gcc/
[official-gcc.git] / gcc / ChangeLog
blobfa0a7907556372927a98a23a6ca0aff394915492
1 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
2             Jakub Jelinek  <jakub@redhat.com>
4         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
5         (BUILT_IN_FABSD64): Likewise.
6         (BUILT_IN_FABSD128): Likewise.
7         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8         (fold_builtin_1): Likewise.
9         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping destination
10         and source operands.
11         (*abstd2_fpr): Likewise.
12         (*nabstd2_fpr): Likewise.
14 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
16         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
17         JUMP_P and INSN_P.
19 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
21         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
22         incomplete.
24 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
26         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
27         * builtins.c (default_libc_has_function): New.
28         (gnu_libc_has_function): Ditto.
29         (no_c99_libc_has_function): Ditto.
30         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
31         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
32         (fold_builtin_sincos): Likewise.
33         (fold_builtin_cexp): Likewise.
34         * builtins.def (DEF_C94_BUILTIN): Likewise.
35         (DEF_C99_BUILTIN): Likewise.
36         (DEF_C99_C90RES_BUILTIN): Likewise.
37         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
38         definitions to using this define.
39         * config/darwin-protos.h (darwin_libc_has_function): New.
40         * config/darwin.c: (darwin_libc_has_function: Ditto.
41         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
42         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
43         * config/darwin.h: Ditto.
44         * config/elfos.h: Ditto.
45         * config/freebsd.h: Ditto.
46         * config/i386/cygming.h: Ditto.
47         * config/i386/djgpp.h: Ditto.
48         * config/i386/i386-interix.h: Ditto.
49         * config/microblaze/microblaze.h: Ditto.
50         * config/mmix/mmix.h: Ditto.
51         * config/gnu-user.h: Ditto.
52         * config/ia64/hpux.h: Ditto.
53         * config/pa/pa-hpux.h: Ditto.
54         * config/pdp11/pdp11.h: Ditto.
55         * config/picochip/picochip.h: Ditto.
56         * config/linux.h: Ditto.
57         * config/netbsd.h: Ditto.
58         * config/openbsd.h: Ditto.
59         * config/rs6000/aix43.h: Ditto.
60         * config/rs6000/aix51.h: Ditto.
61         * config/rs6000/aix52.h: Ditto.
62         * config/rs6000/aix53.h: Ditto.
63         * config/rs6000/aix61.h: Ditto.
64         * config/rs6000/darwin.h: Ditto.
65         * config/rs6000/linux.h: Ditto.
66         * config/rs6000/linux64.h: Ditto.
67         * config/s390/tpf.h: Ditto.
68         * config/sol2-10.h: Ditto.
69         * config/sol2.h: Ditto.
70         * config/vms/vms.h: Ditto.
71         * config/vxworks.h: Ditto.
72         * config/linux-android.c (linux_android_libc_has_function):
73         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
74         * config/linux-protos.h (linux_android_libc_has_function):
75         New declaration.
76         * config/i386/i386.c (ix86_libc_has_function): New.
77         * config/i386/i386-protos.h
78         (ix86_libc_has_function): New declaration.
79         * config/i386/i386.md
80         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
81         ("isinf<mode>2): Likewise.
82         * convert.c (convert_to_integer): Using new target hook
83         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
84         TARGET_C99_FUNCTIONS.
85         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
86         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
87         * coretypes.h (function_class): New enum for different
88         classes of functions.
89         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
90         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
91         (TARGET_HAS_SINCOS): Likewise.
92         (TARGET_LIBC_HAS_FUNCTION): New.
93         * doc/tm.texi: Regenerated.
94         * targhooks.h (default_libc_has_function): New declaration.
95         (no_c99_libc_has_function): Ditto.
96         (gnu_libc_has_function): Ditto.
97         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
98         and TARGET_HAS_SINCOS.
100 2013-08-18  Jan Hubicka  <jh@suse.cz>
102         * Makeifle-in (ipa-devirt.o): New.
103         (GTFILES): Add ipa-utils.h and ipa-devirt.c
104         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
105         (analyze_functions): Look into possible targets of polymorphic call.
106         * dumpfile.c (dump_files): Add type-inheritance dump.
107         * dumpfile.h (TDI_inheritance): New.
108         * ipa-devirt.c: New file.
109         * ipa-utils.h (odr_type_d): Forward declare.
110         (odr_type): New type.
111         (build_type_inheritance_graph): Declare.
112         (possible_polymorphic_call_targets): Declare and introduce inline
113         variant when only edge is pased.
114         (dump_possible_polymorphic_call_targets): Likewise.
115         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
116         * tree.c (type_in_anonymous_namespace_p): Break out from ...
117         (types_same_for_odr): ... here.
118         * tree.h (type_in_anonymous_namespace_p): Declare.
120 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
122         PR tree-optimization/58006
123         * tree-parloops.c (take_address_of): Don't ICE if get_name
124         returns NULL.
125         (eliminate_local_variables_stmt): Remove clobber stmts.
127 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
129         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
130         error message is issued for an alias to undefined symbol.
132 2013-08-18  Jan Hubicka  <jh@suse.cz>
134         * cgraph.c (cgraph_create_indirect_edge): Discover
135         polymorphic calls and record basic info into indirect_info.
136         * gimple-fold.c (gimple_fold_call): When doing BINFO based
137         devirtualization, ignore objc function calls.
138         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
139         call with no parm index info.
140         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
141         * tree.c (virtual_method_call_p): New function.
142         * tree.h (virtual_method_call_p): Declare.
144 2013-08-16  Jan Hubicka  <jh@suse.cz>
146         PR middle-end/58179
147         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
149 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
151         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
152         attribute.
154 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
156         * gengtype.c (type_for_name): Add special-case support for
157         locating types within the "gcc::" namespace.
158         (open_base_files): Emit a "using namespace gcc" directive.
160 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
162         PR target/58160
163         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
164         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
166         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
167         array instead of each individual operand as a separate argument.
168         (emit_fusion_gpr_load): Likewise.
169         (expand_fusion_gpr_load): Add new function declaration.
171         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
172         signature to have the operands passed as an array, instead of as
173         separate arguments.  Allow ZERO_EXTEND to be in the memory
174         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
175         depend on the register live/dead flags when peepholes are run.
176         (expand_fusion_gpr_load): New function to be called from the
177         peephole2 pass, to change the register that addis sets to be the
178         target register.
