zero_extendsidi2_lfiwzx): Likewise.
5548         (extendsidi2_lfiwax): Likewise.
5549         (extendsidi2_nocell): Likewise.
5550         (floatsi<mode>2_lfiwax): Likewise.
5551         (lfiwax): Likewise.
5552         (floatunssi<mode>2_lfiwzx): Likewise.
5553         (lfiwzx): Likewise.
5554         (fix_trunc<mode>_stfiwx): Likewise.
5555         (fixuns_trunc<mode>_stfiwx): Likewise.
5556         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5557         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5558         (parity<mode>2_cmpb): Set length/type attr.
5559         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5560         for 'mr.' to fast_compare.
5561         (bpermd_<mode>): Change type attr to popcnt.
5562         (p8_fmrgow_<mode>): New insns for power8 direct move support.
5563         (p8_mtvsrwz_1): Likewise.
5564         (p8_mtvsrwz_2): Likewise.
5565         (reload_fpr_from_gpr<mode>): Likewise.
5566         (p8_mtvsrd_1): Likewise.
5567         (p8_mtvsrd_2): Likewise.
5568         (p8_xxpermdi_<mode>): Likewise.
5569         (reload_vsx_from_gpr<mode>): Likewise.
5570         (reload_vsx_from_gprsf): Likewise.
5571         (p8_mfvsrd_3_<mode>): LIkewise.
5572         (reload_gpr_from_vsx<mode>): Likewise.
5573         (reload_gpr_from_vsxsf): Likewise.
5574         (p8_mfvsrd_4_disf): Likewise.
5575         (multi-word GPR splits): Do not split direct moves or quad memory
5576         operations.
5578 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
5580         * tree-into-ssa.c (interesting_blocks): Make static.
5582 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
5584         PR target/56564
5585         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5586         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5587         Use DATA_ABI_ALIGNMENT for that case instead if defined.
5588         (get_variable_align): New function.
5589         (get_variable_section, emit_bss, emit_common,
5590         assemble_variable_contents, place_block_symbol): Use
5591         get_variable_align instead of DECL_ALIGN.
5592         (assemble_noswitch_variable): Add align argument, use it
5593         instead of DECL_ALIGN.
5594         (assemble_variable): Adjust caller.  Use get_variable_align
5595         instead of DECL_ALIGN.
5596         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5597         caller.
5598         (DATA_ABI_ALIGNMENT): Define.
5599         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5600         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
5601         opt is false, only return the psABI mandated alignment increase.
5602         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5603         (DATA_ABI_ALIGNMENT): ... this.
5604         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5605         (DATA_ABI_ALIGNMENT): ... this.
5606         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5607         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5608         (DATA_ABI_ALIGNMENT): ... this.
5609         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5610         * doc/tm.texi: Regenerated.
5612 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
5614         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5615         cmp_code to construct REG_EQUAL note.
5617 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
5619         PR target/57568
5620         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
5621         that operands[2] doesn't overlap with operands[0].
5623 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
5624             Jan Hubicka  <jh@suse.cz>
5626         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
5627         hack to mark symbols as used.
5629 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
5631         PR rtl-optimization/57559
5632         * lra-constraints.c (process_alt_operands): Don't discourage
5633         memory with known offset for offsetable memory constraint.
5634         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
5636 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
5638         * varasm.c (struct oc_local_state): Reorder fields.
5639         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
5640         and adjust accordingly.
5641         (output_constructor): Reorder initialization code and adjust call to
5642         output_constructor_bitfield.
5644 2013-06-07  Jan Hubicka  <jh@suse.cz>
5646         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
5648 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
5650         * tree-object-size.c (unknown): Make const.
5652 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5654         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
5655         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
5656         for last alternative in the cpu_facility attribute.
5658 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5660         PR target/56315
5661         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
5662         (xordi3): Change operand 2 constraint to arm_xordi_operand.
5663         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
5664         * config/arm/constraints.md (Dg): New constraint.
5665         * config/arm/neon.md (xordi3_neon): Remove.
5666         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
5667         * config/arm/predicates.md (arm_xordi_operand): New predicate.
5669 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5671         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
5672         Clean up alternatives.
5674 2013-06-07  Alan Modra  <amodra@gmail.com>
5676         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
5677         va_list_gpr_size.
5679 2013-06-07  Alan Modra  <amodra@gmail.com>
5681         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
5683 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5685         * config/arm/constraints.md (Df): New constraint.
5686         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
5687         Correct length attribute for last two alternatives.
5689 2013-06-07  Alan Modra  <amodra@gmail.com>
5691         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5692         override user -mfp-in-toc.
5693         (offsettable_ok_by_alignment): Consider just the current access
5694         rather than the whole object, unless BLKmode.  Handle
5695         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
5696         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
5697         for -mcmodel=medium.
5698         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
5699         override user -mfp-in-toc or -msum-in-toc.  Default to
5700         -mno-fp-in-toc for -mcmodel=medium.
5702 2013-06-06  DJ Delorie  <dj@redhat.com>
5704         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
5705         TARGET_VALID_POINTER_MODE.
5707 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
5708             Pat Haugen <pthaugen@us.ibm.com>
5709             Peter Bergner <bergner@vnet.ibm.com>
5711         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5712         Document new power8 builtins.
5714         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
5715         condition code register, to allow 128-bit logical operations to be
5716         done in the VSX or GPR registers.
5717         (nor<mode>3): Use the canonical form for nor.
5718         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
5719         vclz*, and vpopcnt* vector instructions.
5720         (nand<mode>3): Likewise.
5721         (orc<mode>3): Likewise.
5722         (clz<mode>2): LIkewise.
5723         (popcount<mode>2): Likewise.
5725         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
5726         that only the GPRs are recognized.
5728         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5729         support for new power8 builtins.
5731         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
5732         builtin functions.
5733         (xscvdpspn): Likewise.
5734         (vclz): Likewise.
5735         (vclzb): Likewise.
5736         (vclzh): Likewise.
5737         (vclzw): Likewise.
5738         (vclzd): Likewise.
5739         (vpopcnt): Likewise.
5740         (vpopcntb): Likewise.
5741         (vpopcnth): Likewise.
5742         (vpopcntw): Likewise.
5743         (vpopcntd): Likewise.
5744         (vgbbd): Likewise.
5745         (vmrgew): Likewise.
5746         (vmrgow): Likewise.
5747         (eqv): Likewise.
5748         (eqv_v16qi3): Likewise.
5749         (eqv_v8hi3): Likewise.
5750         (eqv_v4si3): Likewise.
5751         (eqv_v2di3): Likewise.
5752         (eqv_v4sf3): Likewise.
5753         (eqv_v2df3): Likewise.
5754         (nand): Likewise.
5755         (nand_v16qi3): Likewise.
5756         (nand_v8hi3): Likewise.
5757         (nand_v4si3): Likewise.
5758         (nand_v2di3): Likewise.
5759         (nand_v4sf3): Likewise.
5760         (nand_v2df3): Likewise.
5761         (orc): Likewise.
5762         (orc_v16qi3): Likewise.
5763         (orc_v8hi3): Likewise.
5764         (orc_v4si3): Likewise.
5765         (orc_v2di3): Likewise.
5766         (orc_v4sf3): Likewise.
5767         (orc_v2df3): Likewise.
5769         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
5770         allow power8 quad mode in 64-bit.
5771         (rs6000_builtin_vectorized_function): Add support to vectorize
5772         ISA 2.07 count leading zeros, population count builtins.
5773         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
5774         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
5775         (builtin_function_type): Add vgbbd builtin function which takes an
5776         unsigned argument.
5777         (altivec_expand_vec_perm_const): Add support for new power8 merge
5778         instructions.
5780         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
5781         that does not include TImdoe for use with 32-bit.
5782         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
5783         instructions.
5784         (UNSPEC_VSX_CVDPSPN): Likewise.
5785         (vsx_xscvdpspn): Likewise.
5786         (vsx_xscvspdpn): Likewise.
5787         (vsx_xscvdpspn_scalar): Likewise.
5788         (vsx_xscvspdpn_directmove): Likewise.
5789         (vsx_and<mode>3): Split logical operations into 32-bit and
5790         64-bit. Add support to do logical operations on TImode as well as
5791         VSX vector types.  Allow logical operations to be done in either
5792         VSX registers or in general purpose registers in 64-bit mode.  Add
5793         splitters if GPRs were used. For AND, add clobber of CCmode to
5794         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
5795         encoding.
5796         (vsx_and<mode>3_32bit): Likewise.
5797         (vsx_and<mode>3_64bit): Likewise.
5798         (vsx_ior<mode>3): Likewise.
5799         (vsx_ior<mode>3_32bit): Likewise.
5800         (vsx_ior<mode>3_64bit): Likewise.
5801         (vsx_xor<mode>3): Likewise.
5802         (vsx_xor<mode>3_32bit): Likewise.
5803         (vsx_xor<mode>3_64bit): Likewise.
5804         (vsx_one_cmpl<mode>2): Likewise.
5805         (vsx_one_cmpl<mode>2_32bit): Likewise.
5806         (vsx_one_cmpl<mode>2_64bit): Likewise.
5807         (vsx_nor<mode>3): Likewise.
5808         (vsx_nor<mode>3_32bit): Likewise.
5809         (vsx_nor<mode>3_64bit): Likewise.
5810         (vsx_andc<mode>3): Likewise.
5811         (vsx_andc<mode>3_32bit): Likewise.
5812         (vsx_andc<mode>3_64bit): Likewise.
5813         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
5814         and xxlorc instructions.
5815         (vsx_eqv<mode>3_64bit): Likewise.
5816         (vsx_nand<mode>3_32bit): Likewise.
5817         (vsx_nand<mode>3_64bit): Likewise.
5818         (vsx_orc<mode>3_32bit): Likewise.
5819         (vsx_orc<mode>3_64bit): Likewise.
5821         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
5823         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
5824         instruction.
5825         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
5826         (p8_vmrgow): Likewise.
5827         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
5828         GPRs to be split under VSX.
5829         (p8v_clz<mode>2): Add power8 count leading zero support.
5830         (p8v_popcount<mode>2): Add power8 population count support.
5831         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
5832         support.
5834         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
5835         instruction.
5837         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
5838         builtin functions.
5839         (vec_nand): Likewise.
5840         (vec_vclz): Likewise.
5841         (vec_vclzb): Likewise.
5842         (vec_vclzd): Likewise.
5843         (vec_vclzh): Likewise.
5844         (vec_vclzw): Likewise.
5845         (vec_vgbbd): Likewise.
5846         (vec_vmrgew): Likewise.
5847         (vec_vmrgow): Likewise.
5848         (vec_vpopcnt): Likewise.
5849         (vec_vpopcntb): Likewise.
5850         (vec_vpopcntd): Likewise.
5851         (vec_vpopcnth): Likewise.
5852         (vec_vpopcntw): Likewise.
5854 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5856         PR rtl-optimization/57468
5857         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
5858         spilled pseudos.
5860 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5862         PR rtl-optimization/57459
5863         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
5864         type when setting live regs.
5866 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5868         * config/s390/s390.opt (mlra): New option.
5869         * config/s390/s390.c (s390_decompose_address): Check displacement
5870         for all registers for LRA.
5871         (s390_secondary_reload): Don't used secondary reloads for LRA.
5872         (s390_lra_p): New function.
5873         (TARGET_LRA_P): Define.
5874         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
5875         of attribute cpu_facility to zarch for the last alternative.
5876         (*cmpmem_short): Ditto.
5878 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
5880         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
5881         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
5882         (arm_expand_prologue): Likewise.
5884 2013-06-06  Teresa Johnson  <tejohnson@google.com>
5886         PR c++/53743
5887         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
5888         as this is now done by redirect_edge_and_branch_force.
5889         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
5890         barriers, and fix interaction with splitting.
5891         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
5892         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
5893         reflect changes made in the routine.
5894         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
5895         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
5896         since this is called in cfglayout mode, and replace partition fixup
5897         with assert as that is now done by force_nonfallthru_and_redirect.
5898         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
5899         already be marked with region crossing note.
5900         (insert_section_boundary_note): Make non-static, gate on flag
5901         has_bb_partition, rewrite to also check for multiple partitions.
5902         (rest_of_handle_reorder_blocks): Remove call to
5903         insert_section_boundary_note, now done later during free_cfg.
5904         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
5905         * bb-reorder.h (insert_section_boundary_note): Declare.
5906         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
5907         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
5908         invoke insert_section_boundary_note.
5909         (try_redirect_by_replacing_jump): Remove unnecessary
5910         check for region crossing note.
5911         (fixup_partition_crossing): New function.
5912         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
5913         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
5914         in non-cfglayout mode.
5915         (force_nonfallthru_and_redirect): Fixup partition boundaries,
5916         remove old code that tried to do this. Emit barrier correctly
5917         when we are in cfglayout mode.
5918         (last_bb_in_partition): New function.
5919         (rtl_split_edge): Correctly fixup partition boundaries.
5920         (commit_one_edge_insertion): Remove old code that tried to
5921         fixup region crossing edge since this is now handled in
5922         split_block, and set up insertion point correctly since
5923         block may now end in a jump.
5924         (verify_hot_cold_block_grouping): Guard against checking when not in
5925         linearized RTL mode.
5926         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
5927         notes.
5928         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
5929         rtl_verify_flow_info, so not called in cfglayout mode.
5930         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
5931         (fixup_reorder_chain): Remove old code that attempted to fixup region
5932         crossing note as this is now handled in force_nonfallthru_and_redirect.
5933         (duplicate_insn_chain): Don't duplicate switch section notes.
5934         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
5935         note.
5936         * basic-block.h (emit_barrier_after_bb): Declare.
5938 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5940         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
5941         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
5942         arm_usatsihi): Adjust alternatives for arm_restrict_it.
5944 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5946         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
5947         where appropriate.
5948         * config/arm/ldmstm.md: Regenerate.
5950 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5952         * config/arm/sync.md (atomic_loaddi_1):
5953         Disable predication for arm_restrict_it.
5954         (arm_load_exclusive<mode>): Likewise.
5955         (arm_load_exclusivesi): Likewise.
5956         (arm_load_exclusivedi): Likewise.
5957         (arm_load_acquire_exclusive<mode>): Likewise.
5958         (arm_load_acquire_exclusivesi): Likewise.
5959         (arm_load_acquire_exclusivedi): Likewise.
5960         (arm_store_exclusive<mode>): Likewise.
5961         (arm_store_exclusive<mode>): Likewise.
5962         (arm_store_release_exclusivedi): Likewise.
5963         (arm_store_release_exclusive<mode>): Likewise.
5965 2013-06-06  Richard Biener  <rguenther@suse.de>
5967         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
5968         after LTO_null.
5969         (lto_tag_is_tree_code_p): Adjust.
5970         (lto_tag_is_gimple_code_p): Likewise.
5971         (lto_gimple_code_to_tag): Likewise.
5972         (lto_tag_to_gimple_code): Likewise.
5973         (lto_tree_code_to_tag): Likewise.
5974         (lto_tag_to_tree_code): Likewise.
5975         * data-streamer.h (streamer_write_hwi_in_range): Use
5976         uhwi streaming to stream the normalized range.
5977         (streamer_read_hwi_in_range): Likewise.
5979 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5981         * config/arm/arm.md (enabled_for_depr_it): New attribute.
5982         (predicable_short_it): Likewise.
5983         (predicated): Likewise.
5984         (enabled): Handle above.
5985         (define_cond_exec): Set predicated attribute to yes.
5987 2013-06-05  Mike Stump  <mikestump@comcast.net>
5989         * gdbinit.in (__FUNCTION__): Add.
5991 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
5993         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
5994         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
5996 2013-06-05  Jan Hubicka  <jh@suse.cz>
5998         * varasm.c (mark_decl_referenced): Revert the removal until targets
5999         are fixed.
6001 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6003         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6004         instead of mark_decl_referenced.
6006 2013-06-05  Jan Hubicka  <jh@suse.cz>
6008         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6009         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6010         and symtab_used_from_object_file_p.
6011         (cgraph_make_node_local_1): Clear forced_by_abi.
6012         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6013         * cgraph.h (symtab_node_base): Add forced_by_abi.
6014         (decide_is_variable_needed): Remove.
6015         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6016         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6017         (decide_is_symbol_needed): ... this one; handle symbols in general;
6018         always analyze virtuals; honnor forced_by_abi.
6019         (cgraph_finalize_function): Update.
6020         (varpool_finalize_decl): Update.
6021         (symbol_defined_and_needed): Remove.
6022         (analyze_functions): Update.
6023         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6024         output_refs, input_overwrite_node): Handle forced_by_abi.
6025         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6026         (address_taken_from_non_vtable_p): ... this one.
6027         (comdat_can_be_unshared_p_1): New function.
6028         (cgraph_comdat_can_be_unshared_p): Rename to ...
6029         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6030         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6031         (function_and_variable_visibility): Clear forced_by_abi as needed.
6032         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6033         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6034         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6035         * varpool.c (decide_is_variable_needed): Remove.
6037 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6039         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6040         (arm_option_override): Override arm_restrict_it where appropriate.
6041         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6042         * config/arm/arm.opt (mrestrict-it): New command-line option.
6043         * doc/invoke.texi: Document -mrestrict-it.
6045 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6047         * tsan.c (tsan_atomic_table): Make const.
6049 2013-06-05  Richard Biener  <rguenther@suse.de>
6051         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6052         index associated with the tree we are supposed to replace.
6053         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6054         TREE_ASM_WRITTEN as zero for everything but SSA names.
6056 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6058         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6060 2013-06-04  Jan Hubicka  <jh@suse.cz>
6062         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6063         (input_node, input_varpool_node): Handle correctly external same
6064         body aliases.
6065         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6066         nodes at ltrans stage.
6068 2013-06-04  Jan Hubicka  <jh@suse.cz>
6070         * ipa-inline.c (update_caller_keys): Fix availability test.
6071         (update_callee_keys): Likewise.
6072         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6073         to follow ELF standard.
6075 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6077         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6078         (mips64r5900el-*-elf*): New configurations.
6079         * config/mips/mips-cpus.def (r5900): New processor.
6080         * config/mips/mips-tables.opt: Regenerate.
6081         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6082         (mips_issue_rate): Handle PROCESSOR_R5900.
6083         (mips_reorg_process_insns): Force reorder mode for the R5900.
6084         * config/mips/mips.h (TARGET_MIPS5900): Define.
6085         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6086         TARGET_MIPS5900.
6087         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6088         TARGET_MIPS5900.
6089         * config/mips/mips.md (processor): Add r5900.
6090         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6092 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6094         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6095         into function to generate MOVI instruction.
6096         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6097         (aarch64_preferred_simd_mode): Turn into wrapper.
6098         (aarch64_output_scalar_simd_mov_immediate): New function.
6099         * config/aarch64/aarch64-protos.h: Add prototype for above.
6101 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6103         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6104         element_char member.
6105         (sizetochar): Return signed char.
6106         (aarch64_simd_valid_immediate): Remove elchar and other
6107         unnecessary variables.
6108         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6109         Calculate element_char as required.
6110         * config/aarch64/aarch64-protos.h: Update and move prototype
6111         for aarch64_output_simd_mov_immediate.
6112         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6113         Update arguments.
6115 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6117         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6118         information completed by aarch64_simd_valid_immediate.
6119         (aarch64_legitimate_constant_p): Update arguments.
6120         (aarch64_simd_valid_immediate): Work with struct rather than many
6121         pointers.
6122         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6123         (aarch64_simd_make_constant): Update arguments.
6124         (aarch64_output_simd_mov_immediate): Work with struct rather than
6125         many pointers.  Output immediate directly rather than as operand.
6126         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6127         Update prototype.
6128         * config/aarch64/constraints.md (Dn): Update arguments.
6130 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6132         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6133         longer static.
6134         (aarch64_simd_immediate_valid_for_move): Remove.
6135         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6136         (aarch64_simd_make_constant): Update call.
6137         (aarch64_output_simd_mov_immediate): Update call.
6138         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6139         Add prototype.
6140         * config/aarch64/constraints.md (Dn): Update call.
6142 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6144         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6145         return type to bool for prototype.
6146         (aarch64_legitimate_constant_p): Check for true instead of not -1.
6147         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6148         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6150 2013-06-04  Catherine Moore  <clm@codesourcery.com>
6152         * config/mips/mips.opt (meva): New.
6153         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6154         (ASM_SPEC): Handle -meva.
6155         * doc/invoke.texi (meva):  Document.
6157 2013-06-04  Alan Modra  <amodra@gmail.com>
6159         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6160         constant output.
6162 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6164         * rtl.def: Add extra fourth optional field to define_cond_exec.
6165         * gensupport.c (process_one_cond_exec): Process attributes from
6166         define_cond_exec.
6167         * doc/md.texi: Document fourth field in define_cond_exec.
6169 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
6171         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6172         out the processing order as in store_bit_field_1.
6174 2013-06-04  Jan Hubicka  <jh@suse.cz>
6176         PR middle-end/57500
6177         * cgraphunit.c (cgraph_process_same_body_aliases): Create
6178         non-VAR_DECL node if it does not exist yet.
6180 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
6182         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6183         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6184         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6185         target_cpu_default setting.
6187 2013-06-03  Teresa Johnson  <tejohnson@google.com>
6189         * dumpfile.c (opt_info_switch_p): Change -fopt-info
6190         default to -fopt-info=optimized instead of all.
6191         * doc/invoke.texi: Ditto.
6192         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6193         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6194         (execute_vect_slp): Emit BB vectorization success under
6195         MSG_OPTIMIZED_LOCATIONS.
6196         * tree-vect-slp.c (vect_slp_transform_bb): Change
6197         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6198         * tree-vect-loop.c (vect_transform_loop): Ditto.
6200 2013-06-03  Jason Merrill  <jason@redhat.com>
6202         PR c++/57415
6203         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6204         Use TARGET_EXPR for C++.
6206 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
6208         PR rtl-optimization/57268
6209         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6210         if DEBUG_INSN_P (insn).
6212         Reapply
6213         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6215         PR rtl-optimization/57268
6216         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6217         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6219 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6221         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6222         (ix86_avoid_lea_for_addr): Likewise.
6223         (exact_dependency_1): Likewise.
6224         (ix86_adjust_cost): Likewise.
6225         (swap_top_of_ready_list): Fix formatting and !reload_completed check
6226         removed.
6227         (do_reorder_for_imul): Fix typo, formatting and
6228         !reload_completed check removed.
6229         (ix86_sched_reorder): Fix typo and formatting.
6230         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6231         list.
6233 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
6235         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6237 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
6239         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6240         <STRING_CST>: Likewise.
6241         <VECTOR_CST>: Likewise.
6243 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
6244             Mikael Morin  <mikael@gcc.gnu.org>
6246         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6247         * config.in: Regenerated.
6248         * configure: Regenerated.
6250 2013-06-01  Jan Hubicka  <jh@suse.cz>
6252         PR middle-end/57366
6253         * cgraphunit.c (compile): When weakref is not supported,
6254         set up transparent aliases before final output pass.
6255         * varasm.c (assemble_alias): Do not try to do it here.
6257 2013-06-01  Jan Hubicka  <jh@suse.cz>
6259         PR middle-end/57467
6260         * passes.c (for_per_function): Skip unanalyzed functions.
6262 2013-06-01  Jan Hubicka  <jh@suse.cz>
6264         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6265         (lto_symtab_merge_symbols_1): ... this one.
6266         (lto_symtab_merge_cgraph_nodes): Rename to ...
6267         (lto_symtab_merge_symbols): ... this one; simplify.
6268         * cgraph.c (same_body_aliases_done): Rename to ...
6269         (cpp_implicit_aliases_done): ... this one.
6270         (cgraph_create_function_alias): Update.
6271         (cgraph_same_body_alias): Update.
6272         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6273         (verify_edge_corresponds_to_fndecl): Simplify.
6274         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6275         (cgraph_node): Remove same_body_alias.
6276         (varpool_node): Remove alias_of and extra_name_alias.
6277         (same_body_aliases_done): Rename to ..
6278         (cpp_implicit_aliases_done): ... this one.
6279         (symtab_alias_ultimate_target): Add default parameter.
6280         (symtab_resolve_alias): New function.
6281         (fixup_same_cpp_alias_visibility): Declare.
6282         (cgraph_function_node): Add default parameter.
6283         (cgraph_node_asm_name): Likewise.
6284         (cgraph_function_or_thunk_node): Add default parameter; do
6285         not ICE when it is NULL.
6286         (varpool_variable_node): Likewise.
6287         * tree-emutls.c (create_emultls_var): Update.
6288         (ipa_lower_emutls): Update.
6289         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6290         (cgraph_reset_node): Reset alias info.
6291         (cgraph_finalize_function): Update.
6292         (fixup_same_cpp_alias_visibility): Move to symtab.c.
6293         (analyze_function): Simplify.
6294         (cgraph_process_same_body_aliases): Simplify.
6295         (analyze_functions): Fixup same body aliases.
6296         (handle_alias_pairs): Simplify.
6297         (assemble_thunk): Update.
6298         (assemble_thunks_and_aliases): Update.
6299         (output_weakrefs): Rewrite.
6300         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6301         (lto_output_varpool_node): Likewise.
6302         (compute_ltrans_boundary): Remve assert.
6303         (get_alias_symbol): New functoin.
6304         (input_node): Rewrite alias handling.
6305         (input_varpool_node): Likewise.
6306         * ipa-pure-const.c (propagate_pure_const): Fix formating.
6307         * ipa.c (process_references): Handle weakrefs correctly.
6308         (symtab_remove_unreachable_nodes): Likewise.
6309         * trans-mem.c (get_cg_data): Update.
6310         (ipa_tm_create_version_alias): Update.
6311         (ipa_tm_execute): Update.
6312         * symtab.c (dump_symtab_base): Dump aliases.
6313         (verify_symtab_base): Verify aliases.
6314         (symtab_node_availability): New function.
6315         (symtab_alias_ultimate_target): Simplify.
6316         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6317         handle all the fixup cases.
6318         (symtab_resolve_alias): New function.
6319         * passes.c (ipa_write_summaries): Handle weakrefs.
6320         * varpool.c (varpool_analyze_node): Simplify.
6321         (assemble_aliases): Update.
6322         (varpool_create_variable_alias): Simplify.
6323         (varpool_extra_name_alias): Simplify.
6324         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6325         (lto_symtab_merge_symbols): ... this one.
6327 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
6329         Revert
6330         PR rtl-optimization/57268
6331         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6332         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6334 2013-06-01  Tobias Burnus  <burnus@net-b.de>
6336         Partially reverted:
6337         2013-05-31  Tobias Burnus  <burnus@net-b.de>
6339         PR middle-end/57073
6340         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6341         further up.
6343 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6345         PR rtl-optimization/57268
6346         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6347         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6349 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6351         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6352         unordered comparison operators when -fno-trapping-math is in effect
6353         on the e500.
6354         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6355         and implement unordered comparison operators properly on the e500.
6357 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6359         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6360         for constant scalar integers.
6361         (simplify_relational_operation_1): Likewise.
6363 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
6365         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6366         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
6367         Fix comment.
6369 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6370             Igor Zamyatin  <igor.zamyatin@intel.com>
6372         Silvermont (SLM) architecture performance tuning.
6373         * config/i386/i386.h (enum ix86_tune_indices): Add
6374         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6375         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6377         * config/i386/i386.c (initial_ix86_tune_features)
6378         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6379         (ix86_lea_outperforms): Handle Silvermont tuning.
6380         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6381         call.
6382         (ix86_use_lea_for_mov): Likewise.
6383         (ix86_avoid_lea_for_addr): Likewise.
6384         (ix86_lea_for_add_ok): Likewise.
6385         (exact_dependency_1): New function.
6386         (exact_store_load_dependency): Likewise.
6387         (ix86_adjust_cost): Handle Silvermont tuning.
6388         (do_reoder_for_imul): Likewise.
6389         (swap_top_of_ready_list): New function.
6390         (ix86_sched_reorder): Changed to handle Silvermont tuning.
6392         * config/i386/i386.md (peepholes that split memory operand in fp
6393         converts): New.
6395 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6397         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6398         Remove un-necessary braces.
6400 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6402         * config/aarch64/aarch64.c (aarch64_classify_symbol):
6403         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6405 2013-05-31  Tobias Burnus  <burnus@net-b.de>
6407         PR middle-end/57073
6408         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6410 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6412         PR target/56315
6413         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6414         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6415         * config/arm/neon.md (iordi3_neon): Remove.
6416         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6417         * config/arm/predicates.md (imm_for_neon_logic_operand):
6418         Move to earlier in the file.
6419         (neon_logic_op2): Likewise.
6420         (arm_iordi_operand_neon): New predicate.
6422 2013-05-31  Richard Biener  <rguenther@suse.de>
6424         PR tree-optimization/57478
6425         PR tree-optimization/57453
6426         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6427         are life as well.
6429 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6431         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6432         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6434 2013-05-30  Tobias Burnus  <burnus@net-b.de>
6435             Thomas Koenig  <tkoenig@gcc.gnu.org>
6437         PR middle-end/57073
6438         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6439         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6441 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
6443         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6445 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
6447         * target.def (register_usage_leveling_p): New hook.
6448         * targhooks.c (default_register_usage_leveling_p): New.
6449         * targhooks.h (default_register_usage_leveling_p): New prototype.
6450         * lra-assigns.c (register_usage_leveling_p): Use the hook.
6451         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6452         * doc/tm.texi: Update.
6453         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6455 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
6457         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6458         (*insv_reg<mode>): New define_insn.
6460 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
6462         PR rtl-optimization/57439
6463         * postreload.c (move2add_valid_value_p): Check that we have
6464         a zero subreg_regno_offset when accessing the register in
6465         the requested mode.
6467 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6468             Igor Zamyatin  <igor.zamyatin@intel.com>
6470         Silvermont (SLM) architecture pipeline model, tuning and
6471         insn selection.
6472         * config.gcc: Add slm config options and target.
6474         * config/i386/slm.md: New.
6476         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6478         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
6479         PROCESSOR_SLM.
6480         (ix86_target_macros_internal): Likewise.
6482         * gcc/config/i386/i386.c (slm_cost): New cost.
6483         (m_SLM): New macro flag.
6484         (initial_ix86_tune_features): Set m_SLM.
6485         (x86_accumulate_outgoing_args): Likewise.
6486         (x86_arch_always_fancy_math_387): Likewise.
6487         (processor_target_table): Add slm cost.
6488         (cpu_names): Add slm cpu name.
6489         (x86_option_override_internal): Set SLM ISA.
6490         (ix86_issue_rate): New case PROCESSOR_SLM.
6491         (ia32_multipass_dfa_lookahead): Likewise.
6492         (fold_builtin_cpu): Add slm.
6494         * config/i386/i386.h (TARGET_SLM): New target macro.
6495         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6496         (processor_type): Add PROCESSOR_SLM.
6498         * config/i386/i386.md (cpu): Add new value "slm".
6499         (slm.md): Include slm.md.
6501 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
6502             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6504         * config/arm/arm-protos.h: Add and update function protos.
6505         * config/arm/arm.c (use_simple_return_p): New added.
6506         (thumb2_expand_return): Check simple_return flag.
6507         * config/arm/arm.md: Add simple_return and conditional simple_return.
6508         * config/arm/iterators.md: Add iterator for return and simple_return.
6510 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6512         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6513         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6514         (arm_emit_vfp_multi_reg_pop): Likewise.
6515         (thumb2_emit_ldrd_pop): Likewise.
6516         (arm_expand_epilogue): Add misc REG_CFA notes.
6517         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6519 2013-05-29  Lawrence Crowl  <crowl@google.com>
6521         * config/arm/t-arm: Update for below.
6523         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6524         Change type to hash_table.  Update dependent calls and types.
6526         * config/i386/t-cygming: Update for below.
6528         * config/i386/t-interix: Update for below.
6530         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6531         Change type to hash_table.  Update dependent calls and types.
6532         (i386_find_on_wrapper_list::wrappers): Likewise.
6534         * config/ia64/t-ia64: Update for below.
6536         * config/ia64/ia64.c (bundle_state_table):
6537         Change type to hash_table.  Update dependent calls and types.
6539         * config/mips/mips.c (mips_reorg_process_insns::htab):
6540         Change type to hash_table.  Update dependent calls and types.
6542         * config/sol2.c (solaris_comdat_htab):
6543         Change type to hash_table.  Update dependent calls and types.
6545         * config/t-sol2: Update for above.
6547 2013-05-29  Teresa Johnson  <tejohnson@google.com>
6549         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6550         functions are not yet marked as defined.
6552 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
6553             Pat Haugen <pthaugen@us.ibm.com>
6554             Peter Bergner <bergner@vnet.ibm.com>
6556         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6557         instructions.
6558         (VEC_A): Likewise.
6559         (VEC_C): Likewise.
6560         (vrotl<mode>3): Likewise.
6561         (vashl<mode>3): Likewise.
6562         (vlshr<mode>3): Likewise.
6563         (vashr<mode>3): Likewise.
6565         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6566         support for power8 V2DI builtins.
6568         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6569         power8 V2DI builtins.
6570         (vupkhsw): Likewise.
6571         (vupklsw): Likewise.
6572         (vaddudm): Likewise.
6573         (vminsd): Likewise.
6574         (vmaxsd): Likewise.
6575         (vminud): Likewise.
6576         (vmaxud): Likewise.
6577         (vpkudum): Likewise.
6578         (vpksdss): Likewise.
6579         (vpkudus): Likewise.
6580         (vpksdus): Likewise.
6581         (vrld): Likewise.
6582         (vsld): Likewise.
6583         (vsrd): Likewise.
6584         (vsrad): Likewise.
6585         (vsubudm): Likewise.
6586         (vcmpequd): Likewise.
6587         (vcmpgtsd): Likewise.
6588         (vcmpgtud): Likewise.
6589         (vcmpequd_p): Likewise.
6590         (vcmpgtsd_p): Likewise.
6591         (vcmpgtud_p): Likewise.
6592         (vupkhsw): Likewise.
6593         (vupklsw): Likewise.
6594         (vaddudm): Likewise.
6595         (vmaxsd): Likewise.
6596         (vmaxud): Likewise.
6597         (vminsd): Likewise.
6598         (vminud): Likewise.
6599         (vpksdss): Likewise.
6600         (vpksdus): Likewise.
6601         (vpkudum): Likewise.
6602         (vpkudus): Likewise.
6603         (vrld): Likewise.
6604         (vsld): Likewise.
6605         (vsrad): Likewise.
6606         (vsrd): Likewise.
6607         (vsubudm): Likewise.
6609         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6610         support for power8 V2DI instructions.
6612         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6613         power8 V2DI instructions.  Combine pack and unpack insns to use an
6614         iterator for each mode.  Check whether a particular mode supports
6615         Altivec instructions instead of just checking TARGET_ALTIVEC.
6616         (UNSPEC_VPKUWUM): Likewise.
6617         (UNSPEC_VPKSHSS): Likewise.
6618         (UNSPEC_VPKSWSS): Likewise.
6619         (UNSPEC_VPKUHUS): Likewise.
6620         (UNSPEC_VPKSHUS): Likewise.
6621         (UNSPEC_VPKUWUS): Likewise.
6622         (UNSPEC_VPKSWUS): Likewise.
6623         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
6624         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
6625         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
6626         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
6627         (UNSPEC_VUPKHSB): Likewise.
6628         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
6629         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
6630         (UNSPEC_VUPKHSH): Likewise.
6631         (UNSPEC_VUPKLSB): Likewise.
6632         (UNSPEC_VUPKLSH): Likewise.
6633         (VI2): Likewise.
6634         (VI_char): Likewise.
6635         (VI_scalar): Likewise.
6636         (VI_unit): Likewise.
6637         (VP): Likewise.
6638         (VP_small): Likewise.
6639         (VP_small_lc): Likewise.
6640         (VU_char): Likewise.
6641         (add<mode>3): Likewise.
6642         (altivec_vaddcuw): Likewise.
6643         (altivec_vaddu<VI_char>s): Likewise.
6644         (altivec_vadds<VI_char>s): Likewise.
6645         (sub<mode>3): Likewise.
6646         (altivec_vsubcuw): Likewise.
6647         (altivec_vsubu<VI_char>s): Likewise.
6648         (altivec_vsubs<VI_char>s): Likewise.
6649         (altivec_vavgs<VI_char>): Likewise.
6650         (altivec_vcmpbfp): Likewise.
6651         (altivec_eq<mode>): Likewise.
6652         (altivec_gt<mode>): Likewise.
6653         (altivec_gtu<mode>): Likewise.
6654         (umax<mode>3): Likewise.
6655         (smax<mode>3): Likewise.
6656         (umin<mode>3): Likewise.
6657         (smin<mode>3): Likewise.
6658         (altivec_vpkuhum): Likewise.
6659         (altivec_vpkuwum): Likewise.
6660         (altivec_vpkshss): Likewise.
6661         (altivec_vpkswss): Likewise.
6662         (altivec_vpkuhus): Likewise.
6663         (altivec_vpkshus): Likewise.
6664         (altivec_vpkuwus): Likewise.
6665         (altivec_vpkswus): Likewise.
6666         (altivec_vpks<VI_char>ss): Likewise.
6667         (altivec_vpks<VI_char>us): Likewise.
6668         (altivec_vpku<VI_char>us): Likewise.
6669         (altivec_vpku<VI_char>um): Likewise.
6670         (altivec_vrl<VI_char>): Likewise.
6671         (altivec_vsl<VI_char>): Likewise.
6672         (altivec_vsr<VI_char>): Likewise.
6673         (altivec_vsra<VI_char>): Likewise.
6674         (altivec_vsldoi_<mode>): Likewise.
6675         (altivec_vupkhsb): Likewise.
6676         (altivec_vupkhs<VU_char>): Likewise.
6677         (altivec_vupkls<VU_char>): Likewise.
6678         (altivec_vupkhsh): Likewise.
6679         (altivec_vupklsb): Likewise.
6680         (altivec_vupklsh): Likewise.
6681         (altivec_vcmpequ<VI_char>_p): Likewise.
6682         (altivec_vcmpgts<VI_char>_p): Likewise.
6683         (altivec_vcmpgtu<VI_char>_p): Likewise.
6684         (abs<mode>2): Likewise.
6685         (vec_unpacks_hi_v16qi): Likewise.
6686         (vec_unpacks_hi_v8hi): Likewise.
6687         (vec_unpacks_lo_v16qi): Likewise.
6688         (vec_unpacks_hi_<VP_small_lc>): Likewise.
6689         (vec_unpacks_lo_v8hi): Likewise.
6690         (vec_unpacks_lo_<VP_small_lc>): Likewise.
6691         (vec_pack_trunc_v8h): Likewise.
6692         (vec_pack_trunc_v4si): Likewise.
6693         (vec_pack_trunc_<mode>): Likewise.
6695         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
6696         V2DI builtins.
6697         (vec_vmaxsd): Likewise.
6698         (vec_vmaxud): Likewise.
6699         (vec_vminsd): Likewise.
6700         (vec_vminud): Likewise.
6701         (vec_vpksdss): Likewise.
6702         (vec_vpksdus): Likewise.
6703         (vec_vpkudum): Likewise.
6704         (vec_vpkudus): Likewise.
6705         (vec_vrld): Likewise.
6706         (vec_vsld): Likewise.
6707         (vec_vsrad): Likewise.
6708         (vec_vsrd): Likewise.
6709         (vec_vsubudm): Likewise.
6710         (vec_vupkhsw): Likewise.
6711         (vec_vupklsw): Likewise.
6713 2013-05-29  Jan Hubicka  <jh@suse.cz>
6715         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
6716         flags; reorder rest of fields in more consistent way.
6717         (varpool_node): Remove analyzed, finalized and alias.
6718         (cgraph_ndoe): Likewise.
6719         (symtab_alias_ultimate_target): New function.
6720         (cgraph_function_node): Move offline.
6721         (cgraph_reset_node): Declare.
6722         (cgraph_comdat_can_be_unshared_p): Remove.
6723         (varpool_remove_initializer): Declare.
6724         (varpool_first_defined_variable, varpool_next_defined_variable
6725         cgraph_first_defined_function, cgraph_next_defined_function): Update.
6726         (cgraph_function_with_gimple_body_p): Update.
6727         (varpool_all_refs_explicit_p): Update.
6728         (symtab_alias_target): New function.
6729         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
6730         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
6731         (cgraph_function_or_thunk_node): Simplify using
6732         symtab_alias_ultimate_target.
6733         (varpool_variable_node): Likewise.
6734         * cgraph.c (cgraph_create_function_alias): Update.
6735         (cgraph_add_thunk): Update.
6736         (cgraph_remove_node): Update.
6737         (dump_cgraph_node): Do not dump removed flags.
6738         (cgraph_function_body_availability): Update.
6739         (cgraph_propagate_frequency): Update.
6740         (verify_cgraph_node): Check sanity of local flag.
6741         (cgraph_function_node): Move here from cgraph.h; revamp for
6742         cgraph_function_or_thunk_node.
6743         * lto-symtab.c (lto_varpool_replace_node): Update.
6744         (lto_symtab_resolve_can_prevail_p): Update.
6745         (lto_symtab_merge_cgraph_nodes): Update.
6746         * ipa-cp.c (determine_versionability, initialize_node_lattices,
6747         propagate_constants_accross_call, devirtualization_time_bonus,
6748         ipcp_propagate_stage): Update.
6749         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
6750         * ipa-inline-transform.c (clone_inlined_nodes,
6751         preserve_function_body_p): Update.
6752         * ipa-reference.c (propagate): Update.
6753         (write_node_summary_p): Update.
6754         * toplev.c (wrapup_global_declaration_2): Update.
6755         * cgraphunit.c (cgraph_analyze_function): Rename to ...
6756         (analyze_function) ... this one.
6757         (cgraph_process_new_functions): Update.
6758         (cgraph_reset_node): Export.
6759         (cgraph_finalize_function): Update.
6760         (cgraph_add_new_function): Update.
6761         (process_function_and_variable_attributes): Update.
6762         (varpool_finalize_decl): Update.
6763         (symbol_finalized): Remove.
6764         (symbol_finalized_and_needed): Rename to ...
6765         (symbol_defined_and_needed): ... update.
6766         (cgraph_analyze_functions): Update.
6767         (handle_alias_pairs): Update.
6768         (mark_functions_to_output): Update.
6769         (assemble_thunk): Update.
6770         (output_in_order): Update.
6771         (output_weakrefs): Update.
6772         (finalize_compilation_unit): Update.
6773         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
6774         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
6775         input_node, input_varpool_node): Update.
6776         * dbxout.c (dbxout_expand_expr): Update.
6777         * cgraphclones.c (cgraph_clone_node): Update.
6778         (cgraph_copy_node_for_versioning): Update.
6779         (cgraph_materialize_clone): Update.
6780         (cgraph_materialize_all_clones): Update.
6781         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
6782         propagate_pure_const, propagate_nothrow): Update.
6783         * lto-streamer-out.c (lto_output, write_symbol): Update.
6784         * ipa-utils.c (ipa_reverse_postorder): Update.
6785         * ipa-inline.c (can_inline_edge_p): Update.
6786         (update_caller_keys, ipa_inline): Update.
6787         * dwarf2out.c (reference_to_unused,
6788         premark_types_used_by_global_vars_helper): Update.
6789         * tree-eh.c (tree_could_trap_p): Update.
6790         * ipa-split.c (consider_split, execute_split_functions): Update.
6791         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
6792         has_addr_references_p): Update; move ahead in file for better
6793         readability.
6794         (process_references): Simplify.
6795         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
6796         bodies are removed.
6797         (cgraph_comdat_can_be_unshared_p): Make static.
6798         (cgraph_externally_visible_p): Update.
6799         (varpool_externally_visible_p): Update.
6800         (function_and_variable_visibility): Update.
6801         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
6802         ipa_tm_mark_force_output_node): Update.
6803         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
6804         estimate_edge_devirt_benefit, inline_generate_summary,
6805         inline_write_summary): Update.
6806         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
6807         * ipa-prop.c (ipa_compute_jump_functions): Update.
6808         (ipa_print_node_params, ipa_prop_read_section,
6809         ipa_update_after_lto_read, read_replacements_section): Update.
6810         * varasm.c (mark_decl_referenced): Update.
6811         (assemble_alias, dump_tm_clone_pairs): Update.
6812         * tree-inline.c (copy_bb): Update.
6813         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
6814         Update.
6815         * symtab.c (dump_symtab_base): Print new flags.
6816         (verify_symtab_base): Verify new flags.
6817         (symtab_alias_ultimate_target): New function.
6818         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
6819         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
6820         Update.
6821         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
6822         Update.
6823         * i386.c (ix86_get_function_versions_dispatcher,
6824         ix86_generate_version_dispatcher_body): Update.
6825         (fold_builtin_cpu): Use varpool_add_new_variable.
6826         * varpool.c (varpool_remove_initializer): Break out from ...
6827         (varpool_remove_node): ... this one.
6828         (dump_varpool_node, varpool_node_for_asm,
6829         cgraph_variable_initializer_availability, varpool_analyze_node,
6830         varpool_assemble_decl, varpool_remove_unreferenced_decls,
6831         varpool_finalize_named_section_flags, varpool_create_variable_alias):
6832         Update.
6834 2013-05-29  Jan Hubicka  <jh@suse.cz>
6836         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
6838 2013-05-29  Easwaran Raman  <eraman@google.com>
6840         PR tree-optimization/57442
6841         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
6842         when control exits the main loop.
6844 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
6846         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
6847         and RX600.
6848         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
6849         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
6850         * rx/t-rx: Add rx100 under multi library matches option for nofpu
6851         option.
6853 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6855         PR tree-optimization/57441
6856         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
6857         Don't limit size of incr_vec to number of candidates.
6859 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
6861         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
6862         and mips16 directories.
6863         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
6864         (MULTILIB_DIRNAMES): Ditto.
6865         (MULTILIB_EXCEPTIONS): Add new exceptions.
6866         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
6867         (MULTILIB_DIRNAMES): Ditto.
6868         (MULTILIB_EXCEPTIONS): Add new exceptions.
6870 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6871             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6873         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
6874         SYMBOL_TINY_ABSOLUTE.
6875         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
6876         SYMBOL_TINY_ABSOLUTE.
6877         (aarch64_expand_mov_immediate): Likewise.
6878         (aarch64_classify_symbol): Likewise.
6879         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
6880         Permit SYMBOL_TINY_ABSOLUTE.
6881         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
6883 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6884             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6886         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
6887         Refactor if/switch.  Replace gcc_assert with if.
6889 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
6891         * config/i386/i386.c (initial_ix86_tune_features): Enable
6892         FP Reassociation for AMD bdver1 and bdver2.
6894 2013-05-29  Martin Jambor  <mjambor@suse.cz>
6896         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
6897         and IMAGPART_EXPR do not occur within other handled_components.
6899 2013-05-29  Richard Biener  <rguenther@suse.de>
6901         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
6902         access on whether the use is in the BB we currently try to
6903         vectorize.
6904         (vect_bb_vectorization_profitable_p): Pass the BB we currently
6905         vectorize to vect_bb_slp_scalar_cost.
6907 2013-05-29  Richard Biener  <rguenther@suse.de>
6909         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
6910         computing scalar cost offsetted by stmts that are kept live
6911         by scalar uses.
6912         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
6913         for computation of scalar cost.
6915 2013-05-28  Steve Ellcey  <sellcey@mips.com>
6917         * config/mips/mips-cpus.def (mips32r2): Change processor type.
6919 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6921         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
6922         array notation built-in reduction functions.
6923         * doc/passes.texi (Passes): Added documentation about changes done
6924         for Cilk Plus.
6925         * doc/invoke.texi (C Dialect Options): Added documentation about
6926         the -fcilkplus flag.
6927         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
6928         (BUILTINS_DEF): Depend on cilkplus.def.
6929         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
6930         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
6931         * cilkplus.def: New file.
6933 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
6935         PR rtl-optimization/57439
6936         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
6938 2013-05-28  Easwaran Raman  <eraman@google.com>
6940         PR tree-optimization/57337
6941         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
6942         (find_insert_point): Correctly identify the insertion point
6943         when two statements with the same UID is compared.
6945 2013-05-28  Richard Biener  <rguenther@suse.de>
6947         PR tree-optimization/56787
6948         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
6949         from the list of data references.
6950         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
6951         clobbers.
6952         (vect_analyze_loop_operations): Likewise.
6953         (vect_transform_loop): Remove clobbers.