179         (emit_fusion_gpr_load): Change the calling signature to have the
180         operands passed as an array, instead of as separate arguments.
181         Allow ZERO_EXTEND to be in the memory address, and also
182         SIGN_EXTEND if -mpower8-fusion-sign.
184         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
185         unspec enumeration.
186         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
187         adjust the register addis loads up in the peephole2 pass.  Do not
188         depend on the register live/dead state when the peephole pass is
189         done.
191 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
193         * gengtype.c (create_user_defined_type): Ensure that the kind
194         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
195         declaration is seen before the GTY((user)) marking.
197 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
199         PR target/58105
200         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
202 2013-08-16  Jan Hubicka  <jh@suse.cz>
204         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
205         arugment expected_type.
206         (gimple_fold_call): Use it.
207         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
208         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
209         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
210         (try_make_edge_direct_virtual_call): Likewise.
211         * tree.c (obj_type_ref_class): New.
212         * tree.h (obj_type_ref_class): Use it.
214 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
216         * sched-vis.c (rtl_slim_pp_initialized): Remove.
217         (rtl_slim_pp): Likewise.
218         (init_rtl_slim_pretty_print): Likewise.
219         (dump_value_slim):  Don't call it.  Use local pretty printer.
220         (dump_insn_slim): Likewise.
221         (dump_rtl_slim): Likewise.
222         (str_pattern_slim): Likewise.
223         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
224         Simplify.
226 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
228         PR tree-optimization/58164
229         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
230         walk gimple_goto_dest of GIMPLE_GOTO.
232         PR tree-optimization/58165
233         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
234         bi_call must be the last stmt in a bb, don't split_block, instead
235         use fallthru edge from it and give up if there is none.
236         Release conds vector when returning early.
238 2013-08-14  Xinliang David Li  <davidxl@google.com>
240         * config/i386/i386.c (ix86_option_override_internal):
241         Remove unused variable and field.
243 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
245         PR target/57949
246         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
247         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
248         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
249         and Linux, correct BLKmode alignment when 128-bit alignment is
250         required and compatibility flag is not set.
251         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
252         for zero-size arguments when compatibility flag is not set.
254 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
256         PR tree-optimization/58145
257         * tree-sra.c (build_ref_for_offset): If prev_base has
258         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
260 2013-08-14  Xinliang David Li  <davidxl@google.com>
262         * config/i386/i386.c (ix86_option_override_internal):
263         Fix uninitialized variable error.
265 2013-08-14  Xinliang David Li  <davidxl@google.com>
267         * config/i386/i386.opt: Define two new options.
268         * config/i386/x86-tune.def: Add arch selector field in macros.
269         * config/i386/i386.h: Adjust macro definition.
270         * config/i386/i386.c (ix86_option_override_internal):
271         Refactor the code.
272         (parse_mtune_ctrl_str): New function.
273         (set_ix86_tune_features): New function.
274         (ix86_function_specific_restore): Call the new helper function.
276 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
278         PR rtl-optimization/57662
279         * sel-sched.c (code_motion_process_successors): When the current insn
280         is removed after the recursive traversal, break from the loop.
281         Add comments and debug printouts.
283 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
284             Alexandre Oliva  <aoliva@redhat.com>
286         PR target/58067
287         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
288         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
289         there also UNSPEC_PLTOFF.
291 2013-08-14  Marek Polacek  <polacek@redhat.com>
293         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
294         NUM_CONDITIONS bit positions.
296 2013-08-13  Cary Coutant  <ccoutant@google.com>
298         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
299         (attr_checksum): Hash vector contents instead of pointer.
300         (attr_checksum_ordered): Likewise.
302 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
304         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
305         when Pmode != word_mode.  Add length_address attribute.
306         (sse3_monitor_<mode>): Merge from sse3_monitor and
307         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
308         Pmode != word_mode.  Update insn length attribute.
309         * config/i386/i386.c (ix86_option_override_internal): Update
310         ix86_gen_monitor selection for merged sse3_monitor insn.
312 2013-08-13  Julian Brown  <julian@codesourcery.com>
314         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
315         perform invalid legitimization on greater-than-word-size modes for
316         TARGET_E500_DOUBLE.
318 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
320         * ira.c (setup_class_translate_array): Use aclass instead of cl
321         for classes not fully covered by allocno classes.
323 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
325         PR tree-optimization/57661
326         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
327         * tree-inline.c (tree_function_versioning): Initialize it.
328         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
329         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
330         that is not being copied.
332         PR sanitizer/56417
333         * asan.c (instrument_strlen_call): Fix typo in comment.
334         Use char * type even for the lhs of POINTER_PLUS_EXPR.
336 2013-08-13  Steve Ellcey  <sellcey@mips.com>
338         * config/mips/mips.md (prefetch): Use lw instead of ld on
339         loongson in 32bit mode.
341 2013-08-13  Nick Clifton  <nickc@redhat.com>
343         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
345 2013-08-13  Jan Hubicka  <jh@suse.cz>
347         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
348         introduced edge; fix typo in sanity check.
349         (cgraph_resolve_speculation): Export; improve diagnostic.
350         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
351         speculation at type mismatch.
352         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
353         (cgraph_resolve_speculation): Declare.
354         (symtab_can_be_discarded): New function.
355         * value-prof.c (gimple_ic_transform): Remove actual transform code.
356         * ipa-inline-transform.c (speculation_removed): New global var.
357         (clone_inlined_nodes): See if speculation can be removed.
358         (inline_call): If speculations was removed, we growths may not match.
359         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
360         (speculation_useful_p): New function.
361         (resolve_noninline_speculation): New function.
362         (inline_small_functions): Resolve useless speculations.
363         * ipa-inline.h (speculation_useful_p): Declare
364         * ipa.c (can_replace_by_local_alias): Simplify.
365         (ipa_profile): Produce speculative calls in non-lto, too;
366         add simple cost model; produce local aliases.
368 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
370         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
371         PASS_MANAGER_H.
373 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
375         * config/i386/i386.c (ix86_function_versions): Use error + inform.
377 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
379         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
380         iterator instead of X87MODEF.
382 2013-08-12  Perez Read  <netfirewall@gmail.com>
384         PR target/58132
385         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
386         operand 0 for intel asm alternative.
387         (*movabs<mode>_2): Ditto for operand 1.