6955 2013-05-28  Martin Jambor  <mjambor@suse.cz>
6957         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
6958         and REALPART_EXPRs have scalar type.
6960 2013-05-28  Richard Biener  <rguenther@suse.de>
6962         PR tree-optimization/57411
6963         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
6964         virtual operands.
6965         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
6966         virtual operand propagation.
6968 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
6970         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
6971         destination register for bmasksi_vis.
6972         (vector_init_bshuffle): Likewise.
6973         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
6975 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
6977         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
6978         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
6979         mode if the instruction isn't available in the original mode.
6980         * config/sparc/sparc.opt (mfix-ut699): New option.
6981         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
6982         (divdf3): Turn into expander.
6983         (divdf3_nofix): New insn.
6984         (divdf3_fix): Likewise.
6985         (divsf3): Disable if -mfix-ut699.
6986         (sqrtdf2): Turn into expander.
6987         (sqrtdf2_nofix): New insn.
6988         (sqrtdf2_fix): Likewise.
6989         (sqrtsf2): Disable if -mfix-ut699.
6991 2013-05-27  Richard Biener  <rguenther@suse.de>
6993         PR middle-end/57412
6994         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
6995         block for the new loop.
6997 2013-05-27  Richard Biener  <rguenther@suse.de>
6999         PR tree-optimization/57343
7000         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7001         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7002         (number_of_iterations_cond): Do not build the folded tree.
7004 2013-05-27  Richard Biener  <rguenther@suse.de>
7006         Revert
7007         PR middle-end/57381
7008         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7009         OEP_CONSTANT_ADDRESS_OF retained.
7011         PR tree-optimization/57417
7012         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7013         for unchanged base.
7014         (set_ssa_val_to): Compare addresses using
7015         get_addr_base_and_unit_offset.
7017 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7019         PR rtl-optimization/56833
7020         * postreload.c (move2add_record_mode): New function.
7021         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7022         (move2add_use_add2_insn): Use move2add_record_sym_value.
7023         (move2add_use_add3_insn): Likewise.
7024         (reload_cse_move2add): Use move2add_valid_value_p and
7025         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7026         affected regs by setting reg_mode to VOIDmode.
7027         (move2add_note_store): Don't pretend the inside of a SUBREG is
7028         the actual destination.  Invalidate single/leading registers by
7029         setting reg_mode to VOIDmode.
7030         Use move2add_record_sym_value, move2add_valid_value_p and
7031         move2add_record_mode.
7033 2013-05-27  Richard Biener  <rguenther@suse.de>
7035         PR tree-optimization/57396
7036         * tree-affine.c (double_int_constant_multiple_p): Properly
7037         return false for val == 0 and div != 0.
7039 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7041         * config/mips/mips.h: Use #elif in preprocessor conditions.
7043 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7045         PR target/53916
7046         * config/mips/constraints.md (kl): New constraint.
7047         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7048         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7049         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7050         from LO for MIPS16.
7051         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7053 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7055         PR target/55777
7056         * config/mips/mips.c (mips_can_inline_p): New function.
7057         (TARGET_CAN_INLINE_P): Define.
7059 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7061         * sched-int.h (ds_t, dw_t): Make unsigned int.
7062         Fix documentation that describes how all the ds_t bits are used.
7063         Reserve the last bit for delayed-branch scheduling.
7064         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7065         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7066         (gen_dep_weak_1): Remove prototype.
7067         * sched-deps.c (get_dep_weak_1): Make static.
7068         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7069         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7070         * doc/tm.texi: Regenerate.
7071         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7073 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7075         PR debug/56950
7076         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7078 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7079             Sandra Loosemore  <sandra@codesourcery.com>
7081         * config.gcc (powerpc-*): Allow native for with-cpu.
7083 2013-05-24  Jeff Law  <law@redhat.com>
7085         PR tree-optimization/57124
7086         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7087         conversion feeding a condition if the range has an overflow
7088         if -fstrict-overflow.  Add warnings for when we do make the
7089         transformation.
7091 2013-05-24  Dehao Chen  <dehao@google.com>
7093         * tree-cfg.c (locus_discrim_map): Fix the typo.
7094         (locus_discrim_hasher): Likewise.
7095         (locus_discrim_hasher::hash): Likewise.
7096         (locus_discrim_hasher::equal): Likewise.
7098 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7100         PR tree-optimization/57294
7101         * cgraph.h (ipa_record_stmt_references): Declare.
7102         * cgraphbuild.c (ipa_record_stmt_references): New function.
7103         (build_cgraph_edges): Use ipa_record_stmt_references.
7104         (rebuild_cgraph_edges): Likewise.
7105         (cgraph_rebuild_references): Likewise.
7106         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7107         associated with the old statement and build references from the
7108         newly built statements.
7109         * ipa-ref.c (ipa_remove_stmt_references): New function.
7110         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7112 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7114         * lra-constraints.c (emit_spill_move): Use smaller mode for
7115         mem-mem moves.
7116         (check_and_process_move): Consider mem-reg moves for secondary
7117         too.
7118         (curr_insn_transform): Don't lose insns emitted before for
7119         secondary memory moves.
7120         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7121         reg set up in the current insn.
7123 2013-05-24  Dehao Chen  <dehao@google.com>
7125         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7126         hash function.
7127         (locus_descrim_hasher::equal): Likewise.
7128         (build_gimple_cfg): New discriminator assignment algorithm.
7129         (make_edges): Likewise.
7130         (next_discriminator_for_locus): Likewise.
7131         (same_line_p): Likewise.
7132         (assign_discriminators): Likewise.
7133         (make_cond_expr_edges): Likewise.
7134         (make_gimple_switch_edges): Likewise.
7135         (make_goto_expr_edges): Likewise.
7136         (make_gimple_asm_edges): Likewise.
7138 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
7140         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7141         X format specifier to only display bottom 16 bits.
7142         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7143         immediate to match for operand 2, since it will be masked.
7145 2013-05-24  Richard Biener  <rguenther@suse.de>
7147         PR tree-optimization/57287
7148         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7149         all SSA names that occur in abnormal PHIs.
7151 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7153         PR tree-ssa/57385
7154         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7155         that index is not negative.
7157 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
7159         PR rtl-optimization/55177
7160         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7161         (simplify_byte_swapping_operation): New.
7162         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7163         (simplify_relational_operation_1): Deal with BSWAP.
7165 2013-05-23  Richard Henderson  <rth@redhat.com>
7167         PR target/56742
7168         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7169         (ix86_reorg): Call it.
7171 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
7173         PR target/57379
7174         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7175         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7176         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7178 2013-05-23  Christian Bruel  <christian.bruel@st.com>
7180         PR debug/57351
7181         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7183 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7184             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7186         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7187         * config/aarch64/constraints.md (Usa): Remove.
7188         * doc/md.texi (AArch64 Usa): Remove.
7190 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7191             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7193         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7194         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7195         * config/aarch64/predicates.md (aarch64_const_address): Remove.
7196         (aarch64_mov_operand): Use aarch64_mov_operand_p.
7198 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
7200         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7201         instruction (AdvSIMD).
7202         * config/aarch64/aarch64-builtins.c
7203         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7204         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7206 2013-05-23  Martin Jambor  <mjambor@suse.cz>
7208         PR middle-end/57347
7209         * tree.h (contains_bitfld_component_ref_p): Declare.
7210         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7211         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
7212         caller.
7213         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7214         not access a bit-field.  Assert all final offsets are byte-aligned.
7216 2013-05-23  Richard Biener  <rguenther@suse.de>
7218         PR tree-optimization/57380
7219         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7220         least one invariant or re-used load.
7221         * passes.c (init_optimization_passes): Move pass_phiprop before
7222         pass_forwprop.
7224 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
7226         * config/aarch64/aarch64-simd.md
7227         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7229 2013-05-23  Richard Biener  <rguenther@suse.de>
7231         PR middle-end/57381
7232         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7233         OEP_CONSTANT_ADDRESS_OF retained.
7235 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
7237         PR middle-end/57344
7238         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7239         don't lower unit.  Handle unit not being always BITS_PER_WORD.
7241 2013-05-23  Richard Biener  <rguenther@suse.de>
7243         PR rtl-optimization/57341
7244         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7245         instead of true_dependence.
7247 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
7249         * bb-reorder.c (branch_threshold): Make const.
7250         (exec_threshold): Ditto.
7252 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7253             Pat Haugen <pthaugen@us.ibm.com>
7254             Peter Bergner <bergner@vnet.ibm.com>
7256         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7257         documentation for the power8 crypto builtins.
7259         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7261         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7262         macros for defining power8 builtin functions.
7263         (BU_P8V_AV_2): Likewise.
7264         (BU_P8V_AV_P): Likewise.
7265         (BU_P8V_VSX_1): Likewise.
7266         (BU_P8V_OVERLOAD_1): Likewise.
7267         (BU_P8V_OVERLOAD_2): Likewise.
7268         (BU_CRYPTO_1): Likewise.
7269         (BU_CRYPTO_2): Likewise.
7270         (BU_CRYPTO_3): Likewise.
7271         (BU_CRYPTO_OVERLOAD_1): Likewise.
7272         (BU_CRYPTO_OVERLOAD_2): Likewise.
7273         (XSCVSPDP): Fix typo, point to the correct instruction.
7274         (VCIPHER): Add power8 crypto builtins.
7275         (VCIPHERLAST): Likewise.
7276         (VNCIPHER): Likewise.
7277         (VNCIPHERLAST): Likewise.
7278         (VPMSUMB): Likewise.
7279         (VPMSUMH): Likewise.
7280         (VPMSUMW): Likewise.
7281         (VPERMXOR_V2DI): Likewise.
7282         (VPERMXOR_V4SI: Likewise.
7283         (VPERMXOR_V8HI: Likewise.
7284         (VPERMXOR_V16QI: Likewise.
7285         (VSHASIGMAW): Likewise.
7286         (VSHASIGMAD): Likewise.
7287         (VPMSUM): Likewise.
7288         (VPERMXOR): Likewise.
7289         (VSHASIGMA): Likewise.
7291         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7292         __CRYPTO__ if the crypto instructions are available.
7293         (altivec_overloaded_builtins): Add support for overloaded power8
7294         builtins.
7296         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7297         support for power8 crypto builtins.
7298         (builtin_function_type): Likewise.
7299         (altivec_init_builtins): Add support for builtins that take vector
7300         long long (V2DI) arguments.
7302         * config/rs6000/crypto.md: New file, define power8 crypto
7303         instructions.
7305 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7306             Pat Haugen <pthaugen@us.ibm.com>
7307             Peter Bergner <bergner@vnet.ibm.com>
7309         * doc/invoke.texi (Option Summary): Add power8 options.
7310         (RS/6000 and PowerPC Options): Likewise.
7312         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7313         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
7314         wm, wn, wr documentation.
7316         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7317         registers if direct move instructions are enabled.
7318         (wn): New constraint for no registers.
7319         (wq): New constraint for quad word even GPR registers.
7320         (wr): New constraint if 64-bit instructions are enabled.
7321         (wv): New constraint if power8 vector instructions are enabled.
7322         (wQ): New constraint for quad word memory locations.
7324         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7325         constraint for 0..15 for crypto instructions.
7326         (gpc_reg_operand): If VSX allow registers in VSX registers as well
7327         as GPR and floating point registers.
7328         (int_reg_operand): New predicate to match only GPR registers.
7329         (base_reg_operand): New predicate to match base registers.
7330         (quad_int_reg_operand): New predicate to match even GPR registers
7331         for quad memory operations.
7332         (vsx_reg_or_cint_operand): New predicate to allow vector logical
7333         operations in both GPR and VSX registers.
7334         (quad_memory_operand): New predicate for quad memory operations.
7335         (reg_or_indexed_operand): New predicate for direct move support.
7337         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7338         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7339         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7340         (POWERPC_MASKS): Add power8 options.
7341         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7342         various options.
7344         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7345         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7347         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7348         (-mpower8-fusion): New power8 options.
7349         (-mpower8-fusion-sign): Likewise.
7350         (-mpower8-vector): Likewise.
7351         (-mcrypto): Likewise.
7352         (-mdirect-move): Likewise.
7353         (-mquad-memory): Likewise.
7355         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7356         power8.
7357         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7358         registers.
7359         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7360         (rs6000_debug_vector_unit): Add p8_vector.
7361         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7362         definitions.  Also print fusion state.
7363         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7364         (rs6000_builtin_mask_calculate): Add power8 builtin support.
7365         (rs6000_option_override_internal): Add support for power8.
7366         (rs6000_common_init_builtins): Add debugging for skipped builtins
7367         if -mdebug=builtin.
7368         (rs6000_adjust_cost): Add power8 support.
7369         (rs6000_issue_rate): Likewise.
7370         (insn_must_be_first_in_group): Likewise.
7371         (insn_must_be_last_in_group): Likewise.
7372         (force_new_group): Likewise.
7373         (rs6000_register_move_cost): Likewise.
7374         (rs6000_opt_masks): Likewise.
7376         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7377         power8 capable assembler, default to power7 options.
7378         (TARGET_DIRECT_MOVE): Likewise.
7379         (TARGET_CRYPTO): Likewise.
7380         (TARGET_P8_VECTOR): Likewise.
7381         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7382         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7383         (VECTOR_MEM_P8_VECTOR_P): Likewise.
7384         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7385         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7386         (TARGET_XSCVDPSPN): Likewise.
7387         (TARGET_XSCVSPDPN): Likewsie.
7388         (TARGET_SYNC_HI_QI): Likewise.
7389         (TARGET_SYNC_TI): Likewise.
7390         (MASK_CRYPTO): Likewise.
7391         (MASK_DIRECT_MOVE): Likewise.
7392         (MASK_P8_FUSION): Likewise.
7393         (MASK_P8_VECTOR): Likewise.
7394         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7395         temporary used by some of the direct move instructions to get two FP
7396         temporary registers does not force creation of a stack frame.
7397         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7398         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7399         that any VSX registers are tieable, even if they are also an
7400         Altivec vector mode.
7401         (r6000_reg_class_enum): Add wm, wr, wv constraints.
7402         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7403         (RS6000_BTM_CRYPTO): Likewise.
7404         (RS6000_BTM_COMMON): Likewise.
7406         * config/rs6000/rs6000.md (cpu attribute): Add power8.
7407         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7408         (enum rs6000_vector): Add power8 vector support.
7410 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7412         PR target/19599
7413         PR target/57340
7414         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7415         (any_sibcall_could_use_r3): this and handle indirect calls.
7416         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7418 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7420         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7422 2013-05-22  Richard Biener  <rguenther@suse.de>
7424         PR middle-end/57349
7425         * profile.c (branch_prob): Do not split blocks that are
7426         abnormally receiving from ECF_RETURNS_TWICE functions.
7428 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7430         * recog.c (offsettable_address_addr_space_p): Fix calculation of
7431         address mode.  Move pointer mode initialization to the same place.
7433 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7435         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7436         while it has any effect.
7438 2013-05-21  Easwaran Raman  <eraman@google.com>
7440         PR tree-optimization/57322
7441         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7442         UID of the statement added to the BB to be 1.
7444 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
7446         PR tree-optimization/57331
7447         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7448         of conversion from pointer type to integral type with integer.
7450 2013-05-21  Martin Jambor  <mjambor@suse.cz>
7452         PR lto/57289
7453         * ipa-prop.c (ipa_read_node_info): Process param_used and
7454         controlled_uses in the same order as when writing.
7456 2013-05-21  Magnus Granberg  <baldrick@free.fr>
7458         PR plugins/56754
7459         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7461 2013-05-21  Richard Biener  <rguenther@suse.de>
7463         PR tree-optimization/57318
7464         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7465         estimate stmts with side-effects as likely eliminated.
7467 2013-05-21  Richard Biener  <rguenther@suse.de>
7469         PR tree-optimization/57330
7470         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7471         preserve the call stmts fntype.
7473 2013-05-21  Richard Biener  <rguenther@suse.de>
7475         PR tree-optimization/57303
7476         * tree-ssa-sink.c (statement_sink_location): Improve killing
7477         stmt detection and properly handle self-assignments.
7479 2013-05-21  Christian Bruel  <christian.bruel@st.com>
7481         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7482         spanning registers. LEAF_REG_REMAP is supported only for contiguous
7483         registers. Set register size out of the PARALLEL loop.
7485 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
7487         PR target/56547
7488         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7489         (*fmasf4, *fmasf4_media): New insns.
7491 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7493         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7494         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7495         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7496         (mips_idiv_insns): Update the comments to say that the returned
7497         instruction counts are in units of BASE_INSN_LENGTH.
7498         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7499         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
7500         using 2 rather than 4 as the length of indirect MIPS16 and
7501         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
7502         length of a NOP.  Don't divide MIPS16 lengths by 2.
7503         (mips16_split_long_branches): Assume a branch is long if the
7504         length is greater than 4 rather than 8.
7505         * config/mips/mips.md (length): Give MIPS16 lengths directly,
7506         rather than multiplying them by 2.  Multiply instruction counts
7507         by BASE_INSN_LENGTH rather than 4.
7508         (*jump_mips16, tls_get_tp_mips16_<mode>)
7509         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7511 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7513         * config/mips/mips.md (extended_mips16): Remove branch case.
7514         (length): Remove duplicated extended_mips16 test.
7516 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7518         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7520 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
7522         * recog.h (Recog_data): Rename to...
7523         (recog_data_d): ...this.
7524         (recog_data): Update accordingly.
7525         * recog.c (recog_data): Likewise.
7526         * reload.c (save_recog_data): Likewise.
7527         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7528         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7530 2013-05-17  Julian Brown  <julian@codesourcery.com>
7532         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7533         found in a REG_EQUAL note, invalidate it.
7535 2013-05-17  Easwaran Raman  <eraman@google.com>
7537         * tree-ssa-reassoc.c (find_insert_point): New function.
7538         (insert_stmt_after): Likewise.
7539         (get_def_stmt): Likewise.
7540         (ensure_ops_are_available): Likewise.
7541         (not_dominated_by): Likewise.
7542         (rewrite_expr_tree): Do not move statements beyond what is
7543         necessary. Remove call to swap_ops_for_binary_stmt...
7544         (reassociate_bb): ... and move it here.
7545         (build_and_add_sum): Assign UIDs for new statements.
7546         (linearize_expr): Likewise.
7547         (do_reassoc): Renumber gimple statement UIDs.
7549 2013-05-17  Jan Hubicka  <jh@suse.cz>
7551         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7552         weakrefs.
7553         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7554         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7555         weakrefs.
7556         (output_weakrefs): Update.
7558 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
7559             Martin Jambor  <mjambor@suse.cz>
7561         PR middle-end/57276
7562         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7563         value that corresponds to the given aggval is found in values vector.
7565 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
7567         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7568         sse, sse2, sse3, ssse3 and sse4a flags to options.
7570 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
7572         * gengtype-state.c: (s_expr_writer): New class, to handle
7573         prettifying of output layout of s-expressions.
7574         (state_writer): New class, to write out gtype.state.
7575         (state_written_type_count): Move this variable into member data of
7576         state_writer.
7577         (s_expr_writer::s_expr_writer): New code: constructor for new class
7578         (state_writer::state_writer(): ditto
7579         (s_expr_writer::write_new_line): New function
7580         (s_expr_writer::write_any_indent): ditto
7581         (s_expr_writer::begin_s_expr): ditto
7582         (s_expr_writer::end_s_expr): ditto
7583         (write_state_fileloc): convert to method of state_writer...
7584         (state_writer:: write_state_fileloc): ...and use methods of
7585         s_expr_writer to write indentation into the gtype.state output file
7586         to visually represent the hierarchical structure of the list
7587         structures
7588         (write_state_fields): ditto, renaming to...
7589         (state_writer::write_state_fields)
7590         (write_state_a_string): ditto, renaming to...
7591         (state_writer::write_state_a_string)
7592         (write_state_string_option): ditto, renaming to...
7593         (state_writer::write_state_string_option)
7594         (write_state_type_option): ditto, renaming to...
7595         (state_writer::write_state_type_option)
7596         (write_state_nested_option): ditto, renaming to...
7597         (state_writer::write_state_nested_option)
7598         (write_state_option): ditto, renaming to...
7599         (state_writer::write_state_option)
7600         (write_state_options): ditto, renaming to...
7601         (state_writer::write_state_options)
7602         (write_state_lang_bitmap): ditto, renaming to...
7603         (state_writer::write_state_lang_bitmap)
7604         (write_state_version): ditto, renaming to...
7605         (state_writer::write_state_version)
7606         (write_state_scalar_type): ditto, renaming to...
7607         (state_writer::write_state_scalar_type)
7608         (write_state_string_type): ditto, renaming to...
7609         (state_writer::write_state_string_type)
7610         (write_state_undefined_type): ditto, renaming to...
7611         (state_writer::write_state_undefined_type)
7612         (write_state_struct_union_type): ditto, renaming to...
7613         (state_writer::write_state_struct_union_type)
7614         (write_state_struct_type): ditto, renaming to...
7615         (state_writer::write_state_struct_type)
7616         (write_state_user_struct_type): ditto, renaming to...
7617         (state_writer::write_state_user_struct_type)
7618         (write_state_lang_struct_type): ditto, renaming to...
7619         (state_writer::write_state_lang_struct_type)
7620         (write_state_param_struct_type): ditto, renaming to...
7621         (state_writer::write_state_param_struct_type)
7622         (write_state_pointer_type): ditto, renaming to...
7623         (state_writer::write_state_pointer_type)
7624         (write_state_array_type): ditto, renaming to...
7625         (state_writer::write_state_array_type)
7626         (write_state_gc_used): ditto, renaming to...
7627         (state_writer::write_state_gc_used)
7628         (write_state_common_type_content): ditto, renaming to...
7629         (state_writer::write_state_common_type_content)
7630         (write_state_type): ditto, renaming to...
7631         (state_writer::write_state_type)
7632         (write_state_pair_list): ditto, renaming to...
7633         (state_writer::write_state_pair_list)
7634         (write_state_pair): ditto, renaming to...
7635         (state_writer::write_state_pair)
7636         (write_state_typedefs): ditto, renaming to...
7637         (state_writer::write_state_typedefs)
7638         (write_state_structures): ditto, renaming to...
7639         (state_writer::write_state_structures)
7640         (write_state_param_structs): ditto, renaming to...
7641         (state_writer::write_state_param_structs)
7642         (write_state_variables): ditto, renaming to...
7643         (state_writer::write_state_variables)
7644         (write_state_srcdir): ditto, renaming to...
7645         (state_writer::write_state_srcdir)
7646         (write_state_files_list): ditto, renaming to...
7647         (state_writer::write_state_files_list)
7648         (write_state_languages): ditto, renaming to...
7649         (state_writer::write_state_languages)
7650         (write_state): create a state_writer instance and use it when
7651         writing out the state file
7653 2013-05-17  Mike Stump  <mikestump@comcast.net>
7655         PR rtl-optimization/57304
7656         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
7657         accessing DF_REF_REAL_LOC.
7659 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7661         PR rtl-optimization/57281
7662         PR rtl-optimization/57300
7663         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
7664         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
7665         what the other splitter did if the registers are dead.
7667 2013-05-17  Richard Biener  <rguenther@suse.de>
7669         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
7670         MEM_REF offsets.
7672 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7674         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
7675         linking.
7677 2013-05-17  Marek Polacek  <polacek@redhat.com>
7679         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
7680         length when doing non-zero store of storing '\0' to '\0'.
7682 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7684         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
7685         vect_external_def oprnd1 with loop_vinfo, try to emit
7686         optional cast, negation and and stmts on the loop preheader
7687         edge instead of into the pattern def seq.
7689         PR tree-optimization/57051
7690         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
7691         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
7693 2013-05-16  Nick Clifton  <nickc@redhat.com>
7695         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
7696         (rl78_is_naked_func): New function.
7697         (rl78_expand_prologue): Skip prologue generation for naked functions.
7698         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
7699         * doc/extend.texi (naked): Add RL78 to the list of processors
7700         that supports this attribute.
7702 2013-05-16  Jeff Law  <law@redhat.com>
7704         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
7706 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7708         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
7709         cache parameters using detect_caches_amd also for CYRIX,
7710         NSC and TM2 signatures.
7712 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7713             Dzianis Kahanovich  <mahatma@eu.by>
7715         PR target/45359
7716         PR target/46396
7717         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7718         VIA/Centaur processors and determine their cache parameters
7719         using detect_caches_amd.
7721 2013-05-16  Teresa Johnson  <tejohnson@google.com>
7723         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
7724         (rtl_verify_edges): New function.
7725         (rtl_verify_bb_insns): Ditto.
7726         (rtl_verify_bb_pointers): Ditto.
7727         (rtl_verify_bb_insn_chain): Ditto.
7728         (rtl_verify_fallthru): Ditto.
7729         (rtl_verify_bb_layout): Ditto.
7730         (rtl_verify_flow_info_1): Outline checks into new functions.
7731         (rtl_verify_flow_info): Ditto.
7733 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
7735         * cfghooks.c (copy_bbs): Add update_dominance argument.
7736         * cfghooks.h (copy_bbs): Update prototype.
7737         * tree-cfg.c (gimple_duplicate_sese_region):
7738         Add update_dominance argument.
7739         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
7740         * tree-ssa-loop-ch.c (copy_loop_headers): Update
7741         gimple_duplicate_sese_region call.
7742         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
7743         Update copy_bbs call.
7744         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
7745         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
7747 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7749         * tree-vectorizer.h (NUM_PATTERNS): Increment.
7750         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
7751         vect_recog_rotate_pattern.
7752         (vect_recog_rotate_pattern): New function.
7754 2013-05-16  Jason Merrill  <jason@redhat.com>
7756         * Makefile.in (LLINKER): New variable.
7757         (mostlyclean): Remove link mutex.
7758         * configure.ac: Handle --enable-link-mutex.
7759         * lock-and-run.sh: New script.
7761 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7763         PR target/19599
7764         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
7765         for NULL decl.
7767 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7769         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
7771 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
7773         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
7774         * config/arm/arm.c (next_consecutive_mem): New function.
7775         (gen_movmem_ldrd_strd): Likewise.
7776         * config/arm/arm.md (movmemqi): Update condition and code.
7777         (unaligned_loaddi, unaligned_storedi): New patterns.
7779 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7781         * config.gcc: Obsolete *-*-solaris2.9*.
7782         * doc/install.texi (Specific, *-*-solaris2*): Document it.
7784 2013-05-16  Richard Biener  <rguenther@suse.de>
7786         * passes.c (init_optimization_passes): Move pass_parallelize_loops
7787         earlier, after GRAPHITE transforms and IV canonicalization.
7789 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7791         * omp-low.c (extract_omp_for_data): For collapsed loops,
7792         if at least one of the loops is known at compile time to
7793         iterate zero times, set count to 0.
7794         (expand_omp_regimplify_p): New function.
7795         (expand_omp_for_generic): For collapsed loops, if at least
7796         one of the loops isn't known to iterate at least once,
7797         add runtime check with setting count to 0.
7798         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
7799         For unsigned types if it isn't known at compile time that
7800         the loop will iterate at least once, add runtime check to bypass
7801         the whole loop if initial condition isn't true.
7803 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
7805         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
7807 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
7809         PR middle-end/57286
7810         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
7811         transformations to avoid an infinite loop.
7813 2013-05-16  Marek Polacek  <polacek@redhat.com>
7815         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
7817 2013-05-15  Leif Ekblad  <leif@rdos.net>
7819         * config/i386/i386.c (ix86_decompose_address): Use
7820         DEFAULT_TLS_SEG_REG to access TLS segment register.
7821         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
7822         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
7823         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
7825 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
7827         PR target/57260
7828         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
7829         sibling calls to functions that would normally be lazily bound,
7830         unless $gp is call-clobbered.
7832 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
7834         * config/i386/i386.c (ix86_option_override_internal): Update
7835         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
7836         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
7837         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
7838         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
7839         of TARGET_3DNOW.
7840         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
7842 2013-05-15  Andreas Schwab  <schwab@suse.de>
7844         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
7845         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
7846         third operand.
7848 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7850         * loop-unroll.c (report_unroll_peel): Check decision before
7851         emitting unroll/peel message.
7853 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7855         * function.h (has_bb_partition): New rtl_data flag.
7856         (bb_reorder_complete): Ditto.
7857         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
7858         instead of flag_reorder_blocks_and_partition.
7859         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
7860         with some enhancements.
7861         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
7862         * bb-reorder.c (connect_traces): Check for has_bb_partition
7863         instead of flag_reorder_blocks_and_partition.
7864         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
7865         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
7866         verify_hot_cold_block_grouping.
7867         (partition_hot_cold_basic_blocks): Set has_bb_partition.
7869 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7871         PR target/19599
7872         * config/arm/predicates.md (call_insn_operand): New predicate.
7873         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
7874         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
7875         if insn is not a tail call.
7876         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
7877         registers.
7878         * config/arm/arm.h (enum reg_class): New caller save register class.
7879         (REG_CLASS_NAMES): Likewise.
7880         (REG_CLASS_CONTENTS): Likewise.
7881         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
7882         without decls.
7884 2013-05-15  Richard Biener  <rguenther@suse.de>
7886         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
7887         of MSG_OPTIMIZED_LOCATIONS.
7888         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
7889         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
7890         message.
7891         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
7892         of MSG_OPTIMIZED_LOCATIONS.
7893         (execute_vect_slp): Likewise.
7894         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
7895         (vect_create_cond_for_alias_checks): Likewise.
7896         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
7897         (vect_recog_widen_mult_pattern): Likewise.
7898         (vect_recog_widen_sum_pattern): Likewise.
7899         (vect_recog_over_widening_pattern): Likewise.
7900         (vect_recog_widen_shift_pattern): Likewise.
7901         (vect_recog_vector_vector_shift_pattern): Likewise.
7902         (vect_recog_divmod_pattern): Likewise.
7903         (vect_recog_mixed_size_cond_pattern): Likewise.
7904         (vect_recog_bool_pattern): Likewise.
7905         (vect_pattern_recog_1): Likewise.
7907 2013-05-15  Martin Jambor  <mjambor@suse.cz>
7909         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
7910         non-functions to builtin_unreachable.
7911         * ipa-inline-transform.c (inline_call): Do not assert estimates were
7912         correct when new direct edges were discovered.
7914 2013-05-15  Martin Jambor  <mjambor@suse.cz>
7916         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
7917         header, print symbol order instead of node uid, print more information
7918         about indirect edge targets.
7919         (ipa_make_edge_direct_to_target): Print symbol order instead of node
7920         uids.
7921         (ipa_make_edge_direct_to_target): Likewise.
7922         (remove_described_reference): Likewise.
7923         (propagate_controlled_uses): Likewise.
7924         (ipa_print_node_params): Also print symbol order.
7925         (ipcp_transform_function): Print symbol order instead of node uids.
7926         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
7927         (cgraph_get_create_real_symbol_node): Likewise.
7928         * ipa-cp.c (print_lattice): Likewise.
7929         (print_all_lattices): Likewise.
7930         (determine_versionability): Likewise.
7931         (initialize_node_lattices): Likewise.
7932         (estimate_local_effects): Likewise.
7933         (update_profiling_info): Likewise.
7934         (create_specialized_node): Likewise.
7935         (perhaps_add_new_callers): Likewise.
7936         (decide_about_value): Likewise.
7937         (decide_whether_version_node): Likewise.
7938         (identify_dead_nodes): Likewise.
7939         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
7940         (dump_inline_summary): Likewise.
7941         (estimate_node_size_and_time): Likewise.
7942         (inline_analyze_function): Likewise.
7943         * ipa-inline.c (report_inline_failed_reason): Likewise.
7944         (want_early_inline_function_p): Likewise.
7945         (edge_badness): Likewise.
7946         (update_edge_key): Likewise.
7947         (inline_small_functions): Likewise.  Add dumping of order to two other
7948         dumps.
7949         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
7950         instead of node uids.
7951         (propagate_pure_const): Likewise.
7952         (propagate_pure_const): Likewise.
7953         * ipa-utils.c (dump_cgraph_node_set): Likewise.
7954         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
7955         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
7956         of node uids.
7957         * tree-pretty-print.c (dump_function_header): Likewise.
7958         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
7959         Print symbol order instead of node uids.
7961 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7963         * config/s390/s390.c (s390_register_move_cost): Don't impose the
7964         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
7966 2013-05-15  Richard Biener  <rguenther@suse.de>
7968         PR tree-optimization/57275
7969         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
7970         return value for fail to do runtime alias checks for gather loads.
7972 2013-05-15  Jan Hubicka  <jh@suse.cz>
7974         PR lto/57038
7975         PR lto/47375
7976         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
7977         weakrefs are not external.
7978         (lto_symtab_merge_decls): Fix thinko when dealing with
7979         non-lto_symtab decls.
7980         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
7981         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
7982         * varpool.c (dump_varpool_node): Dump more flags.
7984 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7986         * config/i386/i386.c (processor_alias_table): Add instruction
7987         FSGSBASE for AMD bdver3 architecture.
7989 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
7991         * tree.c (warn_deprecated_use): Print file:line using locus color.
7992         * diagnostic.c (diagnostic_report_current_module): Print file:line
7993         and file:line:column using locus color.
7995 2013-05-14  Mike Stump  <mikestump@comcast.net>
7997         * gdbinit.in: Add __null.
7999 2013-05-14  Mike Stump  <mikestump@comcast.net>
8001         * recog.h: Rename struct recog_data to Recog_data.
8002         * recog.c: Likewise.
8003         * reload.c (can_reload_into): Likewise.
8004         * config/picochip/picochip.c: Likewise.
8006 2013-05-14  Mike Stump  <mikestump@comcast.net>
8008         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8010 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8012         * resource.h (struct resources): Remove unch_memory member.
8013         (CLEAR_RESOURCE): Don't clear unch_memory.
8014         * resource.c (mark_referenced_resources): Don't set it.
8015         (mark_set_resources): Likewise.
8016         (mark_target_live_regs): Don't clear it.
8017         (init_resource_info): Likewise.
8018         * reorg.c (resource_conflicts_p): Don't compare it.
8019         (redundant_insn): Don't set it.
8021         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8022         Remove prototypes.
8023         * emit-rtl.c (next_label): Remove unused function.
8024         (skip_consecutive_labels, link_cc0_insns): Move to ...
8025         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8026         only place where these functions are used, and make them static.
8028 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8030         * fold-const.c (fold_negate_expr): Handle vectors.
8031         (fold_truth_not_expr): Make it static.
8032         (fold_invert_truthvalue): New static function.
8033         (invert_truthvalue_loc): Handle vectors. Do not call
8034         fold_truth_not_expr directly.
8035         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8036         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8037         (fold_comparison): Handle vector constants.
8038         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8039         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8040         * tree.h (fold_truth_not_expr): Remove declaration.
8042 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8044         * config/aarch64/aarch64-simd.md
8045         (aarch64_vcond_internal<mode>): Rename to...
8046         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8047         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8048         float modes. Clarify all iterator modes.
8049         (vcond<mode><mode>): Use new name for vcond expanders.
8050         (vcond<v_cmp_result><mode>): Likewise.
8051         (vcondu<mode><mode>: Likewise.
8052         * config/aarch64/iterators.md (VDQF_COND): New.
8054 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8056         PR bootstrap/57266
8057         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8058         variable for the shift amount. Check that we shift by non-negative
8059         amounts.
8061 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8063         PR target/42017
8064         * config/arm/arm.h (EPILOGUE_USES): Only return true
8065         for LR_REGNUM after epilogue_completed.
8067 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8069         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8070         is error_mark_node.
8072 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8074         PR target/57261
8075         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8076         and Solaris 11+/x86 with gld.
8077         * configure: Regenerate.
8079 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8081         * expmed.c (expand_shift_1): Canonicalize rotates by
8082         constant bitsize / 2 to bitsize - 1.
8083         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8084         case ROTATERT>: Likewise.
8086         Revert:
8087         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8089         * config/i386/i386.md (rotateinv): New code attr.
8090         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8091         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8092         roll $31, %eax, etc.
8094 2013-05-14  Richard Biener  <rguenther@suse.de>
8096         PR middle-end/57235
8097         * tree-eh.c (sink_clobbers): Give up for successors with
8098         multiple predecessors and no virtual uses.
8100 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8102         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8103         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8105 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8107         PR middle-end/57251
8108         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8109         the case when both op0 and op1 have VOIDmode.
8111 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8113         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8114         in multiply-accumulate mode.
8116 2013-05-13  Guozhi Wei  <carrot@google.com>
8118         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8120 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8122         PR target/56975
8123         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8124         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8125         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8126         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8127         (ix86_expand_prologue): Likewise.
8128         (ix86_expand_split_stack_prologue): Likewise.
8129         (legitimate_pic_address_disp_p): Likewise.
8130         (legitimize_pic_address): Likewise.
8131         (legitimize_tls_address): Likewise.
8132         (legitimize_pe_coff_symbol): Likewise.
8133         (output_pic_addr_const): Likewise.
8134         (construct_plt_address): Likewise.
8135         (ix86_expand_call): Likewise.
8136         (x86_output_mi_thunk): Likewise.
8137         (x86_function_profiler): Likewise.
8139 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
8141         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8142         similar switch cases.
8143         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8144         (aarch64_simd_mov_to_<mode>low): Delete.
8145         (aarch64_simd_mov_to_<mode>high): Delete.
8146         (move_lo_quad_<mode>): Add w<-r alternative.
8147         (aarch64_simd_move_hi_quad_<mode>): Likewise.
8148         (aarch64_simd_mov_from_*): Update type attribute.
8149         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8150         statement.
8152 2013-05-13  Jan Hubicka  <jh@suse.cz>
8154         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8155         * config/i386/i386.c (ix86_compute_frame_layout,
8156         ix86_expand_epilogue, emit_i387_cw_initialization,
8157         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8158         ix86_local_alignment): Fix use of size/speed predicates.
8160 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8162         PR tree-optimization/45216
8163         PR tree-optimization/57157
8164         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8165         the (-Y) & (B - 1) variant if OP is |.
8166         * expmed.c (expand_shift_1): For rotations by const0_rtx just
8167         return shifted.  Use (-op1) & (prec - 1) as other_amount
8168         instead of prec - op1.
8170 2013-05-13  Martin Jambor  <mjambor@suse.cz>
8172         PR middle-end/42371
8173         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8174         (ipa_constant_data): New type.
8175         (ipa_jump_func): Use ipa_constant_data to hold information about
8176         constant jump functions.
8177         (ipa_get_jf_constant): Adjust to jump function type changes.
8178         (ipa_get_jf_constant_rdesc): New function.
8179         (ipa_param_descriptor): New field controlled_uses.
8180         (ipa_get_controlled_uses): New function.
8181         (ipa_set_controlled_uses): Likewise.
8182         * ipa-ref.h (ipa_find_reference): Declare.
8183         * ipa-prop.c (ipa_cst_ref_desc): New type.
8184         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8185         changes.
8186         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
8187         New parameter cs.  Adjust all callers.
8188         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8189         (remove_described_reference): New function.
8190         (jfunc_rdesc_usable): Likewise.
8191         (try_make_edge_direct_simple_call): Decrement controlled use count,
8192         attempt to remove reference if it hits zero.
8193         (combine_controlled_uses_counters): New function.
8194         (propagate_controlled_uses): Likewise.
8195         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8196         (ipa_edge_duplication_hook): Duplicate reference descriptions.
8197         (ipa_print_node_params): Print described use counter.
8198         (ipa_write_jump_function): Adjust to jump function type changes.
8199         (ipa_read_jump_function): New parameter CS, pass it to
8200         ipa_set_jf_constant.  Adjust caller.
8201         (ipa_write_node_info): Stream controlled use count
8202         (ipa_read_node_info): Likewise.
8203         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8204         asserting.
8205         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8206         count.  Remove cloning-added reference if it reaches zero.
8207         * ipa-ref.c (ipa_find_reference): New function.
8209 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8211         * config/i386/i386.c (processor_target_table): Modified default
8212         alignment values for AMD BD and BT architectures.
8214 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
8216         * tree-vect-generic.c (uniform_vector_p): Move ...
8217         * tree.c (uniform_vector_p): ... here.
8218         * tree.h (uniform_vector_p): Declare it.
8219         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8220         into a scalar.
8222 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8224         PR tree-optimization/57230
8225         * tree-ssa-strlen.c (handle_char_store): Record length for
8226         array store from STRING_CST.
8228         PR tree-optimization/57230
8229         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8230         check.
8232 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
8234         * config/epiphany/epiphany.c (epiphany_init): Check size of
8235         NUM_MODES_FOR_MODE_SWITCHING.
8236         (epiphany_expand_prologue):
8237         Remove CONFIG_REGNUM initial value handling code.
8238         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8239         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8240         (emit_set_fp_mode, epiphany_mode_after): Likewise.
8241         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8242         Don't return 1 for FP_MODE_NONE.
8243         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8244         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8245         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8246         * config/epiphany/epiphany.md (save_config): New pattern.
8248 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
8250         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8252 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
8254         * config/i386/i386.md (memory): Handle sseishft1.
8255         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8256         (*vec_extractv2di_1): Ditto.
8258 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
8260         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8261         saved registers.
8263 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8265         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8266         Add mthumb/march=armv7-a multilib.
8267         Add mthumb/march=armv7-r multilib.
8268         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8270 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
8272         * config/v850/t-rtems: Add more multilibs.
8274 2013-05-10  Richard Biener  <rguenther@suse.de>
8276         PR tree-optimization/57214
8277         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8278         not propagate from SSA names that occur in abnormal PHI nodes.
8280 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
8282         * stor-layout.c (element_precision): New function.
8283         * machmode.h (element_precision): Declare it.
8284         * tree.c (build_minus_one_cst): New function.
8285         (element_precision): Likewise.
8286         * tree.h (build_minus_one_cst): Declare new function.
8287         (element_precision): Likewise.
8288         * fold-const.c (operand_equal_p): Use element_precision.
8289         (fold_binary_loc): Handle vector types.
8290         * convert.c (convert_to_integer): Use element_precision.
8291         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8292         separately.
8294 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
8296         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8297         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8298         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8299         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8300         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8301         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8302         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8303         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8304         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8305         (Uuw8): New constraints.
8306         (Usb4): Move into alphabetical order.
8307         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8308         (sd8_operand, ub8_operand, uw8_operand): New predicates.
8309         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8310         previously unnamed patterns.
8311         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8312         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8313         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8314         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8315         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
8316         of specific lengths.
8318 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8320         * config/i386/i386.md (rotateinv): New code attr.
8321         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8322         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8323         roll $31, %eax, etc.
8325         PR tree-optimization/45216
8326         PR tree-optimization/57157
8327         * tree-ssa-forwprop.c (simplify_rotate): New function.
8328         (ssa_forward_propagate_and_combine): Call it.
8330 2013-05-10  Richard Biener  <rguenther@suse.de>
8332         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8333         disable peeling when we version for aliasing.
8334         (vector_alignment_reachable_p): Honor explicit user alignment.
8335         (vect_supportable_dr_alignment): Likewise.
8336         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8337         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8338         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8339         then peeling to arrange for the cost-model check to come first.
8341 2013-05-10  Alan Modra  <amodra@gmail.com>
8343         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8344         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8345         * configure: Regenerate.
8347 2013-05-10  Alan Modra  <amodra@gmail.com>
8349         PR target/55033
8350         * varasm.c (default_elf_select_section): Move !DECL_P check..
8351         (get_named_section): ..to here before calling get_section_name.
8352         Adjust assertion.
8353         (default_section_type_flags): Add DECL_P check.
8354         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8355         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8357 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
8359         * config/epiphany/epiphany.c (epiphany_expand_prologue):
8360         When using gen_stack_adjust_str with a register offset, add a
8361         REG_FRAME_RELATED_EXPR note.
8363 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
8365         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8366         (*vec_extractv4si_zext_mem): Ditto.
8367         (*vec_extractv2di): Add 0->x and x->x alternatives.
8368         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8369         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8371 2013-05-09  Jason Merrill  <jason@redhat.com>
8373         N3639 C++1y VLA support
8374         * gimplify.c (gimplify_vla_decl): Don't touch an existing
8375         DECL_VALUE_EXPR.
8377         * tree.c (build_constructor_va): New.
8378         * tree.h: Declare it.
8380 2013-05-09  Martin Jambor  <mjambor@suse.cz>
8382         PR lto/57084
8383         * gimple-fold.c (canonicalize_constructor_val): Call
8384         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8386 2013-05-09  Jan Hubicka  <jh@suse.cz>
8387             Richard Biener  <rguenther@suse.de>
8389         PR lto/54095
8390         * symtab.c (symtab_make_decl_local): Do not add private names.
8392 2013-05-09  Jan Hubicka  <jh@suse.cz>
8394         PR lto/54095
8395         * symtab.c (insert_to_assembler_name_hash): Handle clones.
8396         (unlink_from_assembler_name_hash): Likewise.
8397         (symtab_prevail_in_asm_name_hash, symtab_register_node,
8398         symtab_unregister_node, symtab_initialize_asm_name_hash,
8399         change_decl_assembler_name): Update.
8401 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
8403         * config/aarch64/aarch64.md: New movtf split.
8404         (*movtf_aarch64): Update.
8405         (aarch64_movdi_tilow): Handle TF modes and rename to
8406         aarch64_movdi_<mode>low.
8407         (aarch64_movdi_tihigh): Handle TF modes and rename to
8408         aarch64_movdi_<mode>high
8409         (aarch64_movtihigh_di): Handle TF modes and rename to
8410         aarch64_mov<mode>high_di
8411         (aarch64_movtilow_di): Handle TF modes and rename to
8412         aarch64_mov<mode>low_di
8413         (aarch64_movtilow_tilow): Remove spurious whitespace.
8414         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8415         splits.
8416         (aarch64_print_operand): Update.
8418 2013-05-09  Alan Modra  <amodra@gmail.com>
8420         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8421         powerpc64le.
8422         * configure: Regenerate.
8424 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8426         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8427         splitter preparation statements.
8428         * config/i386/sse.md (*vec_extract* splitters): Ditto.
8429         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8430         adjust_address_nv.
8432 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8434         * gimple-ssa-strength-reduction.c (count_candidates): Change
8435         return value to int.
8436         (analyze_candidates_and_replace): Change type of length to int.
8438 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8440         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8441         (*vec_extract<mode>): Use VI12_128 mode iterator.
8442         (*vec_extract<mode>_mem): Ditto.
8443         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8444         attribute.
8446 2013-05-08  Diego Novillo  <dnovillo@google.com>
8448         PR bootstrap/54659
8450         Revert:
8451         2012-08-17  Diego Novillo  <dnovillo@google.com>
8453         PR bootstrap/54281
8454         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8455         * config.in: Regenerate.
8456         * configure: Regenerate.
8457         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8459 2013-05-08  Jan Hubicka  <jh@suse.cz>
8461         PR lto/54095
8462         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8463         * cgraph.h (symtab_node_base): Add unique_name.
8464         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8465         input_overwrite_node, input_varpool_node): Stream unique_name.
8466         * cgraphclones.c (cgraph_create_virtual_clone,
8467         cgraph_function_versioning): Set unique_name.
8468         * ipa.c (function_and_variable_visibility): Set unique_name.
8470 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8472         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8473         (alloc_cand_and_find_basis): Restrict conditional candidate
8474         processing to CAND_MULTs.
8476 2013-05-08  Jan Hubicka  <jh@suse.cz>
8478         PR lto/54095
8479         lto-symtab.c (lto_symtab_symbol_p): New function.
8480         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8481         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8482         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8483         Skip static symbols.
8485 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
8487         PR tree-optimization/57200
8488         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8489         Only call inform if the preceding warning_at returns true.
8491 2013-05-07  Han Shen  <shenhan@google.com>
8493         * cfgexpand.c (record_or_union_type_has_array_p): New function.
8494         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8495         * common.opt (fstack-protector-strong): New option.
8496         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8497         * doc/invoke.texi (Optimization Options): Document
8498         "-fstack-protector-strong".
8499         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8501 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8503         * config/mips/mips.c (mips_machine_reorg2): Return 0.
8505 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
8507         * ira.c (update_equiv_regs): Add insn having equiv memory even if
8508         it is not lhs of the insn.
8509         (setup_reg_equiv): Remove insn having equiv memory which it is not
8510         lhs of the insn.
8511         * lra-constraints.c (process_address): Try to improve generation
8512         code for address base + disp.
8513         (lra_constraints): Make correct the code for checking insn setting
8514         up backward equivalence.  Remove insn only if it is in the init
8515         insn list.