389 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
391         * config/aarch64/arm_none.h
392         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
394 2013-08-12  Nick Clifton  <nickc@redhat.com>
396         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
398 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
400         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
401         expand for QI/HImode operand to produce more effictive code for
402         unsigned char(short) --> float(double) conversion.
404 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
406         * doc/invoke.texi: Mention that -ftls-model does not force the final
407         model.
409 2013-08-12  Marek Polacek  <polacek@redhat.com>
410             Marc Glisse  <marc.glisse@inria.fr>
412         PR tree-optimization/57980
413         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
414         when creating -1 constant.
416 2013-08-10  Jan Hubicka  <jh@suse.cz>
418         Workaround binutils PR14342.
419         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
420         (gimple_init_edge_profiler): Likewise.
421         (gimple_gen_ic_func_profiler): Likewise.
423 2013-08-09  Jan Hubicka  <jh@suse.cz>
425         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
427 2013-08-09  Xinliang David Li  <davidxl@google.com>
429         * config/i386/stringop.def: New file.
430         * config/i386/stringop.opt: New file.
431         * config/i386/i386-opts.h: Include stringopt.def.
432         * config/i386/i386.opt: Include stringopt.opt.
433         * config/i386/i386.c (ix86_option_override_internal):
434         Override default size based stringop inline strategies with options.
435         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
436         New function.
438 2013-08-09  Jan Hubicka  <jh@suse.cz>
440         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
442 2013-08-09  Jan Hubicka  <jh@suse.cz>
444         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
445         CGRAPH_FREQ_MAX.
446         (dump_cgraph_node): Dump profile-id.
447         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
448         and common_target_probability.
449         * lto-cgraph.c (lto_output_edge): Stream common targets.
450         (lto_output_node): Stream profile ids.
451         (input_node): Stream profile ids.
452         (input_edge): Stream common targets.
453         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
454         * ipa.c: Include value-prof.h
455         (ipa_profile_generate_summary): Turn indirect call statement histograms
456         into common targets.
457         (ipa_profile): Turn common targets into speculative edges.
459 2013-08-09  Jan Hubicka  <jh@suse.cz>
461         * cgraph.h (cgraph_node): Add profile_id.
462         * value-prof.c (cgraph_node_map): Turn into pointer_map.
463         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
464         (del_node_map): Update.
465         (find_func_by_funcdef_no): Replace by ...
466         (find_func_by_profile_id): ... this one.
467         (gimple_ic_transform): Do not remove useful histograms when
468         speculation is not done; dump info when indirect call removal
469         can happen at LTO.
470         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
471         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
472         (__gcov_indirect_call_profiler_v2): .. this one.
473         * profile.h (init_node_map): Update.
474         * coverage.c (coverage_compute_profile_id): New function.
475         * coverage.h (coverage_compute_profile_id): Declare.
476         * tree-profile.c (init_ic_make_global_vars): Make
477         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
478         (gimple_init_edge_profiler): Update prototype of
479         __gcov_indirect_call_profiler.
480         (gimple_gen_ic_func_profiler): Simplify.
481         (tree_profiling): Use init_node_map
483 2013-08-09  Jan Hubicka  <jh@suse.cz>
485         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
486         non-speculative refs.
487         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
488         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
489         (cgraph_set_call_stmt): Likewise.
490         (cgraph_create_edge_1): Fix release checking compilatoin;
491         clear lto_stmt_uid.
492         (cgraph_free_edge): Free indirect info.
493         (cgraph_turn_edge_to_speculative): New function.
494         (cgraph_speculative_call_info): New function.
495         (cgraph_make_edge_direct): Return direct edge; handle speculation.
496         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
497         (dump_cgraph_node): Dump speculation.
498         (verify_edge_count_and_frequency): Accept speculative edges.
499         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
500         (verify_cgraph_node): Handle speculation.
501         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
502         (cgraph_set_call_stmt): Update prototype.
503         (cgraph_make_edge_direct): Update prototype.
504         (cgraph_speculative_call_info): Declare.
505         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
506         to change; update call of ipa_find_references.
507         * ipa-ref.c (ipa_record_reference): Fix return value; clear
508         lto_stmt_uid and speculative flags.
509         (ipa_dump_references): Dump speculation.
510         (ipa_clone_references): Clone speculative flag.
511         (ipa_clone_referring): Likewise.
512         (ipa_clone_ref): New function.
513         (ipa_find_reference): Look into lto_stmt_uids
514         (ipa_clear_stmts_in_references): Do not clear speculative calls.
515         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
516         (ipa_find_reference): Update declaration.
517         (ipa_clone_ref): Declare.
518         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
519         stream speculative flag.
520         (lto_output_ref): Stream statements uids and speculation.
521         (input_ref): Likewise.
522         (input_edge): Stream speuclation.
523         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
524         (cgraph_set_call_stmt_including_clones): Handle speculation.
525         * ipa-inline.c (heap_edge_removal_hook): New function.
526         (inline_small_functions): Register it.
527         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
528         also initialize refs.
529         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
530         edge to change.
531         (try_make_edge_direct_simple_call): Likewise.
532         (try_make_edge_direct_simple_call): Likewise.
533         (update_indirect_edges_after_inlining): Likewise.
534         (remove_described_reference): Look proper lto_stmt_uid.
535         (propagate_controlled_uses): Likewise.
536         (propagate_controlled_uses): Liekwise.
537         * tree-inline.c (copy_bb): Copy speculative edges.
538         (redirect_all_calls): New function.
539         (copy_cfg_body): Do redirection after loop info is updated.
540         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
542 2013-08-09  Jan Hubicka  <jh@suse.cz>
544         * lto-streamer-out.c (output_function): Renumber PHIs.
545         * lto-streamer-in.c (input_function): Likewise.
547 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
549         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
550         (get_lane_unsigned): Likewise.
551         (dup_lane_scalar): Likewise.
552         (get_lane): enable for VALL.
553         * config/aarch64/aarch64-simd.md
554         (aarch64_dup_lane_scalar<mode>): Remove.
555         (aarch64_get_lane_signed<mode>): Likewise.
556         (aarch64_get_lane_unsigned<mode>): Likewise.
557         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
558         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
559         (aarch64_get_lane<mode>): Enable for all vector modes.
560         (aarch64_get_lanedi): Remove misleading constraints.
561         * config/aarch64/arm_neon.h
562         (__aarch64_vget_lane_any): Define.
563         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
564         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
565         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
566         * config/aarch64/iterators.md (VDQQH): New.
567         (VDQQHS): Likewise.
568         (vwcore): Likewise.
570 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
572         * configure.ac: Add GAS check for LEON instructions on SPARC.
573         * configure: Regenerate.
574         * config.in: Likewise.
575         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
576         sparc*-*-* block.
577         * config/sparc/sparc.opt (LEON, LEON3): New masks.
578         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
579         for LEON or LEON3.
580         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
581         (AS_LEON_FLAG): New macro.
582         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
583         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
584         Deal with LEON and LEON3 for the memory model.
585         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
586         (atomic_compare_and_swap<mode>_1): Likewise.
587         (*atomic_compare_and_swap<mode>_1): Likewise.
589 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
591         * config/arm/neon.md (vcond): Fix floating-point vector
592         comparisons against 0.
594 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
596         * lra-constraints.c (emit_spill_move): Remove assert.
597         (process_alt_operands): Add more debugging
598         output.  Increase reject for spilling into memory.  Decrease
599         reject for reloading scratch.
600         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
602 2013-08-08  Steve Ellcey  <sellcey@mips.com>
604         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
605         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
606         micromips incompatible.  Add nan2008.
607         (MULTILIB_DIRNAMES): Add nan2008.
608         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
609         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
610         and micromips incompatible.  Add nan2008.
611         (MULTILIB_DIRNAMES): Add nan2008.
612         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
614 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
616         PR rtl-optimization/58079
617         * combine.c (combine_simplify_rtx): Avoid using SUBST if
618         simplify_comparison has widened a comparison with an integer.
620 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
622         * config/arm/neon.md (movmisalign<mode>): Disable when we
623         don't allow unaligned accesses.
624         (*movmisalign<mode>_neon_store): Likewise.
625         (*movmisalign<mode>_neon_load): Likewise.
626         (*movmisalign<mode>_neon_store): Likewise.
627         (*movmisalign<mode>_neon_load): Likewise.
629 2013-08-08  Jan Hubicka  <jh@suse.cz>
631         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
632         (make_pass_rebuild_cgraph_edges): Also clear references.
633         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
634         * ipa-inline-transform.c (inline_transform): Remove all references
635         after inlining.
636         * cgraphunit.c (expand_function): Remove all references after
637         expansion.
638         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
639         (ipa_find_reference): Rewrite to iterator.
640         (remove_stmt_references): Likewise.
641         (ipa_clear_stmts_in_references): New function.
642         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
643         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
644         clear references.
645         * ipa-split.c (split_function): Remove references in split function.
647 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
649         PR target/57431
650         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
651         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
653 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
655         PR target/56979
656         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
657         suggested mode for the assignment isn't compatible with the
658         registers required.
660 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
662         PR target/58065
663         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
665 2013-08-07  Xinliang David Li  <davidxl@google.com>
667         * config/i386/i386.opt: New option -mtune-ctrl=.
668         * config/i386/x86-tune.def: New file.
669         * config/i386/i386.h: include x86-tune.def.
670         * config/i386/i386.c (ix86_option_override_internal):
671         Parsing -mtune-ctrl= option and set tune features.
673 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
675         PR other/12081
676         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
677         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
678         to gen_2arg_fn_t.
680 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
682         * rtl.h (update_alignments): Declare.
683         * final.c (grow_label_align): New function extracted from...
684         (shorten_branches): ...here.  Call it.
685         (update_alignments): New function.
686         * reorg.c (sibling_labels): New variable.
687         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
688         the new label along with it onto the sibling_labels vector.
689         (fill_simple_delay_slots): Adjust call to get_label_before.
690         (fill_slots_from_thread): Likewise.
691         (relax_delay_slots): Likewise.
692         (make_return_insns): Likewise.
693         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
695 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
697         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
698         document its semantics.
699         (diagnostic_report_diagnostic): Adjust accordingly.
701 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
703         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
704         (sparc_option_override): ...and port to new C++ pass API.
705         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
707 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
709         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
711 2013-08-06  Caroline Tice  <cmtice@google.com>
713         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
714         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
715         * tree-pass.h: Add pass_vtable_verify.
716         * varasm.c (assemble_variable): Add code to properly set the comdat
717         section and name for the .vtable_map_vars section.
718         (assemble_vtyv_preinit_initializer): New function.
719         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
720         LINK_ONCE flag.
721         * output.h: Add function decl for assemble_vtv_preinit_initializer.
722         * vtable-verify.c: New file.
723         * vtable-verify.h: New file.
724         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
725         initialiation levels.
726         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
727         * passes.def: Insert pass_vtable_verify.
728         * aclocal.m4: Reorder includes.
729         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
730         -fvtv-counts options.
731         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
732         as appropriate, if -fvtable-verify=... is used.
733         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
734         -fvtable-verify=... is used.
735         * Makefile.in (OBJS):  Add vtable-verify.o to list.
736         (vtable-verify.o): Add new build rule.
737         (GTFILES): Add vtable-verify.c to list.
738         * common.opt (fvtable-verify=): New flag.
739         (vtv_priority): Values for fvtable-verify= flag.
740         (fvtv-counts): New flag.
741         (fvtv-debug): New flag.
742         * tree.h (save_vtable_map_decl): New extern function decl.
744 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
746         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
747         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
748         (pass_data_rl78_devirt): ...new pass_data instance and...
749         (make_pass_rl78_devirt): ...new function.
750         (rl78_asm_file_start): Port pass registration to new C++ API.
752 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
754         * coretypes.h (rtl_opt_pass): Add.
755         (gcc::context): Add.
756         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
757         (epiphany_init): Port to new C++ pass API.
758         (epiphany_optimize_mode_switching): Likewise.
759         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
760         (pass_manager::get_pass_mode_switching): New.
761         (pass_manager::get_pass_peephole2): New.
762         * mode-switching.c (pass_mode_switching): Add clone method.
763         * recog.c (pass_peephole2): Add clone method.
764         (pass_split_all_insns): Add clone method.
766 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
768         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
769         (mips_option_override): ...here, porting to new C++ API for passes.
771 2013-08-06  Jan Hubicka  <jh@suse.cz>
773         * cgraph.c (cgraph_get_body): New function based on lto.c
774         implementation.
775         * cgraph.h (cgraph_get_body): Declare.