8516         * lra-eliminations.c (update_reg_eliminate): Change return value.
8517         (lra_eliminate): Use the result.
8519 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
8521         * config/i386/sse.md (ssescalarnummask): New mode attribute.
8522         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8523         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8524         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8525         register target operands.
8526         (*vec_extractv8hi_sse2): New pattern.
8527         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8528         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8529         (*vec_extract<mode>_mem): New insn and split pattern.
8531 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
8533         * config/arm/arm.c (arm_asan_shadow_offset): New function.
8534         (TARGET_ASAN_SHADOW_OFFSET): Define.
8535         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8536         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8538 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8540         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8541         (incr_vec_index): Return -1 if increment not found.
8542         (create_add_on_incoming_edge): Assert if increment not found.
8543         (record_increment): Limit number of increments recorded.
8544         (all_phi_incrs_profitable): Return false if an increment not found.
8545         (replace_profitable_candidates): Don't process increments that were
8546         not recorded.
8547         (analyze_candidates_and_replace): Limit size of incr_vec.
8549 2013-05-07  Richard Biener  <rguenther@suse.de>
8551         * calls.c (special_function_p): setjmp-like functions are leaf.
8552         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8553         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8555 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
8557         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8558         (aarch64_simd_mov<mode>): New expander.
8559         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8560         (aarch64_simd_mov_to_<mode>high): Likewise.
8561         (aarch64_simd_mov_from_<mode>low): Likewise.
8562         (aarch64_simd_mov_from_<mode>high): Likewise.
8563         (aarch64_dup_lane<mode>): Update.
8564         (aarch64_dup_lanedi): New instruction pattern.
8565         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8566         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8568 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8570         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8571         (replace_mult_candidate): Remove unnecessary argument; remove
8572         unnecessary parameter from call to introduce_cast_before_cand.
8573         (replace_unconditional_candidate): Remove unnecessary parameter
8574         from call to replace_mult_candidate.
8575         (replace_conditional_candidate): Likewise.
8576         (insert_initializers): Use make_temp_ssa_name.
8577         (introduce_cast_before_cand): Remove unnecessary argument; use
8578         make_temp_ssa_name.
8579         (replace_one_candidate): Remove unnecessary argument; remove
8580         unnecessary parameter from calls to introduce_cast_before_cand.
8581         (replace_profitable_candidates): Remove unnecessary parameters
8582         from calls to replace_one_candidate.
8584 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8586         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8587         phi def as possibly hiding a basis for a CAND_ADD whose operands
8588         have been commuted in the analysis.
8589         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8591 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
8593         * config/aarch64/aarch64.md
8594         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8595         shift value between 0-4.
8597 2013-05-07  Richard Biener  <rguenther@suse.de>
8599         * double-int.h (rshift): New overload.
8600         * double-int.c (rshift): New function.
8601         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8602         (create_reference_ops_from_ref): Remove.
8603         (vn_reference_insert): Use shared ops for constructing the
8604         reference and copy it.
8606 2013-05-07  Richard Biener  <rguenther@suse.de>
8608         PR middle-end/57190
8609         * tree-eh.c (sink_clobbers): Properly propagate
8610         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8612 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
8614         PR tree-optimization/57149
8615         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
8616         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
8617         collect_phi_def_edges, execute_late_warn_uninitialized): Use
8618         uninit_undefined_value_p instead of ssa_undefined_value_p.
8620         PR debug/57184
8621         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
8622         for modifier == EXPAND_INITIALIZER.
8624 2013-05-07  Anton Blanchard  <anton@samba.org>
8626         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
8627         for powerpc64 little endian.
8628         * configure: Regenerate.
8630 2013-05-06  Graham Stott  <grahams@btinternet.com>
8632         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
8633         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
8634         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
8635         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
8637 2013-05-06  Graham Stott  <grahams@btinternet.com>
8639         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
8640         codes which allow non-lvalues.
8642 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
8644         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
8645         components are all 1s.
8646         (integer_minus_onep): New function.
8647         * tree.h (integer_minus_onep): Declare it.
8648         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
8649         integer_minus_onep instead of integer_all_onesp.
8651 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8653         PR target/52933
8654         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
8655         variations of these patterns.
8657 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8659         * config/i386/i386.md (isa): Add x64_sse4 member.
8660         (enabled): Handle x64_sse4.
8661         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
8662         instruction for 64bit SSE4_1 targets.  Update insn attributes.
8663         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
8664         instruction for SSE4_1 targets.  Update insn attributes.
8665         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8666         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
8667         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
8668         const_1 selector.
8669         (*vec_extractv4si): Rename from *sse4_1_pextrd.
8670         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
8671         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
8673 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8675         PR target/57108
8676         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
8678 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
8680         * final.c (do_assembler_dialects): Don't handle curly braces and
8681         vertical bar escaped by % as dialect delimiters.
8682         (output_asm_insn): Print curly braces and vertical bar if escaped
8683         by % and ASSEMBLER_DIALECT defined.
8684         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
8685         * doc/tm.texi: Regenerated.
8687 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8689         * config/mips/mips.c: Include tree-pass.h.
8690         (mips_reorg): Split in pre- and post-dbr_schedule parts.
8691         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
8692         (pass_mips_machine_reorg2): New machine specific pass.
8693         (insert_pass_mips_machine_reorg2): New pass plugin definition.
8694         (mips_option_override): Register the new pass.
8695         * rtl.h (cleanup_barriers): Remove prototype.
8696         (dbr_schedule): Likewise.
8697         * jump.c (cleanup_barriers): Make static.
8698         * reorg.c (dbr_schedule): Likewise.
8700 2013-05-06  Richard Biener  <rguenther@suse.de>
8702         PR tree-optimization/57185
8703         * tree-parloops.c (add_field_for_reduction): Handle anonymous
8704         SSA names properly.
8706 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8708         PR target/57106
8709         * config/i386/i386.c (add_parameter_dependencies): Add dependence
8710         between "first_arg" and "insn", not "last" and "insn".
8712 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8714         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
8715         (find_candidates_in_block): Re-enable slsr_process_phi.
8716         (create_phi_basis): Fix double counting of candidate adjustment.
8718 2013-05-06  Richard Biener  <rguenther@suse.de>
8720         PR middle-end/57147
8721         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
8722         the edge is also fallthru, preserve it and just clear the
8723         abnormal flag.
8724         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
8725         also complex, preserve that and just clear the fallthru flag.
8726         * tree-inline.c (update_ssa_across_abnormal_edges): Also
8727         update virtual operands.
8729 2013-05-06  Alan Modra  <amodra@gmail.com>
8731         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
8732         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
8733         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
8734         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
8735         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8736         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
8738 2013-05-06  Alan Modra  <amodra@gmail.com>
8740         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
8741         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
8742         (DEFAULT_ASM_ENDIAN): Define.
8743         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8744         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
8745         Update -K PIC clause from sysv4.h.
8746         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
8747         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
8749 2013-05-06  Alan Modra  <amodra@gmail.com>
8751         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
8752         twice for little-endian.
8753         (ashrdi3_no_power, ashrdi3): Support little-endian.
8755 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8757         PR target/55303
8758         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
8759         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
8760         related expanders.
8761         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
8762         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
8763         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
8764         New predicates.
8766 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
8767             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
8769         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
8770         * config/pa/pa.opt: Make mbig-switch a no-op.
8771         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
8772         (CASE_VECTOR_MODE): Always return SImode.
8773         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
8774         for the !TARGET_BIG_SWITCH case.
8775         * config/pa/pa-linux.h: Likewise.
8776         * config/pa/pa-openbsd.h: Likewise.
8777         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
8778         * config/pa/pa.md (short_jump): Remove define_insn.
8779         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
8780         (casesi0): Remove define_insn.
8781         (type): Remove btable_branch.
8782         (pa_combine_type): Likewise.
8783         (in_nullified_branch_delay): Likewise.
8784         (in_call_delay): Likewise.
8785         (define_delay): Likewise.
8786         (define_insn_reservation "Z3"): Likewise.
8787         (define_insn_reservation "Z4"): Likewise.
8788         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
8789         (pa_adjust_insn_length): Remove adjustment for btable branches.
8790         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
8791         and mno-big-switch
8793 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
8795         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8796         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
8797         Add m->r,x alternatives.
8798         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
8799         splitters using SWI48x mode iterator.
8800         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
8801         TARGET_64BIT.  Add m->x alternative.
8802         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
8803         Add o->x alternative.  Enable for TARGET_SSE.
8804         (sse_storeq): Remove expander.
8805         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
8806         with memory input operand.
8807         (*vec_extractv2di_1 splitter): New.
8808         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
8809         * config/i386/i386.md (ssevecmodelower): New mode attribute.
8811 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8813         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
8814         (INT_LOWPART): Delete.
8815         (extract_MB): Adjust.
8816         (extract_ME): Adjust.
8817         (print_operand): Adjust.
8819 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8821         * config/rs6000/predicates.md (reg_or_add_cint_operand,
8822         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8823         (reg_or_logical_cint_operand, easy_fp_constant,
8824         logical_const_operand): Delete "CONST_DOUBLE" case.
8825         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
8826         "HOST_BITS_PER_WIDE_INT == 64" test.
8827         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
8828         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
8829         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
8830         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
8831         test.
8832         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
8833         CONST_DOUBLE DImode/VOIDmode case.
8834         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
8835         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8836         CONST_DOUBLE VOIDmode case.
8837         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8838         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
8839         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
8840         Delete CONST_DOUBLE case.
8841         (splitters for mov FMOVE64 const_double): Delete
8842         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8843         "HOST_BITS_PER_WIDE_INT >= 64" test.
8844         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
8845         case.
8846         (mov DI const_double): Delete.
8848 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
8850         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
8851         on op shows all bits zero in mode of a lowpart subreg, return zero.
8853 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
8855         PR target/57150
8856         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
8857         to save TFmode registers and DImode to save TImode registers for
8858         caller save operations.
8859         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
8860         mark being partially clobbered since they only use the first
8861         double word.
8863         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
8864         and TDmode only use the upper 64-bits of each VSX register.
8866 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8868         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
8869         (find_candidates_in_block): Disable slsr_process_phi.
8871 2013-05-03  Guozhi Wei  <carrot@google.com>
8873         * coverage.c (coverage_obj_init): Move the construction of gcov
8874         constructor to ...
8875         (build_init_ctor): ... here.
8877 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8879         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
8880         (slsr_cand_d): Redefine def_phi.
8881         (stride_status, phi_adjust_status, count_phis_status): New enums.
8882         (find_phi_def): New.
8883         (find_basis_for_base_expr): New.
8884         (find_basis_for_candidate): Handle hidden bases.
8885         (alloc_cand_and_find_basis): Handle phi candidates.
8886         (slsr_process_phi): New.
8887         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
8888         (create_mul_imm_cand): Likewise.
8889         (create_add_ssa_cand): Exclude phi base candidates.
8890         (create_add_imm_cand): Likewise.
8891         (slsr_process_cast): Likewise.
8892         (slsr_process_copy): Likewise.
8893         (find_candidates_in_block): Handle phi candidates.
8894         (dump_candidate): Likewise.
8895         (unconditional_cands): Delete.
8896         (unconditional_cands_with_known_stride_p): Delete.
8897         (phi_dependent_cand_p): New.
8898         (cand_increment): Handle phi-dependent candidates.
8899         (replace_dependent): Delete.
8900         (replace_mult_candidate): New.
8901         (replace_unconditional_candidate): New.
8902         (incr_vec_index): Move to avoid forward reference.
8903         (create_add_on_incoming_edge): New.
8904         (create_phi_basis): New.
8905         (replace_dependents): Delete.
8906         (replace_conditional_candidate): New.
8907         (phi_add_costs): New.
8908         (replace_uncond_cands_and_profitable_phis): New.
8909         (record_increment): Handle phi adjustments.
8910         (record_phi_increments): New.
8911         (record_increments): Handle phi adjustments.
8912         (phi_incr_cost): New.
8913         (lowest_cost_path): Handle phis.
8914         (total_savings): Likewise.
8915         (analyze_increments): Likewise.
8916         (ncd_with_phi): New.
8917         (ncd_of_cand_and_phis): New.
8918         (nearest_common_dominator_for_cands): Handle phi increments.
8919         (all_phi_incrs_profitable): New.
8920         (replace_profitable_candidates): Handle phi-dependent candidates.
8921         (analyze_candidates_and_replace): Likewise.
8923 2013-05-03  Teresa Johnson  <tejohnson@google.com>
8925         PR bootstrap/57154
8926         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
8927         do not exceed REG_BR_PROB_BASE.
8929 2013-05-03  Jeff Law  <law@redhat.com>
8931         PR tree-optimization/57144
8932         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
8933         operand of the condition will bit into the new type when eliminating
8934         a cast feeding a condition.
8936 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
8938         PR rtl-optimization/57130
8939         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
8940         of COMPARE as in_code to the recursive call if needed.
8942 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
8944         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
8945         (enabled): Handle new members.
8946         * config/i386/sse.md (*vec_concatv2si): Merge from
8947         *vec_concatv2si_sse2 and vec_concatv2si_sse.
8948         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
8950 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
8952         PR tree-optimization/57027
8953         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
8954         for fnms opportunity, check we got the prerequisite kind
8955         of tree / gimple before using accessor functions.
8957 2013-05-03  Richard Biener  <rguenther@suse.de>
8959         * double-int.h (lshift): New overload without precision
8960         and arith argument.
8961         (operator *=, operator +=, operator -=): Move ...
8962         * double-int.c (operator *=, operator +=, operator -=): ... here
8963         and implement more efficiently.
8964         (mul_double_with_sign): Remove.
8965         (lshift_double): Adjust to take unsinged shift argument, push
8966         dispatching code to callers.
8967         (mul_double_wide_with_sign): Add early out for callers that
8968         are not interested in high parts or overflow.
8969         (lshift): New function.
8970         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
8971         dispatch code here.
8972         (lrotate, rrotate): Use logical shifts.
8973         * expr.c (get_inner_reference): Use lshift.
8974         * fixed-value.c (do_fixed_divide): Likewise.
8975         * tree-dfa.c (get_ref_base_and_extent): Likewise.
8976         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
8977         (indirect_refs_may_alias_p): Likewise.
8978         (stmt_kills_ref_p_1): Likewise.
8980 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
8982         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
8984 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
8986         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
8987         scalar form of FABD instruction.
8989 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
8991         * lra-constraints.c (process_alt_operands): Add checking alt
8992         number to choose the best alternative.
8994 2013-05-02  Richard Biener  <rguenther@suse.de>
8996         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
8997         bitmap and its handling.
8998         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9000 2013-05-02  Richard Biener  <rguenther@suse.de>
9002         PR middle-end/57140
9003         * tree-inline.c (copy_loops): Properly handle removed loops.
9004         (copy_cfg_body): Mark destination loops for fixup if source
9005         loops needed fixup.
9007 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9009         PR target/56732
9010         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9011         generating simple_return for naked functions.
9013 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9015         PR middle-end/56988
9016         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9017         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9018         flags match.
9019         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9020         ipa_agg_replacement_value structures.
9021         (known_aggs_to_agg_replacement_list): Likewise.
9022         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9023         (read_agg_replacement_chain): Likewise.
9024         (ipcp_transform_function): Also check that by_ref flags match.
9026 2013-05-02  Richard Biener  <rguenther@suse.de>
9028         * graphds.h (struct graph): Add obstack member.
9029         * graphds.c (new_graph): Initialize obstack and allocate
9030         vertices from it.
9031         (add_edge): Allocate edge from the obstack.
9032         (free_graph): Free the obstack instead of all edges and vertices.
9034 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9036         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9037         divides.
9038         * cfg.c (update_bb_profile_for_threading): Ditto.
9039         * tree-inline.c (copy_bb): Ditto.
9040         (copy_edges_for_bb): Ditto.
9041         (initialize_cfun): Ditto.
9042         (copy_cfg_body): Ditto.
9043         (expand_call_inline): Ditto.
9044         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9045         (estimate_node_size_and_time): Ditto.
9046         (inline_merge_summary): Ditto.
9047         * cgraphclones.c (cgraph_clone_edge): Ditto.
9048         (cgraph_clone_node): Ditto.
9049         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9050         (compute_trg_info): Ditto.
9052 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9054         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9055         S reg when fp attribute set.
9056         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9058 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9060         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9061         New pattern.
9062         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9063         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9064         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9066 2013-05-02  Richard Biener  <rguenther@suse.de>
9068         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9069         (struct instantiate_cache_entry): New type.
9070         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9071         (struct instantiate_cache_type): New type.
9072         (set_instantiated_value, get_instantiated_value): Remove.
9073         (get_instantiated_value_entry): New function.
9074         (instantiate_scev_name): Use the new cache and adjust.
9075         (instantiate_scev_poly): Adjust.
9076         (instantiate_scev_binary): Likewise.
9077         (instantiate_array_ref): Likewise.
9078         (instantiate_scev_convert): Likewise.
9079         (instantiate_scev_not): Likewise.
9080         (instantiate_scev_3): Likewise.
9081         (instantiate_scev_2): Likewise.
9082         (instantiate_scev_r): Likewise.
9083         (instantiate_scev): Likewise.
9084         (resolve_mixers): Likewise.
9086 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9088         PR target/57091
9089         * lra-constraints.c (best_small_class_operands_num): Remove.
9090         (process_alt_operands): Remove small_class_operands_num.  Take
9091         small classes operands into losers and only if the operand is not
9092         matched.  Modify debugging output.
9093         (curr_insn_transform): Remove best_small_class_operands_num.
9094         Print insn name.
9096 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9098         * config/aarch64/aarch64-builtins.c
9099         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9100         * config/aarch64/aarch64-simd-builtins.def
9101         (reduc_splus_): Add new modes.
9102         (reduc_uplus_): New.
9103         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9104         (reduc_uplus_v4sf): Likewise.
9105         (reduc_splus_v4sf): Likewise.
9106         (aarch64_addv<mode>): Likewise.
9107         (reduc_uplus_<mode>): Likewise.
9108         (reduc_splus_<mode>): Likewise.
9109         (aarch64_addvv2di): Likewise.
9110         (reduc_uplus_v2di): Likewise.
9111         (reduc_splus_v2di): Likewise.
9112         (aarch64_addvv2si): Likewise.
9113         (reduc_uplus_v2si): Likewise.
9114         (reduc_splus_v2si): Likewise.
9115         (reduc_<sur>plus_<mode>): New.
9116         (reduc_<sur>plus_v2di): Likewise.
9117         (reduc_<sur>plus_v2si): Likewise.
9118         (reduc_<sur>plus_v4sf): Likewise.
9119         (aarch64_addpv4sf): Likewise.
9120         * config/aarch64/arm_neon.h
9121         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9122         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9123         add UNSPEC_SADDV, UNSPEC_UADDV.
9124         (SUADDV): New.
9125         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9127 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9129         * config/aarch64/arm_neon.h
9130         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9132 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9134         * config/aarch64/aarch64-builtins
9135         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9137 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9139         * config/aarch64/aarch64-simd-builtins.def
9140         (reduc_smax_): New.
9141         (reduc_smin_): Likewise.
9142         (reduc_umax_): Likewise.
9143         (reduc_umin_): Likewise.
9144         (reduc_smax_nan_): Likewise.
9145         (reduc_smin_nan_): Likewise.
9146         (fmax): Remove.
9147         (fmin): Likewise.
9148         (smax): Update for V2SF, V4SF and V2DF modes.
9149         (smin): Likewise.
9150         (smax_nan): New.
9151         (smin_nan): Likewise.
9152         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9153         (<su><maxmin><mode>3): ...This, refactor.
9154         (s<maxmin><mode>3): New.
9155         (<maxmin_uns><mode>3): Likewise.
9156         (reduc_<maxmin_uns>_<mode>): Refactor.
9157         (reduc_<maxmin_uns>_v4sf): Likewise.
9158         (reduc_<maxmin_uns>_v2si): Likewise.
9159         (aarch64_<fmaxmin><mode>: Remove.
9160         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9161         new builtin names.
9162         (vmin<q>_f<32,64>): Likewise.
9163         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9164         (FMAXMIN): New.
9165         (su): Add mappings for smax, smin, umax, umin.
9166         (maxmin): New.
9167         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9168         (FMAXMIN): Rename as...
9169         (FMAXMIN_UNS): ...This.
9170         (maxminv): Remove.
9171         (fmaxminv): Likewise.
9172         (fmaxmin): Likewise.
9173         (maxmin_uns): New.
9174         (maxmin_uns_op): Likewise.
9176 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9178         * config/aarch64/arm_neon.h
9179         (vac<ge, gt><sd>_f<32, 64>): Rename to...
9180         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9181         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9183 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9185         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9186         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9188 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9190         * config/aarch64/aarch64-simd.md
9191         (vcond<mode>_internal): Handle special cases for constant masks.
9192         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9193         (vcondu<mode><mode>): Likewise.
9194         (vcond<v_cmp_result><mode>): New.
9196 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9198         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9199         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9200         * config/aarch64/aarch64-simd-builtins.def
9201         (cmeq): Update to BUILTIN_VALLDI.
9202         (cmgt): Likewise.
9203         (cmge): Likewise.
9204         (cmle): Likewise.
9205         (cmlt): Likewise.
9206         * config/aarch64/arm_neon.h
9207         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9208         to builtins or C as appropriate.
9210 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9212         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9213         (cmgeu): ...This.
9214         (cmhi): Rename to...
9215         (cmgtu): ...This.
9216         * config/aarch64/aarch64-simd.md
9217         (simd_mode): Add SF.
9218         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9219         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9220         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9221         (cstore<mode>_neg): ...This.
9222         * config/aarch64/iterators.md
9223         (VALLF): new.
9224         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9225         (COMPARISONS): New.
9226         (UCOMPARISONS): Likewise.
9227         (optab): Add missing comparisons.
9228         (n_optab): New.
9229         (cmp_1): Likewise.
9230         (cmp_2): Likewise.
9231         (CMP): Likewise.
9232         (cmp): Remove.
9233         (VCMP_S): Likewise.
9234         (VCMP_U): Likewise.
9235         (V_cmp_result): Add DF, SF modes.
9236         (v_cmp_result): Likewise.
9237         (v): Likewise.
9238         (vmtype): Likewise.
9239         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9241 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
9243         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9244         define_insn to define_insn_and_split.
9245         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9246         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9247         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9248         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9249         (thumb2_negscc): Likewise.
9251 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9253         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9255 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9257         * config/arm/thumb2.md: Remove trailing whitespaces.
9259 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9261         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9262         Use gen_int_mode rather than GEN_INT.
9264 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
9266         * value-prof.c (stream_in_histogram_value): Remove the strayed
9267         debug_gimple_stmt.
9269 2013-04-30  Richard Biener  <rguenther@suse.de>
9271         PR middle-end/57122
9272         * cfghooks.c (split_edge): Properly check for the loop latch edge.
9274 2013-04-30  Richard Biener  <rguenther@suse.de>
9276         PR middle-end/57107
9277         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9279 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
9281         PR rtl-optimization/56957
9282         PR rtl-optimization/57105
9283         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9284         variable.  Use just INSN_UID for determining whether an insn
9285         should be only disconnected from the insn stream.
9286         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9288 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
9290         PR tree-optimization/57104
9291         * tsan.c (instrument_expr): Don't instrument accesses to
9292         DECL_HARD_REGISTER VAR_DECLs.
9294 2013-04-30  Richard Biener  <rguenther@suse.de>
9296         * function.h (loops_for_fn): New inline function.
9297         (set_loops_for_fn): Likewise.
9298         * cfgloop.h (place_new_loop): Add struct function parameter.
9299         (get_loop): Likewise.
9300         (get_loops): Likewise.
9301         (number_of_loops): Likewise.
9302         (fel_next): Adjust.
9303         (fel_init): Likewise.
9304         * cfg.c (get_loop_copy): Adjust.
9305         * cfgloop.c (flow_loops_dump): Likewise.
9306         (record_loop_exits): Likewise.
9307         (verify_loop_structure): Likewise.
9308         * cfgloopanal.c (mark_irreducible_loops): Likewise.
9309         (estimate_reg_pressure_cost): Likewise.
9310         (mark_loop_exit_edges): Likewise.
9311         * cfgloopmanip.c (place_new_loop): Likewise.
9312         (add_loop): Likewise.
9313         (duplicate_loop): Likewise.
9314         * graph.c (draw_cfg_nodes): Likewise.
9315         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9316         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9317         (extract_affine_chrec): Likewise.
9318         (build_scop_iteration_domain): Likewise.
9319         * graphite.c (graphite_initialize): Likewise.
9320         * ira-build.c (create_loop_tree_nodes): Likewise.
9321         (more_one_region_p): Likewise.
9322         (rebuild_regno_allocno_maps): Likewise.
9323         (mark_loops_for_removal): Likewise.
9324         (mark_all_loops_for_removal): Likewise.
9325         (remove_unnecessary_regions): Likewise.
9326         (ira_build): Likewise.
9327         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9328         * loop-init.c (fix_loop_structure): Likewise.
9329         (gate_rtl_move_loop_invariants): Likewise.
9330         (gate_rtl_unswitch): Likewise.
9331         (gate_rtl_unroll_and_peel_loops): Likewise.
9332         (rtl_doloop): Likewise.
9333         * lto-streamer-in.c (input_cfg): Likewise.
9334         * lto-streamer-out.c (output_cfg): Likewise.
9335         * modulo-sched.c (sms_schedule): Likewise.
9336         * predict.c (tree_estimate_probability): Likewise.
9337         (tree_estimate_probability_driver): Likewise.
9338         (estimate_loops): Likewise.
9339         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9340         (move_sese_region_to_fn): Likewise.
9341         (debug_loop_num): Likewise.
9342         * tree-chrec.c (chrec_evaluate): Likewise.
9343         (hide_evolution_in_other_loops_than_loop): Likewise.
9344         (chrec_component_in_loop_num): Likewise.
9345         (reset_evolution_in_loop): Likewise.
9346         (evolution_function_is_invariant_rec_p): Likewise.
9347         * tree-if-conv.c (main_tree_if_conversion): Likewise.
9348         * tree-inline.c (copy_loops): Likewise.
9349         (copy_cfg_body): Likewise.
9350         (tree_function_versioning): Likewise.
9351         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9352         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9353         Likewise.
9354         (add_to_evolution_1): Likewise.
9355         (scev_const_prop): Likewise.
9356         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9357         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9358         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9359         (tree_ssa_lim_initialize): Likewise.
9360         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9361         (verify_loop_closed_ssa): Likewise.
9362         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
9363         (tree_ssa_loop_im): Likewise.
9364         (tree_ssa_loop_unswitch): Likewise.
9365         (tree_vectorize): Likewise.
9366         (check_data_deps): Likewise.
9367         (tree_ssa_loop_ivcanon): Likewise.
9368         (tree_ssa_loop_bounds): Likewise.
9369         (tree_complete_unroll): Likewise.
9370         (tree_complete_unroll_inner): Likewise.
9371         (tree_parallelize_loops): Likewise.
9372         (tree_ssa_loop_prefetch): Likewise.
9373         (tree_ssa_loop_ivopts): Likewise.
9374         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9375         * tree-vectorizer.c (vectorize_loops): Likewise.
9377 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
9379         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9380         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9381         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9382         with EABI_LINK_SPEC.
9384 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9386         PR target/44578
9387         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9388         alternative.
9390 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
9392         PR target/57097
9393         * lra-constraints.c (process_alt_operands): Discourage a bit more
9394         using memory for pseudos.  Print cost dump for alternatives.
9395         Modify cost values for conflicts with early clobbers.
9396         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9398 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9400         PR target/57098
9401         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9403 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9405         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9406         from/to S register.
9407         (movdi_aarch64): Support LDR/STR from/to D register.
9409 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9411         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9412         or higher by default.
9414 2013-04-29  Richard Biener  <rguenther@suse.de>
9416         PR middle-end/57075
9417         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9418         even if not adding abnormal edges for calls that can make
9419         abnormal gotos.
9421 2013-04-29  Richard Biener  <rguenther@suse.de>
9423         PR middle-end/57103
9424         * tree-cfg.c (move_stmt_op): Fix condition under which to update
9425         TREE_BLOCK.
9426         (move_stmt_r): Remove redundant checking.
9428 2013-04-29  Teresa Johnson  <tejohnson@google.com>
9430         PR bootstrap/57077
9431         * basic-block.h (apply_scale): New function.
9432         (apply_probability): Use apply_scale.
9433         * gimple-streamer-in.c (input_bb): Ditto.
9434         * lto-streamer-in.c (input_cfg): Ditto.
9435         * lto-cgraph.c (merge_profile_summaries): Ditto.
9436         * tree-optimize.c (execute_fixup_cfg): Ditto.
9437         * tree-inline.c (copy_bb): Update comment to use apply_scale.
9438         (copy_edges_for_bb): Ditto.
9439         (copy_cfg_body): Ditto.
9441 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9443         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9444         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9445         (tail_merge_optimize): Handle current_loops == NULL.
9447 2013-04-26  Jeff Law  <law@redhat.com>
9449         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9450         (simplify_cond_using_ranges): Generalize code to simplify
9451         COND_EXPRs where one argument is a constant and the other
9452         is an SSA_NAME created by an integral type conversion.
9454 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9456         * config/arm/arm.md (store_minmaxsi): Use only when
9457         optimize_insn_for_size_p.
9459 2013-04-29  Christian Bruel  <christian.bruel@st.com>
9461         PR target/57108
9462         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9464 2013-04-29  Richard Biener  <rguenther@suse.de>
9466         PR middle-end/57089
9467         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9468         loop tree make sure to schedule a fixup for the child as well.
9469         (expand_omp_for_generic): Properly add loops.
9470         (expand_omp_for_static_nochunk): Likewise.
9471         (expand_omp_for_static_chunk): Likewise.
9472         (expand_omp_for): For the degenerate case fixup loops.
9473         (expand_omp_sections): Fix default bb placement in loops.
9474         (expand_omp_atomic_pipeline): Properly add loops.
9476 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9478         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9480 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9482         * tree-ssa-tail-merge.c: Update header comment.
9484 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9486         * config/aarch64/arm_neon.h
9487         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9488         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9489         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9490         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9491         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9492         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9493         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9494         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9496 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9498         * config/aarch64/aarch64-simd.md
9499         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9500         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9501         fix_trunc, fixuns_trunc.
9502         (ftrunc<VDQF:mode>2): New.
9503         * config/aarch64/iterators.md (optab): Add fix, fixuns.
9504         (fix_trunc_optab): New.
9506 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9508         * config/aarch64/aarch64-builtins.c
9509         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9510         iceilf, lround, iroundf.
9512 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9514         PR target/54349
9515         * config/i386/i386.h (enum ix86_tune_indices)
9516         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9517         New, split from X86_TUNE_INTER_UNIT_MOVES.
9518         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9519         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9520         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9521         (TARGET_INTER_UNIT_MOVES): Remove.
9522         * config/i386/i386.c (initial_ix86_tune_features): Update.
9523         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9524         (ix86_expand_convert_uns_didf_sse): Use
9525         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9526         (ix86_expand_vector_init_one_nonzero): Ditto.
9527         (ix86_expand_vector_init_interleave): Ditto.
9528         (inline_secondary_memory_needed): Return true for moves from SSE class
9529         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9530         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9531         * config/i386/constraints.md (Yi, Ym): Depend on
9532         TARGET_INTER_UNIT_MOVES_TO_VEC.
9533         (Yj, Yn): New constraints.
9534         * config/i386/i386.md (*movdi_internal): Change constraints of
9535         operand 1 from Yi to Yj and from Ym to Yn.
9536         (*movsi_internal): Ditto.
9537         (*movdf_internal): Ditto.
9538         (*movsf_internal): Ditto.
9539         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9540         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9541         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9542         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9543         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9544         * config/i386/sse.md (movdi_to_sse): Ditto.
9545         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9546         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9547         TARGET_INTER_UNIT_MOVES.
9548         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9549         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9550         instead of TARGET_INTER_UNIT_MOVES.
9551         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9552         operand 1 from Yi to Yj and from Ym to Yn.
9554 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9556         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9557         (float_truncate_hi_): Likewise.
9558         (float_extend_lo_): Likewise.
9559         (float_truncate_lo_): Likewise.
9560         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9561         (aarch64_float_extend_lo_v2df): Likewise.
9562         (vec_unpacks_hi_v4sf): Likewise.
9563         (aarch64_float_truncate_lo_v2sf): Likewise.
9564         (aarch64_float_truncate_hi_v4sf): Likewise.
9565         (vec_pack_trunc_v2df): Likewise.
9566         (vec_pack_trunc_df): Likewise.
9568 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9570         * config/aarch64/aarch64-builtins.c
9571         (aarch64_fold_builtin): Fold float conversions.
9572         * config/aarch64/aarch64-simd-builtins.def
9573         (floatv2si, floatv4si, floatv2di): New.
9574         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9575         * config/aarch64/aarch64-simd.md
9576         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9577         * config/aarch64/iterators.md (FLOATUORS): New.
9578         (optab): Add float, floatuns.
9579         (su_optab): Likewise.
9581 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9583         * config/aarch64/aarch64-builtins.c
9584         (aarch64_builtin_vectorized_function): Use new names for
9585         fcvt builtins.
9586         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9587         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9588         (fcvtzu): Split as...
9589         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9590         (fcvtas): Split as...
9591         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9592         (fcvtau): Split as...
9593         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9594         (fcvtps): Split as...
9595         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9596         (fcvtpu): Split as...
9597         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9598         (fcvtms): Split as...
9599         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9600         (fcvtmu): Split as...
9601         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9602         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9603         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9604         (lfrintnusf, lfrintnudf): Likewise.
9605         * config/aarch64/aarch64-simd.md
9606         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9607         define_insn.
9608         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9609         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9610         (fcvt_pattern): Likewise.
9612 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9614         * config/aarch64/aarch64-simd.md
9615         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
9616         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
9618 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9620         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
9621         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
9622         (vrnd<a,m,n,p>_f32): Implement using builtins.
9623         (vrnd<i,x><q>_f<32, 64>): New.
9625 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9627         * config/aarch64/aarch64-builtins.c
9628         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
9629         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
9630         (frintz): Rename to...
9631         (btrunc): ...this.
9632         (frintp): Rename to...
9633         (ceil): ...this.
9634         (frintm): Rename to...
9635         (floor): ...this.
9636         (frinti): Rename to...
9637         (nearbyint): ...this.
9638         (frintx): Rename to...
9639         (rint): ...this.
9640         (frinta): Rename to...
9641         (round): ...this.
9642         * config/aarch64/aarch64-simd.md
9643         (aarch64_frint<frint_suffix><mode>): Delete.
9644         (<frint_pattern><mode>2): Convert to insn.
9645         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
9646         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
9647         (frint_pattern): Likewise.
9648         (frint_suffix): Likewise.
9650 2013-04-29  Richard Biener  <rguenther@suse.de>
9652         PR tree-optimization/57081
9653         * loop-init.c: Include tree-flow.h.
9654         (loop_optimizer_finalize): Free number of iteration estimates.
9655         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
9657 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
9659         PR tree-optimization/57083
9660         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
9661         non-singleton shift count range, zero extend low_bound for uns case.
9663         * config/i386/predicates.md (general_vector_operand): New predicate.
9664         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
9665         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
9666         if they aren't nonimmediate operands.  If their original values
9667         satisfy const_vector_equal_evenodd_p, don't shift them.
9668         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
9669         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
9670         if not nonimmediate_operand.
9671         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
9672         instead of register_operand.
9673         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
9675 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
9677         * stor-layout.c (finalize_size_functions): Allocate a structure and
9678         reset cfun before dumping the functions.
9680 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
9682         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
9684         PR target/56866
9685         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
9686         use xop_pmacsdqh if uns_p.
9687         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
9688         the immediate rotate count.
9690 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
9692         * rtl.h (struct rtx_def): Add comment for field jump.
9693         (LRA_SUBREG_P): New macro.
9694         * recog.c (register_operand): Check LRA_SUBREG_P.
9695         * lra.c (lra): Add note at the end of RTL code. Align non-empty
9696         stack frame.
9697         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9698         (lra_final_code_change): Skip subreg change for operators.
9699         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9700         if there are no operand changes.
9701         * lra-constraints.c (curr_insn_set): New.
9702         (match_reload): Set LRA_SUBREG_P.
9703         (emit_spill_move): Ditto.
9704         (check_and_process_move): Use curr_insn_set. Process only single
9705         set insns.  Don't initialize sec_mem_p and change_p.
9706         (simplify_operand_subreg): Use LRA_SUBREG_P.
9707         (reg_in_class_p): New function.
9708         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
9709         of #ifdef.  Add code to remove cycling.
9710         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9711         non-null disp.  Reload inner instead of disp when base and index
9712         are null.  Try to put lo_sum into register.
9713         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
9714         (check_and_process_move): Move code for move cost check to
9715         simple_move_p.  Remove equiv_substitution.
9716         (simple_move_p): New function.
9717         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
9718         curr_insn_set.  Call check_and_process_move only for single set
9719         insns.  Use the new function.  Move call of check_and_process_move
9720         after operand equiv substitution and address process.
9722 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9724         PR go/57045
9725         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
9726         with nonlocal goto receivers or returns twice calls, ignore
9727         unininitialized values from abnormal edges to nl goto receiver
9728         or returns twice call.
9730 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9732         PR tree-optimization/57051
9733         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
9734         and VEC_RSHIFT_EXPR if shift count is a multiple of element
9735         bitsize.
9737 2013-04-26  Richard Biener  <rguenther@suse.de>
9739         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
9740         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
9741         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
9742         (fixup_loop_arrays_after_move): New function.
9743         (move_sese_region_to_fn): Properly outline the loop tree parts
9744         of the SESE region.
9746 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
9748         * config/i386/i386.md (type, unit): Fix long lines.
9750 2013-04-26  Richard Biener  <rguenther@suse.de>
9752         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
9753         (lto-streamer-out.o): Likewise.
9754         * cfgloop.c (init_loops_structure): Export, add struct function
9755         argument and adjust.
9756         (flow_loops_find): Adjust.
9757         * cfgloop.h (enum loop_estimation): Add EST_LAST.
9758         (init_loops_structure): Declare.
9759         * lto-streamer-in.c: Include cfgloop.h.
9760         (input_cfg): Input the loop tree.
9761         * lto-streamer-out.c: Include cfgloop.h.
9762         (output_cfg): Output the loop tree.
9763         (output_struct_function_base): Do not drop PROP_loops.
9765 2013-03-26  Richard Biener  <rguenther@suse.de>
9767         * tree-cfg.c (execute_build_cfg): Build the loop tree.
9768         (pass_build_cfg): Provide PROP_loops.
9769         (move_sese_region_to_fn): Remove loops that are outlined into fn
9770         for now.
9771         * tree-inline.c: Include cfgloop.h.
9772         (initialize_cfun): Do not drop PROP_loops.
9773         (copy_loops): New function.
9774         (copy_cfg_body): Copy loop structure.
9775         (tree_function_versioning): Initialize destination loop tree.
9776         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
9777         (pass_parallelize_loops): Do IL verification.
9778         * loop-init.c (loop_optimizer_init): Fixup loops if required.
9779         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
9780         the CFG make sure we fixup loops as well.
9781         * tree-ssa-tail-merge.c: Include cfgloop.h.
9782         (replace_block_by): When merging loop latches mark loops for fixup.
9783         * lto-streamer-out.c (output_struct_function_base): Drop
9784         PROP_loops for now.
9785         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
9786         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
9787         * ipa-split.c: Include cfgloop.h.
9788         (split_function): Add the new return block to the loop tree root.
9789         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
9790         whether we have removed the forwarder block.
9791         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
9792         * cfgloop.h (place_new_loop): Declare.
9793         * cfgloopmanip.c (place_new_loop): Export.
9794         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
9795         (tree-switch-conversion.o): Likewise.
9796         (tree-complex.o): Likewise.
9797         (tree-inline.o): Likewise.
9798         (tree-ssa-tailmerge.o): Likewise.
9799         (ipa-split.o): Likewise.
9800         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
9801         (tree-ssa-copy.o): Likewise.
9802         * tree-switch-conversion.c: Include cfgloop.h
9803         (process_switch): If we emit a bit-test cascade, schedule loops
9804         for fixup.
9805         * tree-complex.c: Include cfgloop.h.
9806         (expand_complex_div_wide): Properly add new basic-blocks to loops.
9807         * asan.c: Include cfgloop.h.
9808         (create_cond_insert_point): Properly add new basic-blocks to
9809         loops, schedule loop fixup.
9810         * cfgloop.c (verify_loop_structure): Check that looks are not
9811         marked for fixup.
9812         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
9813         to loops.
9814         (expand_omp_for_generic): Likewise.
9815         (expand_omp_sections): Likewise.
9816         (expand_omp_atomic_pipeline): Schedule loops for fixup.
9817         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
9818         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
9819         is initialized, not when loops are present.
9820         * tree-parloops.c (parallelize_loops): Remove checking here.
9821         * passes.c (init_optimization_passes): Schedule a copy-propagation
9822         pass before complete unrolling of inner loops.
9824 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9826         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
9827         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
9828         (colorize_init): Add argument to _WIN32 version.
9829         * toplev.c: Include diagnostic-color.h.
9830         (process_options): Default to -fdiagnostics-color=auto if
9831         GCC_COLORS env var is in the environment.
9832         * common.opt (fdiagnostics-color=): Add Var and Init.
9833         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
9834         env var is in the environment, the default is auto rather than never.
9836         * diagnostic.h (file_name_as_prefix): Add context argument.
9837         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
9838         the string as locus.
9839         * langhooks.c (lhd_print_error_function): Adjust caller.
9841 2013-04-25  Lawrence Crowl  <crowl@google.com>
9843         * var-tracking.c (shared_hash_def::htab):
9844         Change type to hash_table.  Update dependent calls and types.
9846 2013-04-25  Lawrence Crowl  <crowl@google.com>
9848         * Makefile.in: Update as needed below.
9850         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
9851         Move declaration to after the type's method definitons.
9853         * attribs.c (htab_t scoped_attributes::attribute_hash):
9854         Change type to hash_table.  Update dependent calls and types.
9856         * bitmap.c (htab_t bitmap_desc_hash):
9857         Change type to hash_table.  Update dependent calls and types.
9859         * cselib.c (htab_t cselib_hash_table):
9860         Change type to hash_table.  Update dependent calls and types.
9862         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
9863         (hash_string_slot_node): Move implementation into lto-streamer.h
9864         struct string_slot_hasher.
9865         (eq_string_slot_node): Likewise.
9867         * data-streamer-out.c: Update output_block::string_hash_table
9868         dependent calls and types.
9870         * dwarf2cfi.c (htab_t trace_index):
9871         Change type to hash_table.  Update dependent calls and types.
9873         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
9874         Change type to hash_table.  Update dependent calls and types.
9875         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
9876         (htab_t optimize_external_refs::map): Likewise.
9877         (htab_t output_comp_unit::extern_map): Likewise.
9878         (htab_t output_comdat_type_unit::extern_map): Likewise.
9879         (htab_t output_macinfo::macinfo_htab): Likewise.
9880         (htab_t optimize_location_lists::htab): Likewise.
9881         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
9883         * except.c (htab_t ehspec_hash_type):
9884         Change type to hash_table.  Update dependent calls and types.
9885         (assign_filter_values::ttypes): Likewise.
9886         (assign_filter_values::ehspec): Likewise.
9887         (sjlj_assign_call_site_values::ar_hash): Likewise.
9888         (convert_to_eh_region_ranges::ar_hash): Likewise.
9890         * gcse.c (htab_t pre_ldst_table):
9891         Change type to hash_table.  Update dependent calls and types.
9893         * ggc-common.c (htab_t saving_htab):
9894         Change type to hash_table.  Update dependent calls and types.
9895         (htab_t loc_hash): Likewise.
9896         (htab_t ptr_hash): Likewise.
9897         (call_count): Rename ggc_call_count.
9898         (call_alloc): Rename ggc_call_alloc.
9899         (loc_descriptor): Rename make_loc_descriptor.
9900         (add_statistics): Rename ggc_add_statistics.
9902         * ggc-common.c (saving_htab):
9903         Change type to hash_table.  Update dependent calls and types.
9905         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
9906         (push_gimplify_context): Likewise.
9907         (pop_gimplify_context): Likewise.
9908         (struct gimple_temp_hash_elt): Added.
9909         (struct gimplify_hasher): Likewise.
9910         (struct gimplify_ctx.temp_htab):
9911         Change type to hash_table.  Update dependent calls and types.
9913         * gimple-fold.c: Include gimplify-ctx.h.
9915         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
9916         Change type to hash_table.  Update dependent calls and types.
9917         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
9918         avoid potential global name collision.
9920         * gimplify.c: Include gimplify-ctx.h.
9921         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
9922         (htab_t gimplify_ctx::temp_htab):
9923         Update dependent calls and types for new type hash_table.
9924         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
9925         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
9927         * gimplify-ctx.h: New.
9928         (struct gimple_temp_hash_elt): Move from gimplify.c.
9929         (class gimplify_hasher): New.
9930         (struct gimplify_ctx): Move from gimple.h.
9931         (htab_t gimplify_ctx::temp_htab):
9932         Change type to hash_table.  Update dependent calls and types.
9934         * graphite-clast-to-gimple.c: Include graphite-htab.h.
9935         (htab_t ivs_params::newivs_index):
9936         Change type to hash_table.  Update dependent calls and types.
9937         (htab_t ivs_params::params_index): Likewise.
9938         (htab_t print_generated_program::params_index): Likewise.
9939         (htab_t gloog::newivs_index): Likewise.
9940         (htab_t gloog::params_index): Likewise.
9942         * graphite.c: Include graphite-htab.h.
9943         4htab_t graphite_transform_loops::bb_pbb_mapping):
9944         Change type to hash_table.  Update dependent calls and types.
9946         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
9947         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
9948         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
9950         * graphite-dependences.c: Include graphite-htab.h.
9951         (loop_is_parallel_p): Change hash table type of parameter.
9953         * graphite-htab.h: New.
9954         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
9955         (extern find_pbb_via_hash): Move from graphite-poly.h.
9956         (extern loop_is_parallel_p): Move from graphite-poly.h.
9957         (extern get_loop_body_pbbs): Move from graphite-poly.h.
9959         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
9960         (extern loop_is_parallel_p): Move to graphite-htab.h.
9961         (extern get_loop_body_pbbs): Move to graphite-htab.h.
9963         * haifa-sched.c (htab_t delay_htab):
9964         Change type to hash_table.  Update dependent calls and types.
9965         (htab_t delay_htab_i2): Likewise.
9967         * ira-color.c (htab_t allocno_hard_regs_htab):
9968         Change type to hash_table.  Update dependent calls and types.
9970         * ira-costs.c (htab_t cost_classes_htab):
9971         Change type to hash_table.  Update dependent calls and types.
9973         * loop-invariant.c (htab_t merge_identical_invariants::eq):
9974         Change type to hash_table.  Update dependent calls and types.
9976         * loop-iv.c (htab_t bivs):
9977         Change type to hash_table.  Update dependent calls and types.
9979         * loop-unroll.c (htab_t opt_info::insns_to_split):
9980         Change type to hash_table.  Update dependent calls and types.
9981         (htab_t opt_info::insns_with_var_to_expand): Likewise.
9983         * lto-streamer.h (struct string_slot): Move from data-streamer.h
9984         (struct string_slot_hasher): New.
9985         (htab_t output_block::string_hash_table):
9986         Change type to hash_table.  Update dependent calls and types.
9988         * lto-streamer-in.c (freeing_string_slot_hasher): New.
9989         (htab_t file_name_hash_table):
9990         Change type to hash_table.  Update dependent calls and types.
9992         * lto-streamer-out.c: Update output_block::string_hash_table dependent
9993         calls and types.
9995         * lto-streamer.c (htab_t tree_htab):
9996         Change type to hash_table.  Update dependent calls and types.
9998         * omp-low.c: Include gimplify-ctx.h.
10000         * passes.c (htab_t name_to_pass_map):
10001         Change type to hash_table.  Update dependent calls and types.
10002         (pass_traverse): Rename to passes_pass_traverse.
10004         * plugin.c (htab_t event_tab):
10005         Change type to hash_table.  Update dependent calls and types.
10007         * postreload-gcse.c (htab_t expr_table):
10008         Change type to hash_table.  Update dependent calls and types.
10009         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10011         * sese.c (debug_rename_map_1): Make extern.
10012         (htab_t copy_bb_and_scalar_dependences::rename_map):
10013         Change type to hash_table.  Update dependent calls and types.
10015         * sese.h (extern debug_rename_map): Move to .c file.
10017         * store-motion.c (htab_t store_motion_mems_table):
10018         Change type to hash_table.  Update dependent calls and types.
10020         * trans-mem.c (htab_t tm_new_mem_hash):
10021         Change type to hash_table.  Update dependent calls and types.
10023         * tree-browser.c (htab_t TB_up_ht):
10024         Change type to hash_table.  Update dependent calls and types.
10026         * tree-cfg.c (htab_t discriminator_per_locus):
10027         Change type to hash_table.  Update dependent calls and types.
10029         * tree-complex.c: Include tree-hasher.h
10030         (htab_t complex_variable_components):
10031         Change type to hash_table.  Update dependent calls and types.
10033         * tree-eh.c (htab_t finally_tree):
10034         Change type to hash_table.  Update dependent calls and types.
10036         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10037         struct int_tree_hasher.
10038         (extern int_tree_map_eq): Likewise.
10039         (uid_decl_map_hash): Removed.
10040         (extern decl_tree_map_eq): Likewise.
10042         * tree-hasher.h: New.
10043         (struct int_tree_hasher): New.
10044         (typedef int_tree_htab_type): New.
10046         * tree-inline.c: Include gimplify-ctx.h.
10048         * tree-mudflap.c: Include gimplify-ctx.h.
10050         * tree-parloops.c: Include tree-hasher.h.
10051         (htab_t eliminate_local_variables_stmt::decl_address):
10052         Change type to hash_table.  Update dependent calls and types.
10053         (htab_t separate_decls_in_region::decl_copies): Likewise.
10055         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10056         Change type to hash_table.  Update dependent calls and types.
10058         * tree-sra.c (candidates):
10059         Change type to hash_table.  Update dependent calls and types.
10061         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10062         in tree-flow.h.
10063         (int_tree_map_hash): Likewise.
10065         * tree-ssa-dom.c (htab_t avail_exprs):
10066         Change type to hash_table.  Update dependent calls and types.
10068         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10069         Change type to hash_table.  Update dependent calls and types.
10071         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10072         Change type to hash_table.  Update dependent calls and types.
10074         * tree-ssa-phiopt.c (seen_ssa_names):
10075         Change type to hash_table.  Update dependent calls and types.
10077         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10078         Change type to hash_table.  Update dependent calls and types.
10080         * tree-ssa-uncprop.c (equiv):
10081         Change type to hash_table.  Update dependent calls and types.
10083 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10085         PR rtl-optimization/57003
10086         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10087         call note_stores with kill_clobbered_value callback again after
10088         killing regs_invalidated_by_call.
10090 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10092         * config/aarch64/aarch64-simd.md
10093         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10094         (aarch64_simd_bsl<mode>): Likewise.
10095         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10097 2013-04-25  Marek Polacek  <polacek@redhat.com>
10099         PR tree-optimization/57066
10100         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10102 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10104         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10106 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10108         * config/aarch64/aarch64-builtins.c
10109         (aarch64_fold_builtin): New.
10110         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10111         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10112         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10113         * config/aarch64/arm_neon.h
10114         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10116 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10117             Tejas Belagod  <tejas.belagod@arm.com>
10119         * config/aarch64/aarch64-builtins.c
10120         (aarch64_gimple_fold_builtin): New.
10121         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10122         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10123         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10124         (addvv4sf): Update.
10125         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10127 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10129         * config/aarch64/aarch64.md
10130         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10132 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10134         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10135         (*ngcsi_uxtw): New pattern.
10137 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10138             Julian Brown  <julian@codesourcery.com>
10140         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10141         (TB_DREG): Add T_V4HF.
10142         (v4hf_UP): New macro.
10143         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10144         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10145         Handle initialisation of V4HF. Adjust initialisation of reinterpret
10146         built-ins.
10147         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10148         (arm_vector_mode_supported_p): Handle V4HF.
10149         (arm_mangle_map): Handle V4HFmode.
10150         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10151         * config/arm/arm_neon_builtins.def: Add entries for
10152         vcvtv4hfv4sf, vcvtv4sfv4hf.
10153         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10154         (neon_vcvtv4hfv4sf): Likewise.
10155         * config/arm/neon-gen.ml: Handle half-precision floating point
10156         features.
10157         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10158         * config/arm/arm_neon.h: Regenerate.
10159         * config/arm/neon.ml (type elts): Add F16.
10160         (type vectype): Add T_float16x4, T_floatHF.
10161         (type vecmode): Add V4HF.
10162         (type features): Add Requires_FP_bit feature.
10163         (elt_width): Handle F16.
10164         (elt_class): Likewise.
10165         (elt_of_class_width): Likewise.
10166         (mode_of_elt): Refactor.
10167         (type_for_elt): Handle F16, fix error messages.
10168         (vectype_size): Handle T_float16x4.
10169         (vcvt_sh): New function.
10170         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10171         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10172         (string_of_mode): Handle V4HF.
10173         * doc/arm-neon-intrinsics.texi: Regenerate.
10175 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10177         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10178         format specifier in 'X' case.
10180 2013-04-25  Alan Modra  <amodra@gmail.com>
10182         PR target/57052
10183         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10184         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10185         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10186         Repeat for many other rotate/shift and mask patterns using subregs.
10187         Name lshiftrt insns.
10188         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10189         on WORDS_BIG_ENDIAN.
10191 2013-04-25  Alan Modra  <amodra@gmail.com>
10193         * config.gcc: Support little-endian powerpc-linux targets.
10194         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10195         (LINK_OS_LINUX_SPEC): Define.
10196         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10197         Preserve MASK_LITTLE_ENDIAN.
10198         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10199         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10200         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10201         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10202         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10203         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10204         Correct fp word order for little-endian.  Don't shift toc entries
10205         smaller than a word for little-endian.
10206         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10207         (bswapdi2 splits): Correct low-part subreg for little-endian.
10208         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10209         low/high where such is correct only for be.
10210         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10211         little-endian for -mcall-aixdesc.
10213 2013-04-25  Alan Modra  <amodra@gmail.com>
10215         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10216         replace_equiv_address_nv.
10218 2013-04-25  Alan Modra  <amodra@gmail.com>
10220         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10222 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10224         Revert:
10225         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10226         * rtl.h (struct rtx_def): ...
10228 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10230         PR rtl-optimizations/57046
10231         * lra-constraints (split_reg): Set up lra_risky_transformations_p
10232         for multi-reg splits.
10234 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
10236         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10238 2013-04-24  Sterling Augustine  <saugustine@google.com>
10240         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10241         (comp_dir_string, debug_str_dwo_section): New.
10242         (DEBUG_STR_DWO_SECTION): Rename to ...
10243         (DEBUG_DWO_STR_SECTION): ... this.
10244         (DEBUG_NORM_STR_SECTION): Delete.
10245         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10246         (DEBUG_STR_DWO_SECTION_FLAGS): New.
10247         (find_AT_string): Move most logic to ...
10248         (find_AT_string_in_table): ... here.  New.
10249         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10250         add_skeleton_AT_string.  Delete logic.
10251         (output_skeleton_debug_sections): Remove call to
10252         add_top_level_skeleton_die_attrs.
10253         (add_comp_dir_attribute): Move logic to comp_dir_string.
10254         (dwarf2out_init): Initialize debug_str_dwo_section.
10255         (output_indirect_string): Call find_string_form.
10256         (output_indirect_strings): Rewrite.
10257         (prune_unused_types): Empty skeleton_debug_str_hash.
10258         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10259         (dwarf2out_finish):  Call output_indirect_strings.
10261 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10263         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10265 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10267         * rtl.h (struct rtx_def): Add comment for field jump.
10268         (LRA_SUBREG_P): New macro.
10269         * recog.c (register_operand): Check LRA_SUBREG_P.
10270         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10271         stack frame.
10272         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10273         (lra_final_code_change): Skip subreg change for operators.
10274         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10275         if there are no operand changes.
10276         * lra-constraints.c (curr_insn_set): New.
10277         (match_reload): Set LRA_SUBREG_P.
10278         (emit_spill_move): Ditto.
10279         (check_and_process_move): Use curr_insn_set. Process only single
10280         set insns.  Don't initialize sec_mem_p and change_p.
10281         (simplify_operand_subreg): Use LRA_SUBREG_P.
10282         (reg_in_class_p): New function.
10283         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10284         of #ifdef.  Add code to remove cycling.
10285         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10286         non-null disp.  Reload inner instead of disp when base and index
10287         are null.  Try to put lo_sum into register.
10288         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10289         (check_and_process_move): Move code for move cost check to
10290         simple_move_p.  Remove equiv_substitution.
10291         (simple_move_p): New function.
10292         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10293         curr_insn_set.  Call check_and_process_move only for single set
10294         insns.  Use the new function.  Move call of check_and_process_move
10295         after operand equiv substitution and address process.
10297 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
10299         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10300         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10301         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10303 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10305         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10307 2013-04-24  Marek Polacek  <polacek@redhat.com>
10309         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10310         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10311         (select_loops_exit_conditions): Likewise.
10312         (number_of_iterations_for_all_loops): Likewise.
10313         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10314         (scev_analysis): Likewise.
10316 2013-04-02  Catherine Moore  <clm@codesourcery.com>
10317             Chao-ying Fu <fu@mips.com>
10319         * config/mips/micromips.md (jraddiusp): New pattern.
10320         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10321         instruction if possible.
10323 2013-04-24  Alan Modra  <amodra@gmail.com>
10325         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10327 2013-04-24  Julian Brown  <julian@codesourcery.com>
10328             Chung-Lin Tang  <cltang@codesourcery.com>
10330         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10331         dependency behavior in enumeration type DIE generation. Add TODO note
10332         to comments about future DW_FORM_sdata/udata re-work of related code.
10334 2013-04-23  Lawrence Crowl  <crowl@google.com>
10336         * Makefile.in: Update as needed below.
10338         * hash-table.h (class hash_table):
10339         Correct many methods with parameter types compare_type to the correct
10340         value_type.  (Correct code was unlikely to notice the change.)
10341         (hash_table::elements_with_deleted) New.
10342         (class hashtable::iterator): New.
10343         (hashtable::begin()): New.
10344         (hashtable::end()): New.
10345         (FOR_EACH_HASH_TABLE_ELEMENT): New.
10347         * statistics.c (statistics_hashes):
10348         Change type to hash_table.  Update dependent calls and types.
10350         * tree-into-ssa.c (var_infos):
10351         Change type to hash_table.  Update dependent calls and types.
10353         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10354         Change type to hash_table.  Update dependent calls and types.
10356         * tree-ssa-loop-im.c (struct mem_ref.refs):
10357         Change type to hash_table.  Update dependent calls and types.
10359         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10360         Change type to hash_table.  Update dependent calls and types.
10362         * tree-ssa-sccvn.c (vn_tables_s::nary):
10363         Change type to hash_table.  Update dependent calls and types.
10364         (vn_tables_s::phis): Likewise.
10365         (vn_tables_s::references): Likewise.
10367         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10368         (vn_reference_eq): Update parameter and return types.
10370         * tree-ssa-structalias.c (pointer_equiv_class_table):
10371         Change type to hash_table.  Update dependent calls and types.
10372         (location_equiv_class_table): Likewise.
10374         * tree-vect-data-refs.c: Consequential changes for making
10375         peeling a hash_table.
10377         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10378         (destroy_loop_vec_info): Dependent hash_table update.
10380         * tree-vectorizer.h (peeling_htab):
10381         Change type to hash_table.  Update dependent calls and types.
10383 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
10385         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10386         to check the register content is equal or not.
10387         * lra-constraints.c (match_reload): Use lra_assign_reg_val
10388         to assign register content record.
10389         * lra-eliminations.c (update_reg_eliminate): Use
10390         lra_update_reg_val_offset to update register content offset.
10391         * lra-int.h (struct lra_reg): Add offset member.
10392         (lra_reg_val_equal_p): New static inline function.
10393         (lra_update_reg_val_offset): New static inline function.
10394         (lra_assign_reg_val): New static inline function.
10395         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10396         to assign register content record.
10397         (initialize_lra_reg_info_element): Initial offset to zero.
10399 2013-04-23  Catherine Moore  <clm@codesourcery.com>
10401         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10402         operands.  Record compression.
10404 2013-04-23  Xinliang David Li  <davidxl@google.com>
10406         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10408 2013-04-23  Richard Biener  <rguenther@suse.de>
10410         PR middle-end/57036
10411         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10412         parameter, only add abnormal goto edges from the copied body
10413         if the call could perform abnormal gotos.
10414         (copy_cfg_body): Adjust.
10416 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
10418         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10420 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
10422         * coretypes.h (gimple_stmt_iterator): Add struct to make
10423         compatible with C.
10425 2013-04-23  Richard Biener  <rguenther@suse.de>
10427         PR tree-optimization/57026
10428         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10429         from SSA names occuring in abnormal PHI nodes.
10431 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10433         * lto/lto.c (print_lto_report_1): Fix LTO report names.
10435 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10437         * lto/lto.c (print_lto_report_1): Declare early.
10438         (read_cgraph_and_symbols): Call print_lto_report_1 early.
10440 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10442         * common.opt (-flto-report-wpa): Add.
10443         * doc/invoke.texi (-flto-report-wpa): Add.
10444         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10445         (lto_main): dito.
10447 2013-04-22  Xinliang David Li  <davidxl@google.com>
10449         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10450         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10451         * Makefile.in: New dependency
10453             David Daney <ddaney.cavm@gmail.com>
10455         * configure.ac (gcc_cv_as_micromips_support):  Use the
10456         --fatal-warnings option.
10457         * configure: Regenerate.
10459 2013-04-22  Marek Polacek  <polacek@redhat.com>
10461         PR sanitizer/56990
10462         * tsan.c (instrument_expr): Don't instrument expression
10463         in case its size is zero.
10465 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
10467         PR target/57032
10468         Revert:
10469         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
10471         * config/alpha/alpha.c (TARGET_LRA_P): New define.
10473 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10475         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10476         (gimple_stmt_iterator): New typedef.
10477         * gimple.h (gimple_stmt_iterator): Rename to...
10478         (gimple_stmt_iterator_d): ... This.
10479         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10480         trees be valid for GIMPLE and GENERIC.
10481         (TARGET_GIMPLE_FOLD_BUILTIN): New.
10482         * gimple-fold.c (gimple_fold_call): Call target hook
10483         gimple_fold_builtin.
10484         * hooks.c (hook_bool_gsiptr_false): New.
10485         * hooks.h (hook_bool_gsiptr_false): New.
10486         * target.def (fold_stmt): New.
10487         * doc/tm.texi: Regenerate.
10489 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
10491         PR target/57018
10492         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10493         a set sp if no stack realignment.
10495 2013-04-22  Nick Clifton  <nickc@redhat.com>
10497         * config.gcc (tilegx-linux): Extend extra_objs rather than
10498         overwriting it.
10499         (tilepro-linux): Likewise.
10501 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10503         * config/aarch64/aarch64-builtins.c
10504         (CF): Remove.
10505         (CF0, CF1, CF2, CF3, CF4, CF10): New.
10506         (VAR<1-12>): Add MAP parameter.
10507         (BUILTIN_*): Likewise.
10508         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10509         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10510         (aarch64_ushl_n<mode>): Likewise.
10511         (aarch64_sshr_n<mode>): Likewise.
10512         (aarch64_ushr_n<mode>): Likewise.
10513         (aarch64_<maxmin><mode>): Likewise.
10514         (aarch64_sqrt<mode>): Likewise.
10515         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10516         (vshr<q>_n_*): Likewise.
10518 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10520         * config/aarch64/aarch64-builtins.c
10521         (aarch64_simd_builtin_type_mode): Handle SF types.
10522         (sf_UP): Define.
10523         (BUILTIN_GPF): Define.
10524         (aarch64_init_simd_builtins): Handle SF types.
10525         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10526         (frecps): Likewise.
10527         (frecpx): Likewise.
10528         * config/aarch64/aarch64-simd.md
10529         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10530         (aarch64_frecpe<mode>): New.
10531         (aarch64_frecps<mode>): Likewise.
10532         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10533         (v8type): Add frecp<esx>.
10534         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10535         (aarch64_frecps<mode>): Likewise.
10536         * config/aarch64/iterators.md (FRECP): New.
10537         (frecp_suffix): Likewise.
10538         * config/aarch64/arm_neon.h
10539         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10541 2013-04-22  Christian Bruel  <christian.bruel@st.com>
10543         PR target/56995
10544         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10545         (REG_CLASS_NAMES): Idem.
10546         (REG_CLASS_CONTENTS): Idem.
10547         (REGCLASS_HAS_FP_REG): Idem.
10548         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10549         (sh_conditional_register_usage): Idem.
10551 2013-04-21  Jeff Law  <law@redhat.com>
10553         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10554         (ssa_forward_propagate_and_combine): Use it.
10556 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10558         * lra.c: Update the flow chart diagram.
10560 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10562         PR rtl-optimization/56847
10563         * lra-constraints.c (process_alt_operands): Discourage alternative
10564         with non-matche doffsettable memory constraint fro memory with
10565         known offset.
10567 2013-04-19  Richard Biener  <rguenther@suse.de>
10569         PR tree-optimization/56982
10570         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10571         function.
10572         * gimplify.c (gimplify_call_expr): Notice special calls.
10573         (gimplify_modify_expr): Likewise.
10574         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10575         abnormal control flow receivers.
10576         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10577         in the same way as cfun->has_nonlocal_labels.
10578         (gimple_purge_dead_abnormal_call_edges): Likewise.
10579         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10580         receivers start a basic-block.
10582 2013-04-19  Richard Biener  <rguenther@suse.de>
10584         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10585         member ...
10586         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
10587         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10588         (SLP_TREE_LOAD_PERMUTATION): Add.
10589         (vect_transform_slp_perm_load): Adjust prototype.
10590         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10591         (vect_free_slp_instance): Likewise.
10592         (vect_create_new_slp_node): Likewise.
10593         (vect_supported_slp_permutation_p): Remove.
10594         (vect_slp_rearrange_stmts): Adjust.
10595         (vect_supported_load_permutation_p): Likewise.  Inline
10596         vect_supported_slp_permutation_p here.
10597         (vect_analyze_slp_instance): Compute load permutations per
10598         slp node instead of per instance.
10599         (vect_get_slp_defs): Adjust.
10600         (vect_transform_slp_perm_load): Likewise.
10601         (vect_schedule_slp_instance): Remove redundant code.
10602         (vect_schedule_slp): Remove hack for PR56270, add it ...
10603         * tree-vect-stmts.c (vectorizable_load): ... here, do not
10604         CSE loads for SLP.  Adjust.
10606 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10608         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10609         spelling in two comments.
10611 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10613         PR target/56797
10614         * config/arm/arm.c (load_multiple_sequence): Require SP
10615         as base register for loads if SP is in the register list.
10617 2013-04-19  Martin Jambor  <mjambor@suse.cz>
10619         PR tree-optimization/56718
10620         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
10621         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
10622         and made public.  Adjusted all callers.
10623         (ipa_intraprocedural_devirtualization): New function.
10624         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
10625         (ipa_intraprocedural_devirtualization): Likewise.
10626         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
10628 2013-04-19  Richard Biener  <rguenther@suse.de>
10630         PR tree-optimization/57000
10631         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
10633 2013-04-19  Terry Guo  <terry.guo@arm.com>
10635         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
10636         Replace with ...
10637         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
10638         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
10639         (cortex_m4_fmacs): Use new reservations.
10640         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
10642 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
10644         PR rtl-optimization/56999
10645         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
10646         related code.
10647         (lra_coalesce): Remove split_origin_bitmap and related code.
10648         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
10649         ranges if necessary.
10651 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
10653         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
10654         New array.
10655         (ix86_expand_call): Remove clobbered_registers array and use
10656         x86_64_ms_sysv_extra_clobbered_registers instead.
10657         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
10658         Declare here.
10659         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
10660         predicate.
10661         * config/i386/i386.md (*call_rex64_ms_sysv): Use
10662         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
10663         (*call_value_rex64_ms_sysv): Ditto.
10665 2013-04-18  Cary Coutant  <ccoutant@google.com>
10667         * dwarf2out.c (output_pubnames): Check die_perennial_p of
10668         parent instead of die_mark.
10670 2013-04-18  Diego Novillo  <dnovillo@google.com>
10672         * gimple.c (create_gimple_tmp): New.
10673         (get_expr_type): New.
10674         (build_assign): New.
10675         (build_type_cast): New.
10676         * gimple.h (enum ssa_mode): Define.
10677         (gimple_seq_set_location): New.
10678         * asan.c (build_check_stmt): Change some gimple_build_* calls
10679         to use build_assign and build_type_cast.
10681 2013-04-18  Richard Biener  <rguenther@suse.de>
10683         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
10684         handle negative step.  Remove redundant checks.
10685         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
10686         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
10687         for negative step and grouped loads fail to vectorize.
10689 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10691         * emit-rtl.c (reset_insn_used_flags): New function.
10692         (reset_all_used_flags): Use it.
10693         (verify_insn_sharing): New function.
10694         (verify_rtl_sharing): Fix verification for SEQUENCEs.
10696 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
10698         PR tree-optimization/56984
10699         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
10700         and (x >> M) >= N don't register any assertion if N << M is the
10701         minimum value.
10703 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10705         * lower-subreg.c (resolve_simple_move): If called self-recursive,
10706         do not delete_insn insns that have not yet been emitted, only
10707         unlink them with remove_insn.
10708         * df-scan.c (df_insn_delete): Revert r197492.
10710 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
10712         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
10713         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
10715 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10717         * config/arm/arm.md (movsicc_insn): Convert define_insn into
10718         define_insn_and_split.
10719         (and_scc,ior_scc,negscc): Likewise.
10720         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
10722 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10724         * config/arm/arm.c (use_return_insn): Return 0 for targets that
10725         can benefit from using a sequence of LDRD instructions in epilogue
10726         instead of a single LDM instruction.
10728 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
10730         PR 45688
10731         * doc/extend.texi: Fix typo.
10733 2013-04-17  Richard Biener  <rguenther@suse.de>
10735         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
10736         (vect_build_slp_tree): ... here.
10737         (vect_build_slp_tree_1): Compute which stmts of the SLP group
10738         match.  Remove special-casing of mismatched complex loads.
10739         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
10740         re-try the match with swapped commutative operands.
10741         (vect_supported_load_permutation_p): Remove special-casing of
10742         mismatched complex loads.
10743         (vect_analyze_slp_instance): Adjust.
10745 2013-04-17  Richard Biener  <rguenther@suse.de>
10747         PR rtl-optimization/56921
10748         * cfgloop.h (struct loop): Add simple_loop_desc member.
10749         (struct niter_desc): Mark with GTY(()).
10750         (simple_loop_desc): Do not use aux field but simple_loop_desc.
10751         * loop-iv.c (get_simple_loop_desc): Likewise.
10752         (free_simple_loop_desc): Likewise.
10754         Revert
10755         2013-04-16  Richard Biener  <rguenther@suse.de>
10757         PR rtl-optimization/56921
10758         * loop-init.c (pass_rtl_move_loop_invariants): Add
10759         TODO_do_not_ggc_collect to todo_flags_finish.
10760         (pass_rtl_unswitch): Same.
10761         (pass_rtl_unroll_and_peel_loops): Same.
10762         (pass_rtl_doloop): Same.
10764 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
10766         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
10767         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
10768         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
10769         references.
10770         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
10771         * tree-streamer.c (record_common_node): Adjust reference in comment.
10773 2013-04-17  Terry Guo  <terry.guo@arm.com>
10775         * config/arm/cortex-m4.md: Add a new bypass.
10777 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10779         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
10780         New pattern.
10781         (*subs_<optab><mode>_multp2): New pattern.
10782         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10783         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10785 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10787         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
10788         (*subs_mul_imm_<mode>): New pattern.
10790 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
10792         PR target/56948
10793         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
10794         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
10795         (vsx_movti_32bit): Change j->wa to O->wa.
10797 2013-04-16  Richard Biener  <rguenther@suse.de>
10799         PR rtl-optimization/56921
10800         * loop-init.c (pass_rtl_move_loop_invariants): Add
10801         TODO_do_not_ggc_collect to todo_flags_finish.
10802         (pass_rtl_unswitch): Same.
10803         (pass_rtl_unroll_and_peel_loops): Same.
10804         (pass_rtl_doloop): Same.
10806 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
10808         * config/arm/arm.c (emit_multi_reg_push): New declaration
10809         for an existing function.
10810         (arm_emit_strd_push): New function.
10811         (arm_expand_prologue): Used here.
10812         (arm_emit_ldrd_pop): New function.
10813         (arm_expand_epilogue): Used here.
10814         (arm_get_frame_offsets): Update condition.
10815         (arm_emit_multi_reg_pop): Add a special case for load of a single
10816         register with writeback.
10818 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
10820         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
10821         description.
10823 2013-04-16  Richard Biener  <rguenther@suse.de>
10825         PR tree-optimization/56756
10826         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
10827         (first_mem_ref_loc): New.
10828         (execute_sm): Place the load temporarily before a previous
10829         access instead of in the latch edge to ensure its SSA dependencies
10830         are defined at points dominating the load.
10832 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
10834         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
10835         correct fix by moving header and footer insn to the footer of
10836         the merged basic block.  Clear BB_END of the merged-away block.
10838         PR middle-end/43631
10839         * emit-rtl.c (make_note_raw): New function.
10840         (link_insn_into_chain): New static inline function.
10841         (add_insn): Use it.
10842         (add_insn_before, add_insn_after): Factor insn chain linking code...
10843         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
10844         using link_insn_into_chain.
10845         (note_outside_basic_block_p): New helper function for emit_note_after
10846         and emit_note_before.
10847         (emit_note_after): Use nobb variant of add_insn_after if the note
10848         should not be contained in a basic block.
10849         (emit_note_before): Use nobb variant of add_insn_before if the note
10850         should not be contained in a basic block.
10851         (emit_note_copy): Use make_note_raw.
10852         (emit_note): Likewise.
10853         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
10854         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
10855         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
10856         the moved barrier the tail of the basic block it follows.
10857         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
10859 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
10861         PR tree-optimization/56962
10862         * gimple-ssa-strength-reduction.c (record_increment): Only set
10863         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
10864         either rhs1 or rhs2 is equal to c->base_expr.
10866 2013-04-15  Richard Biener  <rguenther@suse.de>
10868         PR tree-optimization/56933
10869         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
10870         member.
10871         (GROUP_READ_WRITE_DEPENDENCE): Remove.
10872         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
10873         * tree-vect-data-refs.c (vect_analyze_group_access): Move
10874         dependence check ...
10875         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
10876         ... here.
10877         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
10878         GROUP_READ_WRITE_DEPENDENCE.
10880 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10882         * emit-rtl.c (reset_all_used_flags): New function.
10883         (verify_rtl_sharing): Call reset_all_used_flags before and after
10884         performing the checks.
10886 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10888         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
10889         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
10890         * config/arm/constraints.md (De): New constraint.
10891         * config/arm/neon.md (anddi3_neon): Delete.
10892         (neon_vand<mode>): Expand to standard anddi3 pattern.
10893         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
10894         Move earlier in the file.
10895         (neon_inv_logic_op2): Likewise.
10896         (arm_anddi_operand_neon): New predicate.
10898 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10900         * configure.ac (gcc_cv_ld_as_needed): Set
10901         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
10902         Use -z ignore, -z record on *-*-solaris2*.
10903         (HAVE_LD_AS_NEEDED): Update comment.
10904         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
10905         * configure: Regenerate.
10906         * config.in: Regenerate.
10907         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
10908         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
10909         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
10910         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
10911         equivalents.  Fix markup.
10912         * doc/tm.texi: Regenerate.
10914 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
10916         * config/i386/i386.opt: New option mstack-protector-guard=.
10917         * config/i386/i386-opts.h: Add enum stack_protector_guard.
10918         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
10919         TARGET_SSP_TLS_GUARD.
10920         * config/i386/i386.c (ix86_option_override_internal): Set
10921         ix86_stack_protector_guard.
10922         * config/i386/i386.md (stack_protect_set): Enable for
10923         TARGET_SSP_TLS_GUARD only.
10924         (stack_protect_set_<mode>): Ditto.
10925         (stack_protect_test): Ditto.
10926         (stack_protect_test_<mode>): Ditto.
10927         * doc/invoke.texi (i386 Option): Document.
10929 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
10931         PR target/56890
10932         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
10933         (S_MODES): Set H_MODE bit.
10934         (SF_MODES): Set only S_MODE and SF_MODE bits.
10935         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
10936         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
10937         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
10938         <MODE_FLOAT>: Likewise.
10940 2013-04-15  Joey Ye  <joey.ye@arm.com>
10942         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
10944 2013-04-15  Joey Ye  <joey.ye@arm.com>
10946         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
10947         for real far jump.
10948         (thumb_far_jump_used_p): Count instruction size and set
10949         far_jump_used.
10951 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
10953         * reorg.c (fill_simple_delay_slots): Reindent block of code.
10954         * resource.c (mark_target_live_regs): Reformat conditional block.
10956 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
10958         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
10959         notes, they are emitted only just before final.
10960         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
10962 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
10964         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
10965         * cfgrtl.c (delete_insn): Call it here instead.
10966         * lra-spills.c (lra_final_code_change): Use delete_insn.
10967         * haifa-sched.c (sched_remove_insn): Likewise.
10968         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
10969         returning to the nop pool.
10970         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
10971         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
10973 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
10975         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
10976         * doc/tm.texi: Regenerated.
10978 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
10980         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
10981         QImode checks.
10983 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
10985         * df-core.c (df_find_def): Compare register numbers.
10986         (df_find_use): Likewise.
10988 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
10990         PR target/56903
10991         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
10992         lra_in_progress for return.
10994 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
10996         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
10997         define_insn into define_insn_and_split and emit movsicc patterns.
10999 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11001         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11003 2013-04-12  Richard Biener  <rguenther@suse.de>
11005         * tree-pass.h (TODO_do_not_ggc_collect): New.
11006         * passes.c (execute_one_ipa_transform_pass): Honor
11007         TODO_do_not_ggc_collect.
11008         (execute_one_pass): Likewise.
11010         Revert
11011         2013-04-10  Richard Biener  <rguenther@suse.de>
11013         * passes.c (init_optimization_passes): Remove reload pass.
11014         * ira.c (do_reload): Merge into ...
11015         (ira): ... this.
11016         (rest_of_handle_reload): Remove.
11017         (pass_reload): Likewise.
11018         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11019         of reload for vzeroupper pass placement.
11021 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11023         PR tree-optimization/56918
11024         PR tree-optimization/56920
11025         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11026         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11027         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11028         use wide_mul_with_sign method.
11030 2013-04-12  Richard Biener  <rguenther@suse.de>
11032         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11033         not be considered a gimple constant.
11035 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11037         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11038         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11039         operations.
11041 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11042             Jakub Jelinek  <jakub@redhat.com>
11044         * opts.c: Include diagnostic-color.h.
11045         (common_handle_option): Handle OPT_fdiagnostics_color_.
11046         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11047         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11048         (diagnostic-color.o): New.
11049         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11050         (diagnostic_color_rule): New enum.
11051         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11052         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11053         the location string.
11054         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11055         either NULL, or color kind.
11056         * diagnostic-color.c: New file.
11057         * diagnostic-color.h: New file.
11058         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11059         arguments.
11060         * doc/invoke.texi (-fdiagnostics-color): Document.
11061         * pretty-print.h (pp_show_color): Define.
11062         (struct pretty_print_info): Add show_color field.
11063         * diagnostic.c: Include diagnostic-color.h.
11064         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11065         macros.  Colorize error:, warning: etc. strings and also the location
11066         string.
11067         (diagnostic_show_locus): Colorize the caret line.
11068         * pretty-print.c: Include diagnostic-color.h.
11069         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11070         inside of %< %> quotes or quoted through q format modifier.
11072 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11074         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11076 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11078         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11079         code in CC_NZ mode.
11080         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11081         pattern.
11083 2013-04-11  Marek Polacek  <polacek@redhat.com>
11085         PR tree-optimization/48184
11086         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11088 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11090         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11091         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11092         (skip_simple_arithmetic): Tidy up.
11093         * tree.h (skip_simple_constant_arithmetic): Declare.
11095 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11097         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11099 2013-04-11  Richard Biener  <rguenther@suse.de>
11101         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11102         generate vector constants.
11104 2013-04-11  Richard Biener  <rguenther@suse.de>
11106         PR tree-optimization/56878
11107         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11108         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11109         New function.
11110         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11111         Prefer to align the DR with the most invariant base address.
11113 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11115         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11116         comment.
11118 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11120         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11121         floating-point vector comparisons against 0.
11123 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11125         PR tree-optimization/56899
11126         * fold-const.c (extract_muldiv_1): Apply distributive law
11127         only if TYPE_OVERFLOW_WRAPS (ctype).
11129 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11131         PR target/56124
11132         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11133         loading has side effect.
11135 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
11137         * config/sparc/sparc.c: Include tree-pass.h.
11138         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11139         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
11140         head of file.  Change return type.  Split off gate function.
11141         (sparc_gate_work_around_errata): New function.
11142         (pass_work_around_errata): New pass definition.
11143         (insert_pass_work_around_errata) New pass insert definition to
11144         insert pass_work_around_errata just after delayed-branch scheduling.
11145         (sparc_option_override): Insert the pass.
11146         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11148 2013-04-10  David S. Miller  <davem@davemloft.net>
11150         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11151         or -mcpu=hypersparc.
11153         * target.def (cstore_mode): New hook.
11154         * target.h: Include insn-codes.h
11155         * targhooks.c: Likewise.
11156         (default_cstore_mode): New function.
11157         * targhooks.h: Declare it.
11158         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11159         * doc/tm.texi: Rebuild.
11160         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11161         target hook, rather than inspecting the insn_data.
11162         * config/sparc/sparc.c (sparc_cstore_mode): New function.
11163         (TARGET_CSTORE_MODE): Redefine.
11164         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11165         result patterns.
11166         * config/sparc/predicates.md (cstore_result_operand): New special
11167         predicate.
11168         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11169         Use it for operand 0.
11170         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11171         (*snesi_special): Likewise.
11172         (*snesi_zero): Likewise.
11173         (*seqsi_zero): Likewise.
11174         (*sltu_insn): Likewise.
11175         (*sgeu_insn): Likewise.
11176         (*seqdi_special): Make operand 0 and comparison operation be of
11177         DImode.
11178         (*snedi_special): Likewise.
11179         (*snedi_special_vis3): Likewise.
11180         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11181         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11182         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11183         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11184         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11185         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11186         (*sltu_extend_sp64): Likewise.
11187         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11188         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11189         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11190         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11191         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11193 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
11195         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11196         (aarch64_start_file): Use the new function.
11198 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11199             Jason Merrill  <jason@redhat.com>
11201         * common.opt: Add -gdwarf.
11202         * opts.c (common_handle_option): Handle it.
11203         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11205 2013-04-10  Richard Biener  <rguenther@suse.de>
11207         * passes.c (execute_todo): Do not call ggc_collect conditional here.
11208         (execute_one_ipa_transform_pass): But unconditionally here.
11209         (execute_one_pass): And here.
11210         (init_optimization_passes): Remove reload pass.
11211         * tree-pass.h (TODO_ggc_collect): Remove.
11212         (pass_reload): Likewise.
11213         * ira.c (do_reload): Merge into ...
11214         (ira): ... this.
11215         (rest_of_handle_reload): Remove.
11216         (pass_reload): Likewise.
11217         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11218         of reload for vzeroupper pass placement.
11219         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11220         and todo_flags_finish of all passes.
11222 2013-04-10  Richard Biener  <rguenther@suse.de>
11224         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11225         first_const_oprnd field, rename first_def_type to first_op_type.
11226         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11227         (vect_get_and_check_slp_defs): Always use the type of the
11228         operand.  Allow mixed vect_external_def, vect_constant_def types.
11229         (vect_get_constant_vectors): Handle mixed vect_external_def,
11230         vect_constant_def types.
11232 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
11234         PR tree-optimization/55524
11235         * tree-ssa-math-opts.c
11236         (convert_mult_to_fma): Don't use an fms construct
11237         when we don't have an fms operation, but fnma, and it looks
11238         likely that we'll be able to use the latter.
11240 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11242         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11243         function.
11244         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11245         inline fail caused by overwritable functions.
11247 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
11249         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11250         unnecessary bits in the constant power of two case.
11252 2013-04-10  Richard Biener  <rguenther@suse.de>
11254         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11255         broken code swapping operands.
11256         (vect_build_slp_tree): Do not compute load permutations here.
11257         (vect_analyze_slp_instance): Compute load permutations here,
11258         after building the SLP tree.
11260 2013-04-09  Christian Bruel  <christian.bruel@st.com>
11262         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11263         of next/prev_real_insn.
11265 2013-04-09  Jan Hubicka  <jh@suse.cz>
11267         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11268         Drop aliased parameter.
11269         (function_and_variable_visibility): Do not handle alias pairs.
11270         * cgraph.c (varpool_externally_visible_p): Update prototype.
11271         * varpool.c (varpool_add_new_variable): Update.
11273 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11275         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11277 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
11279         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11281         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11283 2013-04-09  Marek Polacek  <polacek@redhat.com>
11285         PR tree-optimization/48762
11286         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11288 2013-04-09  Richard Biener  <rguenther@suse.de>
11290         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11291         dealing with cost.
11292         (vect_build_slp_tree): Likewise.
11293         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11294         calculating the cost of a SLP instance.
11295         (vect_analyze_slp_instance): Use it from here, after building
11296         the SLP tree.
11298 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11300         PR middle-end/56883
11301         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11302         expand_omp_for_static_chunk): Use simple_p = true in
11303         force_gimple_operand_gsi calls when assigning to addressable decls.
11305 2013-04-09  Jeff Law  <law@redhat.com>
11307         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11308         when the boolean was created by converting a wider object which
11309         had a boolean range.
11311 2013-04-09  Richard Biener  <rguenther@suse.de>
11313         * tree-vectorizer.h (slp_void_p): Remove.
11314         (slp_tree): Typedef before _slp_tree declaration.
11315         (struct _slp_tree): Use a vector of slp_tree as children.
11316         (vect_get_place_in_interleaving_chain): Remove.
11317         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11318         Move ...
11319         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11320         and make static.
11321         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11322         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11323         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11324         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11325         Use slp_node instead of slp_void_p and adjust.
11327 2013-04-09  Richard Biener  <rguenther@suse.de>
11329         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11330         work that is not necessary.
11332 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11334         PR tree-optimization/56854
11335         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11336         forward into clobber stmts if it would change MEM_REF lhs into
11337         non-MEM_REF.
11339 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
11341         * tree.c (type_hash_lookup, type_hash_add): Make static.
11342         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11344 2013-04-09  Richard Biener  <rguenther@suse.de>
11346         * tree.h (unsave_expr_now): Remove.
11347         * tree-inline.c (mark_local_for_remap_r): Remove.
11348         (unsave_expr_1): Likewise.
11349         (unsave_r): Likewise.
11350         (unsave_expr_now): Likewise.
11351         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11352         (propagate_tree_value): Likewise.
11354 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
11356         * doc/rtl.texi (sequence): Rewrite documentation to match the
11357         current use of SEQUENCE rtl objects.
11358         * rtl.def (SEQUENCE): Likewise.
11360         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11361         Update documentation.
11362         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11363         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11365         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11367 2013-04-08  Teresa Johnson  <tejohnson@google.com>
11369         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11370         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11371         methods.
11372         (estimate_edge_size_and_time): Add comment to suggest using rounding
11373         methods.
11374         (estimate_node_size_and_time): Ditto.
11375         (remap_edge_change_prob): Use helper rounding divide methods.
11376         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11377         (gimple_mod_pow2_value_transform): Ditto.
11378         (gimple_mod_subtract_transform): Ditto.
11379         (gimple_ic_transform): Ditto.
11380         (gimple_stringops_transform): Ditto.
11381         * stmt.c (conditional_probability): Ditto.
11382         (emit_case_dispatch_table): Ditto.
11383         * lto-cgraph.c (merge_profile_summaries): Ditto.
11384         * tree-optimize.c (execute_fixup_cfg): Ditto.
11385         * cfgcleanup.c (try_forward_edges): Ditto.
11386         * cfgloopmanip.c (scale_loop_profile): Ditto.
11387         (loopify): Ditto.
11388         (duplicate_loop_to_header_edge): Ditto.
11389         (lv_adjust_loop_entry_edge): Ditto.
11390         * tree-vect-loop.c (vect_transform_loop): Ditto.
11391         * profile.c (compute_branch_probabilities): Ditto.
11392         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11393         * lto-streamer-in.c (input_cfg): Ditto.
11394         * gimple-streamer-in.c (input_bb): Ditto.
11395         * ipa-cp.c (update_profiling_info): Ditto.
11396         (update_specialized_profile): Ditto.
11397         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11398         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11399         rounding methods.
11400         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11401         (compute_trg_info): Ditto.
11402         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11403         (purge_dead_edges): Ditto.
11404         * loop-unswitch.c (unswitch_loop): Ditto.
11405         * cgraphclones.c (cgraph_clone_edge): Ditto.
11406         (cgraph_clone_node): Ditto.
11407         * tree-inline.c (copy_bb): Ditto.
11408         (copy_edges_for_bb): Ditto.
11409         (initialize_cfun): Ditto.
11410         (copy_cfg_body): Ditto.
11411         (expand_call_inline): Ditto.
11413 2013-04-08  Kai Tietz  <ktietz@redhat.com>
11415         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11416         TARGET_CYGWIN64 by TARGET_64BIT.
11418 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11420         * config/epiphany/epiphany.md (GPR_1): New constant.
11421         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11422         * config/epiphany/epiphany.c (gen_compare_reg):
11423         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11424         is already in place.
11425         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11426         Don't require being called during rtl expansion; If y operlaps r0,
11427         return 0.
11428         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11429         (epiphany_expand_epilogue): Likewise.
11431         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11432         Don't use CC_FPmode for ORDERED / UNORDERED.
11433         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11435         * config/epiphany/constraints.md (CnL): New constraint.
11436         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11437         * config/epiphany/predicates.md (add_operand): Allow 1024.
11439         * config/epiphany/epiphany.md (logical_op): New code iterator.
11440         (op_mnc): New code attribute.
11441         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11442         (mov_f+1, mov_f+2): New peephole2 patterns.
11444         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11445         (cstoresi4): Also allow re-use of zero result when doing a NE
11446         comparison to a non-zero operand.
11447         Use (clobber (scratch)) for first insn if the gpr output is not needed.
11449         * config/epiphany/epiphany.md (<insn_opname>v2si3):
11450         Use gen_addsi3_i / gen_subsi3_i.
11452 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11454         PR c++/34949
11455         PR c++/50243
11456         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11457         contain anything but clobbers, at most one __builtin_stack_restore,
11458         optionally debug stmts and final resx, and if it has at least one
11459         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
11460         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11461         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11462         which isn't defaut definition, remove them.
11463         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11464         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11465         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11466         with MEM_REF LHS with SSA_NAME address.
11468 2013-04-08  Jeff Law  <law@redhat.com>
11470         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11472 2013-04-08  Richard Biener  <rguenther@suse.de>
11474         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11475         extra newline.
11476         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11477         determined vector type.
11478         (vect_analyze_data_refs): Likewise.
11479         (vect_get_new_vect_var): Adjust.
11480         (vect_create_destination_var): Preserve SSA name versions.
11481         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11482         not dump anything here.
11484 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11486         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11487         Add member lr_slot_known.
11488         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11489         if necessary.
11490         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11491         Remove code that sets lr_slot_offset according to what a previous
11492         version of epiphany_emit_save_restore used to do.
11493         (epiphany_emit_save_restore): When doing an lr save or restore,
11494         set/verify lr_slot_known and lr_slot_offset.
11496 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
11498         PR target/54338
11499         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11500         in ALL_REGS.
11502 2013-04-08  Richard Biener  <rguenther@suse.de>
11504         * alias.c (find_base_term): Fix thinko in previous change.
11506 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11508         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11509         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11510         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
11511         if possible to compute val.
11512         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11513         For QImode integers don't require anything about precision.  Use
11514         const_with_all_bytes_same to find out if the constant doesn't have
11515         repeated bytes in it.
11517 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11519         * config/s390/s390.c (s390_expand_insv): Only accept insertions
11520         within mode size.
11522 2013-04-08  Marek Polacek  <polacek@redhat.com>
11524         PR rtl-optimization/48182
11525         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11526         value to 1.
11528 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11530         PR target/55487
11531         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11532         nuses, make sure we have a label.
11534 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11536         PR target/56843
11537         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11538         (rs6000_emit_swdiv_low_precision): Remove.
11539         (rs6000_emit_swdiv): Rewrite to handle between one and four
11540         iterations of Newton-Raphson generally; modify required number of
11541         iterations for some cases.
11542         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11544 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
11546         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11547         set-but-unused variable.
11549         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11550         basic blocks of released function bodies garbage-collectable.
11552         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11553         (struct rtl_opt_pass): Add TODO_df_finish.
11555         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11557 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11559         * config/arm/constraints.md (q): New constraint.
11560         * config/arm/ldrdstrd.md: New file.
11561         * config/arm/arm.md (ldrdstrd.md) New include.
11562         (arm_movdi): Use "q" instead of "r" constraint
11563         for double-word memory access.
11564         (movdf_soft_insn): Likewise.
11565         * config/arm/vfp.md (movdi_vfp): Likewise.
11566         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11567         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11568         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11569         (mem_ok_for_ldrd_strd): Likewise.
11570         (output_move_double): Update assertion.
11572 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11574         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11576 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11578         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11579         define_insn_and_split.
11580         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11582 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11584         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11585         define_insn_and_split.
11586         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11587         (shiftsi3_compare): New pattern.
11588         (rrx): New pattern.
11589         * config/arm/unspecs.md (UNSPEC_RRX): New.
11591 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11593         * config/arm/arm.md (negdi_extendsidi): New pattern.
11594         (negdi_zero_extendsidi): Likewise.
11596 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11598         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11599         define_insn_and_split.
11600         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11601         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11603 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11605         * config/arm/arm.md (arm_subdi3): Convert define_insn into
11606         define_insn_and_split.
11607         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11608         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11610 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11612         * config/arm/arm.md (subsi3_carryin): New pattern.
11613         (subsi3_carryin_const): Likewise.
11614         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11615         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
11617 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11619         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
11621 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11623         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
11624         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
11626 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11628         * config/arm/arm.c (arm_expand_builtin): Change fcode
11629         type to unsigned int.
11631 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11633         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
11635 2013-04-04  Ian Lance Taylor  <iant@google.com>
11637         * doc/standards.texi (Standards): The Go frontend supports the Go 1
11638         language standard.
11640 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
11642         PR middle-end/56729
11643         * df-scan.c (df_insn_delete): Disable failing assert.
11645 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11647         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
11648         New function prototype.
11649         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
11650         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
11651         (arm_builtin_vectorized_function): New function.
11653 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11655         * config/arm/arm_neon_builtins.def: New file.
11656         * config/arm/arm.c (neon_builtin_data): Move contents to
11657         arm_neon_builtins.def.
11658         (enum arm_builtins): Include neon builtin definitions.
11659         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
11660         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
11662 2013-04-04  Marek Polacek  <polacek@redhat.com>
11664         PR tree-optimization/48186
11665         * predict.c (maybe_hot_frequency_p): Return false if
11666         HOT_BB_FREQUENCY_FRACTION is 0.
11667         (cgraph_maybe_hot_edge_p): Likewise.
11669 2013-04-04  Richard Biener  <rguenther@suse.de>
11671         PR tree-optimization/56826
11672         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
11673         more accurately.
11675 2013-04-04  Richard Biener  <rguenther@suse.de>
11677         PR tree-optimization/56213
11678         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
11679         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
11681 2013-04-04  Richard Biener  <rguenther@suse.de>
11683         PR tree-optimization/56837
11684         * tree-loop-distribution.c (classify_partition): For non-zero
11685         values require that the value has the same precision as its
11686         mode to be useful as memset value.
11688 2013-04-03  Nick Clifton  <nickc@redhat.com>
11690         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
11691         (fmssf4): Use fmsf.s on E3V5 architectures.