776         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
777         LTO paths.
778         * cgraphunit.c (expand_function): Get body prior expanding.
779         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
780         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
781         really need.
782         * passes.c (do_per_function_toporder): Get body.
783         * tree-inline.c (expand_call_inline): Get body prior inlining it.
784         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
786 2013-08-06  Martin Jambor  <mjambor@suse.cz>
788         PR fortran/57987
789         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
790         is not re-finalized.  Rename second parameter to no_collect.
792 2013-08-06  Martin Jambor  <mjambor@suse.cz>
794         PR middle-end/58041
795         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
796         MEM_REF has proper alignment information.
798 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
800         PR other/12081
801         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
802         class insn_gen_fn.
803         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
804         rtx (*) (rtx, ...) with insn_gen_fn.
805         * genoutput.c (output_insn_data): Cast gen_? function pointers to
806         insn_gen_fn::stored_funcptr.  Add initializer braces.
808 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
810         Rewrite how instances of passes are cloned to remove assumptions
811         about their sizes (thus allowing pass subclasses to have
812         additional data fields, albeit non-GC-managed ones at this point).
814         * passes.c (make_pass_instance): Now that passes have clone
815         methods, rewrite this function to eliminate XNEW and memcpy
816         calls that used hardcoded sizes.  Since this function no longer
817         creates pass instances, rename it to...
818         (add_pass_instance): ...this.  Document the old way that passes were
819         numbered and flagged, and rework this function to continue using it.
820         (next_pass_1): Add an initial_pass argument for use by
821         add_pass_instance.
822         (position_pass): When adding multiple instances of a pass, use
823         the pass's clone method, rather than relying on the XNEW/memcpy
824         within the former make_pass_instance (now add_pass_instance).
825         (pass_manager::pass_manager): When invoking next_pass_1, also supply
826         the initial instance of the current pass within the pass manager.
828 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
830         This is the automated part of the conversion of passes from C
831         structs to C++ classes.
833         Patch autogenerated by refactor_passes.py from
834         https://github.com/davidmalcolm/gcc-refactoring-scripts
835         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
837         * asan.c (pass_asan): Convert from a global struct to a subclass of
838         gimple_opt_pass along with...
839         (pass_data_asan): ...new pass_data instance and...
840         (make_pass_asan): ...new function.
841         (pass_asan_O0): Convert from a global struct to a subclass of
842         gimple_opt_pass along with...
843         (pass_data_asan_O0): ...new pass_data instance and...
844         (make_pass_asan_O0): ...new function.
845         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
846         subclass of rtl_opt_pass along with...
847         (pass_data_inc_dec): ...new pass_data instance and...
848         (make_pass_inc_dec): ...new function.
849         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
850         a subclass of rtl_opt_pass along with...
851         (pass_data_reorder_blocks): ...new pass_data instance and...
852         (make_pass_reorder_blocks): ...new function.
853         (pass_duplicate_computed_gotos): Convert from a global struct to a
854         subclass of rtl_opt_pass along with...
855         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
856         (make_pass_duplicate_computed_gotos): ...new function.
857         (pass_partition_blocks): Convert from a global struct to a subclass of
858         rtl_opt_pass along with...
859         (pass_data_partition_blocks): ...new pass_data instance and...
860         (make_pass_partition_blocks): ...new function.
861         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
862         struct to a subclass of rtl_opt_pass along with...
863         (pass_data_branch_target_load_optimize1): ...new pass_data instance
864         and...
865         (make_pass_branch_target_load_optimize1): ...new function.
866         (pass_branch_target_load_optimize2): Convert from a global struct to a
867         subclass of rtl_opt_pass along with...
868         (pass_data_branch_target_load_optimize2): ...new pass_data instance
869         and...
870         (make_pass_branch_target_load_optimize2): ...new function.
871         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
872         of rtl_opt_pass along with...
873         (pass_data_jump): ...new pass_data instance and...
874         (make_pass_jump): ...new function.
875         (pass_jump2): Convert from a global struct to a subclass of
876         rtl_opt_pass along with...
877         (pass_data_jump2): ...new pass_data instance and...
878         (make_pass_jump2): ...new function.
879         * cfgexpand.c (pass_expand): Convert from a global struct to a
880         subclass of rtl_opt_pass along with...
881         (pass_data_expand): ...new pass_data instance and...
882         (make_pass_expand): ...new function.
883         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
884         of rtl_opt_pass along with...
885         (pass_data_free_cfg): ...new pass_data instance and...
886         (make_pass_free_cfg): ...new function.
887         (pass_into_cfg_layout_mode): Convert from a global struct to a
888         subclass of rtl_opt_pass along with...
889         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
890         (make_pass_into_cfg_layout_mode): ...new function.
891         (pass_outof_cfg_layout_mode): Convert from a global struct to a
892         subclass of rtl_opt_pass along with...
893         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
894         (make_pass_outof_cfg_layout_mode): ...new function.
895         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
896         struct to a subclass of gimple_opt_pass along with...
897         (pass_data_build_cgraph_edges): ...new pass_data instance and...
898         (make_pass_build_cgraph_edges): ...new function.
899         (pass_rebuild_cgraph_edges): Convert from a global struct to a
900         subclass of gimple_opt_pass along with...
901         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
902         (make_pass_rebuild_cgraph_edges): ...new function.
903         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
904         subclass of gimple_opt_pass along with...
905         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
906         and...
907         (make_pass_remove_cgraph_callee_edges): ...new function.
908         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
909         struct to a subclass of rtl_opt_pass along with...
910         (pass_data_stack_adjustments): ...new pass_data instance and...
911         (make_pass_stack_adjustments): ...new function.
912         * combine.c (pass_combine): Convert from a global struct to a subclass
913         of rtl_opt_pass along with...
914         (pass_data_combine): ...new pass_data instance and...
915         (make_pass_combine): ...new function.
916         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
917         global struct to a subclass of rtl_opt_pass along with...
918         (pass_data_compare_elim_after_reload): ...new pass_data instance
919         and...
920         (make_pass_compare_elim_after_reload): ...new function.
921         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
922         of rtl_opt_pass along with...
923         (pass_data_rtl_cprop): ...new pass_data instance and...
924         (make_pass_rtl_cprop): ...new function.
925         * cse.c (pass_cse): Convert from a global struct to a subclass of
926         rtl_opt_pass along with...
927         (pass_data_cse): ...new pass_data instance and...