11692         (fnmasf4): Use fnmaf.s on E3V5 architectures.
11693         (fnmssf4): Use fnmsf.s on E3V5 architectures.
11695 2013-04-03  Jeff Law  <law@redhat.com>
11697         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
11698         (lra-eliminations.o): Likewise.
11700 2013-04-03  Teresa Johnson  <tejohnson@google.com>
11702         * gcov-io.c (compute_working_sets): Moved most of body of old
11703         compute_working_sets here from profile.c.
11704         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
11705         (gcov_working_set_t): Moved typedef here from basic-block.h
11706         (compute_working_set): Declare.
11707         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
11708         (get_working_sets): Renamed from compute_working_set,
11709         replace most of body with call to new compute_working_sets.
11710         (get_exec_counts): Replace call to compute_working_sets
11711         to get_working_sets.
11712         * profile.h (get_working_sets): Renamed from compute_working_set.
11713         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
11714         to get_working_sets.
11715         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
11716         * gcov-dump.c (dump_working_sets): New function.
11718 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
11720         * hwint.c (sext_hwi, zext_hwi): New functions.
11721         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
11722         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
11723         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
11724         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
11725         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
11726         (sext_hwi, zext_hwi): New functions.
11728 2013-04-03  Jeff Law  <law@redhat.com>
11730         PR tree-optimization/56799
11731         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
11732         back test for widening conversion erroneously dropped in prior change.
11734 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11736         PR target/56809
11737         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
11738         instead of next_real_insn.
11740 2013-04-03  Marek Polacek  <polacek@redhat.com>
11742         PR sanitizer/55702
11743         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
11745 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11747         PR target/56809
11748         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
11749         next_real_insn.
11750         (thumb1_output_casesi): Likewise.
11751         (thumb2_output_casesi): Likewise.
11753 2013-04-03  Richard Biener  <rguenther@suse.de>
11755         PR tree-optimization/56817
11756         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
11757         Split out ...
11758         (tree_unroll_loops_completely_1): ... new function to manually
11759         walk the loop tree, properly defering outer loops of unrolled
11760         loops to later iterations.
11762 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11764         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
11765         (vectorizable_load): Likewise.
11766         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
11767         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
11769 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11771         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
11772         BIT_FIELD_REF.
11774 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
11776         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
11778 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
11780         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
11782 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11784         PR tree-optimization/56790
11785         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
11786         folding.
11788 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11790         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
11791         Handle VEC_MERGE.
11792         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
11793         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
11794         equal arguments.
11796 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
11798         PR c/19449
11799         * tree.h (force_folding_builtin_constant_p): New decl.
11800         * builtins.c (force_folding_builtin_constant_p): New variable.
11801         (fold_builtin_constant_p): Fold immediately also if
11802         force_folding_builtin_constant_p.
11804 2013-04-03  Richard Biener  <rguenther@suse.de>
11806         PR tree-optimization/56812
11807         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
11808         DRs of the same interleaving chain are independent.
11810 2013-04-02  Jason Merrill  <jason@redhat.com>
11812         * gdbinit.in (pbb): Use debug fn.
11814 2013-04-02  Lawrence Crowl  <crowl@google.com>
11816         * sese.h (struct ivtype_map_elt_s): Remove unused.
11817         (extern debug_ivtype_map): Remove unused.
11818         (extern eq_ivtype_map_elts): Remove unused.
11819         * sese.c (debug_ivtype_map): Removed unused.
11820         (debug_ivtype_map_1): Removed unused.
11821         (debug_ivtype_elt): Remove unused.
11822         (eq_ivtype_map_elts): Remove unused.
11825 2013-04-02  Kai Tietz  <ktietz@redhat.com>
11827         PR target/52790
11828         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
11829         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
11830         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
11831         function.
11832         (legitimize_pe_coff_symbol): Likewise.
11833         (is_imported_p): New helper-function.
11834         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
11835         for Windows x64 targets.
11836         (ix86_expand_prologue): Optimize for pe-coff targets.
11837         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
11838         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
11839         medium/large code-model.
11840         (legitimize_pic_address): Likewise.
11841         (legitimize_tls_address): Likewise.
11842         (ix86_expand_call): Likewise.
11843         (x86_output_mi_thunk): Likewise.
11844         (get_dllimport_decl): Add new beimport argument.
11845         (construct_plt_address): Don't assert for x64 pe-coff targets.
11846         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
11847         targets.
11848         (SYMBOL_FLAG_STUBVAR): New macro.
11849         (SYMBOL_REF_STUBVAR_P): Likewise.
11850         * config/i386/winnt.c (stub_list): New structure.
11851         (stub_head): New local variable.
11852         (i386_pe_record_stub): New function.
11853         (i386_pe_file_end): Emit refptr-stubs.
11855 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
11857         PR rtl-optimization/56745
11858         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
11859         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
11861         PR c++/34949
11862         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
11863         and both of them are MEM_REFs, just compare first argument for
11864         equality and attempt to deal even with differing offsets.
11866         PR c++/34949
11867         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
11868         of gimple_clobber_p to be MEM_REF.
11869         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
11870         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
11871         after gimplification.
11872         * asan.c (get_mem_ref_of_assignment): Don't instrument
11873         gimple_clobber_p stmts.
11874         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
11875         gimple_clobber_p stmt if they have MEM_REF lhs and
11876         are dead because of another gimple_clobber_p stmt.
11877         * tree-ssa-live.c (clear_unused_block_pointer): Treat
11878         gimple_clobber_p stmts like debug stmts.
11879         (remove_unused_locals): Remove clobbers with MEM_REF lhs
11880         that refer to unused VAR_DECLs or uninitialized values.
11881         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
11882         gimple_clobber_p stmts if they refer to removed parameters.
11883         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
11884         formatting.
11886 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
11888         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
11889         using SWI48 mode attribute.
11891 2013-04-02  Wei Mi  <wmi@google.com>
11893         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
11894         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
11895         *<rotate_insn><mode>3_mask in i386.md.
11897 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11899         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
11901 2013-04-02  Richard Biener  <rguenther@suse.de>
11903         PR tree-optimization/56778
11904         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11905         Runtime alias tests are not supported for gather loads.
11906         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
11907         stmts referenced from SSA operands before updating SSA form.
11909 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
11910             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11912         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
11913         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
11914         * config/arm/cortex-a53.md: New file.
11915         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
11916         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
11917         * config/arm/arm.c (arm_issue_rate): Likewise.
11918         * config/arm/arm-tune.md: Regenerate
11919         * config/arm/arm-tables.opt: Regenerate.
11920         * config/arm/arm-cores.def: Add cortex-a53.
11922 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
11924         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
11925         non-static link.
11927 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
11929         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
11930         scalar load/store operations using B/H registers.
11931         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
11933 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
11935         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
11936         scalar move.
11937         * config/aarch64/aarch64.c
11938         (aarch64_simd_scalar_immediate_valid_for_move): New.
11939         * config/aarch64/aarch64-protos.h
11940         (aarch64_simd_scalar_immediate_valid_for_move): New.
11941         * config/aarch64/constraints.md (Dh, Dq): New.
11942         * config/aarch64/iterators.md (hq): New.
11944 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
11946         * reorg.c (get_branch_condition): Deal with conditional returns.
11947         (fill_simple_delay_slots): Remove dead code dealing with jumps.
11949 2013-04-01  Wei Mi  <wmi@google.com>
11951         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
11952         Truncate operand 2 using %b asm operand modifier.
11953         (*<shift_insn><mode>3_mask): Ditto.
11954         (*<rotate_insn><mode>3_mask): Ditto.
11956 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
11958         PR middle-end/56798
11959         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
11961 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
11963         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
11964         of next_real_insn.
11965         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
11967 2013-03-30  Lawrence Crowl  <crowl@google.com>
11969         * dse.c (clear_alias_sets): Remove never set.
11970         (disqualified_clear_alias_sets): Remove never set.
11971         (clear_alias_mode_pool): Remove never set.
11972         (dse_step0): Remove condition that is never true.
11973         (canon_address): Remove condition that is never true.
11974         (dse_step7): Remove condition that is never true.
11975         (rest_of_handle_dse): Remove condition that is never true.
11976         (rest_of_handle_dse::did_global): Remove never read from above.
11977         (dse_step2_spill): Remove never called from above.
11978         (dse_step5_spill): Remove never called from above.
11980 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
11982         * doc/md.texi (Standard Names) <casesi>: Update documentation for
11983         JUMP_TABLE_DATA changes.
11984         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
11985         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
11986         (Insns) <jump_table_data>: New entry.
11987         * doc/tm.texi: Regenerate.
11989         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
11991         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
11992         for table jump at the end of a basic block using tablejump_p.
11993         * targhooks.c (default_invalid_within_doloop): Likewise.
11994         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
11995         target hook implementation that is identical to the default hook.
11996         (rs6000_invalid_within_doloop): Remove.
11998         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
11999         unused variable from tablejump_p call.
12001         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12002         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12003         (INSN_DELETED_P): Likewise.
12004         (emit_jump_table_data): New prototype.
12005         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12006         after 4th as unused.
12007         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12008         * sched-vis.c (print_insn): Likewise.
12009         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12010         insn for compatibility with back ends that use next_active_insn to
12011         identify jump table data.
12012         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12013         (remove_insn): Likewise.
12014         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12015         to be emitted.
12016         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12017         (emit_jump_table_data): New function.
12019         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12020         basic block, a JUMP_TABLE_DATA never is.
12021         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12022         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12023         off from code handling real insns.
12024         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12025         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12026         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12027         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12028         is not a NONDEBUG_INSN_P.
12029         * ira-costs.c (scan_one_insn): Likewise.
12030         * jump.c (mark_all_labels): Likewise.
12031         (mark_jump_label_1): Likewise.
12032         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12033         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12034         a basic block.
12035         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12036         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12037         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12038         JUMP_TABLE_DATA_P insns.
12039         (calculate_elim_costs_all_insns): Likewise.
12040         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12041         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12042         (delete_output_reload): Code style fixups.
12043         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12044         insn flags on this non-insn.
12045         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12046         as scheduling barriers, for pre-change compatibility.
12047         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12048         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12050         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12051         redundant JUMP_TABLE_DATA_P test.
12052         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12053         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12054         (frv_for_each_packet): Likewise.
12055         * config/i386/i386.c (min_insn_size): Likewise.
12056         (ix86_avoid_jump_mispredicts): Likewise.
12057         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12058         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12059         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12060         (mips16_insn_length): Robustify.
12061         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12062         (mips16_split_long_branches): Likewise.
12063         * config/pa/pa.c (pa_combine_instructions): Likewise.
12064         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12065         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12066         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12067         as contributing to pool range lengths.
12068         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12069         Remove redundant JUMP_TABLE_DATA_P test.
12070         (sh_loop_align): Likewise.
12071         (split_branches): Likewise.
12072         (sh_insn_length_adjustment): Likewise.
12073         * config/spu/spu.c (get_branch_target): Likewise.
12075 2013-03-29  Jan Hubicka  <jh@suse.cz>
12077         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12078         gcov streaming; stream hot bb threshold to ltrans.
12079         * predict.c (get_hot_bb_threshold): Break out from ....
12080         (maybe_hot_count_p): ... here.
12081         (set_hot_bb_threshold): New function.
12082         * lto-section-in.c (lto_section_name): Add profile.
12083         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12084         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12085         and data-streamer.h
12086         (histogram_entry): New structure.
12087         (histogram, histogram_pool): New global vars.
12088         (histogram_hash): New structure.
12089         (histogram_hash::hash): New method.
12090         (histogram_hash::equal): Likewise.
12091         (account_time_size): New function.
12092         (cmp_counts): New function.
12093         (dump_histogram): New function.
12094         (ipa_profile_generate_summary): New function.
12095         (ipa_profile_write_summary): New function.
12096         (ipa_profile_read_summary): New function.
12097         (ipa_profile): Decide on threshold.
12098         (pass_ipa_profile): Add ipa_profile_write_summary and
12099         ipa_profile_read_summary.
12100         * Makefile.in (ipa.o): Update dependencies.
12101         * lto-streamer.h (LTO_section_ipa_profile): New section.
12103 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12105         * tree.h (VAR_P): New.
12107 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12109         PR lto/56777
12110         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12112 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12114         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12115         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12116         (control_flow_insn_p): Likewise.
12117         * cfgrtl.c (duplicate_insn_chain): Likewise.
12118         * final.c (get_attr_length_1): Likewise.
12119         (shorten_branches): Likewise.
12120         (final_scan_insn): Likewise.
12121         * function.c (instantiate_virtual_regs): Likewise.
12122         * gcse.c (insert_insn_end_basic_block): Likewise.
12123         * ira-costs.c (scan_one_insn): Likewise.
12124         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12125         * lra.c (check_rtl): Likewise.
12126         * reload1.c (elimination_costs_in_insn): Likewise.
12127         * reorg.c (follow_jumps): Likewise.
12129         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12130         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12131         (thumb_far_jump_used_p): Likewise.
12132         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12133         (workaround_speculation): Likewise.
12134         (add_sched_insns_for_speculation): Likewise.
12135         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12136         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12137         (frv_for_each_packet): Likewise.
12138         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12139         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12140         (final_emit_insn_group_barriers): Likewise.
12141         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12142         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12143         (mips16_insn_length): Likewise.
12144         * config/pa/pa.c (pa_reorg): Likewise.
12145         (pa_combine_instructions): Likewise.
12146         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12147         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12148         (sh_reorg): Likewise.
12149         (split_branches): Likewise.
12150         * config/spu/spu.c (get_branch_target): Likewise.
12152         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12153         JUMP_TABLE_DATA_P.
12155 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
12157         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12158         Fix declaration name.
12160 2013-03-28  Lawrence Crowl  <crowl@google.com>
12162         * graphds.h (struct graph.indicies): Remove unused.
12163         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12164         (SCOP_ORIGINAL_PDDRS): Remove unused.
12165         * sese.h (extern insert_loop_close_phis): Removed unused.
12166         (extern insert_guard_phis): Removed unused.
12167         (extern ivtype_map_elt_info): Removed unused.
12168         (new_ivtype_map_elt): Removed unused.
12169         * sese.c (ivtype_map_elt_info): Removed unused.
12171 2013-03-28  Lawrence Crowl  <crowl@google.com>
12173         * Makefile.in: Add several missing include dependences.
12174         (DUMPFILE_H): New.
12175         (test-dump.o): New.  This object is not added to any executable,
12176         but is present for ad-hoc testing.
12177         * bitmap.c
12178         (debug (const bitmap_head_def &)): New.
12179         (debug (const bitmap_head_def *)): New.
12180         * bitmap.h
12181         (extern debug (const bitmap_head_def &)): New.
12182         (extern debug (const bitmap_head_def *)): New.
12183         * cfg.c
12184         (debug (edge_def &)): New.
12185         (debug (edge_def *)): New.
12186         * cfghooks.c
12187         (debug (basic_block_def &)): New.
12188         (debug (basic_block_def *)): New.
12189         * dumpfile.h
12190         (dump_node (const_tree, int, FILE *)): Correct source file.
12191         * dwarf2out.c
12192         (debug (die_struct &)): New.
12193         (debug (die_struct *)): New.
12194         * dwarf2out.h
12195         (extern debug (die_struct &)): New.
12196         (extern debug (die_struct *)): New.
12197         * gimple-pretty-print.c
12198         (debug (gimple_statement_d &)): New.
12199         (debug (gimple_statement_d *)): New.
12200         * gimple-pretty-print.h
12201         (extern debug (gimple_statement_d &)): New.
12202         (extern debug (gimple_statement_d *)): New.
12203         * ira-build.c
12204         (debug (ira_allocno_copy &)): New.
12205         (debug (ira_allocno_copy *)): New.
12206         (debug (ira_allocno &)): New.
12207         (debug (ira_allocno *)): New.
12208         * ira-int.h
12209         (extern debug (ira_allocno_copy &)): New.
12210         (extern debug (ira_allocno_copy *)): New.
12211         (extern debug (ira_allocno &)): New.
12212         (extern debug (ira_allocno *)): New.
12213         * ira-lives.c
12214         (debug (live_range &)): New.
12215         (debug (live_range *)): New.
12216         * lra-int.h
12217         (debug (lra_live_range &)): New.
12218         (debug (lra_live_range *)): New.
12219         * lra-lives.c
12220         (debug (lra_live_range &)): New.
12221         (debug (lra_live_range *)): New.
12222         * omega.c
12223         (debug (omega_pb_d &)): New.
12224         (debug (omega_pb_d *)): New.
12225         * omega.h
12226         (extern debug (omega_pb_d &)): New.
12227         (extern debug (omega_pb_d *)): New.
12228         * print-rtl.c
12229         (debug (const rtx_def &)): New.
12230         (debug (const rtx_def *)): New.
12231         * print-tree.c
12232         (debug_tree (tree): Move within file.
12233         (debug_raw (const tree_node &)): New.
12234         (debug_raw (const tree_node *)): New.
12235         (dump_tree_via_hooks (const tree_node *, int)): New.
12236         (debug (const tree_node &)): New.
12237         (debug (const tree_node *)): New.
12238         (debug_verbose (const tree_node &)): New.
12239         (debug_verbose (const tree_node *)): New.
12240         (debug_head (const tree_node &)): New.
12241         (debug_head (const tree_node *)): New.
12242         (debug_body (const tree_node &)): New.
12243         (debug_body (const tree_node *)): New.
12244         (debug_vec_tree (tree): Move and reimplement in terms of dump.
12245         (debug (vec<tree, va_gc> &)): New.
12246         (debug (vec<tree, va_gc> *)): New.
12247         * rtl.h
12248         (extern debug (const rtx_def &)): New.
12249         (extern debug (const rtx_def *)): New.
12250         * sbitmap.c
12251         (debug_raw (simple_bitmap_def &)): New.
12252         (debug_raw (simple_bitmap_def *)): New.
12253         (debug (simple_bitmap_def &)): New.
12254         (debug (simple_bitmap_def *)): New.
12255         * sbitmap.h
12256         (extern debug (simple_bitmap_def &)): New.
12257         (extern debug (simple_bitmap_def *)): New.
12258         (extern debug_raw (simple_bitmap_def &)): New.
12259         (extern debug_raw (simple_bitmap_def *)): New.
12260         * sel-sched-dump.c
12261         (debug (vinsn_def &)): New.
12262         (debug (vinsn_def *)): New.
12263         (debug_verbose (vinsn_def &)): New.
12264         (debug_verbose (vinsn_def *)): New.
12265         (debug (expr_def &)): New.
12266         (debug (expr_def *)): New.
12267         (debug_verbose (expr_def &)): New.
12268         (debug_verbose (expr_def *)): New.
12269         (debug (vec<rtx> &)): New.
12270         (debug (vec<rtx> *)): New.
12271         * sel-sched-dump.h
12272         (extern debug (vinsn_def &)): New.
12273         (extern debug (vinsn_def *)): New.
12274         (extern debug_verbose (vinsn_def &)): New.
12275         (extern debug_verbose (vinsn_def *)): New.
12276         (extern debug (expr_def &)): New.
12277         (extern debug (expr_def *)): New.
12278         (extern debug_verbose (expr_def &)): New.
12279         (extern debug_verbose (expr_def *)): New.
12280         (extern debug (vec<rtx> &)): New.
12281         (extern debug (vec<rtx> *)): New.
12282         * sel-sched-ir.h
12283         (_list_iter_cond_expr): Make inline instead of static.
12284         * sreal.c
12285         (debug (sreal &)): New.
12286         (debug (sreal *)): New.
12287         * sreal.h
12288         (extern debug (sreal &)): New.
12289         (extern debug (sreal *)): New.
12290         * tree.h
12291         (extern debug_raw (const tree_node &)): New.
12292         (extern debug_raw (const tree_node *)): New.
12293         (extern debug (const tree_node &)): New.
12294         (extern debug (const tree_node *)): New.
12295         (extern debug_verbose (const tree_node &)): New.
12296         (extern debug_verbose (const tree_node *)): New.
12297         (extern debug_head (const tree_node &)): New.
12298         (extern debug_head (const tree_node *)): New.
12299         (extern debug_body (const tree_node &)): New.
12300         (extern debug_body (const tree_node *)): New.
12301         (extern debug (vec<tree, va_gc> &)): New.
12302         (extern debug (vec<tree, va_gc> *)): New.
12303         * tree-cfg.c
12304         (debug (struct loop &)): New.
12305         (debug (struct loop *)): New.
12306         (debug_verbose (struct loop &)): New.
12307         (debug_verbose (struct loop *)): New.
12308         * tree-dump.c: Add header dependence.
12309         * tree-flow.h
12310         (extern debug (struct loop &)): New.
12311         (extern debug (struct loop *)): New.
12312         (extern debug_verbose (struct loop &)): New.
12313         (extern debug_verbose (struct loop *)): New.
12314         * tree-data-ref.c
12315         (debug (data_reference &)): New.
12316         (debug (data_reference *)): New.
12317         (debug (vec<data_reference_p> &)): New.
12318         (debug (vec<data_reference_p> *)): New.
12319         (debug (vec<ddr_p> &)): New.
12320         (debug (vec<ddr_p> *)): New.
12321         * tree-data-ref.h
12322         (extern debug (data_reference &)): New.
12323         (extern debug (data_reference *)): New.
12324         (extern debug (vec<data_reference_p> &)): New.
12325         (extern debug (vec<data_reference_p> *)): New.
12326         (extern debug (vec<ddr_p> &)): New.
12327         (extern debug (vec<ddr_p> *)): New.
12328         * tree-ssa-alias.c
12329         (debug (pt_solution &)): New.
12330         (debug (pt_solution *)): New.
12331         * tree-ssa-alias.h
12332         (extern debug (pt_solution &)): New.
12333         (extern debug (pt_solution *)): New.
12334         * tree-ssa-alias.c
12335         (debug (_var_map &)): New.
12336         (debug (_var_map *)): New.
12337         (debug (tree_live_info_d &)): New.
12338         (debug (tree_live_info_d *)): New.
12339         * tree-ssa-alias.h
12340         (extern debug (_var_map &)): New.
12341         (extern debug (_var_map *)): New.
12342         (extern debug (tree_live_info_d &)): New.
12343         (extern debug (tree_live_info_d *)): New.
12345 2013-03-28  Jan Hubicka  <jh@suse.cz>
12347         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12349 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
12351         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12352         record only when desired or required.
12354 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
12356         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12357         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
12359 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12361         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12362         (*andsi3_compare0_uxtw): New pattern.
12363         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12364         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12366 2013-03-28  Jan Hubicka  <jh@suse.cz>
12368         * data-streamer-in.c (streamer_read_gcov_count): New function.
12369         * gimple-streamer-out.c: Include value-prof.h.
12370         (output_gimple_stmt): Output histogram.
12371         (output_bb): Use streamer_write_gcov_count.
12372         * value-prof.c: Include data-streamer.h
12373         (dump_histogram_value): Add HIST_TYPE_MAX.
12374         (stream_out_histogram_value): New function.
12375         (stream_in_histogram_value): New function.
12376         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12377         (stream_out_histogram_value, stream_in_histogram_value): Declare.
12378         * data-streamer-out.c (streamer_write_gcov_count): New function.
12379         (streamer_write_gcov_count_stream): New function.
12380         * lto-cgraph.c (lto_output_edge): Update counter streaming.
12381         (lto_output_node): Likewise.
12382         (input_node, input_edge): Likewise.
12383         * lto-streamer-out.c (output_cfg): Update streaming.
12384         * lto-streamer-in.c (input_cfg): Likewise.
12385         * data-streamer.h (streamer_write_gcov_count,
12386         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12387         * gimple-streamer-in.c: Include value-prof.h
12388         (input_gimple_stmt): Input histograms.
12389         (input_bb): Update profile streaming.
12391 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12393         * genmodes.c (emit_max_int): New function.
12394         (emit_insn_modes_h): Added call to emit_max_function.
12395         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12396         Added doc.
12397         * machmode.def: Fixed comment.
12399 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12401         * combine.c (try_combine): Removed useless assert.
12402         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12404 2013-03-28  Marek Polacek  <polacek@redhat.com>
12405             Richard Biener  <rguenther@suse.de>
12407         PR tree-optimization/56695
12408         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12409         build signed result of a vector comparison.
12410         * tree-cfg.c (verify_gimple_comparison): Check that a result
12411         of a vector comparison has signed type.
12413 2013-03-28  Richard Biener  <rguenther@suse.de>
12415         PR tree-optimization/37021
12416         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12417         do not restrict gaps between groups.
12418         * tree-vect-stmts.c (vectorizable_load): Properly account for
12419         a gap between groups.
12421 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
12423         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12424         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12425         is not enabled.
12427 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
12429         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12430         * doc/extend.texi (Named Address Spaces): Ditto.
12431         (Variable Attributes): Ditto.
12433 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12435         * config.build: Add support for cygwin x64 target.
12436         * config.gcc: Likewise.
12437         * config.host: Likewise.
12438         * configure.ac: Likewise
12439         * configure: Regenerated.
12441 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12443         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12444         * config/i386/t-cygwin-w64: New file.
12445         * config/i386/cygwin-w64.h: New file.
12446         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12447         and add support for x64-cygwin target.
12448         (CPP_SPEC): Likewise.
12449         (CXX_WRAP_SPEC_LIST): Undefine before define.
12450         (LIBGCJ_SONAME): Use 15 as version.
12452 2013-03-27  Richard Biener  <rguenther@suse.de>
12454         PR tree-optimization/56716
12455         * tree-ssa-structalias.c (perform_var_substitution): Adjust
12456         dumping for ref nodes.
12458 2013-03-27  Martin Jambor  <mjambor@suse.cz>
12460         PR tree-optimization/55334
12461         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12462         restricted pointers to arrays.
12464 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12466         * Makefile.in (.SUFFIXES): Add .cc.
12467         (.c.o): Apply same recipe for implicit rule .cc.o.
12469 2013-03-27  Richard Biener  <rguenther@suse.de>
12471         PR tree-optimization/37021
12472         * tree-vect-data-refs.c (vect_check_strided_load): Allow
12473         REALPART/IMAGPART_EXPRs around the supported refs.
12474         * tree-ssa-structalias.c (find_func_aliases): Assume that
12475         floating-point values are not used to transfer pointers.
12477 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12479         * target.def (TARGET_HAS_IFUNC_P): New target hook.
12480         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12481         * doc/tm.texi: Regenerate.
12482         * targhooks.h (default_has_ifunc_p): New.
12483         * targhooks.c (default_has_ifunc_p): Ditto.
12484         * config/linux-protos.h: New file.
12485         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12486         this hook for linux which disables support of indirect functions in
12487         android.
12488         * config/linux-android.c: New file.
12489         * config/t-linux-android.c: Ditto.
12490         * config.gcc: Added new object file linux-android.o.
12491         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
12492         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12493         * varasm.c (do_assemble_alias): Likewise.
12494         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12495         doesn't support indirect functions.
12496         * configure: Regenerate.
12498 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
12500         PR target/56102
12501         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12502         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12503         mult-word mode.
12505 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12507         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12509 2013-03-27  Terry Guo  <terry.guo@arm.com>
12511         * config/arm/arm-cores.def: Added core cortex-r7.
12512         * config/arm/arm-tune.md: Regenerated.
12513         * config/arm/arm-tables.opt: Regenerated.
12514         * doc/invoke.texi: Added entry for core cortex-r7.
12516 2013-03-27  Walter Lee  <walt@tilera.com>
12518         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12519         double-decrement of next_scratch_regno.
12521 2013-03-27  Walter Lee  <walt@tilera.com>
12523         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12524         input operands.
12525         (insn_v1mulus): Ditto.
12526         (insn_v2muls): Ditto.
12528 2013-03-27  Walter Lee  <walt@tilera.com>
12530         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12531         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12533 2013-03-27  Walter Lee  <walt@tilera.com>
12535         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12536         (*sibcall_value): Ditto.
12538 2013-03-27  Walter Lee  <walt@tilera.com>
12540         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12541         (insn_mnz_v8qi): ... this ...
12542         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
12543         vector equivalent.
12544         (insn_v<n>mnz): Replaced by ...
12545         (insn_v1mnz): ... this ...
12546         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
12547         equivalent.
12548         (insn_mz_<mode>): Replaced by ...
12549         (insn_mz_v8qi): ... this ...
12550         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
12551         vector equivalent.
12552         (insn_v<n>mz): Replaced by ...
12553         (insn_v1mz): ... this ...
12554         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
12555         equivalent.
12557 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
12559         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12561 2013-03-26  Roland McGrath  <mcgrathr@google.com>
12563         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12564         than fprintf with a non-constant, non-format string.
12566 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12568         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12569         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
12570         operand 0 predicate.
12571         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12572         attribute.  Use general_x64nomem_operand as operand 1 predicate.
12573         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12574         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12575         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12576         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12577         (mov<mode>_insv_1): Remove expander.  Merge insn with
12578         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12579         Use general_x64nomem_operand as operand 1 predicate.
12580         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12581         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12582         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12583         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12584         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
12585         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12586         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12587         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12588         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
12589         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12590         (general_x64nomem_operand): Ditto.
12592 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12594         * config/rtems.opt: Add -pthread option.
12596 2013-03-26  Richard Biener  <rguenther@suse.de>
12598         * alias.c (find_base_term): Avoid redundant and not used recursion.
12599         (base_alias_check): Get the initial base term from the caller.
12600         (true_dependence_1): Compute and pass base terms to base_alias_check.
12601         (write_dependence_p): Likewise.
12602         (may_alias_p): Likewise.
12604 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
12606         * config/aarch64/aarch64.c (aarch64_classify_address): Support
12607         PC-relative load in SI modes and above only.
12609 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
12611         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12612         * config/arm/iwmmxt.md (WCGR0): Update.
12613         (WCGR1, WCGR2, WCGR3): Likewise.
12615 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12617         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
12618         Use x64 and nox64 isa attributes.
12620 2013-03-26  Richard Biener  <rguenther@suse.de>
12622         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
12623         alignment computations and rely on get_object_alignment_1
12624         for the !TYPE_P case.
12625         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
12627 2013-03-26  Walter Lee  <walt@tilera.com>
12629         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
12630         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
12632 2013-03-25  Jeff Law  <law@redhat.com>
12634         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
12635         check for INTEGRAL_TYPE_P that was missing due to checking in
12636         wrong version of prior patch.
12638 2013-03-25  Walter Lee  <walt@tilera.com>
12640         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
12641         TILEGX_INSN_SHUFFLEBYTES1.
12642         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
12643         shufflebytes1.
12644         (tilegx_builtins): Ditto.
12645         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
12647 2013-03-25  Walter Lee  <walt@tilera.com>
12649         * config/tilegx/tilegx.md (floatsisf2): New pattern.
12650         (floatunssisf2): New pattern.
12651         (floatsidf2): New pattern.
12652         (floatunssidf2): New pattern.
12654 2013-03-25  Walter Lee  <walt@tilera.com>
12656         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
12657         tests for constraint J, K, N, P.
12659 2013-03-25  Walter Lee  <walt@tilera.com>
12661         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
12662         Use indirect/pcrel encoding.
12663         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
12664         Ditto.
12666 2013-03-25  Steve Ellcey  <sellcey@mips.com>
12668         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
12669         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
12670         * config/mips/mips.c (mips_option_override): Set IMADD default.
12671         * config/mips/mips.h (PTF_AVOID_IMADD): New.
12672         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
12673         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
12674         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
12675         * doc/invoke.texi (-mimadd/-mno-imadd): New.
12677 2013-03-25  Jeff Law  <law@redhat.com>
12679         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
12680         slightly to avoid creating and folding useless trees.  Simplify
12681         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
12683 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
12685         * config/i386/i386.md (*zero_extendsidi2): Merge with
12686         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
12687         * config/i386/predicates.md (x86_64_zext_operand): Rename from
12688         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
12689         targets.  Clarify comment.
12691 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12693         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
12694         pass-through jump functions differently.
12695         (ipa_read_jump_function): Likewise.  Also use setter functions to set
12696         up jump functions.
12698 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12700         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
12701         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
12702         process it.
12703         (ipa_get_indirect_edge_target): New function.
12704         (devirtualization_time_bonus): New parameter known_aggs, pass it to
12705         ipa_get_indirect_edge_target.  Update all callers.
12706         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
12707         ipa_get_indirect_edge_target_1 instead of calling
12708         ipa_get_indirect_edge_target.
12709         (create_specialized_node): Pass aggvlas to
12710         ipcp_discover_new_direct_edges.
12712 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12714         * config/arm/arm.md (f_sels, f_seld): New types.
12715         (*cmov<mode>): New pattern.
12716         * config/arm/predicates.md (arm_vsel_comparison_operator): New
12717         predicate.
12719 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12721         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
12722         POSIX-printf for mingw-hosted  builds.
12724 2013-03-25  Richard Biener  <rguenther@suse.de>
12726         PR middle-end/56694
12727         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
12728         must-not-throw stmt location.
12730 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12732         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
12733         Emit load-acquire versions when acq is true.
12734         (arm_emit_store_exclusive): Add rel parameter.
12735         Emit store-release versions when rel is true.
12736         (arm_split_compare_and_swap): Use acquire-release instructions
12737         instead.
12738         of barriers when appropriate.
12739         (arm_split_atomic_op): Likewise.
12740         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
12741         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
12742         (VUNSPEC_SLX): Likewise.
12743         (VUNSPEC_LDA): Likewise.
12744         (VUNSPEC_STL): Likewise.
12745         * config/arm/sync.md (atomic_load<mode>): New pattern.
12746         (atomic_store<mode>): Likewise.
12747         (arm_load_acquire_exclusive<mode>): Likewise.
12748         (arm_load_acquire_exclusivesi): Likewise.
12749         (arm_load_acquire_exclusivedi): Likewise.
12750         (arm_store_release_exclusive<mode>): Likewise.
12752 2013-03-25  Catherine Moore  <clm@codesourcery.com>
12754         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
12755         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
12756         * config/mip/predicates.md (lwsp_swsp_operand,
12757         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
12758         sb16_operand, db4_operand, db7_operand, ib3_operand,
12759         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
12760         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
12761         andi16_operand): New predicates.
12762         * config/mips/mips.md (compression): New attribute.
12763         (enabled): New attribute.
12764         (length): Consider compression in computing length.
12765         (shift_compression): New code attribute.
12766         (*add<mode>3): New operands. Record compression.
12767         (sub<mode>3): Likewise.
12768         (one_cmpl<mode>2): Likewise.
12769         (*and<mode>3): Likewise.
12770         (*ior<mode>3): Likewise.
12771         (unnamed pattern for xor): Likewise.
12772         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
12773         (*<optab><mode>3): Likewise.
12774         (*mov<mode>_internal: Likewise.
12775         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
12776         (mips_unsigned_immediate_p): New.
12777         (umips_lwsp_swsp_address_p): New.
12778         (m16_based_address_p): New.
12779         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
12780         (mips_unsigned_immediate_p): New prototype.
12781         (lwsp_swsp_address_p): New prototype.
12782         (m16_based_address_p): New prototype.
12783         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
12784         (mips_signed_immediate_p): New function.
12785         (m16_based_address_p): New function.
12786         (lwsp_swsp_address_p): New function.
12787         (mips_print_operand_punctuation): Recognize short delay slot insns
12788         for microMIPS.add<mode>3"
12790 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12792         PR target/56720
12793         * config/arm/iterators.md (v_cmp_result): New mode attribute.
12794         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
12796 2013-03-25  Richard Biener  <rguenther@suse.de>
12798         PR tree-optimization/56689
12799         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
12800         any edge.
12802 2013-03-25  Richard Biener  <rguenther@suse.de>
12804         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
12805         of bitmap.
12806         (memory_references): Likewise.
12807         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
12808         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
12809         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
12810         (gather_mem_refs_in_loops): Fold into ...
12811         (analyze_memory_references): ... this.  Move initialization
12812         to tree_ssa_lim_initialize.
12813         (fill_always_executed_in): Rename to ...
12814         (fill_always_executed_in_1): ... this.
12815         (fill_always_executed_in): Move contains_call computation to
12816         this new function from ...
12817         (tree_ssa_lim_initialize): ... here.
12818         (tree_ssa_lim): Call fill_always_executed_in.
12820 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12822         * postreload.c (reload_combine): Fix code detecting returns.
12824 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12826         * function.c (emit_use_return_register_into_block): On cc0 targets,
12827         do not emit the sequence between cc0 setter and user.
12829 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12831         * config/i386/predicates.md (local_symbolic_operand): Interpret
12832         dll-imported symbols as none-local.
12834 2013-03-25  Richard Biener  <rguenther@suse.de>
12836         * tree-ssa-loop-im.c (struct depend): Remove.
12837         (struct lim_aux_data): Make depends a vec of gimples.
12838         (free_lim_aux_data): Adjust.
12839         (add_dependency): Likewise.
12840         (set_level): Likewise.
12842 2013-03-25  Richard Biener  <rguenther@suse.de>
12844         PR middle-end/56434
12845         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
12846         the pointer returned by calls with ECF_MALLOC set.
12848 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12850         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
12852 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12854         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
12855         using MMXMODE mode iterator.
12856         (*move<mode>_internal): Merge with *movv2sf_internal and
12857         *movv2sf_internal_rex64 using MMXMODE mode iterator.
12859 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12861         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
12862         (record_last_mem_set_info): Likewise.
12864         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
12865         of XNEWVEC followed by memset.
12866         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
12868 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12870         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
12871         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
12872         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
12873         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
12874         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
12875         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
12876         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
12877         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
12878         BARRIER_P instead of GET_CODE.
12880 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12882         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
12883         inaccuracy in the probing code.
12885         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
12886         (ctrapdi4): Likewise.
12888 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12890         * calls.c (expand_call): Add missing guard to code handling return
12891         of non-BLKmode structures in MSB.
12892         * function.c (expand_function_end): Likewise.
12894 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12896         * combine.c (try_combine): Adjust comment.  Do not add the set of
12897         insn #0 if the destination indirectly is set or dies in insn #2.
12898         Tidy up code to distribute a new note.
12900 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
12902         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
12903         also for alternatives 16 and 17.
12905 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
12907         * config/i386/sse.md (*mov<mode>_internal): Merge with
12908         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
12909         Emit insn template depending on type attribute.  Use
12910         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
12911         movd instead of movq mnemonic for interunit moves.  Rewrite mode
12912         attribute calculation.  Remove unit attribute calculation.
12913         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
12914         Set prefix_data16 attribute for DImode ssemov types.
12915         Use Ym instead of y for SSE-MMX conversion alternatives.
12916         Reorder operand constraints.
12918 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
12920         * df.h (df_insn_delete): Adjust prototype.
12921         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
12922         and let it decide whether mark the basic block dirty.
12923         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
12924         * df-scan.c (df_insn_info_delete): New helper function, split
12925         off from df_insn_delete.
12926         (df_scan_free_bb_info): Use it.
12927         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
12928         Likewise.
12929         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
12930         that the insn is actually an insn and it has a non-NULL basic block.
12931         Do not mark basic block dirty if only deleting a DEBUG_INSN.
12933 2013-03-22  Richard Biener  <rguenther@suse.de>
12935         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
12936         dep_ref members.
12937         (mem_ref_alloc): Do not allocate them.
12938         (refs_independent_p): Do not query or maintain a cache.
12940 2013-03-22  Richard Biener  <rguenther@suse.de>
12942         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
12943         (gather_mem_refs_in_loops): Do not compute it.
12944         (analyze_memory_references): Do not allocate it.
12945         (tree_ssa_lim_finalize): Do not free it.
12946         (for_all_locs_in_loop): Do not query all_refs_in_loop.
12948 2013-03-22  Richard Biener  <rguenther@suse.de>
12950         * is-a.h (as_a): Use gcc_checking_assert.
12952 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
12954         * config/aarch64/aarch64.c (aarch64_print_operand): New
12955         format specifier for printing a constant in hex.
12956         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
12957         format specifier for printing second operand.
12959 2013-03-22  Richard Biener  <rguenther@suse.de>
12961         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
12962         bitmaps.
12963         (gather_mem_refs_in_loops): Perform store accumulation here.
12964         (create_vop_ref_mapping_loop): Remove.
12965         (create_vop_ref_mapping): Likewise.
12966         (analyze_memory_references): Initialize refs_stored_in_loop.
12967         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
12968         (record_indep_loop): Remove.
12969         (record_dep_loop): New function.
12970         (ref_indep_loop_p_1): Adjust to only walk over references
12971         in the loop, not its subloops.
12972         (ref_indep_loop_p): Rename to ...
12973         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
12974         maintaining a more fine-grained cache.
12975         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
12976         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
12978 2013-03-22  Richard Biener  <rguenther@suse.de>
12980         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
12981         (struct mem_ref): Make accesses_in_loop a vec of a vec of
12982         aggregate mem_ref_loc.
12983         (free_mem_ref_locs): Inline into ...
12984         (memref_free): ... this and adjust.
12985         (mem_ref_alloc): Adjust.
12986         (mem_ref_locs_alloc): Remove.
12987         (record_mem_ref_loc): Adjust.
12988         (get_all_locs_in_loop): Rewrite into ...
12989         (for_all_locs_in_loop): ... this iterator.
12990         (rewrite_mem_ref_loc): New functor.
12991         (rewrite_mem_refs): Use for_all_locs_in_loop.
12992         (sm_set_flag_if_changed): New functor.
12993         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
12994         (ref_always_accessed): New functor.
12995         (ref_always_accessed_p): Use for_all_locs_in_loop.
12997 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
12999         * tree-pass.h (PROP_gimple_lvec): New.
13000         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13001         (init_optimization_passes): Move pass_lower_vector.
13002         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13003         PROP_gimple_lvec.
13004         (pass_lower_vector): Provide PROP_gimple_lvec.
13005         (pass_lower_vector_ssa): Likewise.
13006         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13008 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13010         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13012 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13014         * config/i386/i386.md (*movdi_internal): Disparage slightly
13015         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13016         conversion alternatives.
13018 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13020         PR middle-end/48087
13021         * diagnostic.def (DK_WERROR): New kind.
13022         * diagnostic.h (werrorcount): Define.
13023         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13024         promoted to DK_ERROR, increment DK_WERROR counter instead of
13025         DK_ERROR counter.
13026         * toplev.c (toplev_main): Call print_ignored_options even if
13027         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13028         even if just werrorcount is non-zero.
13030         PR debug/55608
13031         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13032         on failure.
13033         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13034         (string_cst_pool_decl): New function.
13035         (optimize_one_addr_into_implicit_ptr): New function.
13036         (resolve_addr_in_expr): Optimize DWARF location expression
13037         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13038         which doesn't live in memory, but has DW_AT_location or
13039         DW_AT_const_value, or refers to a string literal, into
13040         DW_OP_GNU_implicit_pointer.
13041         (optimize_location_into_implicit_ptr): New function.
13042         (resolve_addr): If removing DW_AT_location of a variable because
13043         it was DW_OP_addr of address of the variable, but the variable doesn't
13044         live in memory, try to emit const value attribute for the initializer.
13046 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13048         * tree.h (VECTOR_TYPE_P): New macro.
13049         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13050         TYPE_MODE): Use it.
13051         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13052         VEC_COND_EXPR cannot be lvalues.
13053         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13055 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13057         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13058         Restrict the transformation to equal modes.
13060 2013-03-21  Richard Biener  <rguenther@suse.de>
13062         PR tree-optimization/39326
13063         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13064         (MEM_ANALYZABLE): Adjust.
13065         (record_mem_ref_loc): Move bitmap ops ...
13066         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13067         unanalyzable refs, do not record locations for it.
13068         (analyze_memory_references): Allocate ref zero as shared
13069         unanalyzable ref.
13070         (refs_independent_p): Do not test for unanalyzed mems here.
13071         (ref_indep_loop_p_1): Special-case disambiguation against
13072         the unanalyzed ref.
13073         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13075 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13077         * config/arm/arm-protos.h (tune_params): Add
13078         prefer_neon_for_64bits field.
13079         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13080         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13081         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13082         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13083         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13084         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13085         (arm_option_override): Handle -mneon-for-64bits new option.
13086         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13087         (prefer_neon_for_64bits): Declare new variable.
13088         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13089         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13090         (arch_enabled): Handle new arch types. Remove support for onlya8
13091         and nota8.
13092         (one_cmpldi2): Use new arch names.
13093         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13094         * config/arm/arm.opt (mneon-for-64bits): Add option.
13095         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13096         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13097         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13098         of onlya8.
13099         * doc/invoke.texi (-mneon-for-64bits): Document.
13101 2013-03-21  Richard Biener  <rguenther@suse.de>
13103         PR tree-optimization/39326
13104         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13105         (sort_bbs_in_loop_postorder_cmp): New function.
13106         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13108 2013-03-21  Richard Biener  <rguenther@suse.de>
13110         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13111         (vect_insert_into_interleaving_chain): Likewise.
13112         (vect_drs_dependent_in_basic_block): Inline ...
13113         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13114         split out from ...
13115         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13116         (vect_check_interleaving): Simplify.
13117         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13118         (vect_slp_analyze_data_ref_dependences): ... this new function.
13119         (dr_group_sort_cmp): New function.
13120         (vect_analyze_data_ref_accesses): Compute data-reference groups
13121         here instead of in vect_analyze_data_ref_dependence.  Use
13122         a more efficient algorithm.
13123         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13124         vect_slp_analyze_data_ref_dependences.  Call
13125         vect_analyze_data_ref_accesses earlier.
13126         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13127         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13128         (vect_slp_analyze_data_ref_dependences): New prototype.
13130 2013-03-21  Richard Biener  <rguenther@suse.de>
13132         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13133         ref is stored in the loop.
13134         (find_refs_for_sm): Walk only over all stores.
13135         (store_motion_loop): Allocate from lim_bitmap_obstack.
13136         (store_motion): Likewise.
13138 2013-03-21  Richard Biener  <rguenther@suse.de>
13140         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13141         Update virtual SSA form.
13143 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13145         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13146         * configure: Regenerate.
13147         * config.in: Regenerate.
13148         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13149         if !HAVE_LD_EH_FRAME_CIEV3.
13151 2013-03-21  Richard Biener  <rguenther@suse.de>
13153         * tree-cfg.c (verify_expr_no_block): New function.
13154         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13155         nor DECL_VALUE_EXPR have locations with associated blocks.
13156         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13157         (clear_unused_block_pointer): Remove code dealing with
13158         blocks in DECL_DEBUG_EXPR locations.
13160 2013-03-21  Richard Biener  <rguenther@suse.de>
13162         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13163         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
13164         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13165         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13166         instead of DECL_DEBUG_EXPR_IS_FROM.
13167         * gimplify.c (gimplify_modify_expr): Likewise.
13168         * tree-cfg.c (verify_expr_location_1): Likewise.
13169         * tree-complex.c (create_one_component_var): Likewise.
13170         * tree-sra.c (create_access_replacement): Likewise.
13171         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13172         (clear_unused_block_pointer): Likewise.
13173         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13174         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13175         * var-tracking.c (var_debug_decl): Likewise.
13176         (track_expr_p): Likewise.
13177         * tree-inline.c (add_local_variables): Likewise.  Set
13178         DECL_HAS_DEBUG_EXPR_P after copying it.
13179         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13180         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
13182 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13184         PR bootstrap/56656
13185         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13186         * configure: Regenerate.
13187         * config.in: Regenerate.
13188         * config/i386/i386.md (*movdf_internal): Use
13189         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13190         movd instead of movq mnemonic for interunit moves.
13191         (*movdi_internal): Ditto.
13193 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13195         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13196         (abd<mode>_3): New pattern.
13197         (aba<mode>_3): New pattern.
13198         (fabd<mode>_3): New pattern.
13200 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13202         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13203         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13204         occurrence of REGISTER_PREFIX as its empty string.
13206 2013-03-20  Jeff Law  <law@redhat.com>
13208         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13209         addititional equivalences for equality comparisons between an SSA_NAME
13210         and a constant where the SSA_NAME was set from a widening conversion.
13212 2013-03-20  Walter Lee  <walt@tilera.com>
13214         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13216 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13218         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13219         depending on type attribute.
13220         (*movti_internal): Ditto.
13221         (*movtf_internal): Ditto.
13222         (*movxf_internal): Ditto.
13223         (*movdf_internal): Ditto.
13224         (*movsf_internal): Ditto.
13226 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13228         * config/i386/i386.md (*movti_internal): Set prefix attribute to
13229         maybe_vex for sselog1 and ssemov types.
13230         (*movdi_internal): Reorder operand constraints.
13231         (*movsi_internal): Ditto.  Set prefix attribute to
13232         maybe_vex for sselog1 and ssemov types.
13233         (*movtf_internal): Set prefix attribute to maybe_vex
13234         for sselog1 and ssemov types.
13235         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
13236         DImode ssemov types.  Reorder operand constraints.
13237         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
13238         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
13239         attribute for SImode ssemov types.  Reorder operand constraints.
13241 2013-03-20  Martin Jambor  <mjambor@suse.cz>
13243         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13244         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13246 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
13248         * config/rs6000/predicates.md (indexed_address, update_address_mem
13249         update_indexed_address_mem): New predicates.
13250         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13251         attribute for load/store instructions.
13252         * config/rs6000/dfp.md (movsd_store): Likewise.
13253         (movsd_load): Likewise.
13254         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13255         (unnamed HI->DI extend define_insn): Likewise.
13256         (unnamed SI->DI extend define_insn): Likewise.
13257         (unnamed QI->SI extend define_insn): Likewise.
13258         (unnamed QI->HI extend define_insn): Likewise.
13259         (unnamed HI->SI extend define_insn): Likewise.
13260         (unnamed HI->SI extend define_insn): Likewise.
13261         (extendsfdf2_fpr): Likewise.
13262         (movsi_internal1): Likewise.
13263         (movsi_internal1_single): Likewise.
13264         (movhi_internal): Likewise.
13265         (movqi_internal): Likewise.
13266         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13267         attribute for load/store instructions.
13268         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13269         instructions.
13270         (mov<mode>_softfloat): Likewise.
13271         (mov<mode>_hardfloat32): Likewise.
13272         (mov<mode>_hardfloat64): Likewise.
13273         (mov<mode>_softfloat64): Likewise.
13274         (movdi_internal32): Likewise.
13275         (movdi_internal64): Likewise.
13276         (probe_stack_<mode>): Likewise.
13278 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
13280         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13281         floating point, and decimal floating point to reload iterator.
13283         * config/rs6000/constraints.md (wl constraint): New constraints to
13284         return FLOAT_REGS if certain options are used to reduce the number
13285         of separate patterns that exist in the file.
13286         (wx constraint): Likewise.
13287         (wz constraint): Likewise.
13289         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13290         -mdebug=reg, print wg, wl, wx, and wz constraints.
13291         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13292         Initialize the reload functions for 64-bit binary/decimal floating
13293         point types.
13294         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13295         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13296         create the buffer on the stack to overcome not having a 32-bit
13297         load and store.
13298         (rs6000_emit_move): Likewise.
13299         (rs6000_secondary_memory_needed_rtx): Likewise.
13300         (rs6000_alloc_sdmode_stack_slot): Likewise.
13301         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13302         via xxlxor, just like DFmode 0.0.
13304         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13305         define as 1 if we are running on a power7 or newer.
13306         (enum r6000_reg_class_enum): Add new constraints.
13308         * config/rs6000/dfp.md (movsd): Delete, combine with binary
13309         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13310         with other moves by using conditional constraits (wg).  Use LFIWZX
13311         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
13312         (movsd splitter): Likewise.
13313         (movsd_hardfloat): Likewise.
13314         (movsd_softfloat): Likewise.
13316         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13317         binary and decimal floating point moves.
13318         (fmove_ok): New attributes to combine binary and decimal floating
13319         point moves, and to combine power6x (mfpgpr) moves along normal
13320         floating moves.
13321         (real_value_to_target): Likewise.
13322         (f32_lr): Likewise.
13323         (f32_lm): Likewise.
13324         (f32_li): Likewise.
13325         (f32_sr): Likewise.
13326         (f32_sm): Likewise.
13327         (f32_si): Likewise.
13328         (movsf): Combine binary and decimal floating point moves.  Combine
13329         power6x (mfpgpr) moves with other moves by using conditional
13330         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
13331         (mov<mode> for SFmode/SDmode); Likewise.
13332         (SFmode/SDmode splitters): Likewise.
13333         (movsf_hardfloat): Likewise.
13334         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13335         (movsf_softfloat): Likewise.
13336         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13338         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13339         wx and wz constraints.
13341         * config/rs6000/constraints.md (wg constraint): New constraint to
13342         return FLOAT_REGS if -mmfpgpr (power6x) was used.
13344         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13345         constraint.
13347         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13348         -mdebug=reg, print wg, wl, wx, and wz constraints.
13349         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13350         Initialize the reload functions for 64-bit binary/decimal floating
13351         point types.
13352         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13353         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13354         create the buffer on the stack to overcome not having a 32-bit
13355         load and store.
13356         (rs6000_emit_move): Likewise.
13357         (rs6000_secondary_memory_needed_rtx): Likewise.
13358         (rs6000_alloc_sdmode_stack_slot): Likewise.
13359         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13360         via xxlxor, just like DFmode 0.0.
13362         * config/rs6000/dfp.md (movdd): Delete, combine with binary
13363         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13364         with other moves by using conditional constraits (wg).  Use LFIWZX
13365         and STFIWX for loading SDmode on power7.
13366         (movdd splitters): Likewise.
13367         (movdd_hardfloat32): Likewise.
13368         (movdd_softfloat32): Likewise.
13369         (movdd_hardfloat64_mfpgpr): Likewise.
13370         (movdd_hardfloat64): Likewise.
13371         (movdd_softfloat64): Likewise.
13373         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13374         64-bit binary and decimal floating point moves.
13375         (FMOVE64X): Likewise.
13376         (movdf): Combine 64-bit binary and decimal floating point moves.
13377         Combine power6x (mfpgpr) moves with other moves by using
13378         conditional constraits (wg).
13379         (mov<mode> for DFmode/DDmode): Likewise.
13380         (DFmode/DDmode splitters): Likewise.
13381         (movdf_hardfloat32): Likewise.
13382         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13383         (movdf_softfloat32): Likewise.
13384         (movdf_hardfloat64_mfpgpr): Likewise.
13385         (movdf_hardfloat64): Likewise.
13386         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13387         (movdf_softfloat64): Likewise.
13388         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13389         (reload_<mode>_load): Move to later in the file so they aren't in
13390         the middle of the floating point move insns.
13391         (reload_<mode>_store): Likewise.
13393         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13394         constraint.
13396         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13397         constraint if -mdebug=reg.
13398         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13399         Enable using dd reload support if needed.
13401         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13402         binary and decimal floating point moves in rs6000.md.
13403         (movtd_internal): Likewise.
13405         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13406         decimal floating point moves.
13407         (movtf): Likewise.
13408         (movtf_internal): Likewise.
13409         (mov<mode>_internal, TDmode/TFmode): Likewise.
13410         (movtf_softfloat): Likewise.
13411         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13413         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13414         movdi_internal64, using wg constraint for move direct operations.
13415         (movdi_internal64): Likewise.
13417         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13418         MODES_TIEABLE_P for selected modes.  Print the numerical value of
13419         the various virtual registers. Use GPR/FPR first/last values,
13420         instead of hard coding the register numbers.  Print which modes
13421         have reload functions registered.
13422         (rs6000_option_override_internal): If -mdebug=reg, trace the options
13423         settings before/after setting cpu, target and subtarget settings.
13424         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13425         and for secondary reload failures in rs6000_secondary_reload_inner.
13426         (rs6000_secondary_reload_fail): Likewise.
13427         (rs6000_secondary_reload_inner): Likewise.
13429         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13430         macros for first/last GPR and FPR registers.
13431         (LAST_GPR_REGNO): Likewise.
13432         (FIRST_FPR_REGNO): Likewise.
13433         (LAST_FPR_REGNO): Likewise.
13435         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13436         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13437         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13438         (vcond<mode><mode>): Likewise.
13439         (vcondu<mode><mode>): Likewise.
13440         (vector_gtu<mode>): Likewise.
13441         (vector_gte<mode>): Likewise.
13442         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13443         to prevent the compiler from converting DImode operations to TImode.
13444         (ior<mode>3): Likewise.
13445         (and<mode>3): Likewise.
13446         (one_cmpl<mode>2): Likewise.
13447         (nor<mode>3): Likewise.
13448         (andc<mode>3): Likewise.
13450         * config/rs6000/constraints.md (wt constraint): New constraint
13451         that returns VSX_REGS if TImode is allowed in VSX registers.
13453         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13454         constant under VSX.
13456         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13457         similar to TImode, but it is restricted to being in the GPRs.
13459         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13460         TImode to occupy a single VSX register.
13462         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13463         -mvsx-timode for power7/power8.
13464         (power7 cpu): Likewise.
13465         (power8 cpu): Likewise.
13467         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13468         sure that TFmode/TDmode take up two registers if they are ever
13469         allowed in the upper VSX registers.
13470         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13471         registers.
13472         (rs6000_init_hard_regno_mode_ok): Likewise.
13473         (rs6000_debug_reg_global): Add debugging for PTImode and wt
13474         constraint.  Print if LRA is turned on.
13475         (rs6000_option_override_internal): Give an error if -mvsx-timode
13476         and VSX is not enabled.
13477         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
13478         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13479         to reg+offset addressing.  Use PTImode when checking offset
13480         addresses for validity.
13481         (reg_offset_addressing_ok_p): Likewise.
13482         (rs6000_legitimate_offset_address_p): Likewise.
13483         (rs6000_legitimize_address): Likewise.
13484         (rs6000_legitimize_reload_address): Likewise.
13485         (rs6000_legitimate_address_p): Likewise.
13486         (rs6000_eliminate_indexed_memrefs): Likewise.
13487         (rs6000_emit_move): Likewise.
13488         (rs6000_secondary_reload): Likewise.
13489         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
13490         reloads to fpr registers to continue to use reg+offset addressing,
13491         but 64-bit reloads to altivec registers need reg+reg addressing.
13492         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13493         it.  Treat LO_SUM like a PLUS operation.
13494         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13495         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13496         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13497         registers to share a register with a smaller sized type, since VSX
13498         puts scalars in the upper 64-bits.
13499         (print_operand): Add support for PTImode.
13500         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13501         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13502         registers, but don't have arithmetic support.
13503         (rs6000_memory_move_cost): Add test for VSX.
13504         (rs6000_opt_masks): Add -mvsx-timode.
13506         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13507         for TImode.
13508         (VSs): Likewise.
13509         (VSr): Use wt constraint for TImode.
13510         (VSv): Drop TImode support.
13511         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13512         (vsx_movti_64bit): Likewise.
13513         (vsx_movti_32bit): Likewise.
13514         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13515         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13516         one '?' on the appropriate output constraint.  Do not allow TImode
13517         logical operations on 32-bit systems.
13518         (vsx_ior<mode>3): Likewise.
13519         (vsx_xor<mode>3): Likewise.
13520         (vsx_one_cmpl<mode>2): Likewise.
13521         (vsx_nor<mode>3): Likewise.
13522         (vsx_andc<mode>3): Likewise.
13523         (vsx_concat_<mode>): Likewise.
13524         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13526         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13527         OPTION_MASK_VSX_TIMODE.
13528         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13529         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13531         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13532         (TI2 iterator): New iterator for TImode, PTImode.
13533         (wd mode attribute): Add values for vector types.
13534         (movti_string): Replace TI move operations with operations for TImode
13535         and PTImode.  Add support for TImode being allowed in VSX registers.
13536         (mov<mode>_string, TImode/PTImode): Likewise.
13537         (movti_ppc64): Likewise.
13538         (mov<mode>_ppc64, TImode/PTImode): Likewise.
13539         (TI mode splitters): Likewise.
13541         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13542         constraint.
13544 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
13546         PR tree-optimization/56355
13547         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13548         Also handle integers with undefined overflow.
13550 2013-03-20  Catherine Moore  <clm@codesourcery.com>
13551             Maciej W. Rozycki  <macro@codesourcery.com>
13552             Tom de Vries  <tom@codesourcery.com>
13553             Nathan Sidwell <nathan@codesourcery.com>
13554             Iain Sandoe  <iain@codesourcery.com>
13555             Nathan Froyd  <froydnj@codesourcery.com>
13556             Chao-ying Fu <fu@mips.com>
13558         * doc/extend.texi: (micromips, nomicromips, nocompression):
13559         Document new function attributes.
13560         * doc/invoke.texi (minterlink-compressed, mmicromips,
13561         m14k, m14ke, m14kec): Document new options.
13562         (minterlink-mips16): Update documentation.
13563         * doc/md.texi (ZC, ZD): Document new constraints.
13564         * configure.ac (gcc_cv_as_micromips): Check if linker
13565         supports the .set micromips directive.
13566         * configure: Regenerate.
13567         * config.in: Regenerate.
13568         * config/mips/mips-tables.opt: Regenerate.
13569         * config/mips/micromips.md: New file.
13570         * constraints.md (ZC, ZD): New constraints.
13571         * config/mips/predicates.md (movep_src_register): New predicate.
13572         (movep_src_operand): New predicate.
13573         (non_volatile_mem_operand): New predicate.
13574         * config/mips/mips.md (multimem): New type.
13575         (length): Differentiate between 17-bit and 18-bit branch offsets.
13576         (MOVEP1, MOVEP2): New mode iterator.
13577         (mov_<load>l): Use ZC constraint.
13578         (mov_<load>r): Likewise.
13579         (mov_<store>l): Likewise.
13580         (mov_<store>r): Likewise.
13581         (*branch_equality<mode>_inverted): Add microMIPS support.
13582         (*branch_equality<mode>): Likewise.
13583         (*jump_absolute): Likewise.
13584         (indirect_jump_<mode>): Likewise.
13585         (tablejump_<mode>): Likewise.
13586         (<optab>_internal): Likewise.
13587         (sibcall_internal): Likewise.
13588         (sibcall_value_internal): Likewise.
13589         (prefetch): Use constraint ZD.
13590         * config/mips/mips.opt (minterlink-compressed): New option.
13591         (minterlink-mips16): Now an alias for minterlink-compressed.
13592         (mmicromips): New option.
13593         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13594         (compare_and_swap_12): Likewise.
13595         (sync_add<mode>): Likewise.
13596         (sync_<optab>_12): Likewise.
13597         (sync_old_<optab>_12): Likewise.
13598         (sync_new_<optab>_12): Likewise.
13599         (sync_nand_12): Likewise.
13600         (sync_old_nand_12): Likewise.
13601         (sync_new_nand_12): Likewise.
13602         (sync_sub<mode>): Likewise.
13603         (sync_old_add<mode>): Likewise.
13604         (sync_old_sub<mode>): Likewise.
13605         (sync_new_add<mode>): Likewise.
13606         (sync_new_sub<mode>): Likewise.
13607         (sync_<optab><mode>): Likewise.
13608         (sync_old_<optab><mode>): Likewise.
13609         (sync_new_<optab><mode>): Likewise.
13610         (sync_nand<mode>): Likewise.
13611         (sync_old_nand<mode>): Likewise.
13612         (sync_new_nand<mode>): Likewise.
13613         (sync_lock_test_and_set<mode>): Likewise.
13614         (test_and_set_12): Likewise.
13615         (atomic_compare_and_swap<mode>): Likewise.
13616         (atomic_exchange<mode>_llsc): Likewise.
13617         (atomic_fetch_add<mode>_llsc): Likewise.
13618         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
13619         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
13620         (umips_save_restore_pattern_p): Likewise.
13621         (umips_load_store_pair_p): Likewise.
13622         (umips_output_load_store_pair): Likewise.
13623         (umips_movep_target_p): Likewise.
13624         (umips_12bit_offset_address_p): Likewise.
13625         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
13626         (mips_base_mips16): Rename this...
13627         (mips_base_compression_flags): ...to this. Update all uses.
13628         (mips_attribute_table): Add micromips, nomicromips and nocompression.
13629         (mips_mips16_decl_p): Delete.
13630         (mips_nomips16_decl_p): Delete.
13631         (mips_get_compress_on_flags): New function.
13632         (mips_get_compress_off_flags): New function.
13633         (mips_get_compress_mode): New function.
13634         (mips_get_compress_on_name): New function.
13635         (mips_get_compress_off_name): New function.
13636         (mips_insert_attributes): Support multiple compression types.
13637         (mips_merge_decl_attributes): Likewise.
13638         (umips_12bit_offset_address_p): New function.
13639         (mips_start_function_definition): Emit .set micromips directive.
13640         (mips_call_may_need_jalx_p): New function.
13641         (mips_function_ok_for_sibcall): Add microMIPS support.
13642         (mips_print_operand_punctuation): Support short delay slots and
13643         compact jumps.
13644         (umips_swm_mask, umips_swm_encoding): New.
13645         (umips_build_save_restore): New function.
13646         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
13647         (was_mips16_p): Remove.
13648         (old_compression_mode): New.
13649         (mips_set_compression_mode): New function.
13650         (mips_set_current_function): Add microMIPS support.
13651         (mips_option_override): Likewise.
13652         (umips_save_restore_pattern_p): New function.
13653         (umips_output_save_restore): New function.
13654         (umips_load_store_pair_p_1): New function.
13655         (umips_load_store_pair_p): New function.
13656         (umips_output_load_store_pair_1): New function.
13657         (umips_output_load_store_pair): New function.
13658         (umips_movep_target_p) New function.
13659         (mips_prepare_pch_save): Add microMIPS support.
13660         * config/mips/mips.h (TARGET_COMPRESSION): New.
13661         (TARGET_CPU_CPP_BUILTINS): Update macro
13662         to use new compression flags and to support microMIPS.
13663         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
13664         (MIPS_ARCH_FLOAT_SPEC): Likewise.
13665         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
13666         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
13667         (ASM_SPEC): Support mmicromips and mno-micromips.
13668         (M16STORE_REG_P): New macro.
13669         (MIPS_CALL): Support TARGET_MICROMIPS.
13670         (MICROMIPS_J): New macro.
13671         (mips_base_mips16): Rename this...
13672         (mips_base_compression_flags): ...to this.
13673         (UMIPS_12BIT_OFFSET_P): New macro.
13674         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
13675         (MULTILIB_DIRNAMES): Likewise.
13676 2013-03-20  Richard Biener  <rguenther@suse.de>
13678         PR tree-optimization/56661
13679         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
13680         the result does not have to be distinct.
13682 2013-03-20  Richard Biener  <rguenther@suse.de>
13684         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
13685         remap_gimple_op_r.
13687 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13688             Steven Bosscher <steven@gcc.gnu.org>
13690         PR rtl-optimization/56605
13691         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
13693 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13695         PR bootstrap/56656
13696         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
13697         that require movd instead of movq.
13699 2013-03-20  Richard Biener  <rguenther@suse.de>
13701         * tree-ssa-structalias.c (struct variable_info): Add pointer
13702         to the first field of an aggregate with sub-vars.  Make
13703         this and the pointer to the next subfield its ID.
13704         (vi_next): New function.
13705         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
13706         storedanything_id, integer_id): Increment by one.
13707         (new_var_info, get_call_vi, lookup_call_clobber_vi,
13708         get_call_clobber_vi): Adjust.
13709         (solution_set_expand): Simplify and speedup.
13710         (solution_set_add): Inline into ...
13711         (set_union_with_increment): ... this.  Adjust accordingly.
13712         (do_sd_constraint): Likewise.
13713         (do_ds_constraint): Likewise.
13714         (do_complex_constraint): Simplify.
13715         (build_pred_graph): Adjust.
13716         (solve_graph): Likewise.  Simplify and speedup.
13717         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
13718         get_constraint_for_component_ref, get_constraint_for_1,
13719         first_vi_for_offset, first_or_preceding_vi_for_offset,
13720         create_function_info_for, create_variable_info_for_1,
13721         create_variable_info_for, intra_create_variable_infos): Adjust.
13722         (init_base_vars): Push NULL for ID zero.
13723         (compute_points_to_sets): Adjust.
13725 2013-03-20  Richard Biener  <rguenther@suse.de>
13727         * cfgloop.c (verify_loop_structure): Streamline and avoid
13728         ICEing on corrupt loop tree.
13729         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
13730         loop tree.
13732 2013-03-20  Richard Biener  <rguenther@suse.de>
13734         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
13735         check whether an SSA update is needed.
13737 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
13739         * config/mips/constraints.md (T): Rename to...
13740         (Yf): ...this.
13741         (U): Rename to...
13742         (Yd): ...this.
13743         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
13744         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
13746 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13748         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
13749         (*subsi3_carryin_uxtw): Likewise.
13751 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13753         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
13754         (*rorsi3_insn_uxtw): Likewise.
13756 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13758         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
13759         (*extrsi5_insn_uxtw): Likewise.
13761 2013-03-19  Richard Biener  <rguenther@suse.de>
13763         PR tree-optimization/56273
13764         * passes.c (init_optimization_passes): Move second VRP after DOM.
13766 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
13768         * config/i386/i386.md (*movti_internal): Merge from
13769         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
13770         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
13771         nox64 isa attributes.
13773 2013-03-18  Richard Biener  <rguenther@suse.de>
13775         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
13776         (unite): Likewise.
13777         (merge_node_constraints): Likewise.
13778         (build_succ_graph): Likewise.
13779         (valid_graph_edge): Inline into single caller.
13780         (unify_nodes): Likewise.  Use bitmap_set_bit return value
13781         and cache varinfo.
13782         (scc_visit): Fix formatting and variable use.
13783         (do_sd_constraint): Use gcc_checking_assert.
13784         (do_ds_constraint): Likewise.
13785         (do_complex_constraint): Likewise.
13786         (condense_visit): Likewise.  Cleanup.
13787         (dump_pred_graph): New function.
13788         (perform_var_substitution): Dump the pred-graph before
13789         variable substitution.
13790         (find_equivalent_node): Use gcc_checking_assert.
13791         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
13793 2013-03-18  Richard Biener  <rguenther@suse.de>
13795         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
13796         Remove cond_expr_stmt_list argument and do not gimplify the
13797         built expression.
13798         (vect_loop_versioning): Adjust.
13799         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
13800         Cleanup to use less temporaries.
13801         (vect_create_data_ref_ptr): Cleanup.
13803 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13805         PR tree-optimization/56635
13806         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
13807         require types_compatible_p types.
13809 2013-03-18  Nick Clifton  <nickc@redhat.com>
13811         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
13812         spurious backslash.
13814         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
13815         Add missing line to comment describing function.
13817 2013-03-18  Richard Biener  <rguenther@suse.de>
13819         PR tree-optimization/56210
13820         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13821         Handle string / character search functions.
13822         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
13824 2013-03-18  Richard Biener  <rguenther@suse.de>
13826         PR middle-end/56483
13827         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
13828         and implement properly.
13829         * gimple.h (gimple_cond_single_var_p): Remove.
13831 2013-03-18  Richard Biener  <rguenther@suse.de>
13833         * tree-data-ref.h (find_data_references_in_loop): Declare.
13834         * tree-data-ref.c (get_references_in_stmt): Use a stack
13835         vector pre-allocated in the callers.
13836         (find_data_references_in_stmt): Adjust.
13837         (graphite_find_data_references_in_stmt): Likewise.
13838         (create_rdg_vertices): Likewise.
13839         (find_data_references_in_loop): Export.
13840         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13841         Compute dependences here...
13842         (vect_analyze_data_refs): ...not here.  When we encounter
13843         a non-vectorizable data reference in basic-block vectorization
13844         truncate the data reference vector.  Do not bother to
13845         fixup data-dependence information for gather loads.
13846         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
13847         of data references, as reported.
13849 2013-03-18  Richard Biener  <rguenther@suse.de>
13851         PR tree-optimization/3713
13852         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
13853         has_constants and expr.
13854         (stmt_has_constants): Properly valueize SSA names when deciding
13855         whether the stmt has constants.
13857 2013-03-18  Richard Biener  <rguenther@suse.de>
13859         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
13860         whole function when there is nothing to do.
13861         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
13862         * tree-vectorizer.c (vectorize_loops): Update virtual and
13863         loop-closed SSA once.
13864         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
13866 2013-03-18  Richard Biener  <rguenther@suse.de>
13868         PR middle-end/56113
13869         * domwalk.c (bb_postorder): New global static.
13870         (cmp_bb_postorder): New function.
13871         (walk_dominator_tree): Replace scheme imposing an order for
13872         visiting dominator sons by one sorting them at the time they
13873         are pushed on the stack.
13875 2013-03-18  Richard Biener  <rguenther@suse.de>
13877         PR tree-optimization/39326
13878         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
13879         (struct mem_ref): Replace mem member with ao_ref typed member.
13880         (MEM_ANALYZABLE): Adjust.
13881         (memref_eq): Likewise.
13882         (mem_ref_alloc): Likewise.
13883         (gather_mem_refs_stmt): Likewise.
13884         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
13885         (execute_sm_if_changed_flag_set): Adjust.
13886         (execute_sm): Likewise.
13887         (ref_always_accessed_p): Likewise.
13888         (refs_independent_p): Likewise.
13889         (can_sm_ref_p): Likewise.
13891 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13893         PR c/56566
13894         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
13895         return 1 even for !unsignedp.
13897 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13899         * config/i386/i386.md (isa): Add x64 and nox64.
13900         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
13901         (*pushtf): Enable *roF alternative for x64 isa only.
13902         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
13903         mode attribute of integer alternatives to DImode for TARGET_64BIT.
13904         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
13905         (*movtf_internal): Merge from *movtf_internal_rex64 and
13906         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
13907         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
13908         nox64 isa attributes.
13909         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
13910         nox64 isa attributes.
13911         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
13913 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13915         * config/alpha/alpha.c (TARGET_LRA_P): New define.
13917 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
13919         PR target/56640
13920         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
13921         class names.  Remove trailing comma after "ALL_REGS".
13923 2013-03-16  Jan Hubicka  <jh@suse.cz>
13925         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
13926         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
13927         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
13928         of cgraph_get_create_node.
13929         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13931 2013-03-16  Jason Merrill  <jason@redhat.com>
13933         PR debug/49090
13934         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
13935         with DW_AT_default_value.
13937 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
13939         * BASE-VER: Set to 4.9.0.
13941 2013-03-14  Andi Kleen  <ak@linux.intel.com>
13943         PR target/56619
13944         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
13945         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
13946         Document _x* TSX intrinsics.
13948 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
13949             David Holsgrove  <david.holsgrove@xilinx.com>
13951         * configure.ac: Add MicroBlaze TLS support detection.
13952         * configure: Regenerate.
13953         * config/microblaze/microblaze-protos.h
13954         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
13955         symbol_mentioned_p, label_mentioned_p): Add prototypes.
13956         * config/microblaze/microblaze.c (microblaze_address_type): Add
13957         ADDRESS_TLS and tls_reloc address types.
13958         (microblaze_address_info): Add tls_reloc.
13959         (TARGET_HAVE_TLS): Define.
13960         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
13961          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
13962          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
13963         load_tls_operand,  microblaze_call_tls_get_addr,
13964         microblaze_legitimize_tls_address): New functions.
13965         (microblaze_classify_unspec): Handle UNSPEC_TLS.
13966         (get_base_reg): Use microblaze_tls_symbol_p.
13967         (microblaze_classify_address): Handle TLS.
13968         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
13969         label_mentioned_p and microblaze_tls_referenced_p.
13970         (microblaze_legitimize_address): Handle TLS.
13971         (microblaze_address_insns): Handle ADDRESS_TLS.
13972         (pic_address_needs_scratch): Handle TLS.
13973         (print_operand_address): Handle TLS.
13974         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
13975         (microblaze_expand_move): Handle TLS.
13976         (microblaze_legitimate_constant_p): Check
13977         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
13978         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
13979         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
13980         (PIC_OFFSET_TABLE_REGNUM): Set.
13981         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
13982         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
13983         (addsi3, movsi_internal2, movdf_internal): Update constraints
13984         * config/microblaze/predicates.md (arith_plus_operand): Define
13985         (move_operand): Redefine as move_src_operand,
13986         check microblaze_tls_referenced_p.
13988 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
13990         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
13991         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
13993 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
13995         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
13996         CC mode for AND.
13998 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14000         PR tree-optimization/53265
14001         * common.opt (Waggressive-loop-optimizations): New option.
14002         * tree-ssa-loop-niter.c: Include tree-pass.h.
14003         (do_warn_aggressive_loop_optimizations): New function.
14004         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14005         if number_of_latch_executions returned constant.
14006         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14007         early.  If number_of_latch_executions returned constant, set
14008         nb_iterations_upper_bound back to it.
14009         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14010         field.
14011         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14012         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14014         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14015         (MULTILIB_OSDIRNAMES): Set.
14016         * genmultilib: If defaultosdirname doesn't start with :: , set
14017         defaultosdirname2 instead, clear it and emit two . multilib_raw
14018         entries instead of just one.
14020 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14022         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14023         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14024         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14025         (SUBTARGET_OVERRIDE_OPTIONS): New.
14027 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14029         PR target/49880
14030         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14031         (musermode): Convert to Var(TARGET_USERMODE).
14032         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14033         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14034         * config/sh/sh.c (sh_option_override): Use
14035         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14036         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14037         condition.
14038         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14039         TARGET_SH4.
14040         (udivsi3_i4_single, divsi3_i4_single): Use
14041         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14043 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14045         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14046         default setting.
14048 2013-03-13  Richard Biener  <rguenther@suse.de>
14050         PR tree-optimization/56608
14051         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14052         calls when vectorizing basic-blocks.
14054 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14056         PR plugins/45078
14057         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14058         tm_file.
14060 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14062         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14064 2013-03-11  Jan Hubicka  <jh@suse.cz>
14066         PR lto/56557
14067         * lto-streamer-out.c (output_symbol_p): Skip references from
14068         constructors of external variables.
14070 2013-03-11  Jan Hubicka  <jh@suse.cz>
14072         PR middle-end/56571
14073         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14074         from pseudos.
14075         * emit-rtl.c (verify_rtx_sharing): Likewise.
14076         (copy_insn_1): Likewise.
14077         * rtl.c (copy_rtx): Likewise.
14079 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14081         PR target/56591
14082         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14083         output_operand_lossage message.
14085 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14087         PR target/56470
14088         * arm.c (shift_op): Validate RTL pattern on the fly.
14089         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14090         the RTL.
14092 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14094         PR target/56347
14095         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14096         new call patterns that clobber %fr12.
14097         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14098         split and postreload patterns.
14099         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14100         registers %fr12 and %fr12R as call used.
14102 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14104         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14105         (canon_address, record_store, replace_read, check_mem_read_rtx,
14106         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14107         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14108         rest_of_handle_dse): Likewise.
14110 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14112         PR middle-end/56524
14113         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14114         Add base_optabs.
14115         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14116         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14117         (save_optabs_if_changed): Replace with...
14118         (init_tree_optimization_optabs): ...this.
14119         * optabs.c (save_optabs_if_changed): Rename to...
14120         (init_tree_optimization_optabs): ...this.  Take the optimization node
14121         as argument.  Do nothing if the base optabs are already correct.
14122         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14123         to recompute optabs.
14124         * function.h (function): Remove optabs field.
14125         * function.c (invoke_set_current_function_hook): Call
14126         init_tree_optimization_optabs.  Use the result to initialize
14127         this_fn_optabs.
14129 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14131         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14132         if GTMA_HAS_NO_INSTRUMENTATION.
14133         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14134         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14135         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14136         * gimple-pretty-print.c (dump_gimple_transaction): Handle
14137         GTMA_HAS_NO_INSTRUMENTATION.
14139 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
14141         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14142         libasan_preinit.o.
14144 2013-03-08  Marek Polacek  <polacek@redhat.com>
14145             Jakub Jelinek  <jakub@redhat.com>
14147         PR tree-optimization/56478
14148         * predict.c (is_comparison_with_loop_invariant_p): Change the
14149         type of loop_step to tree.
14150         (predict_loops): Adjust.
14151         (predict_iv_comparison): Perform the computations on double_ints.
14153 2013-03-08  Richard Biener  <rguenther@suse.de>
14155         PR tree-optimization/56570
14156         * tree-cfg.c (verify_expr_location_1): Verify locations for
14157         DECL_DEBUG_EXPR.
14158         * tree-sra.c (create_access_replacement): Strip locations
14159         from DECL_DEBUG_EXPRs.
14161 2013-03-08  Richard Biener  <rguenther@suse.de>
14163         * tree-inline.c (expand_call_inline): Do not associate
14164         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14165         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14167 2013-03-08  Richard Biener  <rguenther@suse.de>
14169         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14170         or block changes with -Og.  Fix for location / block encoding
14171         changes and PHI arguments with locations.
14173 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
14175         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14176         for all counters.
14177         (struct output_info): Likewise.
14178         (register_overhead): Remove bad gcc_assert.
14179         (bitmap_find_bit): If there is only a single bitmap element, do not
14180         count a miss as a search.
14181         (print_statistics): Update for counter type changes.
14182         (dump_bitmap_statistics): Likewise.  Print headers such that they
14183         are properly lined up with the printed counters.
14185 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
14187         PR tree-optimization/56559
14188         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14189         check that it has only a single use.
14191 2013-03-07  Richard Biener  <rguenther@suse.de>
14193         * doc/invoke.texi (fwhole-program): Discourage use in combination
14194         with -flto.
14196 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
14198         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14200         PR tree-optimization/56539
14201         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14202         instead of GSI_CONTINUE_LINKING as last argument to
14203         force_gimple_operand_gsi.  Adjust function comment.
14205         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14206         aarch64-cores.def.
14208         PR middle-end/56548
14209         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14210         promoted mode, convert the result back to the original mode.
14212 2013-03-06  Richard Biener  <rguenther@suse.de>
14214         PR middle-end/56294
14215         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14216         (insert_updated_phi_nodes_compare_uids): New function.
14217         (update_ssa): Sort symbols_to_rename after UID before
14218         traversing it to insert PHI nodes.
14220 2013-03-06  Richard Biener  <rguenther@suse.de>
14222         PR middle-end/50494
14223         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14224         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14226         Revert
14227         2013-02-13  Richard Biener  <rguenther@suse.de>
14229         PR lto/50494
14230         * varasm.c (output_constant_def_1): Get the decl representing
14231         the constant as argument.
14232         (output_constant_def): Wrap output_constant_def_1.
14233         (make_decl_rtl): Use output_constant_def_1 with the decl
14234         representing the constant.
14235         (build_constant_desc): Optionally re-use a decl already
14236         representing the constant.
14237         (tree_output_constant_def): Adjust.
14239 2013-03-06  Joey Ye  <joey.ye@arm.com>
14241         PR lto/50293
14242         * gcc.c (convert_white_space): New function.
14243         (main): Handles white space in function name.
14245 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
14247         PR target/56529
14248         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14249         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
14250         to SH_DIV_CALL_TABLE for TARGET_SH2.
14251         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14252         list.
14253         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14254         call-table options.
14256 2013-03-05  Sterling Augustine  <saugustine@google.com>
14257             Cary Coutant  <ccoutant@google.com>
14259         PR debug/55364
14260         * dwarf2out.c (resolve_addr): Don't call
14261         remove_loc_list_addr_table_entries a second time for the same
14262         expression.
14264 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14266         PR debug/56510
14267         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14268         (avoid_complex_debug_insns): New function.
14269         (expand_debug_locations): Call it.
14271         PR rtl-optimization/56484
14272         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14273         lifetimes of hard registers on small register class machines.
14275 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14277         * config/microblaze/microblaze-protos.h: Rename
14278         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14279         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14280         fast_interrupt.
14281         (microblaze_fast_interrupt_function_p): New function.
14282         (microblaze_is_interrupt_handler): Rename to
14283         microblaze_is_interrupt_variant and add fast_interrupt check.
14284         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14285         (save_restore_insns): Likewise.
14286         (compute_frame_size): Likewise.
14287         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14288         (microblaze_globalize_label): Likewise.
14289         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14290         * config/microblaze/microblaze.md: Use wrapper
14291         microblaze_is_interrupt_variant.
14293 2013-03-05  Kai Tietz  <ktietz@redhat.com>
14295         * sdbout.c (sdbout_one_type): Switch to current function's section
14296         supporting cold/hot.
14298 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14300         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14301         -mxl-reorder.
14303 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14305         PR middle-end/56461
14306         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14307         if VALGRIND_GET_VBITS is defined, temporarily make object
14308         memory all defined, and restore previous valgrind addressability
14309         and definability afterwards.  Free this_object at the end.
14311         PR middle-end/56461
14312         * lra.c (lra): Call lra_clear_live_ranges if live_p,
14313         right before calling lra_create_live_ranges, also call it
14314         when clearing live_p.  Only call lra_clear_live_ranges
14315         at the end if live_p.
14317         PR middle-end/56461
14318         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14320 2013-03-05  Richard Biener  <rguenther@suse.de>
14322         PR tree-optimization/56521
14323         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14324         value-id.
14326 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
14328         PR c++/55135
14329         * except.h (remove_unreachable_eh_regions): New prototype.
14330         * except.c (remove_eh_handler_splicer): New function, split out
14331         of remove_eh_handler.
14332         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
14333         warning about running it on many EH regions one at a time.
14334         (remove_unreachable_eh_regions_worker): New function, walk the
14335         EH tree in depth-first order and remove non-marked regions.
14336         (remove_unreachable_eh_regions): New function.
14337         * tree-eh.c (mark_reachable_handlers): New function, split out
14338         from remove_unreachable_handlers.
14339         (remove_unreachable_handlers): Use mark_reachable_handlers and
14340         remove_unreachable_eh_regions.
14341         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14342         and remove_unreachable_eh_regions.
14344 2013-03-05  Richard Biener  <rguenther@suse.de>
14346         PR middle-end/56525
14347         * loop-init.c (fix_loop_structure): Remove loops in two stages,
14348         not freeing them until the end.
14350 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14352         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14354 2013-03-05  Richard Biener  <rguenther@suse.de>
14356         PR tree-optimization/56270
14357         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14358         of loads after scheduling an SLP instance.
14360 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14362         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14363         tic6x.exp.
14364         (check_gcc_parallelize): Run guality.exp as a separate job from
14365         vect.exp with unsorted.exp and $(dg_target_exps) separately from
14366         struct-layout-1.exp with stackalign.exp.
14368         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14370         PR middle-end/56461
14371         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14372         load_index sbitmap even if some bit in it isn't set.
14374         PR middle-end/56461
14375         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14376         (discover_iteration_bound_by_body_walk): Change queues to
14377         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
14378         spelling in comment.  Call safe_push on queues[bound_index] directly.
14379         Release queues[queue_index] in every iteration unconditionally.
14380         Release bounds vector.
14382         PR middle-end/56461
14383         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14384         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14385         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14386         inner_phis vector.
14388 2013-03-05  Richard Biener  <rguenther@suse.de>
14390         PR lto/56515
14391         * tree-inline.c (remap_blocks_to_null): New function.
14392         (expand_call_inline): When expanding a call stmt without
14393         an associated block inline remap all callee blocks to NULL.
14395 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14397         PR rtl-optimization/56494
14398         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14399         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14400         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14402         PR middle-end/56461
14403         * sel-sched-ir.c (free_sched_pools): Release
14404         succs_info_pool.stack[succs_info_pool.max_top] vectors too
14405         if succs_info_pool.max_top isn't -1.
14407         PR bootstrap/56509
14408         * opts.c (opts_obstack, opts_concat): Moved to...
14409         * opts-common.c (opts_obstack, opts_concat): ... here.
14411 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14413         PR middle-end/56461
14414         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14416 2013-03-04  Martin Jambor  <mjambor@suse.cz>
14418         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14419         all appropriate places.
14421 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
14423         PR tree-optimization/56424
14424         * ipa-split.c (split_function): Do not set the RSO flag if result is
14425         not by reference and its type is a register type.
14427 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
14429         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14430         (microblaze_legitimate_pic_operand): Likewise
14431         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14432         new function microblaze_legitimate_pic_operand
14433         * config/microblaze/microblaze-protos.h
14434         (microblaze_legitimate_pic_operand): Declare.
14436 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14438         * config/microblaze/predicates.md (call_insn_simple_operand):
14439         New predicate for supported rtx code types.
14440         * config/microblaze/microblaze.md (call_internal1): Use
14441         call_insn_simple_operand predicate.
14443 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14445         PR middle-end/56461
14446         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14447         partitions.ordered_remove.
14449         PR middle-end/56461
14450         * tree-vect-stmts.c (vectorizable_conversion): Don't call
14451         vec_oprnds0.create (1) for modifier == NONE.
14453         PR middle-end/56461
14454         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14455         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14456         vec_oprnds1 right before pushing anything to it for
14457         scalar_shift_arg.
14459         PR middle-end/56461
14460         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14461         set nbbs to 0 instead of having separate code path.
14462         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14463         instead of false as last argument if returning NULL.
14465 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
14467         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14468         the attribute is now called "target" instead of "option".
14469         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14470         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
14471         attribute/pragma name for TARGET_OPTION_VALID_P and
14472         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
14473         * doc/tm.texi: Regenerated.
14475 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
14477         * config/microblaze/microblaze.c:
14478         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14479         * config/microblaze/microblaze.h: Add -mxl-reorder to
14480         DRIVER_SELF_SPECS.
14481         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14482         instructions emitted if TARGET_REORDER.
14483         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14484         or 0 for -m/-mno case, but initialises as 2 to detect default use case
14485         separately.
14487 2013-03-01  Xinliang David Li  <davidxl@google.com>
14489         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14490         walk length.
14492 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
14494         PR middle-end/56461
14495         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14496         vector even when returning true.  Fix up function comment formatting.
14498         PR middle-end/56461
14499         * ira-build.c (ira_loop_nodes_count): New variable.
14500         (create_loop_tree_nodes): Initialize it.
14501         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14503         PR middle-end/56461
14504         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14505         method on dr_chain and result_chain.
14506         * tree-vect-stmts.c (vectorizable_store): Only call
14507         result_chain.create if j == 0.
14509         PR middle-end/56461
14510         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14511         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14512         before overwriting it.
14514 2013-03-01  Tobias Burnus  <burnus@net-b.de>
14516         * doc/extended.texi (C Extensions): Change order in @menu
14517         to match @node.
14518         (Other MIPS Built-in Functions): Move last MIPS entry before
14519         "picoChip Built-in Functions".
14520         (SH Built-in Functions): Move after RX Built-in Functions.
14521         * doc/gcc.texi (Introduction): Change order in @menu
14522         to match @node.
14523         * doc/md.texi (Constraints): Ditto.
14524         * gty.texi (Type Information): Ditto.
14525         (User-provided marking routines for template types): Make
14526         subsection.
14527         * doc/invoke.texi (AArch64 Options): Move before
14528         "Adapteva Epiphany Options".
14530 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
14531             Jakub Jelinek  <jakub@redhat.com>
14533         PR sanitizer/56454
14534         * asan.c (gate_asan): Lookup no_sanitize_address instead of
14535         no_address_safety_analysis attribute.
14536         * doc/extend.texi (no_address_safety_attribute): Rename to
14537         no_sanitize_address attribute, mention no_address_safety_analysis
14538         attribute as deprecated alias.
14540 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
14542         PR middle-end/56461
14543         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14544         type to vec<vec<tree> > *.
14545         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
14546         to be vec<tree> instead of vec<tree> *, set vec_defs
14547         to vNULL and call vec_defs.create (number_of_vects), adjust other
14548         uses of vec_defs.
14549         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14550         vectorizable_condition): Adjust vect_get_slp_defs callers.
14552 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14554         * config/aarch64/aarch64.c
14555         (aarch64_float_const_representable): Remove unused variable.
14557 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14559         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14561 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14563         * config/aarch64/aarch64-builtins.c
14564         (aarch64_init_simd_builtins): Make static.
14566 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14568         * config/aarch64/aarch64.c
14569         (aarch64_simd_make_constant): Make static.
14571 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14573         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14574         with no initialization to the RHS of debug statements.
14576 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14578         PR tree-optimization/56294
14579         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14580         Adjust dumping.
14581         (get_access_replacement): Do not call create_access_replacement.
14582         Assert a replacement exists.
14583         (get_repl_default_def_ssa_name): Create the replacement declaration
14584         itself.
14586 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14588         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14589         final_end_function.
14591 2013-02-28  Marek Polacek  <polacek@redhat.com>
14593         PR rtl-optimization/56466
14594         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14595         if we're changing a loop.
14596         (peel_loops_completely): Likewise.
14598 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
14600         PR c++/55813
14601         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14603 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14605         PR target/56445
14606         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14607         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14608         INTX_FTYPE_FX, FX_FTYPE_INTX.
14609         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14611 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14613         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14614         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14615         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
14616         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
14617         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
14618         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
14619         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
14620         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
14621         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
14622         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
14623         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14624         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
14625         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
14626         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
14627         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
14628         (avrxmega6): Increase max flash segments from 5 to 6.
14629         * config/avr/t-multilib: Regenerate.
14630         * config/avr/avr-tables.opt: Regenerate.
14631         * doc/avr-mmcu.texi: Regenerate.
14633 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14635         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
14636         (avr_device_to_arch): Rename to avr_device_to_ld.
14637         (avr_device_to_as): New prototype.
14638         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
14639         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
14640         * config/avr/driver-avr.c (avr_device_to_as): New.
14641         (avr_device_to_arch): Rename to avr_device_to_ld.
14643 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14645         PR middle-end/56461
14646         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
14647         method on dr_chain and result_chain.
14649         PR middle-end/56461
14650         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
14651         pointer_set_destroy on not_executed_last_iteration.
14653         PR middle-end/56461
14654         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
14656         PR middle-end/56461
14657         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
14658         FOR_EACH_DEFINED_FUNCTION when freeing state.
14660         PR middle-end/56461
14661         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
14662         pool_free.
14663         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
14664         overwriting it.
14666         PR middle-end/56461
14667         * ipa-cp.c (decide_whether_version_node): Call vec_free on
14668         known_aggs[i].items and release known_aggs vector.
14670         PR middle-end/56461
14671         * ipa-reference.c (propagate): Free node_info even for alias nodes.
14673 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14675         * config/microblaze/microblaze.c (microblaze_emit_compare):
14676         Use xor for EQ/NE comparisions.
14677         * config/microblaze/microblaze.md (cstoresf4): Add constraints
14678         (cbranchsf4): Adjust operator to comparison_operator.
14680 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14682         PR middle-end/56461
14683         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
14684         vector.
14685         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
14686         vec_safe_push, always update *slot.
14687         (redirect_edge_var_map_clear): Use vec_free.
14688         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
14689         (free_var_map_entry): Use vec_free.
14690         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
14691         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
14693 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
14695         PR middle-end/45472
14696         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
14697         when the may_trap_p bit of the exprs being merged differs.
14698         Reorder tests for speculativeness in the logical and operator.
14700 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14702         * incpath.c (add_standard_paths): Use reconcat instead of concat
14703         where appropriate and avoid leaking memory.
14705         * opts.h: Include obstack.h.
14706         (opts_concat): New prototype.
14707         (opts_obstack): New declaration.
14708         * opts.c (opts_concat): New function.
14709         (opts_obstack): New variable.
14710         (init_options_struct): Call gcc_init_obstack on opts_obstack.
14711         (finish_options): Use opts_concat instead of concat
14712         and XOBNEWVEC instead of XNEWVEC.
14713         * opts-common.c (generate_canonical_option, decode_cmdline_option,
14714         generate_option): Likewise.
14715         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14716         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
14718         PR target/56455
14719         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
14720         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
14722 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14724         PR middle-end/56461
14725         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
14727 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14729         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
14730         (arm_block_move_unaligned_straight): Likewise.
14731         (arm_adjust_block_mem): Likewise.
14733 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14735         PR target/48901
14736         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
14737         temp, cond and label.
14738         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
14740         PR target/52500
14741         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
14742         * config/c6x/c6x.h (dbx_register_map): Update declaration.
14744         PR target/52501
14745         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
14746         of prologue/epilogue functions.
14748         PR target/52550
14749         * config/tilegx/tilegx.c (tilegx_expand_prologue):
14750         Remove unused variable cfa_offset.
14751         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
14753         PR target/54639
14754         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
14755         type promotion to unsigned.
14757         PR target/54640
14758         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
14759         for HOST_WIDE_INT of 32 bit / same size as int.
14760         (arm_block_move_unaligned_straight): Likewise.
14761         (arm_adjust_block_mem): Likewise.
14763         PR target/54662
14764         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
14765         ALL_CFLAGS.
14767 2013-02-26  Marek Polacek  <polacek@redhat.com>
14769         PR tree-optimization/56426
14770         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
14772 2013-02-26  Richard Biener  <rguenther@suse.de>
14774         PR target/56444
14775         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
14776         unused variable loops.