928         (make_pass_cse): ...new function.
929         (pass_cse2): Convert from a global struct to a subclass of
930         rtl_opt_pass along with...
931         (pass_data_cse2): ...new pass_data instance and...
932         (make_pass_cse2): ...new function.
933         (pass_cse_after_global_opts): Convert from a global struct to a
934         subclass of rtl_opt_pass along with...
935         (pass_data_cse_after_global_opts): ...new pass_data instance and...
936         (make_pass_cse_after_global_opts): ...new function.
937         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
938         of rtl_opt_pass along with...
939         (pass_data_ud_rtl_dce): ...new pass_data instance and...
940         (make_pass_ud_rtl_dce): ...new function.
941         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
942         rtl_opt_pass along with...
943         (pass_data_fast_rtl_dce): ...new pass_data instance and...
944         (make_pass_fast_rtl_dce): ...new function.
945         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
946         a subclass of rtl_opt_pass along with...
947         (pass_data_df_initialize_opt): ...new pass_data instance and...
948         (make_pass_df_initialize_opt): ...new function.
949         (pass_df_initialize_no_opt): Convert from a global struct to a
950         subclass of rtl_opt_pass along with...
951         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
952         (make_pass_df_initialize_no_opt): ...new function.
953         (pass_df_finish): Convert from a global struct to a subclass of
954         rtl_opt_pass along with...
955         (pass_data_df_finish): ...new pass_data instance and...
956         (make_pass_df_finish): ...new function.
957         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
958         rtl_opt_pass along with...
959         (pass_data_rtl_dse1): ...new pass_data instance and...
960         (make_pass_rtl_dse1): ...new function.
961         (pass_rtl_dse2): Convert from a global struct to a subclass of
962         rtl_opt_pass along with...
963         (pass_data_rtl_dse2): ...new pass_data instance and...
964         (make_pass_rtl_dse2): ...new function.
965         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
966         subclass of rtl_opt_pass along with...
967         (pass_data_dwarf2_frame): ...new pass_data instance and...
968         (make_pass_dwarf2_frame): ...new function.
969         * except.c (pass_set_nothrow_function_flags): Convert from a global
970         struct to a subclass of rtl_opt_pass along with...
971         (pass_data_set_nothrow_function_flags): ...new pass_data instance
972         and...
973         (make_pass_set_nothrow_function_flags): ...new function.
974         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
975         subclass of rtl_opt_pass along with...
976         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
977         and...
978         (make_pass_convert_to_eh_region_ranges): ...new function.
979         * final.c (pass_compute_alignments): Convert from a global struct to a
980         subclass of rtl_opt_pass along with...
981         (pass_data_compute_alignments): ...new pass_data instance and...
982         (make_pass_compute_alignments): ...new function.
983         (pass_final): Convert from a global struct to a subclass of
984         rtl_opt_pass along with...
985         (pass_data_final): ...new pass_data instance and...
986         (make_pass_final): ...new function.
987         (pass_shorten_branches): Convert from a global struct to a subclass of
988         rtl_opt_pass along with...
989         (pass_data_shorten_branches): ...new pass_data instance and...
990         (make_pass_shorten_branches): ...new function.
991         (pass_clean_state): Convert from a global struct to a subclass of
992         rtl_opt_pass along with...
993         (pass_data_clean_state): ...new pass_data instance and...
994         (make_pass_clean_state): ...new function.
995         * function.c (pass_instantiate_virtual_regs): Convert from a global
996         struct to a subclass of rtl_opt_pass along with...
997         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
998         (make_pass_instantiate_virtual_regs): ...new function.
999         (pass_leaf_regs): Convert from a global struct to a subclass of
1000         rtl_opt_pass along with...
1001         (pass_data_leaf_regs): ...new pass_data instance and...
1002         (make_pass_leaf_regs): ...new function.
1003         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1004         subclass of rtl_opt_pass along with...
1005         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1006         and...
1007         (make_pass_thread_prologue_and_epilogue): ...new function.
1008         (pass_match_asm_constraints): Convert from a global struct to a
1009         subclass of rtl_opt_pass along with...
1010         (pass_data_match_asm_constraints): ...new pass_data instance and...
1011         (make_pass_match_asm_constraints): ...new function.
1012         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1013         subclass of rtl_opt_pass along with...
1014         (pass_data_rtl_fwprop): ...new pass_data instance and...
1015         (make_pass_rtl_fwprop): ...new function.
1016         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1017         rtl_opt_pass along with...
1018         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1019         (make_pass_rtl_fwprop_addr): ...new function.
1020         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1021         rtl_opt_pass along with...
1022         (pass_data_rtl_pre): ...new pass_data instance and...
1023         (make_pass_rtl_pre): ...new function.
1024         (pass_rtl_hoist): Convert from a global struct to a subclass of
1025         rtl_opt_pass along with...
1026         (pass_data_rtl_hoist): ...new pass_data instance and...
1027         (make_pass_rtl_hoist): ...new function.
1028         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1029         subclass of gimple_opt_pass along with...
1030         (pass_data_lower_cf): ...new pass_data instance and...
1031         (make_pass_lower_cf): ...new function.
1032         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1033         from a global struct to a subclass of gimple_opt_pass along with...
1034         (pass_data_strength_reduction): ...new pass_data instance and...
1035         (make_pass_strength_reduction): ...new function.
1036         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1037         of rtl_opt_pass along with...
1038         (pass_data_rtl_ifcvt): ...new pass_data instance and...
1039         (make_pass_rtl_ifcvt): ...new function.
1040         (pass_if_after_combine): Convert from a global struct to a subclass of
1041         rtl_opt_pass along with...
1042         (pass_data_if_after_combine): ...new pass_data instance and...
1043         (make_pass_if_after_combine): ...new function.
1044         (pass_if_after_reload): Convert from a global struct to a subclass of
1045         rtl_opt_pass along with...
1046         (pass_data_if_after_reload): ...new pass_data instance and...
1047         (make_pass_if_after_reload): ...new function.
1048         * init-regs.c (pass_initialize_regs): Convert from a global struct to
1049         a subclass of rtl_opt_pass along with...
1050         (pass_data_initialize_regs): ...new pass_data instance and...
1051         (make_pass_initialize_regs): ...new function.
1052         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1053         of ipa_opt_pass_d along with...
1054         (pass_data_ipa_cp): ...new pass_data instance and...