14778 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14780         PR tree-optimization/56448
14781         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
14782         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
14783         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
14784         later operands of the references, or even first operand for
14785         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
14787         PR tree-optimization/56443
14788         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
14789         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
14790         to type_for_mode langhook.
14792 2013-02-25  Matt Turner  <mattst88@gmail.com>
14794         * doc/invoke.texi: Document r4700.
14796 2013-02-25  Richard Biener  <rguenther@suse.de>
14798         PR tree-optimization/56175
14799         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
14800         split out from ...
14801         (simplify_bitwise_binary): ... here.  Also guard the conversion
14802         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
14804 2013-02-25  Catherine Moore  <clm@codesourcery.com>
14806         Revert:
14807         2013-02-24  Catherine Moore  <clm@codesourcery.com>
14808             Maciej W. Rozycki  <macro@codesourcery.com>
14809             Tom de Vries  <tom@codesourcery.com>
14810             Nathan Sidwell  <nathan@codesourcery.com>
14811             Iain Sandoe  <iain@codesourcery.com>
14812             Nathan Froyd  <froydnj@codesourcery.com>
14813             Chao-ying Fu  <fu@mips.com>
14815         * doc/extend.texi: (micromips, nomicromips, nocompression):
14816         Document new function attributes.
14817         * doc/invoke.texi (minterlink-compressed, mmicromips,
14818         m14k, m14ke, m14kec): Document new options.
14819         (minterlink-mips16): Update documentation.
14820         * doc/md.texi (ZC, ZD): Document new constraints.
14821         * configure.ac (gcc_cv_as_micromips): Check if linker
14822         supports the .set micromips directive.
14823         * configure: Regenerate.
14824         * config.in: Regenerate.
14825         * config/mips/mips-tables.opt: Regenerate.
14826         * config/mips/micromips.md: New file.
14827         * constraints.md (ZC, AD): New constraints.
14828         * config/mips/predicates.md (movep_src_register): New predicate.
14829         (movep_src_operand): New predicate.
14830         (non_volatile_mem_operand): New predicate.
14831         * config/mips/mips.md (multimem): New type.
14832         (length): Differentiate between 17-bit and 18-bit branch offsets.
14833         (MOVEP1, MOVEP2): New mode iterator.
14834         (mov_<load>l): Use ZC constraint.
14835         (mov_<load>r): Likewise.
14836         (mov_<store>l): Likewise.
14837         (mov_<store>r): Likewise.
14838         (*branch_equality<mode>_inverted): Add microMIPS support.
14839         (*branch_equality<mode>): Likewise.
14840         (*jump_absolute): Likewise.
14841         (indirect_jump_<mode>): Likewise.
14842         (tablejump_<mode>): Likewise.
14843         (<optab>_internal): Likewise.
14844         (sibcall_internal): Likewise.
14845         (sibcall_value_internal): Likewise.
14846         (prefetch): Use constraint ZD.
14847         * config/mips/mips.opt (minterlink-compressed): New option.
14848         (minterlink-mips16): Now an alias for minterlink-compressed.
14849         (mmicromips): New option.
14850         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14851         (compare_and_swap_12): Likewise.
14852         (sync_add<mode>): Likewise.
14853         (sync_<optab>_12): Likewise.
14854         (sync_old_<optab>_12): Likewise.
14855         (sync_new_<optab>_12): Likewise.
14856         (sync_nand_12): Likewise.
14857         (sync_old_nand_12): Likewise.
14858         (sync_new_nand_12): Likewise.
14859         (sync_sub<mode>): Likewise.
14860         (sync_old_add<mode>): Likewise.
14861         (sync_old_sub<mode>): Likewise.
14862         (sync_new_add<mode>): Likewise.
14863         (sync_new_sub<mode>): Likewise.
14864         (sync_<optab><mode>): Likewise.
14865         (sync_old_<optab><mode>): Likewise.
14866         (sync_new_<optab><mode>): Likewise.
14867         (sync_nand<mode>): Likewise.
14868         (sync_old_nand<mode>): Likewise.
14869         (sync_new_nand<mode>): Likewise.
14870         (sync_lock_test_and_set<mode>): Likewise.
14871         (test_and_set_12): Likewise.
14872         (atomic_compare_and_swap<mode>): Likewise.
14873         (atomic_exchange<mode>_llsc): Likewise.
14874         (atomic_fetch_add<mode>_llsc): Likewise.
14875         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14876         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14877         (umips_save_restore_pattern_p): Likewise.
14878         (umips_load_store_pair_p): Likewise.
14879         (umips_output_load_store_pair): Likewise.
14880         (umips_movep_target_p): Likewise.
14881         (umips_12bit_offset_address_p): Likewise.
14882         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14883         (mips_base_mips16): Rename this...
14884         (mips_base_compression_flags): ...to this. Update all uses.
14885         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14886         (mips_mips16_decl_p): Delete.
14887         (mips_nomips16_decl_p): Delete.
14888         (mips_get_compress_on_flags): New function.
14889         (mips_get_compress_off_flags): New function.
14890         (mips_get_compress_mode): New function.
14891         (mips_get_compress_on_name): New function.
14892         (mips_get_compress_off_name): New function.
14893         (mips_insert_attributes): Support multiple compression types.
14894         (mips_merge_decl_attributes): Likewise.
14895         (umips_12bit_offset_address_p): New function.
14896         (mips_start_function_definition): Emit .set micromips directive.
14897         (mips_call_may_need_jalx_p): New function.
14898         (mips_function_ok_for_sibcall): Add microMIPS support.
14899         (mips_print_operand_punctuation): Support short delay slots and
14900         compact jumps.
14901         (umips_swm_mask, umips_swm_encoding): New.
14902         (umips_build_save_restore): New function.
14903         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14904         (was_mips16_p): Remove.
14905         (old_compression_mode): New.
14906         (mips_set_compression_mode): New function.
14907         (mips_set_current_function): Add microMIPS support.
14908         (mips_option_override): Likewise.
14909         (umips_save_restore_pattern_p): New function.
14910         (umips_output_save_restore): New function.
14911         (umips_load_store_pair_p_1): New function.
14912         (umips_load_store_pair_p): New function.
14913         (umips_output_load_store_pair_1): New function.
14914         (umips_output_load_store_pair): New function.
14915         (umips_movep_target_p) New function.
14916         (mips_prepare_pch_save): Add microMIPS support.
14917         * config/mips/mips.h (TARGET_COMPRESSION): New.
14918         (TARGET_CPU_CPP_BUILTINS): Update macro
14919         to use new compression flags and to support microMIPS.
14920         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14921         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14922         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14923         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14924         (ASM_SPEC): Support mmicromips and mno-micromips.
14925         (M16STORE_REG_P): New macro.
14926         (MIPS_CALL): Support TARGET_MICROMIPS.
14927         (MICROMIPS_J): New macro.
14928         (mips_base_mips16): Rename this...
14929         (mips_base_compression_flags): ...to this.
14930         (UMIPS_12BIT_OFFSET_P): New macro.
14931         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14932         (MULTILIB_DIRNAMES): Likewise.
14934 2013-02-25  Tom de Vries  <tom@codesourcery.com>
14936         PR rtl-optimization/56131
14937         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
14938         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
14939         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
14941 2013-02-25  Tobias Burnus  <burnus@net-b.de>
14943         * doc/invoke.texi (-fsanitize=): Move from optimization
14944         to debugging options.
14946 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
14948         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
14950 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
14951             Alexander Monakov  <amonakov@ispras.ru>
14953         PR middle-end/56077
14954         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
14955         flush pending lists also on non-jumps.  Adjust comment.
14957 2013-02-24  Catherine Moore  <clm@codesourcery.com>
14958             Maciej W. Rozycki  <macro@codesourcery.com>
14959             Tom de Vries  <tom@codesourcery.com>
14960             Nathan Sidwell  <nathan@codesourcery.com>
14961             Iain Sandoe  <iain@codesourcery.com>
14962             Nathan Froyd  <froydnj@codesourcery.com>
14963             Chao-ying Fu  <fu@mips.com>
14965         * doc/extend.texi: (micromips, nomicromips, nocompression):
14966         Document new function attributes.
14967         * doc/invoke.texi (minterlink-compressed, mmicromips,
14968         m14k, m14ke, m14kec): Document new options.
14969         (minterlink-mips16): Update documentation.
14970         * doc/md.texi (ZC, ZD): Document new constraints.
14971         * configure.ac (gcc_cv_as_micromips): Check if linker
14972         supports the .set micromips directive.
14973         * configure: Regenerate.
14974         * config.in: Regenerate.
14975         * config/mips/mips-tables.opt: Regenerate.
14976         * config/mips/micromips.md: New file.
14977         * constraints.md (ZC, AD): New constraints.
14978         * config/mips/predicates.md (movep_src_register): New predicate.
14979         (movep_src_operand): New predicate.
14980         (non_volatile_mem_operand): New predicate.
14981         * config/mips/mips.md (multimem): New type.
14982         (length): Differentiate between 17-bit and 18-bit branch offsets.
14983         (MOVEP1, MOVEP2): New mode iterator.
14984         (mov_<load>l): Use ZC constraint.
14985         (mov_<load>r): Likewise.
14986         (mov_<store>l): Likewise.
14987         (mov_<store>r): Likewise.
14988         (*branch_equality<mode>_inverted): Add microMIPS support.
14989         (*branch_equality<mode>): Likewise.
14990         (*jump_absolute): Likewise.
14991         (indirect_jump_<mode>): Likewise.
14992         (tablejump_<mode>): Likewise.
14993         (<optab>_internal): Likewise.
14994         (sibcall_internal): Likewise.
14995         (sibcall_value_internal): Likewise.
14996         (prefetch): Use constraint ZD.
14997         * config/mips/mips.opt (minterlink-compressed): New option.
14998         (minterlink-mips16): Now an alias for minterlink-compressed.
14999         (mmicromips): New option.
15000         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15001         (compare_and_swap_12): Likewise.
15002         (sync_add<mode>): Likewise.
15003         (sync_<optab>_12): Likewise.
15004         (sync_old_<optab>_12): Likewise.
15005         (sync_new_<optab>_12): Likewise.
15006         (sync_nand_12): Likewise.
15007         (sync_old_nand_12): Likewise.
15008         (sync_new_nand_12): Likewise.
15009         (sync_sub<mode>): Likewise.
15010         (sync_old_add<mode>): Likewise.
15011         (sync_old_sub<mode>): Likewise.
15012         (sync_new_add<mode>): Likewise.
15013         (sync_new_sub<mode>): Likewise.
15014         (sync_<optab><mode>): Likewise.
15015         (sync_old_<optab><mode>): Likewise.
15016         (sync_new_<optab><mode>): Likewise.
15017         (sync_nand<mode>): Likewise.
15018         (sync_old_nand<mode>): Likewise.
15019         (sync_new_nand<mode>): Likewise.
15020         (sync_lock_test_and_set<mode>): Likewise.
15021         (test_and_set_12): Likewise.
15022         (atomic_compare_and_swap<mode>): Likewise.
15023         (atomic_exchange<mode>_llsc): Likewise.
15024         (atomic_fetch_add<mode>_llsc): Likewise.
15025         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15026         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15027         (umips_save_restore_pattern_p): Likewise.
15028         (umips_load_store_pair_p): Likewise.
15029         (umips_output_load_store_pair): Likewise.
15030         (umips_movep_target_p): Likewise.
15031         (umips_12bit_offset_address_p): Likewise.
15032         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15033         (mips_base_mips16): Rename this...
15034         (mips_base_compression_flags): ...to this. Update all uses.
15035         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15036         (mips_mips16_decl_p): Delete.
15037         (mips_nomips16_decl_p): Delete.
15038         (mips_get_compress_on_flags): New function.
15039         (mips_get_compress_off_flags): New function.
15040         (mips_get_compress_mode): New function.
15041         (mips_get_compress_on_name): New function.
15042         (mips_get_compress_off_name): New function.
15043         (mips_insert_attributes): Support multiple compression types.
15044         (mips_merge_decl_attributes): Likewise.
15045         (umips_12bit_offset_address_p): New function.
15046         (mips_start_function_definition): Emit .set micromips directive.
15047         (mips_call_may_need_jalx_p): New function.
15048         (mips_function_ok_for_sibcall): Add microMIPS support.
15049         (mips_print_operand_punctuation): Support short delay slots and
15050         compact jumps.
15051         (umips_swm_mask, umips_swm_encoding): New.
15052         (umips_build_save_restore): New function.
15053         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15054         (was_mips16_p): Remove.
15055         (old_compression_mode): New.
15056         (mips_set_compression_mode): New function.
15057         (mips_set_current_function): Add microMIPS support.
15058         (mips_option_override): Likewise.
15059         (umips_save_restore_pattern_p): New function.
15060         (umips_output_save_restore): New function.
15061         (umips_load_store_pair_p_1): New function.
15062         (umips_load_store_pair_p): New function.
15063         (umips_output_load_store_pair_1): New function.
15064         (umips_output_load_store_pair): New function.
15065         (umips_movep_target_p) New function.
15066         (mips_prepare_pch_save): Add microMIPS support.
15067         * config/mips/mips.h (TARGET_COMPRESSION): New.
15068         (TARGET_CPU_CPP_BUILTINS): Update macro
15069         to use new compression flags and to support microMIPS.
15070         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15071         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15072         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15073         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15074         (ASM_SPEC): Support mmicromips and mno-micromips.
15075         (M16STORE_REG_P): New macro.
15076         (MIPS_CALL): Support TARGET_MICROMIPS.
15077         (MICROMIPS_J): New macro.
15078         (mips_base_mips16): Rename this...
15079         (mips_base_compression_flags): ...to this.
15080         (UMIPS_12BIT_OFFSET_P): New macro.
15081         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15082         (MULTILIB_DIRNAMES): Likewise.
15084 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15086         PR target/52555
15087         * target-globals.c (save_target_globals): For init_reg_sets and
15088         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15090 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15092         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15093         * config/aarch64/t-aarch64
15094         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15096 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15098         PR inline-asm/56148
15099         * lra-constraints.c (process_alt_operands): Reload operand
15100         conflicting with earlier clobber only if no more other conflicting
15101         operands.
15103 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15105         PR sanitizer/56393
15106         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15107         if not linking a shared library.
15109 2013-02-22  Seth LaForge  <sethml@google.com>
15111         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15113 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15115         * config/arm/arm.md (split for extendsidi): Update condition.
15116         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15117         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15118         (qhs_zextenddi_cstr): Likewise.
15120 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15122         PR middle-end/56420
15123         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15124         avoid signed wrapping.
15125         (expand_mult): Handle properly multiplication by
15126         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15127         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15128         in the compiler if coeff is HOST_WIDE_INT_MIN.
15129         (expand_divmod): Don't make ext_op1 static, change it's type to
15130         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15132         PR rtl-optimization/50339
15133         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15134         field.
15135         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15136         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15137         into splitting_ashiftrt field.
15138         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15139         ASHIFTRT.
15140         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15141         choices.
15143 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
15145         PR middle-end/56108
15146         * trans-mem.c (execute_tm_mark): Do not expand transactions that
15147         are sure to go irrevocable.
15149 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
15151         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15152         scalars are valid operands.
15154 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15156         PR tree-optimization/56310
15157         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15158         only matching indices and non-negative final offsets.
15159         (intersect_aggregates_with_edge): Pass src_idx to
15160         agg_replacements_to_vector.  Pass src_idx insstead of index to
15161         intersect_with_agg_replacements.
15163 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15165         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15166         instead of hard-wired defaults.
15168 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
15170         * doc/invoke.texi (MIPS Options): Update documentation of the
15171         floating-point multiply-accumulate instruction restrictions.
15173 2013-02-21  Kostya Serebryany  <kcc@google.com>
15175         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15176         asan_shadow_offset on x86_64 linux.
15178 2013-02-21  Richard Biener  <rguenther@suse.de>
15180         PR tree-optimization/56415
15181         Revert
15182         2013-02-11  Richard Biener  <rguenther@suse.de>
15184         PR tree-optimization/56273
15185         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15186         first VRP run.
15188 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15190         PR bootstrap/56258
15191         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15192         instead of @itemx.
15194         PR inline-asm/56405
15195         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15196         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15198 2013-02-20  Jan Hubicka  <jh@suse.cz>
15200         PR tree-optimization/56265
15201         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15202         when target is referenced for first time.
15204 2013-02-20  Richard Biener  <rguenther@suse.de>
15206         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15207         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15208         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15209         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15210         not return anything.
15211         (rename_ssa_copies): Do not remove unused locals.
15212         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15213         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15214         * passes.c (execute_function_todo): Do not schedule unused locals
15215         removal if cleanup_tree_cfg did something.
15216         * tree-ssa-live.c (remove_unused_locals): Dump statistics
15217         about the number of removed locals.
15219 2013-02-20  Richard Biener  <rguenther@suse.de>
15221         PR tree-optimization/56398
15222         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15224 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15226         PR tree-optimization/55334
15227         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15228         restricted pointers to arrays.
15230 2013-02-20  Richard Biener  <rguenther@suse.de>
15231             Jakub Jelinek  <jakub@redhat.com>
15233         PR tree-optimization/56396
15234         * tree-ssa-ccp.c (n_const_val): New static variable.
15235         (get_value): Return NULL for SSA names we don't have a lattice
15236         entry for.
15237         (ccp_initialize): Initialize n_const_val.
15238         * tree-ssa-copy.c (n_copy_of): New static variable.
15239         (init_copy_prop): Initialize n_copy_of.
15240         (get_value): Return NULL_TREE for SSA names we don't have a
15241         lattice entry for.
15243 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15245         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15247 2013-02-20  Richard Biener  <rguenther@suse.de>
15249         * genpreds.c (write_lookup_constraint): Do not compare first
15250         letter of the constraint again.
15252 2013-02-20  Richard Biener  <rguenther@suse.de>
15254         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15255         and ceil_log2.
15256         (get_use_iv_cost): Terminate hashtable walk when coming across
15257         an empty entry.
15259 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
15261         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15262         reassociation for avx2 targets.
15264 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15266         * config/microblaze/microblaze.c: microblaze_has_clz = 0
15267         Add version check for v8.10.a to enable microblaze_has_clz
15268         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15269         version and TARGET_PATTERN_COMPARE check
15270         * config/microblaze/microblaze.md: New clzsi2 instruction
15272 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15274         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15275         function before branching.
15277 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15279         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15280         DUMP_INSN_RTX_UID.
15281         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15283 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15285         PR middle-end/55889
15286         * sel-sched.c: Include ira.h.
15287         (implicit_clobber_conflict_p): New function.
15288         (moveup_expr): Use it.
15289         * Makefile.in (sel-sched.o): Depend on ira.h.
15291 2013-02-19  Richard Biener  <rguenther@suse.de>
15293         PR tree-optimization/56384
15294         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15295         (vn_hash_type): Split out from ...
15296         (vn_hash_constant_with_type): ... here.
15297         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15298         (vn_phi_eq): Compare types from vn_phi_s structure.
15299         (vn_phi_lookup): Populate vn_phi_s type.
15300         (vn_phi_insert): Likewise.
15302 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
15304         PR tree-optimization/56350
15305         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15306         if haven't found reduction or nested cycle operand, rather than
15307         asserting we must find it.
15309         PR tree-optimization/56381
15310         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15311         to fold_build3.
15313 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
15314             Jakub Jelinek  <jakub@redhat.com>
15316         PR target/52555
15317         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15318         (swap_optab_enable): Same.
15319         (init_all_optabs): Use argument instead of global.
15320         * tree.h (struct tree_optimization_option): New field target_optabs.
15321         * expr.h (init_all_optabs): Add argument to prototype.
15322         (TREE_OPTIMIZATION_OPTABS): New.
15323         (save_optabs_if_changed): Protoize.
15324         * optabs.h: Declare this_fn_optabs.
15325         * optabs.c (save_optabs_if_changed): New.
15326         Declare this_fn_optabs.
15327         (init_optabs): Add argument to init_all_optabs() call.
15328         * function.c (invoke_set_current_function_hook): Handle per
15329         function optabs.
15330         * function.h (struct function): New field optabs.
15331         * config/mips/mips.c (mips_set_mips16_mode): Handle when
15332         optimization_current_node has changed.
15333         * target-globals.h (save_target_globals_default_opts): Protoize.
15334         * target-globals.c (save_target_globals_default_opts): New.
15336 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15338         PR target/56347
15339         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15340         registers %fr12 and %fr12R as call used.
15342         PR target/56214
15343         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15344         and HImode, require all displacements to be an integer multiple of
15345         their mode size.
15346         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15347         only allow QImode and HImode when reload is in progress and strict is
15348         true.  Likewise for symbolic addresses.  Use base14_operand to check
15349         displacements in REG+BASE addresses.
15351 2013-02-18  Richard Biener  <rguenther@suse.de>
15353         PR tree-optimization/56366
15354         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15355         handle sign-conversion of outer-loop initial induction value.
15357 2013-02-18  Richard Biener  <rguenther@suse.de>
15359         PR middle-end/56349
15360         * cfghooks.c (merge_blocks): If we merge a latch into another
15361         block adjust references to it.
15362         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15363         (verify_loop_structure): Verify that a recorded latch is in fact
15364         a latch.
15366 2013-02-18  Richard Biener  <rguenther@suse.de>
15368         PR tree-optimization/56321
15369         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15370         order SSA name release and virtual operand unlinking.
15372 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15374         * config/microblaze/microblaze.md (save_stack_block): Define.
15375         (restore_stack_block): Likewise.
15377 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15379         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15380         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15381         * config/microblaze/microblaze.c (microblaze_option_override):
15382         Bail out early for PIC modes when target does not support PIC.
15384 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15386         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15387         Replace with a microblaze version.
15388         (microblaze_trampoline_init): Adapt for microblaze.
15389         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15390         microblaze.
15392 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
15393             Dodji Seketeli  <dodji@redhat.com>
15395         PR asan/56330
15396         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15397         (instrument_mem_region_access): Do not forget to always put
15398         instrumentation of the of 'base' and 'base + len' in a "if (len !=
15399         0) statement, even for cases where either 'base' or 'base + len'
15400         are not instrumented -- because they have been previously
15401         instrumented.  Simplify the logic by putting all the statements
15402         instrument 'base + len' inside a sequence, and then insert that
15403         sequence right before the current insertion point.  Then, to
15404         instrument 'base + len', just get an iterator on that statement.
15405         And do not forget to update the pointer to iterator the function
15406         received as argument.
15408 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
15410         PR rtl-optimization/56348
15411         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15413 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
15415         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15416         (clean_graph_dump_file): Pass base to start_graph_dump.
15418 2013-02-14  Richard Henderson  <rth@redhat.com>
15420         PR target/55941
15421         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15423 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
15425         * collect2-aix.h: Define F_LOADONLY.
15427 2013-02-14  Richard Biener  <rguenther@suse.de>
15429         PR lto/50494
15430         * varasm.c (output_constant_def_1): Get the decl representing
15431         the constant as argument.
15432         (output_constant_def): Wrap output_constant_def_1.
15433         (make_decl_rtl): Use output_constant_def_1 with the decl
15434         representing the constant.
15435         (build_constant_desc): Optionally re-use a decl already
15436         representing the constant.
15437         (tree_output_constant_def): Adjust.
15439 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
15441         Fix an asan crash
15442         * asan.c (instrument_builtin_call):  Really put the length of the
15443         second source argument into src1_len.
15445 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15447         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15448         argument.  If it is false, don't create edge from then_bb to
15449         fallthru_bb.
15450         (insert_if_then_before_iter): Pass true to it.
15451         (build_check_stmt): Pass false to it.
15452         (transform_statements): Flush hash table only on extended basic
15453         block boundaries, rather than at the beginning of every bb.
15454         Don't flush hash table on nonfreeing_call_p calls.
15455         * tree-flow.h (nonfreeing_call_p): New prototype.
15456         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15458 2013-02-13  David S. Miller  <davem@davemloft.net>
15460         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15462 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
15464         PR target/56184
15465         * ira.c (max_regno_before_ira): Move from ...
15466         (ira): ... here.
15467         (fix_reg_equiv_init): Use max_regno_before_ira instead of
15468         vec_safe_length.
15470 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15472         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15474 2013-02-13  Richard Biener  <rguenther@suse.de>
15476         PR lto/56295
15477         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15478         globals in MEM_REFs.
15480 2013-02-13  Richard Biener  <rguenther@suse.de>
15482         * loop-init.c (loop_optimizer_init): Clear loop state when
15483         re-initializing preserved loops.
15484         * loop-unswitch.c (unswitch_single_loop): Return whether
15485         we unswitched the loop.  Do not verify loop state here.
15486         (unswitch_loops): When we unswitched a loop discover new loops.
15488 2013-02-13  Kostya Serebryany  <kcc@google.com>
15490         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15491         on x86_64 linux.
15492         * sanitizer.def: Rename __asan_init to __asan_init_v1.
15494 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
15496         Avoid instrumenting duplicated memory access in the same basic block
15497         * Makefile.in (asan.o): Add new dependency on hash-table.h
15498         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15499         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15500         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15501         (free_mem_ref_resources, has_mem_ref_been_instrumented)
15502         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15503         (get_mem_ref_of_assignment): New functions.
15504         (get_mem_refs_of_builtin_call): Extract from
15505         instrument_builtin_call and tweak a little bit to make it fit with
15506         the new signature.
15507         (instrument_builtin_call): Use the new
15508         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
15509         of is_gimple_builtin_call.
15510         (instrument_derefs, instrument_mem_region_access): Insert the
15511         instrumented memory reference into the hash table.
15512         (maybe_instrument_assignment): Renamed instrument_assignment into
15513         this, and change it to advance the iterator when instrumentation
15514         actually happened and return true in that case.  This makes it
15515         homogeneous with maybe_instrument_assignment, and thus give a
15516         chance to callers to be more 'regular'.
15517         (transform_statements): Clear the memory reference hash table
15518         whenever we enter a new BB, when we cross a function call, or when
15519         we are done transforming statements.  Use
15520         maybe_instrument_assignment instead of instrumentation.  No more
15521         need to special case maybe_instrument_assignment and advance the
15522         iterator after calling it; it's now handled just like
15523         maybe_instrument_call.  Update comment.
15525 2013-02-13  Richard Biener  <rguenther@suse.de>
15527         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15528         Fix loop discovery code.
15530 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
15532         PR inline-asm/56148
15533         * lra-constraints.c (process_alt_operands): Match early clobber
15534         operand with itself.  Check conflicts with earlyclobber only if
15535         the operand is not reloaded.  Prefer to reload conflicting operand
15536         if earlyclobber and matching operands are the same.
15538 2013-02-12  Richard Biener  <rguenther@suse.de>
15540         PR lto/56297
15541         * lto-streamer-out.c (write_symbol): Do not output symbols
15542         for hard register variables.
15544 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
15546         PR target/54222
15547         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15548         (umulsidi3_insn, mulsidi3_insn): New insns.
15550 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
15552         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15553         (struct tune_params): Add vec_costs field.
15554         * config/arm/arm.c (arm_builtin_vectorization_cost)
15555         (arm_add_stmt_cost): New functions.
15556         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15557         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15558         (arm_default_vec_cost): New struct of type cpu_vec_costs.
15559         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15560         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15561         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15562         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15564 2013-02-12  Richard Biener  <rguenther@suse.de>
15566         PR lto/56295
15567         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15568         decls again if possible.
15570 2013-02-12  Richard Biener  <rguenther@suse.de>
15572         PR middle-end/56288
15573         * tree-ssa.c (verify_ssa_name): Fix check, move
15574         SSA_NAME_IN_FREE_LIST check up.
15576 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
15577             Steven Bosscher   <steven@gcc.gnu.org>
15579         PR rtl-optimization/56151
15580         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15581         equal to op0 or op1, and last_insn pattern is CODE operation
15582         with MEM dest and one of the operands matches that MEM.
15584 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
15586         * doc/extend.texi: Document Function Multiversioning and "default"
15587         parameter string to target attribute.
15588         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15589         target attribute parameter is "default".
15590         (ix86_compare_version_priority): Remove checks for target attribute.
15591         (ix86_mangle_function_version_assembler_name): Change error to sorry.
15592         Remove check for target attribute equal to NULL. Add assert.
15593         (ix86_generate_version_dispatcher_body): Change error to sorry.
15595 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
15596             Jack Howarth  <howarth@bromo.med.uc.edu>
15597             Patrick Marlier  <patrick.marlier@gmail.com>
15599         PR libitm/55693
15600         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15601         define ENDFILE_SPEC as TM_DESTRUCTOR.
15602         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15604 2013-02-11  Alexander Potapenko  <glider@google.com>
15605             Jack Howarth  <howarth@bromo.med.uc.edu>
15606             Jakub Jelinek  <jakub@redhat.com>
15608         PR sanitizer/55617
15609         * config/darwin.c (cdtor_record): Rename ctor_record.
15610         (sort_cdtor_records): Rename sort_ctor_records.
15611         (finalize_dtors): New routine to sort destructors by
15612         priority before use in assemble_integer.
15613         (machopic_asm_out_destructor): Use finalize_dtors if needed.
15615 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
15617         PR rtl-optimization/56275
15618         * simplify-rtx.c (avoid_constant_pool_reference): Check that
15619         offset is non-negative and less than cmode size before
15620         calling simplify_subreg.
15622 2013-02-11  Richard Biener  <rguenther@suse.de>
15624         PR tree-optimization/56264
15625         * cfgloop.h (fix_loop_structure): Adjust prototype.
15626         * loop-init.c (fix_loop_structure): Return the number of
15627         newly discovered loops.
15628         * tree-cfgcleanup.c (repair_loop_structures): When new loops
15629         are discovered, do a full loop-closed SSA rewrite.
15631 2013-02-11  Richard Biener  <rguenther@suse.de>
15633         PR tree-optimization/56273
15634         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15635         first VRP run.
15636         (check_array_ref): Fix missing newline in dumps.
15637         (search_for_addr_array): Likewise.
15639 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
15641         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
15643 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
15645         PR target/56256
15646         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
15648 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
15650         PR rtl-optimization/56246
15651         * lra-constraints.c (simplify_operand_subreg): Try to reuse
15652         reload pseudo.
15653         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
15654         constraints are satisfied.
15656 2013-02-08  Jeff Law  <law@redhat.com>
15658         PR debug/53948
15659         * emit-rtl.c (reg_is_parm_p): New function.
15660         * regs.h (reg_is_parm_p): New prototype.
15661         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
15662         callee-clobbered registers.
15664 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
15666         PR target/56043
15667         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
15668         If there is no implicit builtin declaration, just return NULL.
15670 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15672         * config/i386/sse.md (FMAMODEM): New mode iterator.
15673         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
15674         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
15676 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15678         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
15679         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
15680         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
15682 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15684         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
15685         (microblaze*-*-elf): Likewise.
15686         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
15687         LINK_SPEC.
15688         * config/microblaze/microblaze-c.c: Add builtin defines for
15689         _LITTLE_ENDIAN and _BIG_ENDIAN.
15690         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
15691         add to TARGET_DEFAULT flags.
15692         Expand ASM_SPEC and LINK_SPEC.
15693         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
15694         * config/microblaze/microblaze.md: Update extendsidi2 and
15695         movdi_internal instructions to use low-order / high-order reg
15696         print_operands.
15697         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
15698         options and inversemask / mask of LITTLE_ENDIAN.
15699         * config/microblaze/t-microblaze: Expand multilib options to
15700         include mlittle-endian (le) and update exceptions patterns.
15702 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
15704         PR rtl-optimization/56195
15705         * lra-constraints.c (get_reload_reg): Don't reuse regs
15706         if they have smaller mode than requested, if they have
15707         wider mode than requested, try to return a SUBREG.
15709         PR tree-optimization/56250
15710         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
15711         if type is unsigned and code isn't MULT_EXPR.
15713 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15715         PR tree-optimization/56064
15716         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
15717         bits according to mode.
15718         * fixed-value.h (fixed_from_double_int)
15719         (const_fixed_from_double_int): Adjust comments.
15721 2013-02-08  Richard Biener  <rguenther@suse.de>
15723         PR lto/56231
15724         * lto-streamer.h (struct data_in): Remove current_file, current_line
15725         and current_col members.
15726         * lto-streamer-out.c (lto_output_location): Stream changed bits
15727         en-block for efficiency.
15728         * lto-streamer-in.c (clear_line_info): Remove.
15729         (lto_input_location): Cache current file, line and column
15730         globally via local statics.  Read changed bits en-block.
15731         (input_function): Do not call clear_line_info.
15732         (lto_read_body): Likewise.
15733         (lto_input_toplevel_asms): Likewise.
15735 2013-02-08  Michael Matz  <matz@suse.de>
15737         PR tree-optimization/52448
15738         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
15739         (nt_call_phase): New static.
15740         (add_or_mark_expr): Only mark accesses with newer phase than any
15741         call seen.
15742         (nonfreeing_call_p): New.
15743         (nt_init_block): Update nt_call_phase, mark blocks as visited.
15744         (nt_fini_block): Keep blocks marked as visited.
15745         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
15747 2013-02-08  Richard Biener  <rguenther@suse.de>
15749         * ira.c (ira): Free broken dominator information.
15751 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15753         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
15755 2013-02-08  Marek Polacek  <polacek@redhat.com>
15757         * cfgloop.c (verify_loop_structure): Add more checking of headers.
15759 2013-02-08  Richard Biener  <rguenther@suse.de>
15761         PR middle-end/56181
15762         * cfgloop.h (flow_loops_find): Adjust.
15763         (bb_loop_header_p): Declare.
15764         * cfgloop.c (bb_loop_header_p): New function split out from ...
15765         (flow_loops_find): ... here.  Adjust function signature,
15766         support incremental loop structure update.
15767         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
15768         * cfgloopmanip.c (fix_loop_structure): Move ...
15769         * loop-init.c (fix_loop_structure): ... here.
15770         (apply_loop_flags): Split out from ...
15771         (loop_optimizer_init): ... here.
15772         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
15773         in incremental mode, only remove dead loops here.
15775 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15777         PR target/54222
15778         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
15779         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
15780         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
15781         (*round<mode>3.libgcc): New insns for fixed-modes.
15782         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
15783         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
15784         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
15785         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
15786         implementations.  Define to __builtin_avr_absFX,
15787         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
15788         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
15789         __builtin_avr_countlsFX, respectively.
15790         * config/avr/avr-c.c (target.h): Include it.
15791         (enum avr_builtin_id): New enum.
15792         (avr_resolve_overloaded_builtin): New static function.
15793         (avr_register_target_pragmas): Use it to set
15794         targetm.resolve_overloaded_builtin.
15795         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
15796         tree nodes used by DEF_BUILTIN.
15797         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
15798         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
15799         <AVR_BUILTIN_xxBITS>: Same.
15801 2013-02-08  Richard Biener  <rguenther@suse.de>
15803         * cfgloop.c (verify_loop_structure): Properly handle
15804         a loop exiting to another loop header.
15805         * ira-int.h (ira_loops): Remove.
15806         * ira.c (ira_loops): Remove.
15807         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
15808         (do_reload): Use loop_optimizer_finalize.
15809         * ira-build.c (create_loop_tree_nodes): Use get_loops and
15810         number_of_loops to access the loop tree.
15811         (more_one_region_p): Likewise.
15812         (finish_loop_tree_nodes): Likewise.
15813         (rebuild_regno_allocno_maps): Likewise.
15814         (mark_loops_for_removal): Likewise.
15815         (mark_all_loops_for_removal): Likewise.
15816         (remove_unnecessary_regions): Likewise.
15817         (ira_build): Likewise.
15818         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
15820 2013-02-08  Richard Biener  <rguenther@suse.de>
15822         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
15823         * ipa-pure-const.c (analyze_function): Avoid calling
15824         mark_irreducible_loops twice.
15825         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
15827 2013-02-07  David S. Miller  <davem@davemloft.net>
15829         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
15830         on 'reg'.
15831         * var-tracking.c (vt_add_function_parameter): Test the presence of
15832         HAVE_window_save properly and do not remap argument registers when
15833         we have a leaf function.
15835 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
15837         PR bootstrap/56227
15838         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
15839         instead of "ll".
15840         * config/i386/i386.c (ix86_print_operand): Ditto.
15842 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15844         * lra-constraints.c (process_alt_operands): Fix recently added comment.
15846 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15848         PR rtl-optimization/56225
15849         * lra-constraints.c (process_alt_operands): Check that reload hard
15850         reg can hold value for strict_low_part.
15852 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15854         PR debug/56154
15855         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
15856         dwarf2out_end_function.
15857         (in_first_function_p, maybe_at_text_label_p,
15858         first_loclabel_num_not_at_text_label): New variables.
15859         (dwarf2out_var_location): In the first function find out
15860         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
15861         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
15862         functions.
15864 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
15866         PR rtl-optimization/56178
15867         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
15868         SUBREG of a register.  Tidy up related block of code.
15869         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
15870         note if the source is a register or a SUBREG of a register.
15872 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15874         PR target/56228
15875         * config/rs6000/rs6000.md (ptrm): New mode attr.
15876         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
15877         call_value_indirect_aix<pttrsize>,
15878         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
15879         m in constraints.
15881 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
15883         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
15884         if -bnortl. Convert to strcmp and strncmp.
15886 2013-02-07  Alan Modra  <amodra@gmail.com>
15888         PR target/54009
15889         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
15890         addresses won't wrap when offsetting.
15891         (rs6000_secondary_reload): Provide secondary reloads needed for
15892         wrapping LO_SUM addresses.
15894 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
15896         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
15897         MACH, just __MACH__.
15899 2013-02-06  Richard Biener  <rguenther@suse.de>
15901         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
15902         instead of calling fix_loop_structure.
15904 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
15906         PR middle-end/56217
15907         * omp-low.c (use_pointer_for_field): Return false if
15908         lower_send_shared_vars doesn't generate any copy-out code.
15910 2013-02-06  Tom de Vries  <tom@codesourcery.com>
15912         PR rtl-optimization/56131
15913         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
15914         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
15915         of the label is NULL.  Add comment.
15917 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
15919         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
15921         PR sanitizer/55374
15922         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
15923         (STATIC_LIBTSAN_LIBS): Likewise.
15924         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
15925         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
15926         is defined, don't add anything else beyond that.
15927         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
15928         (LINK_COMMAND_SPEC): Use them.
15930         PR tree-optimization/56205
15931         * tree-stdarg.c (check_all_va_list_escapes): Return true if
15932         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
15933         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
15935 2013-02-05  Richard Biener  <rguenther@suse.de>
15937         PR tree-optimization/53342
15938         PR tree-optimization/53185
15939         * tree-vectorizer.h (vect_check_strided_load): Remove.
15940         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
15941         not disallow peeling for vectorized strided loads.
15942         (vect_check_strided_load): Make static and simplify.
15943         (vect_analyze_data_refs): Adjust.
15944         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
15945         correctly when vectorizing strided loads.
15947 2013-02-05  Richard Biener  <rguenther@suse.de>
15949         * doc/install.texi: Refer to ISL, not PPL.
15951 2013-02-05  Jan Hubicka  <jh@suse.cz>
15953         PR tree-optimization/55789
15954         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
15956 2013-02-05  Jan Hubicka  <jh@suse.cz>
15958         PR tree-optimization/55789
15959         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
15960         the dead call anyway.
15962 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
15964         PR sanitizer/55374
15965         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
15967 2013-02-04  Alexander Potapenko  <glider@google.com>
15968             Jack Howarth  <howarth@bromo.med.uc.edu>
15969             Jakub Jelinek  <jakub@redhat.com>
15971         PR sanitizer/55617
15972         * config/darwin.c (sort_ctor_records): Stabilized qsort
15973         on constructor priority by using original position.
15974         (finalize_ctors): New routine to sort constructors by
15975         priority before use in assemble_integer.
15976         (machopic_asm_out_constructor): Use finalize_ctors if needed.
15978 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
15980         PR libstdc++/54314
15981         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
15982         about visibility on artificial decls.
15983         * config/sol2.c (solaris_assemble_visibility): Likewise.
15985 2013-02-04  Kai Tietz  <ktietz@redhat.com>
15987         PR target/56186
15988         * config/i386/i386.c (function_value_ms_64): Add additional valtype
15989         argument and improve checking of return-argument types for 16-byte
15990         modes.
15991         (ix86_function_value_1): Add additional valtype argument on call
15992         of function_value_64.
15993         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
15994         handling infunction_value_64 function.
15996 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
15998         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16000 2013-02-04  Richard Biener  <rguenther@suse.de>
16002         PR tree-optimization/56188
16003         * tree-ssa-structalias.c (label_visit): Consider case with
16004         initially non-empty points-to set.
16005         (perform_var_substitution): Dump node mapping and clean up.
16007 2013-02-04  Richard Guenther  <rguenther@suse.de>
16009         PR lto/56168
16010         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16011         node prevail as last resort.
16012         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16013         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16015 2013-02-04  Richard Biener  <rguenther@suse.de>
16017         PR tree-optimization/56113
16018         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16019         Merge into ...
16020         (equiv_class_lookup_or_add): ... this.
16021         (label_visit): Adjust and fix error in previous patch.
16022         (perform_var_substitution): Adjust.
16024 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16026         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16027         * config/sh/sh4-300.md: Likewise.
16028         * config/sh/sh4a.md: Likewise.
16029         * config/sh/constraints.md: Likewise.
16030         * config/sh/sh.md: Likewise.
16031         * config/sh/netbsd-elf.h: Likewise.
16032         * config/sh/predicates.md: Likewise.
16033         * config/sh/sh-protos.h: Likewise.
16034         * config/sh/ushmedia.h: Likewise.
16035         * config/sh/linux.h: Likewise.
16036         * config/sh/sh.c: Likewise.
16037         * config/sh/superh.h: Likewise.
16038         * config/sh/elf.h: Likewise.
16039         * config/sh/sh4.md: Likewise.
16040         * config/sh/sh.h: Likewise.
16042 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16044         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16045         constraint to match_test floating_point_store_memory_operand().
16046         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16047         (base14_operand): New.
16048         (floating_point_store_memory_operand): New.
16049         (integer_store_memory_operand): Revise to use base14_operand and
16050         reg_plus_base_memory_operand.
16051         (move_dest_operand): Allow symbolic_memory_operands.
16052         (symbolic_memory_operand): Check for LO_SOM.
16053         (symbolic_operand): Change default case to break.
16054         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16055         CONST_DOUBLE values to be reloaded by putting them into memory when
16056         the destination is a floating point register.
16057         (movdf): Remove code to handle CONST_DOUBLE.
16058         (movsf): Likewise.
16059         (reload_indf_r1): New.
16060         (reload_insf_r1): New.
16061         Consistently use "Q" and "T" constraints with integer and floating
16062         point move instructions, respectively.
16063         (movdi): Remove FAIL.
16064         Change predicate for source operand unamed DImode move from
16065         general_operand to move_src_operand.
16066         (umulsidi3): Change predicate for destination operand to
16067         register_operand.
16068         Likewise for similar unamed patterns.
16069         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16070         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16071         (hppa_legitimize_address): Simplify mask calculation.
16072         (pa_emit_move_sequence): Revised handling of secondary reloads from
16073         REG+D addresses for floating point loads and stores.  Directly handle
16074         loading CONST0_RTX (mode) to a floating point register.
16075         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16076         to floating point registers.  Don't restrict secondary reloads to
16077         floating point registers to integer modes.  Revise some comments and
16078         cleanup some code.
16079         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16080         (pa_legitimate_address_p): New.
16081         (pa_legitimize_reload_address): New.
16082         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16083         (STRICT_REG_OK_FOR_BASE_P): New.
16084         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16085         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16087 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16088             Andrew Dixie  <andrewd@gentrack.com>
16090         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16091         flag set.
16093 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16095         * expmed.c (extract_bit_field_1): Pass the full width of the
16096         structure to get_best_reg_extraction_insn.
16098 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16100         PR target/54601
16101         * configure.ac (use_cxa_atexit): Add AIX.
16102         * configure: Regenerate.
16104         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16106 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16108         PR debug/54793
16109         * final.c (need_profile_function): New variable.
16110         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16111         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16112         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16113         notes, targetm.asm_out.function_prologue doesn't emit anything,
16114         HAVE_prologue and profiler should be emitted before prologue,
16115         set need_profile_function instead of emitting it.
16116         (final_scan_insn): If need_profile_function, emit
16117         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16118         NOTE_INSN_FUNCTION_BEG note.
16120 2013-02-01  Richard Henderson  <rth@redhat.com>
16122         * config/rs6000/rs6000.md (smulditi3): New.
16123         (umulditi3): New.
16125         * config/alpha/alpha.md (umulditi3): New.
16127 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16129         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16130         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16132 2013-02-01  Richard Biener  <rguenther@suse.de>
16134         PR tree-optimization/56113
16135         * tree-ssa-structalias.c (label_visit): Reduce work for
16136         single-predecessor nodes.
16138 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
16140         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16141         range isn't testing for zero.
16143 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
16145         PR middle-end/56113
16146         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16148 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
16149             Nick Clifton  <nickc@redhat.com>
16151         * config/v850/constraints.md (Q): Define as a memory constraint.
16152         * config/v850/predicates.md (label_ref_operand): New predicate.
16153         (e3v5_shift_operand): New predicate.
16154         (ior_operator): New predicate.
16155         * config/v850/t-v850: Add e3v5 multilib.
16156         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16157         (v850_gen_movdi): Prototype.
16158         * config/v850/v850.c: Add support for e3v5 architecture.
16159         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16160         TARGET_V850E_UP.
16161         (construct_save_jarl): Add e3v5 long JARL support.
16162         (v850_adjust_insn_length): New function.  Adjust length of call
16163         insns when using e3v5 instructions.
16164         (v850_gen_movdi): New function: Generate instructions to move a
16165         DImode value.
16166         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16167         (CPP_SPEC): Define __v850e3v5__ as appropriate.
16168         (TARGET_USE_FPU): Enable for e3v5.
16169         (CONST_OK_FOR_W): New macro.
16170         (ADJUST_INSN_LENGTH): Define.
16171         * config/v850/v850.md (UNSPEC_LOOP): Define.
16172         (attr cpu): Add v850e3v5.
16173         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16174         (movdi): New pattern.
16175         (movdi_internal): New pattern.
16176         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16177         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16178         (cstoresf4): Likewise.
16179         (cstoredf4): Likewise.
16180         (insv): New pattern.
16181         (rotlso3_a): New pattern.
16182         (rotlsi3_b): New pattern
16183         (rotlsi3_v850e3v5): New pattern.
16184         (doloop_begin): New pattern.
16185         (fix_loop_counter): New pattern.
16186         (doloop_end): New pattern.
16187         (branch_normal): Add e3v5 long branch support.
16188         (branch_invert): Likewise.
16189         (branch_z_normal): Likewise.
16190         (branch_z_invert): Likewise.
16191         (branch_nz_normal): Likewise.
16192         (branch_nz_invert): Likewise.
16193         (call_internal_short): Add e3v5 register-indirect JARL support.
16194         (call_internal_long): Likewise.
16195         (call_value_internal_short): Likewise.
16196         (call_value_internal_long): Likewise.
16197         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16198         (mloop): New option.
16199         * config.gcc: Add support for configuring v840e3v5 target.
16200         * doc/invoke.texi: Document new v850 specific command line options.
16202 2013-01-31  Paul Koning  <ni1d@arrl.net>
16204         PR debug/55059
16205         PR debug/54508
16206         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16207         children if parent is a class.
16208         (prune_unused_types_prune): Don't add DW_AT_declaration.
16210 2013-01-31  Richard Biener  <rguenther@suse.de>
16212         PR tree-optimization/56157
16213         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16214         match up operand with SLP child.
16216 2013-01-31  Jason Merrill  <jason@redhat.com>
16218         PR debug/54410
16219         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16220         parameters the first time.
16221         (gen_scheduled_generic_parms_dies): Check completeness here.
16223 2013-01-31  Richard Biener  <rguenther@suse.de>
16225         PR middle-end/53073
16226         * common.opt (faggressive-loop-optimizations): New flag,
16227         enabled by default.
16228         * doc/invoke.texi (faggressive-loop-optimizations): Document.
16229         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16230         infer_loop_bounds_from_undefined by it.
16232 2013-01-31  Richard Biener  <rguenther@suse.de>
16234         PR tree-optimization/56150
16235         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16236         visit virtual operands.
16237         (find_uses_to_rename_bb): Likewise.