1055         (make_pass_ipa_cp): ...new function.
1056         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1057         global struct to a subclass of gimple_opt_pass along with...
1058         (pass_data_inline_parameters): ...new pass_data instance and...
1059         (make_pass_inline_parameters): ...new function.
1060         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1061         subclass of gimple_opt_pass along with...
1062         (pass_data_early_inline): ...new pass_data instance and...
1063         (make_pass_early_inline): ...new function.
1064         (pass_ipa_inline): Convert from a global struct to a subclass of
1065         ipa_opt_pass_d along with...
1066         (pass_data_ipa_inline): ...new pass_data instance and...
1067         (make_pass_ipa_inline): ...new function.
1068         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1069         struct to a subclass of gimple_opt_pass along with...
1070         (pass_data_local_pure_const): ...new pass_data instance and...
1071         (make_pass_local_pure_const): ...new function.
1072         (pass_ipa_pure_const): Convert from a global struct to a subclass of
1073         ipa_opt_pass_d along with...
1074         (pass_data_ipa_pure_const): ...new pass_data instance and...
1075         (make_pass_ipa_pure_const): ...new function.
1076         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1077         to a subclass of ipa_opt_pass_d along with...
1078         (pass_data_ipa_reference): ...new pass_data instance and...
1079         (make_pass_ipa_reference): ...new function.
1080         * ipa-split.c (pass_split_functions): Convert from a global struct to
1081         a subclass of gimple_opt_pass along with...
1082         (pass_data_split_functions): ...new pass_data instance and...
1083         (make_pass_split_functions): ...new function.
1084         (pass_feedback_split_functions): Convert from a global struct to a
1085         subclass of gimple_opt_pass along with...
1086         (pass_data_feedback_split_functions): ...new pass_data instance and...
1087         (make_pass_feedback_split_functions): ...new function.
1088         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1089         global struct to a subclass of simple_ipa_opt_pass along with...
1090         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1091         instance and...
1092         (make_pass_ipa_function_and_variable_visibility): ...new function.
1093         (pass_ipa_free_inline_summary): Convert from a global struct to a
1094         subclass of simple_ipa_opt_pass along with...
1095         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1096         (make_pass_ipa_free_inline_summary): ...new function.
1097         (pass_ipa_whole_program_visibility): Convert from a global struct to a
1098         subclass of ipa_opt_pass_d along with...
1099         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1100         and...
1101         (make_pass_ipa_whole_program_visibility): ...new function.
1102         (pass_ipa_profile): Convert from a global struct to a subclass of
1103         ipa_opt_pass_d along with...
1104         (pass_data_ipa_profile): ...new pass_data instance and...
1105         (make_pass_ipa_profile): ...new function.
1106         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1107         ipa_opt_pass_d along with...
1108         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1109         (make_pass_ipa_cdtor_merge): ...new function.
1110         * ira.c (pass_ira): Convert from a global struct to a subclass of
1111         rtl_opt_pass along with...
1112         (pass_data_ira): ...new pass_data instance and...
1113         (make_pass_ira): ...new function.
1114         (pass_reload): Convert from a global struct to a subclass of
1115         rtl_opt_pass along with...
1116         (pass_data_reload): ...new pass_data instance and...
1117         (make_pass_reload): ...new function.
1118         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1119         subclass of rtl_opt_pass along with...
1120         (pass_data_cleanup_barriers): ...new pass_data instance and...
1121         (make_pass_cleanup_barriers): ...new function.
1122         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1123         of rtl_opt_pass along with...
1124         (pass_data_loop2): ...new pass_data instance and...
1125         (make_pass_loop2): ...new function.
1126         (pass_rtl_loop_init): Convert from a global struct to a subclass of
1127         rtl_opt_pass along with...
1128         (pass_data_rtl_loop_init): ...new pass_data instance and...
1129         (make_pass_rtl_loop_init): ...new function.
1130         (pass_rtl_loop_done): Convert from a global struct to a subclass of
1131         rtl_opt_pass along with...
1132         (pass_data_rtl_loop_done): ...new pass_data instance and...
1133         (make_pass_rtl_loop_done): ...new function.
1134         (pass_rtl_move_loop_invariants): Convert from a global struct to a
1135         subclass of rtl_opt_pass along with...
1136         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1137         (make_pass_rtl_move_loop_invariants): ...new function.
1138         (pass_rtl_unswitch): Convert from a global struct to a subclass of
1139         rtl_opt_pass along with...
1140         (pass_data_rtl_unswitch): ...new pass_data instance and...
1141         (make_pass_rtl_unswitch): ...new function.
1142         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1143         subclass of rtl_opt_pass along with...
1144         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1145         and...
1146         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1147         (pass_rtl_doloop): Convert from a global struct to a subclass of
1148         rtl_opt_pass along with...
1149         (pass_data_rtl_doloop): ...new pass_data instance and...
1150         (make_pass_rtl_doloop): ...new function.
1151         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1152         a subclass of rtl_opt_pass along with...
1153         (pass_data_lower_subreg): ...new pass_data instance and...
1154         (make_pass_lower_subreg): ...new function.
1155         (pass_lower_subreg2): Convert from a global struct to a subclass of
1156         rtl_opt_pass along with...
1157         (pass_data_lower_subreg2): ...new pass_data instance and...
1158         (make_pass_lower_subreg2): ...new function.
1159         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1160         struct to a subclass of ipa_opt_pass_d along with...
1161         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1162         (make_pass_ipa_lto_gimple_out): ...new function.
1163         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1164         of ipa_opt_pass_d along with...
1165         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1166         (make_pass_ipa_lto_finish_out): ...new function.
1167         * mode-switching.c (pass_mode_switching): Convert from a global struct
1168         to a subclass of rtl_opt_pass along with...
1169         (pass_data_mode_switching): ...new pass_data instance and...
1170         (make_pass_mode_switching): ...new function.
1171         * modulo-sched.c (pass_sms): Convert from a global struct to a
1172         subclass of rtl_opt_pass along with...
1173         (pass_data_sms): ...new pass_data instance and...
1174         (make_pass_sms): ...new function.
1175         * omp-low.c (pass_expand_omp): Convert from a global struct to a
1176         subclass of gimple_opt_pass along with...
1177         (pass_data_expand_omp): ...new pass_data instance and...
1178         (make_pass_expand_omp): ...new function.