16239 2013-01-31  Richard Biener  <rguenther@suse.de>
16241         PR tree-optimization/56150
16242         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16243         mixed store non-store stmts.
16245 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
16247         PR sanitizer/55374
16248         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16249         LIBASAN_EARLY_SPEC is defined.
16250         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16251         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16252         before %o.
16253         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16255         PR c++/55742
16256         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16257         invalid args instead of ICEing on it.
16258         (ix86_valid_target_attribute_tree): Return error_mark_node if
16259         ix86_valid_target_attribute_inner_p failed.
16260         (ix86_valid_target_attribute_p): Return false only if
16261         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
16262         target("default") attribute.
16263         (sorted_attr_string): Change argument from const char * to tree,
16264         merge in all target attribute arguments rather than just one.
16265         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
16266         instead of free.  Avoid using strcat.
16267         (ix86_mangle_function_version_assembler_name): Mangle
16268         target("default") as if no target attribute is present.  Adjust
16269         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
16270         instead of xmalloc and XDELETEVEC instead of free.
16271         (ix86_function_versions): Don't return true if one of the decls
16272         doesn't have target attribute.  If they don't and one of the decls
16273         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
16274         sorted_attr_string caller.  Use XDELETEVEC instead of free.
16275         (ix86_supports_function_versions): Remove.
16276         (make_name): Fix up formatting.
16277         (make_dispatcher_decl): Remove resolver_name and its initialization.
16278         Avoid leaking memory.
16279         (is_function_default_version): Return true if there is
16280         target("default") attribute rather than no target attribute at all.
16281         (make_resolver_func): Avoid leaking memory.
16282         (ix86_generate_version_dispatcher_body): Likewise.
16283         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16284         * target.def (supports_function_versions): Remove.
16285         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16286         * doc/tm.texi: Regenerated.
16288 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
16290         PR rtl-optimization/56144
16291         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16292         for values with side effects.
16294 2013-01-30  Richard Biener  <rguenther@suse.de>
16296         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16297         (sparseset_pop): Likewise.
16298         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
16299         to be able to use quick_push in the worker loop.
16301 2013-01-30  Marek Polacek  <polacek@redhat.com>
16303         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16305 2013-01-30  Richard Biener  <rguenther@suse.de>
16307         PR lto/56147
16308         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16310 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
16312         PR tree-optimization/56064
16313         * fixed-value.c (fixed_from_double_int): New function.
16314         * fixed-value.h (fixed_from_double_int): New prototype.
16315         (const_fixed_from_double_int): New static inline function.
16316         * fold-const.c (native_interpret_fixed): New static function.
16317         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16318         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16319         (native_encode_fixed): New static function.
16320         (native_encode_expr) <FIXED_CST>: Use it.
16321         (native_interpret_int): Move double_int worker code to...
16322         * double-int.c (double_int::from_buffer): ...this new static method.
16323         * double-int.h (double_int::from_buffer): Prototype it.
16325 2013-01-30  Richard Biener  <rguenther@suse.de>
16327         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16328         New pointer-map and obstack.
16329         (init_alias_vars): Allocate pointer-map and obstack.
16330         (delete_points_to_sets): Free them.
16331         (find_what_var_points_to): Cache result.
16332         (find_what_p_points_to): Adjust for changed interface of
16333         find_what_var_points_to.
16334         (compute_points_to_sets): Likewise.
16335         (ipa_pta_execute): Likewise.
16337 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16339         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16340         * configure: Regenerate.
16341         * config.in: Regenerate.
16342         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16343         #nobits/#progbits if supported.
16345 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
16347         PR target/56121
16348         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16349         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16350         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16352 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16354         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16355         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16357 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16359         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16360         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16362 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16364         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16365         declaration.
16366         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16367         * config/arm/cortex-a7.md: New bypasses using
16368         arm_mac_accumulator_is_result.
16370 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16372         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
16373         (cortex_a7_neon_mla): Likewise.
16374         (cortex_a7_fpfmad): New reservation.
16375         (cortex_a7_fpmacs): Use ffmas and update required units.
16376         (cortex_a7_fpmuld): Update required units and latency.
16377         (cortex_a7_fpmacd): Likewise.
16378         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16379         (cortex_a7_neon). Likewise.
16380         (bypass) Update participating units.
16382 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16384         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16385         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16386         from fmac to ffma.
16387         * config/arm/vfp11.md (vfp_farith): Use ffmas.
16388         (vfp_fmul): Use ffmad.
16389         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16390         (cortex_r4_fmacd): Use ffmad.
16391         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16392         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
16393         (cortex_a9_fmacd): Use ffmad.
16394         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16395         (cortex_a8_vfp_macd): Use ffmad.
16396         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16397         (cortex_a5_fpmacd): Use ffmad.
16398         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16399         (cortex_a15_vfp_macd): Use ffmad.
16400         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16402 2013-01-29  Jason Merrill  <jason@redhat.com>
16404         PR libstdc++/54314
16405         * varasm.c (default_assemble_visibility): Don't warn about
16406         visibility on artificial decls.
16408 2013-01-29  Richard Biener  <rguenther@suse.de>
16410         PR tree-optimization/56113
16411         * tree-ssa-structalias.c (equiv_class_lookup): Also return
16412         the bitmap leader.
16413         (label_visit): Free duplicate bitmaps and record the leader instead.
16414         (perform_var_substitution): Adjust.
16416 2013-01-29  Richard Biener  <rguenther@suse.de>
16418         PR tree-optimization/55270
16419         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16420         the CFG, schedule loops for fixup.
16422 2013-01-29  Nick Clifton  <nickc@redhat.com>
16424         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16425         SP_REG.
16427 2013-01-28  Leif Ekblad  <leif@rdos.net>
16429         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16430         * config/i386/i386.h (TARGET_RDOS): New macro.
16431         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16432         * config/i386/i386.c (ix86_option_override_internal): For 64bit
16433         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16434         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16435         DEFAULT_LARGE_SECTION_THRESHOLD.
16436         * config/i386/i386.md (R14_REG, R15_REG): New constants.
16437         * config/i386/rdos.h: New file.
16438         * config/i386/rdos64.h: New file.
16440 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
16442         PR other/54814
16443         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16444         TEST_HARD_REG_BIT.
16446 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16448         PR rtl-optimization/56117
16449         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16450         call cselib_lookup_from_insn on the MEM before calling
16451         add_insn_mem_dependence.
16453 2013-01-28  Richard Biener  <rguenther@suse.de>
16455         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16456         to a stmt that didn't have one.
16457         (copy_phis_for_bb): Likewise for PHI arguments.
16458         (copy_debug_stmt): Likewise for debug stmts.
16460 2013-01-28  Richard Biener  <rguenther@suse.de>
16462         PR tree-optimization/56034
16463         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16464         (partition_builtin_p): Adjust.
16465         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
16466         it is the last partition.
16467         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16468         up the vertex for the definition.
16469         (classify_partition): Classify whether a partition is a
16470         PKIND_REDUCTION, thus has uses outside of the loop.
16471         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16472         Merge all PKIND_REDUCTION partitions into the last partition.
16473         (tree_loop_distribution): Seed partitions from reductions as well.
16475 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16477         PR tree-optimization/56125
16478         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16479         pow(x,c) into sqrt(x) * powi(x, n/2) or
16480         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16481         optimizing for size.
16482         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16483         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16484         integer.
16486         PR tree-optimization/56094
16487         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16488         to UNKNOWN_LOCATION while gimplifying expr.
16490 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
16492         PR target/56114
16493         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16494         operand 0 in movabs insn template for -masm=intel asm alternative.
16495         (*movabs<mode>_2): Ditto for operand 1.
16497 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
16499         PR target/54663
16500         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16501         of microblaze-c.o
16503 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16505         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16506         tm_file.
16508 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16510         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16511         Undef to avoid warning.
16513 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16515         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16516         * configure: Regenerate.
16518 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
16520         PR tree-optimization/56098
16521         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16522         for stmts with volatile ops.
16523         (cond_store_replacement): Don't optimize if assign has volatile ops.
16524         (cond_if_else_store_replacement_1): Don't optimize if either
16525         then_assign or else_assign have volatile ops.
16526         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16527         volatile ops.
16529 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16531         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16533 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16535         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16536         missing ':' in asm example.
16538 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
16540         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16541         entries into lane and laneq entries.
16542         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16543         Remove AdvSIMD scalar modes.
16544         (aarch64_sq<r>dmulh_laneq<mode>): New.
16545         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16546         modes.
16547         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16548         builtin implementations to relfect changes in RTL in aarch64-simd.md.
16549         * config/aarch64/iterators.md (VCOND): New.
16550         (VCONQ): New.
16552 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16554         PR target/54222
16555         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16556         Add NULL LIBNAME argument to existing definitions.
16557         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16558         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16559         * config/avr/avr.c (DEF_BUILTIN): Same.
16560         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16561         (avr_expand_builtin): Expand to a vanilla call if a libgcc
16562         implementation is available (DECL_ASSEMBLER_NAME is set).
16563         (avr_fold_absfx): New static function.
16564         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16565         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16566         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16567         AVR_BUILTIN_ABSLLK.
16568         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16569         (abshk, absk, abslk, absllk): Provide as static inline functions.
16571 2013-01-25  Marek Polacek  <polacek@redhat.com>
16573         PR tree-optimization/56035
16574         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16576 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
16578         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16579         (*movtf_internal_rex64): Add (!o,C) alternative
16580         (*movxf_internal_rex64): Ditto.
16581         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16583 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
16585         * doc/invoke.texi: fix typo.
16586         * doc/objc.texi: fix typo.
16588 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
16590         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16591         for the first two alternatives.
16593 2013-01-24  Diego Novillo  <dnovillo@google.com>
16595         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
16596         (ggc-zone.o): Remove.
16597         * configure.ac: Remove option --with-gc.
16598         * configure: Re-generate.
16599         * doc/install.texi: Remove documentation for --with-gc.
16600         * gengtype.c (write_enum_defn): Remove.  Update all users.
16601         (write_Types_process_field): Remove generation of gt_e_* argument.
16602         (output_type_enum): Remove.  Update all users.
16603         (write_enum_defn): Remove.  Update all users.
16604         (enum alloc_zone): Remove.  Update all users.
16605         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16606         * ggc-common.c (ggc_splay_alloc): Remove first argument.
16607         Update all callers.
16608         (struct ptr_data): Remove field TYPE.  Update all users.
16609         (gt_pch_note_object): Remove argument TYPE.  Update all users.
16610         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16611         Update all users.
16612         * ggc-none.c (ggc_alloc_typed_stat): Remove.
16613         (struct alloc_zone): Remove.
16614         (ggc_internal_alloc_zone_stat): Remove.
16615         (ggc_internal_cleared_alloc_zone_stat): Remove.
16616         * ggc-page.c (ggc_alloc_typed_stat): Remove.
16617         (ggc_pch_count_object): Remove last argument.  Update all users.
16618         (ggc_pch_alloc_object): Remove last argument.  Update all users.
16619         (struct alloc_zone): Remove.
16620         * ggc-zone.c: Remove.
16621         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
16622         (struct alloc_zone): Remove.
16623         (ggc_alloc_typed_stat): Remove.
16624         (ggc_alloc_typed): Remove.
16625         (ggc_splay_alloc): Remove first argument.
16626         (rtl_zone): Remove.  Update all users.
16627         (tree_zone): Remove.  Update all users.
16628         (tree_id_zone): Remove.  Update all users.
16629         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
16630         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
16631         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
16632         * tree-ssanames.c: Remove references to zone allocator in comments.
16634 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
16636         * config/avr/avr.c (avr_out_fract): Make register numbers that
16637         might be outside of source operand signed.
16639 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
16641         * config/i386/constraints.md (Yf): New constraint.
16642         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
16643         of f constraint to conditionaly disable x87 register preferences.
16644         (*movdf_internal): Ditto.
16645         (*movsf_internal): Ditto.
16647 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
16649         PR inline-asm/55934
16650         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
16651         that have operands with impossible constraints.
16652         Add a FIXME for a speed-up opportunity.
16653         * lra-constraints.c (process_alt_operands): Verify that a class
16654         selected from constraints on asms is valid for the operand mode.
16655         (curr_insn_transform): Remove incorrect comment.
16657 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
16659         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
16660         TOC operand is a valid symbol ref in the constant pool.
16662 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16664         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
16666 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
16668         PR target/54222
16669         * config/avr/stdfix.h: New file.
16670         * t-avr (stdfix-gcc.h): New rule to build it.
16671         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
16673 2013-01-23  Kostya Serebryany  <kcc@google.com>
16675         * config/darwin.h: remove dependency on
16676         CoreFoundation (asan on Mac OS).
16678 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
16680         PR target/49069
16681         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
16682         instead of cmpdi_operand for first comparison operand.
16683         Don't assert that comparison operands aren't both constants.
16685 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16687         * doc/install.texi (Downloading the Source): Update references to
16688         downloading separate components.
16690 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16692         * doc/extend.texi (__int128): Improve grammar.
16694 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
16696         PR target/56028
16697         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
16698         alternative to (o,r).
16699         (*movdi_internal_rex64): Remove (!o,n) alternative.
16700         (DImode immediate->memory splitter): Remove.
16701         (DImode immediate->memory peephole2): Remove.
16702         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
16703         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
16704         alternative to (!o,*r).
16705         (*movtf_internal_sse): New pattern.
16706         (*movxf_internal_rex64): New pattern.
16707         (*movxf_internal): Disable for TARGET_64BIT.
16708         (*movdf_internal_rex64): Remove (!o,F) alternative.
16710 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
16712         PR middle-end/56074
16713         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
16714         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
16715         * tree-vect-loop-manip.c (find_loop_location): Also ignore
16716         stmt locations where LOCATION_LOCUS of the stmt location is
16717         UNKNOWN_LOCATION or BUILTINS_LOCATION.
16719         PR target/55686
16720         * config/i386/i386.md (UNSPEC_STOS): New.
16721         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
16722         *strsetqi_1): Add UNSPEC_STOS.
16724 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
16726         PR c++/56067
16727         * doc/invoke.texi: Remove left over -Wsynth example.
16729 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16731         PR tree-optimization/56051
16732         * fold-const.c (fold_binary_loc): Don't fold
16733         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
16734         a narrowing conversion, or widening conversion from signed
16735         to unsigned.
16737 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16739         PR rtl-optimization/56023
16740         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
16741         dependent on debug instruction.
16743 2013-01-21  Martin Jambor  <mjambor@suse.cz>
16745         PR middle-end/56022
16746         * function.c (allocate_struct_function): Call
16747         invoke_set_current_function_hook earlier.
16749 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16751         * reload1.c (init_reload): Only initialize reload_obstack
16752         during the first call.
16754 2013-01-21  Marek Polacek  <polacek@redhat.com>
16756         * cfgloop.c (verify_loop_structure): Fix up grammar.
16758 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
16760         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
16761         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
16763 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16765         PR target/56058
16766         * config/arm/marvell-pj4.md: Update copyright year.
16767         Fix up use of alu to alu_reg and simple_alu_imm.
16769 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16771         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
16773 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
16775         PR target/55433
16776         * lra-constraints.c (curr_insn_transform): Don't reuse original
16777         insn for secondary memory move when memory mode should be different.
16779 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16781         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
16782         atomic_storedi_1): New patterns.
16784 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
16786         btver2 pipeline descriptions.
16787         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
16788         descriptions.
16789         * config/i386/i386.md (btver2_decode): New type attributes.
16790         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
16791         type attributes.
16792         * config/i386/btver2.md: New file describing btver2 pipelines.
16794 2013-01-19  Andrew Pinski  <apinski@cavium.com>
16796         PR tree-optimization/52631
16797         * tree-ssa-sccvn (visit_use): Before looking up the original
16798         statement, try looking up the simplified expression.
16800 2013-01-19  Anthony Green  <green@moxielogic.com>
16802         * config/moxie/moxie.c (moxie_expand_prologue): Set
16803         current_function_static_stack_size.
16805 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16807         PR tree-optimization/56029
16808         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
16809         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
16811 2013-01-18  Sharad Singhai  <singhai@google.com>
16813         PR tree-optimization/55995
16814         * dumpfile.c (dump_loc): Print location only if available.
16815         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
16817 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
16819         PR target/55433
16820         * lra-constraints.c (curr_insn_transform): Reuse original insn for
16821         secondary memory move.
16822         (inherit_reload_reg): Use rclass instead of cl for
16823         check_secondary_memory_needed_p.
16825 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16827         PR middle-end/56015
16828         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
16829         the case where writing real complex part of target modifies op1.
16831 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
16833         * config/aarch64/aarch64-simd.md
16834         (aarch64_vcond_internal<mode>): Handle unordered cases.
16835         * config/aarch64/iterators.md (v_cmp_result): New.
16837 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
16838             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16840         * config/arm/marvell-pj4.md: New file.
16841         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
16842         * config/arm/arm.md (generic_sched): Add marvell_pj4.
16843         (generic_vfp): Likewise.
16844         * config/arm/arm-cores.def: Add marvell-pj4.
16845         * config/arm/arm-tune.md: Regenerate.
16846         * config/arm/arm-tables.opt: Regenerate.
16847         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
16848         * doc/invoke.texi: Document marvell-pj4.
16850 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
16852         * config/aarch64/arm_neon.h: Map scalar types to standard types.
16854 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16856         PR debug/54114
16857         PR debug/54402
16858         PR debug/49888
16859         * var-tracking.c (negative_power_of_two_p): New.
16860         (global_get_addr_cache, local_get_addr_cache): New.
16861         (get_addr_from_global_cache, get_addr_from_local_cache): New.
16862         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
16863         heading comment.
16864         (vt_stack_offset_p): Remove.
16865         (vt_canon_true_dep): Always canonicalize loc's address.
16866         (clobber_overlapping_mems): Make sure we have a MEM.
16867         (local_get_addr_clear_given_value): New.
16868         (val_reset): Clear local cached entries.
16869         (compute_bb_dataflow): Create and release the local cache.
16870         Disable duplicate MEMs clobbering.
16871         (emit_notes_in_bb): Clobber MEMs likewise.
16872         (vt_emit_notes): Create and release the local cache.
16873         (vt_initialize, vt_finalize): Create and release the global
16874         cache, respectively.
16875         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
16877 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16879         PR libmudflap/53359
16880         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
16881         not found in the symtab.
16883 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16885         PR debug/56006
16886         PR rtl-optimization/55547
16887         PR rtl-optimization/53827
16888         PR debug/53671
16889         PR debug/49888
16890         * alias.c (offset_overlap_p): New, factored out of...
16891         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
16892         the conservative special case for symbolic constants.  Don't
16893         adjust zero sizes on alignment.
16895 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
16897         PR rtl-optimization/52573
16898         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
16899         REG_UNUSED for the same register.
16901 2013-01-17  Richard Biener  <rguenther@suse.de>
16902             Marek Polacek  <polacek@redhat.com>
16904         PR rtl-optimization/55833
16905         * loop-unswitch.c (unswitch_loops): Move loop verification...
16906         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
16907         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
16908         Set it to true when we're removing a loop from hierarchy tree in
16909         an irreducible region.
16910         (fix_bb_placements): Adjust caller.
16911         (fix_loop_placements): Likewise.
16913 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
16915         * config/avr/builtins.def (DEF_BUILTIN): Factor out
16916         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
16917         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
16918         Remove ID.  Adjust comments.
16919         * config/avr/avr-c.c (avr_builtin_name): Remove.
16920         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
16921         * config/avr/avr.c (avr_tolower): New static function.
16922         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
16923         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
16924         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
16925         default expansion.
16927 2013-01-17  Jan Hubicka  <jh@suse.cz>
16929         PR tree-optimization/55273
16930         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
16932 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
16934         PR target/55981
16935         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
16936         store through atomic_store<mode>_1.
16937         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
16939 2013-01-17  Martin Jambor  <mjambor@suse.cz>
16941         PR tree-optimizations/55264
16942         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
16943         for virtual methods.
16944         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
16945         virtual methods before inlining is over.
16946         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
16947         virtual functions.
16948         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
16949         non-virtual.
16951 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
16953         PR rtl-optimization/56005
16954         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
16955         pending reads for prefetch.
16957 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
16959         * config/aarch64/aarch64.md
16960         (*cstoresi_neg_uxtw): New pattern.
16961         (*cmovsi_insn_uxtw): New pattern.
16962         (*<optab>si3_uxtw): New pattern.
16963         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
16964         (*<optab>si3_insn_uxtw): New pattern.
16965         (*bswapsi2_uxtw): New pattern.
16967 2013-01-16  Richard Biener  <rguenther@suse.de>
16969         * tree-inline.c (tree_function_versioning): Remove set but
16970         never used variable.
16972 2013-01-16  Richard Biener  <rguenther@suse.de>
16974         PR tree-optimization/55964
16975         * tree-flow.h (rename_variables_in_loop): Remove.
16976         (rename_variables_in_bb): Likewise.
16977         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
16978         (copy_loop_before): Adjust and delete update-ssa status.
16979         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
16980         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
16981         (rename_variables_in_loop): Remove.
16982         (slpeel_update_phis_for_duplicate_loop): Likewise.
16983         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
16984         use available cfg machinery instead of duplicating it.
16985         Update PHI nodes and perform poor-mans SSA update here.
16986         (slpeel_tree_peel_loop_to_edge): Adjust.
16988 2013-01-16  Richard Biener  <rguenther@suse.de>
16990         PR tree-optimization/54767
16991         PR tree-optimization/53465
16992         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
16993         (vrp_visit_phi_node): For PHI arguments coming via backedges
16994         drop all symbolical range information.
16995         (execute_vrp): Compute backedges.
16997 2013-01-16  Richard Biener  <rguenther@suse.de>
16999         * doc/install.texi: Update CLooG and ISL requirements to
17000         0.18.0 and 0.11.1.
17002 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17004         PR target/55301
17005         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17006         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17007         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17009 2013-01-16  DJ Delorie  <dj@redhat.com>
17011         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17012         (UNSPECV_SP_SWITCH_E): New.
17013         (sp_switch_1): Change to an unspec.
17014         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17015         replace $r15.
17017 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17019         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17020         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17021         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17022         (expand_mem_thread_fence): Ditto.
17023         (expand_mem_signal_fence): Ditto.
17024         (expand_atomic_load): Ditto.
17025         (expand_atomic_store): Ditto.
17027 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17029         PR rtl-optimization/55547
17030         PR rtl-optimization/53827
17031         PR debug/53671
17032         PR debug/49888
17033         * alias.c (memrefs_conflict_p): Set sizes to negative after
17034         AND adjustments.
17036 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17038         PR target/55940
17039         * function.c (thread_prologue_and_epilogue_insns): Always
17040         add crtl->drap_reg to set_up_by_prologue.set, even if
17041         stack_realign_drap is false.
17043 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17045         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17046         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17047         *call): Fix indention.
17049 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17051         PR target/55876
17052         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17053         Update comment.
17055 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17057         PR rtl-optimization/55153
17058         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17060 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17062         PR tree-optimization/55920
17063         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17064         accesses as grp_to_be_debug_replaced.
17066 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17068         PR tree-optimization/55920
17069         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17070         there is non-useless type conversion needed from debug rhs to lhs,
17071         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17073 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17074             Mikael Pettersson  <mikpe@it.uu.se>
17076         PR target/43961
17077         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17078         Thumb.
17079         (ASM_OUTPUT_CASE_LABEL): Remove.
17080         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17081         * final.c (shorten_branches): Update alignment of labels before
17082         jump tables if CASE_VECTOR_SHORTEN_MODE.
17084 2013-01-15  Richard Biener  <rguenther@suse.de>
17086         PR bootstrap/55961
17087         * system.h: Do not include gmp.h for building host tools.
17089 2013-01-15  Richard Biener  <rguenther@suse.de>
17091         PR middle-end/55882
17092         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17093         account for bitpos when computing alignment.
17095 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17097         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17098         (ix86_target_macros_internal): Likewise.
17100         * config/i386/i386.c (m_CORE2I7): Removed.
17101         (m_CORE_HASWELL): New macro.
17102         (m_CORE_ALL): Likewise.
17103         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17104         (initial_ix86_arch_features): Likewise.
17105         (processor_target_table): Initializations for Core avx2.
17106         (cpu_names): New names "core-avx2".
17107         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17108         PROCESSOR_CORE_HASWELL.
17109         (ix86_issue_rate): New case.
17110         (ia32_multipass_dfa_lookahead): Likewise.
17111         (ix86_sched_init_global): Likewise.
17113         * config/i386/i386.h (TARGET_HASWELL): New macro.
17114         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17115         (processor_type): New PROCESSOR_HASWELL.
17117 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17119         PR tree-optimization/55955
17120         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17121         *SHIFT_EXPR and *ROTATE_EXPR codes.
17123         PR tree-optimization/48766
17124         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17125         -ftrapv disable -fwrapv.
17127 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17129         PR target/55974
17130         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17131         etc. to 1 and not to __flash.
17132         Use LL suffix for __INT24_MAX__ with -mint8.
17133         Use ULL suffix for __UINT24_MAX__ with -mint8.
17135 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17137         * config/avr/avr-arch.h
17138         (struct base_arch_s): Use typedef avr_arch_t instead.
17139         (struct arch_info_s): Use typedef avr_arch_info_t instead.
17140         (struct mcu_type_s): Use typedef avr_mcu_t instead.
17141         * config/avr/avr.c: Same.
17142         * config/avr/avr-devices.c: Same.
17143         * config/avr/driver-avr.c: Same.
17144         * config/avr/gen-avr-mmcu-texi.c: Same.
17145         * config/avr/avr-mcus.def: Adjust comment.
17147 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
17149         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17150         * config/aarch64/iterators.md (VALLDI): New.
17152 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
17153             Andi Kleen  <ak@linux.intel.com>
17155         PR target/55948
17156         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17157         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17158         memmodel flag.
17160 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17162         * config/avr/avr-stdint.h: Remove trailing blanks.
17163         * config/avr/avr-log.h: Same.
17164         * config/avr/avr-arch.h: Same.
17165         * config/avr/avr-devices.c: Same.
17166         * config/avr/avr-dimode.md: Same.
17167         * config/avr/predicates.md: Same.
17168         * config/avr/avr-c.c: Same.  And fix typo.
17170         * config/avr/avr-protos.h: Same.  And:
17171         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17172         (init_cumulative_args): Rename to avr_init_cumulative_args.
17173         (expand_prologue): Rename to avr_expand_prologue.
17174         (expand_epilogue): Rename to avr_expand_epilogue.
17175         (adjust_insn_length): Rename to avr_adjust_insn_length.
17176         (notice_update_cc): Rename to avr_notice_update_cc.
17177         (final_prescan_insn): Rename to avr_final_prescan_insn.
17178         * config/avr/avr.c: Same.
17179         * config/avr/avr.h: Same.
17180         * config/avr/avr.md: Remove trailing blanks.
17181         (prologue): Use avr_expand_prologue.
17182         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17184 2013-01-14  Richard Biener  <rguenther@suse.de>
17186         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17187         verify_location, collect_subblocks): New functions.
17188         (verify_gimple_in_cfg): Verify that locations only reference
17189         BLOCKs in the functions BLOCK tree.
17191 2013-01-14  Richard Biener  <rguenther@suse.de>
17193         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17194         PHI argument.
17195         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17196         unshare reference.
17197         (insert_out_of_ssa_copy_on_edge): Likewise.
17198         (rewrite_close_phi_out_of_ssa): Likewise.
17199         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17200         debug expressions.
17201         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17202         propagated constants.
17203         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17204         can not be shared.
17206 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17208         * config/avr/avr-modes.def: Add GPL copyright notice.
17210 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
17212         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17213         MEMMODEL_MASK to determine memory model.
17214         (atomic_store<mode>): Ditto from operands[2].
17215         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17217 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
17219         PR fortran/55935
17220         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17221         (fold_gimple_assign): Don't call unshare_expr here.
17222         (fold_ctor_reference): Call unshare_expr.
17224 2013-01-13  Terry Guo  <terry.guo@arm.com>
17226         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17227         * doc/fragments.texi: Document MULTILIB_REUSE.
17228         * gcc.c (multilib_reuse): New internal spec.
17229         (set_multilib_dir): Also search multilib from multilib_reuse.
17230         * genmultilib (tmpmultilib3): Refactor code.
17231         (tmpmultilib4): Ditto.
17232         (multilib_reuse): New multilib argument.
17234 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
17236         * Makefile.in: Update copyright.
17238 2013-01-12  Tom de Vries  <tom@codesourcery.com>
17240         PR middle-end/55890
17241         * calls.c (expand_call): Check if arg_nr is valid.
17243 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17245         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17246         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17247         documentation.  Add missing '__' in front of
17248         __builtin_ia32_packssdw256.
17250 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17252         PR target/55719
17253         * config/s390/s390.c (s390_preferred_reload_class): Do not return
17254         NO_REGS for larl operands.
17255         (s390_reload_larl_operand): Use s390_load_address instead of
17256         emit_move_insn.
17258 2013-01-11  Richard Biener  <rguenther@suse.de>
17260         * tree-cfg.c (verify_node_sharing_1): Split out from ...
17261         (verify_node_sharing): ... here.
17262         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17264 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
17266         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17267         Substitute TREECHECKING.
17268         * configure: Regenerate.
17269         * Makefile.in (TREECHECKING): New.
17271 2013-01-11  Richard Guenther  <rguenther@suse.de>
17273         PR tree-optimization/44061
17274         * tree-vrp.c (extract_range_basic): Compute zero as
17275         value-range for __builtin_constant_p of function parameters.
17277 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
17279         Update copyright years.
17281 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
17283         PR rtl-optimization/55672
17284         * lra-eliminations.c (mark_not_eliminable): Permit addition with
17285         const to be eliminable.
17287 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
17289         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17290         * configure: Regenerate.
17292 2013-01-10  Richard Biener  <rguenther@suse.de>
17294         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17296 2013-01-10  Richard Biener  <rguenther@suse.de>
17298         PR bootstrap/55792
17299         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17300         locations for virtual PHI arguments.
17301         (rewrite_update_phi_arguments): Likewise.
17303 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
17305         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17306         on to assembler.
17308 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
17310         PR tree-optimization/55921
17311         * tree-complex.c (expand_complex_asm): New function.
17312         (expand_complex_operations_1): Call it for GIMPLE_ASM.
17314 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17316         PR target/55718
17317         * config/s390/s390.c (s390_symref_operand_p)
17318         (s390_loadrelative_operand_p): Merge the two functions.
17319         (s390_check_qrst_address, print_operand_address): Add parameters
17320         to s390_loadrelative_operand_p invokation.
17321         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17322         (s390_reload_larl_operand, s390_secondary_reload): Use
17323         s390_loadrelative_operand_p instead of s390_symref_operand_p.
17324         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17326 2013-01-09  Mike Stump  <mikestump@comcast.net>
17328         * dse.c (record_store): Remove unnecessary assert.
17330 2013-01-09  Jan Hubicka  <jh@suse.cz>
17332         PR tree-optimization/55569
17333         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17334         * cfgloop.h (scale_loop_profile): Likewise.
17336 2013-01-09  Jan Hubicka  <jh@suse.cz>
17338         PR lto/45375
17339         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17340         functions.
17341         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17343 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
17345         PR middle-end/55114
17346         * expr.h (maybe_emit_group_store): Declare.
17347         * expr.c (maybe_emit_group_store): New function.
17348         * builtins.c (expand_builtin_int_roundingfn): Call it.
17349         (expand_builtin_int_roundingfn_2): Likewise.
17351 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
17353         PR rtl-optimization/55829
17354         * lra-constraints.c (match_reload): Add code for absent output.
17355         (curr_insn_transform): Add code for reloads of matched inputs
17356         without output.
17358 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
17360         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17361         attribute of movddup insn to DF.
17362         (*vec_interleave_lowv2df): Ditto.
17363         (vec_dupv2df): Ditto.
17365 2013-01-09  Jan Hubicka  <jh@suse.cz>
17367         PR tree-optimiation/55875
17368         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17369         EVERY_ITERATION parameter.
17370         (number_of_iterations_exit): Check if exit is executed every iteration.
17371         (idx_infer_loop_bounds): Similarly here.
17372         (n_of_executions_at_most): Simplify
17373         to only test for cases where statement is dominated by the
17374         particular bound; handle correctly the "postdominance" test.
17375         (scev_probably_wraps_p): Use max loop iterations info
17376         as a global bound first.
17378 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
17379             Nick Clifton  <nickc@redhat.com>
17381         * config/v850/v850.md (cbranchsf4): New pattern.
17382         (cstoresf4): New pattern.
17383         (cbranchdf4): New pattern.
17384         (cstoredf4): New pattern.
17385         (movsicc): Disallow floating point comparisons.
17386         (cmpsf_le_insn): Fix order of operators.
17387         (cmpsf_lt_insn): Likewise.
17388         (cmpsf_eq_insn): Likewise.
17389         (cmpdf_le_insn): Likewise.
17390         (cmpdf_lt_insn): Likewise.
17391         (cmpdf_eq_insn): Likewise.
17392         (cmpsf_ge_insn): Use LE comparison.
17393         (cmpdf_ge_insn): Likewise.
17394         (cmpsf_gt_insn): Use LT comparison.
17395         (cmpdf_gt_insn): Likewise.
17396         (cmpsf_ne_insn): Delete pattern.
17397         (cmpdf_ne_insn): Delete pattern.
17398         * config/v850/v850.c (v850_gen_float_compare): Use
17399         gen_cmpdf_eq_insn for NE comparison.
17400         (v850_float_z_comparison_operator)
17401         (v850_float_nz_comparison_operator): Move from here ...
17402         * config/v850/predicates.md: ... to here.  Move GT and GE
17403         comparisons into v850_float_z_comparison_operator.
17404         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17405         Delete prototype.
17406         (v850_float_nz_comparison_operator): Likewise.
17408 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17410         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17411         with calls to gen_insvsi/gen_insvdi.
17413 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17415         * config/i386/i386.c (initial_ix86_tune_features): Set up
17416         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17418 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
17419             Jakub Jelinek  <jakub@redhat.com>
17421         PR tree-optimization/48189
17422         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17423         If nitercst is 0, don't predict the exit edge.
17425 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17427         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17428         in asm_fprintf with reg_names.
17429         (aarch64_print_operand_address): Likewise.
17430         (aarch64_return_addr): Likewise.
17431         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17433 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17435         * config/pa/pa.h (VAL_U6_BITS_P): Define.
17436         (INT_U6_BITS): Likewise.
17437         * config/pa/predicates.md (uint6_operand): New predicate.
17438         (shift5_operand, shift6_operand): Likewise.
17439         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17440         arith32_operand.
17441         (lshrdi3): Use shift6_operand.
17442         (shrpsi4, shrpdi4): New insn patterns.
17443         (extzv): Delete expander.
17444         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
17445         predicates in unamed zero extract patterns.  Tighten common constraint.
17446         (extv): Delete expander.
17447         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
17448         predicates in unamed sign extract patterns.  Tighten common constraint.
17449         (insv): Delete expander.
17450         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
17451         predicates in unamed insert patterns.  Tighten common constraint.
17452         Change uint32_operand predicate to uint6_operand predicate in unamed
17453         DImode pattern to insert constant values of type 1...1xxxx.
17455 2013-01-04  Jan Hubicka  <jh@suse.cz>
17457         PR tree-optimization/55823
17458         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17459         issue.
17461 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17462             Uros Bizjak  <ubizjak@gmail.com>
17464         PR rtl-optimization/55845
17465         * df-problems.c (can_move_insns_across): Stop scanning at
17466         volatile_insn_p source instruction or give up if
17467         across_from .. across_to range contains any volatile_insn_p
17468         instructions.
17470 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17472         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17473         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17474         Declare.
17475         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17476         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17478 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17480         PR fortran/55341
17481         * asan.c (asan_clear_shadow): New function.
17482         (asan_emit_stack_protection): Use it.
17484 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17486         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17487         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17488         with tab instead of space.
17490 2013-01-08  Nick Clifton  <nickc@redhat.com>
17492         * config/rl78/rl78.c (rl78_expand_prologue): Always select
17493         register bank 0 at the start of an interrupt handler.
17494         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17495         MDBH registers.
17497 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17499         * config/aarch64/aarch64-simd.md
17500         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17501         (aarch64_simd_bsl): Likewise.
17502         (aarch64_vcond_internal<mode>): Likewise.
17503         (vcond<mode><mode>): Likewise.
17504         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17505         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17507 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17509         * config/aarch64/aarch64-builtins.c
17510         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17512 2013-01-08  Martin Jambor  <mjambor@suse.cz>
17514         PR debug/55579
17515         * tree-sra.c (analyze_access_subtree): Return true also after
17516         potentially creating a debug-only replacement.
17518 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17520         PR middle-end/55890
17521         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17523         PR tree-optimization/54120
17524         * tree-vrp.c (range_fits_type_p): Don't allow
17525         src_precision < precision from signed vr to unsigned_p
17526         if vr->min or vr->max is negative.
17527         (simplify_float_conversion_using_ranges): Test can_float_p
17528         against CODE_FOR_nothing.
17530 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17531             Richard Biener  <rguenther@suse.de>
17533         PR middle-end/55851
17534         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17535         types instead of just INTEGER_TYPE types.
17537 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
17539         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17540         TF_SIZE): Define.
17542 2013-01-07  Steve Ellcey  <sellcey@mips.com>
17544         PR target/42661
17545         * config/mips/mips.opt: Change mad to mmad to match documentation.
17547 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17549         PR target/55897
17550         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17551         .progmemx.data now.
17553 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17555         PR target/55897
17556         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17557         (avr_addrspace_t): Add .section_name field.
17558         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17559         array size.
17560         (avr_addrspace): Same.  Initialize .section_name.  Remove last
17561         NULL entry.  Put __memx into .progmemx.data.
17562         (progmem_section_prefix): Remove.
17563         (avr_asm_init_sections): No need to initialize progmem_section.
17564         (avr_asm_named_section): Use avr_addrspace[].section_name to get
17565         section name prefix.
17566         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
17567         retrieve the progmem section.
17568         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17569         boundary to run over avr_addrspace[].
17570         (avr_register_target_pragmas): Ditto.
17572 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
17574         * varasm.c (output_constant_def_contents): For asan_protect_global
17575         protected strings, adjust DECL_ALIGN if needed, before testing for
17576         anchored symbols.
17577         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17578         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
17579         normal decls.
17580         (output_object_block): For asan protected decls, emit asan padding
17581         after their contents.
17582         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17583         (asan_finish_file): Test it here instead.
17585 2013-01-07  Nick Clifton  <nickc@redhat.com>
17586             Matthias Klose  <doko@debian.org>
17587             Doug Kwan  <dougkwan@google.com>
17588             H.J. Lu  <hongjiu.lu@intel.com>
17590         PR driver/55470
17591         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17593         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17595         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17597         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17599         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17601 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17603         PR target/54461
17604         * doc/install.texi (Cross-Compiler-Specific Options): Document
17605         --with-avrlibc.
17607 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
17609         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17610         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17611         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17612         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17613         vqmovun_high_s64): Fix source operand number and update copyright.
17615 2013-01-07  Richard Biener  <rguenther@suse.de>
17617         PR middle-end/55890
17618         * gimple.h (gimple_call_builtin_p): New overload.
17619         * gimple.c (validate_call): New function.
17620         (gimple_call_builtin_p): Likewise.
17621         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17622         Use gimple_call_builtin_p.
17623         (find_func_clobbers): Likewise.
17624         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
17625         (strlen_optimize_stmt): Likewise.
17627 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17629         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
17630         (vld1q_dup_*): Likewise.
17631         (vld1_*): Likewise.
17632         (vld1q_*): Likewise.
17633         (vld1_lane_*): Likewise.
17634         (vld1q_lane_*): Likewise.
17636 2013-01-07  Richard Biener  <rguenther@suse.de>
17638         * lto-streamer.h (LTO_minor_version): Bump to 2.
17640 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17642         * config/aarch64/aarch64-protos.h
17643         (aarch64_const_double_zero_rtx_p): Rename to...
17644         (aarch64_float_const_zero_rtx_p): ...this.
17645         (aarch64_float_const_representable_p): New.
17646         (aarch64_output_simd_mov_immediate): Likewise.
17647         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
17648         move immediate case.
17649         * config/aarch64/aarch64.c
17650         (aarch64_const_double_zero_rtx_p): Rename to...
17651         (aarch64_float_const_zero_rtx_p): ...this.
17652         (aarch64_print_operand): Allow printing of new constants.
17653         (aarch64_valid_floating_const): New.
17654         (aarch64_legitimate_constant_p): Check for valid floating-point
17655         constants.
17656         (aarch64_simd_valid_immediate): Likewise.
17657         (aarch64_vect_float_const_representable_p): New.
17658         (aarch64_float_const_representable_p): Likewise.
17659         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
17660         (aarch64_output_simd_mov_immediate): New.
17661         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
17662         (*movdf_aarch64): Likewise.
17663         * config/aarch64/constraints.md (Ufc): New.
17664         (Y): call aarch64_float_const_zero_rtx.
17665         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
17667 2013-01-07  Richard Biener  <rguenther@suse.de>
17669         PR tree-optimization/55888
17670         PR tree-optimization/55862
17671         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
17672         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
17673         not if it is contained therein.
17675 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17677         * config/avr/t-avr: Typo.
17679 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17681         PR55243
17682         * config/avr/t-avr: Don't automatically rebuild
17683         $(srcdir)/config/avr/t-multilib
17684         $(srcdir)/config/avr/avr-tables.opt
17685         $(srcdir)/doc/avr-mmcu.texi
17686         (avr-mcus): New phony target to build them on request.
17687         (s-avr-mlib, s-avr-mmcu-texi): Remove.
17688         * avr/avr-mcus.def: Adjust comments.
17690 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
17692         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
17694 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17696         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
17698 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17700         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
17702 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
17704         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
17705         to generate profiling.
17706         * config/rs6000/aix64.h (LIB_SPEC): Same.
17708 2013-01-04  Andrew Pinski  <apinski@cavium.com>
17710         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
17711         New function.
17712         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
17714 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
17716         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
17717         unconditionally.
17718         (ix86_expand_move): Ditto.
17719         (ix86_zero_extend_to_Pmode): Ditto.
17720         (ix86_expand_call): Ditto.
17721         (ix86_expand_special_args_builtin): Ditto.
17722         (ix86_expand_builtin): Ditto.
17724 2013-01-04  Richard Biener  <rguenther@suse.de>
17726         PR tree-optimization/55862
17727         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
17728         translating them through PHI nodes.
17730 2013-01-04  Martin Jambor  <mjambor@suse.cz>
17732         PR tree-optimization/55755
17733         * tree-sra.c (sra_modify_assign): Do not check that an access has no
17734         children when trying to avoid producing a VIEW_CONVERT_EXPR.
17736 2013-01-04  Marek Polacek  <polacek@redhat.com>
17738         PR middle-end/55859
17739         * opts.c (default_options_optimization): Clarify error message.
17741 2013-01-04  Richard Biener  <rguenther@suse.de>
17743         PR middle-end/55863
17744         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
17745         reassociation.
17747 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17749         PR target/53789
17750         * config/pa/pa.md (movsi): Revert previous change.
17751         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
17752         references.
17754 2013-01-03  Richard Henderson  <rth@redhat.com>
17756         * config/i386/i386.c (ix86_expand_move): Always assign to op1
17757         after eliminating TLS symbols.
17759 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
17761         PR bootstrap/50167
17762         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
17763         * graphite-poly.c (debug_gmp_value): Likewise.
17765 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
17767         PR target/55712
17768         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
17769         selected code model, define __code_mode_small__, __code_model_medium__,
17770         __code_model_large__, __code_model_32__ or __code_model_kernel__.
17771         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
17772         xchg temporary register with %k.  Declare temporary register as
17773         early clobbered.
17774         [__x86_64__]: For medium and large code models, preserve %rbx register.
17776 2013-01-03  Richard Biener  <rguenther@suse.de>
17778         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
17779         (dump_subscript): Adjust.
17780         (finalize_ddr_dependent): Do not dump redundant info.
17781         (analyze_siv_subscript): Adjust.
17782         (subscript_dependence_tester): Likewise.
17783         (compute_affine_dependence): Likewise.
17785 2013-01-03  Richard Biener  <rguenther@suse.de>
17787         Revert
17788         2013-01-03  Richard Biener  <rguenther@suse.de>
17790         PR tree-optimization/55857
17791         * tree-vect-stmts.c (vectorizable_load): Do not setup
17792         re-alignment for invariant loads.
17794         2013-01-02  Richard Biener  <rguenther@suse.de>
17796         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17797         invariant load do not generate a vector load from the scalar location.
17799 2013-01-03  Richard Biener  <rguenther@suse.de>
17801         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
17802         for not vectorizing.
17803         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
17804         not build INDIRECT_REFs, call get_name once only.
17805         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
17806         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
17808 2013-01-03  Richard Biener  <rguenther@suse.de>
17810         PR tree-optimization/55857
17811         * tree-vect-stmts.c (vectorizable_load): Do not setup
17812         re-alignment for invariant loads.
17814 2013-01-03  Richard Biener  <rguenther@suse.de>
17816         PR lto/55848
17817         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
17818         prefer a built-in decl.
17820 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17822         * gcc.c (process_command): Update copyright notice dates.
17823         * gcov.c (print_version): Likewise.
17824         * gcov-dump.c (print_version): Likewise.
17826         PR rtl-optimization/55838
17827         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
17828         iv0.step, iv1.step and step.
17830 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17831             Marc Glisse  <marc.glisse@inria.fr>
17833         PR tree-optimization/55832
17834         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
17835         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
17836         integer_{one,zero}_node.
17838 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17840         PR debug/54402
17841         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
17842         * var-tracking.c (reverse_op): Don't add reverse ops to
17843         VALUEs that have already
17844         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
17846 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
17848         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
17850 2013-01-02  Teresa Johnson  <tejohnson@google.com>
17852         * dumpfile.c (dump_loc): Print filename with location.
17853         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
17854         new location_t parameter to emit complete unroll message with
17855         new dump framework.
17856         (canonicalize_loop_induction_variables): Compute loops location
17857         and pass to try_unroll_loop_completely.
17858         * loop-unroll.c (report_unroll_peel): New function.
17859         (peel_loops_completely): Use new dump format with location
17860         for main dumpfile message, and invoke report_unroll_peel on success.
17861         (decide_unrolling_and_peeling): Ditto.
17862         (decide_peel_once_rolling): Remove old dumpfile message subsumed
17863         by report_unroll_peel.
17864         (decide_peel_completely): Ditto.
17865         (decide_unroll_constant_iterations): Ditto.
17866         (decide_unroll_runtime_iterations): Ditto.
17867         (decide_peel_simple): Ditto.
17868         (decide_unroll_stupid): Ditto.
17869         * cfgloop.c (get_loop_location): New function.
17870         * cfgloop.h (get_loop_location): Declare.
17872 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17874         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
17875         NULL.
17877 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17879         PR middle-end/55198
17880         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
17881         BLKmode objects when EXPAND_MEMORY is specified.
17883 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17885         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
17886         in loop predicate.
17887         (fold_builtin_cpu): Do not share cpu model decls across statements.
17889 2013-01-02  Jason Merrill  <jason@redhat.com>
17891         PR c++/55804
17892         * tree.c (build_array_type_1): Revert earlier change.
17894 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
17896         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
17897         "cortex-a57".
17898         * config/aarch64/aarch64-tune.md: Re-generate.
17900 2013-01-02  Richard Biener  <rguenther@suse.de>
17902         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17903         invariant load do not generate a vector load from the scalar location.
17905 2013-01-02  Richard Biener  <rguenther@suse.de>
17907         PR bootstrap/55784
17908         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
17909         * configure: Regenerate.
17911 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
17913         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
17914         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
17915         (expand_builtin_int_roundingfn_2): Keep the original target around
17916         for the fallback case.
17918 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
17920         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
17921         to be clear for sign changes.
17923 2013-01-01  Jan Hubicka  <jh@suse.cz>
17925         * ipa-inline-analysis.c: Fix formatting.
17927 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
17929         PR tree-optimization/55831
17930         * tree-vect-loop.c (get_initial_def_for_induction): Use
17931         gsi_after_labels instead of gsi_start_bb.
17933 Copyright (C) 2013 Free Software Foundation, Inc.
17935 Copying and distribution of this file, with or without modification,
17936 are permitted in any medium without royalty provided the copyright
17937 notice and this notice are preserved.