1179         (pass_lower_omp): Convert from a global struct to a subclass of
1180         gimple_opt_pass along with...
1181         (pass_data_lower_omp): ...new pass_data instance and...
1182         (make_pass_lower_omp): ...new function.
1183         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1184         of gimple_opt_pass along with...
1185         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1186         (make_pass_diagnose_omp_blocks): ...new function.
1187         * passes.c (pass_early_local_passes): Convert from a global struct to
1188         a subclass of simple_ipa_opt_pass along with...
1189         (pass_data_early_local_passes): ...new pass_data instance and...
1190         (make_pass_early_local_passes): ...new function.
1191         (pass_all_early_optimizations): Convert from a global struct to a
1192         subclass of gimple_opt_pass along with...
1193         (pass_data_all_early_optimizations): ...new pass_data instance and...
1194         (make_pass_all_early_optimizations): ...new function.
1195         (pass_all_optimizations): Convert from a global struct to a subclass
1196         of gimple_opt_pass along with...
1197         (pass_data_all_optimizations): ...new pass_data instance and...
1198         (make_pass_all_optimizations): ...new function.
1199         (pass_all_optimizations_g): Convert from a global struct to a subclass
1200         of gimple_opt_pass along with...
1201         (pass_data_all_optimizations_g): ...new pass_data instance and...
1202         (make_pass_all_optimizations_g): ...new function.
1203         (pass_rest_of_compilation): Convert from a global struct to a subclass
1204         of rtl_opt_pass along with...
1205         (pass_data_rest_of_compilation): ...new pass_data instance and...
1206         (make_pass_rest_of_compilation): ...new function.
1207         (pass_postreload): Convert from a global struct to a subclass of
1208         rtl_opt_pass along with...
1209         (pass_data_postreload): ...new pass_data instance and...
1210         (make_pass_postreload): ...new function.
1211         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1212         subclass of rtl_opt_pass along with...
1213         (pass_data_gcse2): ...new pass_data instance and...
1214         (make_pass_gcse2): ...new function.
1215         * postreload.c (pass_postreload_cse): Convert from a global struct to
1216         a subclass of rtl_opt_pass along with...
1217         (pass_data_postreload_cse): ...new pass_data instance and...
1218         (make_pass_postreload_cse): ...new function.
1219         * predict.c (pass_profile): Convert from a global struct to a subclass
1220         of gimple_opt_pass along with...
1221         (pass_data_profile): ...new pass_data instance and...
1222         (make_pass_profile): ...new function.
1223         (pass_strip_predict_hints): Convert from a global struct to a subclass
1224         of gimple_opt_pass along with...
1225         (pass_data_strip_predict_hints): ...new pass_data instance and...
1226         (make_pass_strip_predict_hints): ...new function.
1227         * recog.c (pass_peephole2): Convert from a global struct to a subclass
1228         of rtl_opt_pass along with...
1229         (pass_data_peephole2): ...new pass_data instance and...
1230         (make_pass_peephole2): ...new function.
1231         (pass_split_all_insns): Convert from a global struct to a subclass of
1232         rtl_opt_pass along with...
1233         (pass_data_split_all_insns): ...new pass_data instance and...
1234         (make_pass_split_all_insns): ...new function.
1235         (pass_split_after_reload): Convert from a global struct to a subclass
1236         of rtl_opt_pass along with...
1237         (pass_data_split_after_reload): ...new pass_data instance and...
1238         (make_pass_split_after_reload): ...new function.
1239         (pass_split_before_regstack): Convert from a global struct to a
1240         subclass of rtl_opt_pass along with...
1241         (pass_data_split_before_regstack): ...new pass_data instance and...
1242         (make_pass_split_before_regstack): ...new function.
1243         (pass_split_before_sched2): Convert from a global struct to a subclass
1244         of rtl_opt_pass along with...
1245         (pass_data_split_before_sched2): ...new pass_data instance and...
1246         (make_pass_split_before_sched2): ...new function.
1247         (pass_split_for_shorten_branches): Convert from a global struct to a
1248         subclass of rtl_opt_pass along with...
1249         (pass_data_split_for_shorten_branches): ...new pass_data instance
1250         and...
1251         (make_pass_split_for_shorten_branches): ...new function.
1252         * ree.c (pass_ree): Convert from a global struct to a subclass of
1253         rtl_opt_pass along with...
1254         (pass_data_ree): ...new pass_data instance and...
1255         (make_pass_ree): ...new function.
1256         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1257         subclass of rtl_opt_pass along with...
1258         (pass_data_stack_regs): ...new pass_data instance and...
1259         (make_pass_stack_regs): ...new function.
1260         (pass_stack_regs_run): Convert from a global struct to a subclass of
1261         rtl_opt_pass along with...
1262         (pass_data_stack_regs_run): ...new pass_data instance and...
1263         (make_pass_stack_regs_run): ...new function.
1264         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1265         subclass of rtl_opt_pass along with...
1266         (pass_data_cprop_hardreg): ...new pass_data instance and...
1267         (make_pass_cprop_hardreg): ...new function.
1268         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1269         subclass of rtl_opt_pass along with...
1270         (pass_data_reginfo_init): ...new pass_data instance and...
1271         (make_pass_reginfo_init): ...new function.
1272         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1273         of rtl_opt_pass along with...
1274         (pass_data_regmove): ...new pass_data instance and...
1275         (make_pass_regmove): ...new function.
1276         * regrename.c (pass_regrename): Convert from a global struct to a
1277         subclass of rtl_opt_pass along with...
1278         (pass_data_regrename): ...new pass_data instance and...
1279         (make_pass_regrename): ...new function.
1280         * reorg.c (pass_delay_slots): Convert from a global struct to a
1281         subclass of rtl_opt_pass along with...
1282         (pass_data_delay_slots): ...new pass_data instance and...
1283         (make_pass_delay_slots): ...new function.
1284         (pass_machine_reorg): Convert from a global struct to a subclass of
1285         rtl_opt_pass along with...
1286         (pass_data_machine_reorg): ...new pass_data instance and...
1287         (make_pass_machine_reorg): ...new function.
1288         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1289         of rtl_opt_pass along with...
1290         (pass_data_sched): ...new pass_data instance and...
1291         (make_pass_sched): ...new function.
1292         (pass_sched2): Convert from a global struct to a subclass of
1293         rtl_opt_pass along with...
1294         (pass_data_sched2): ...new pass_data instance and...
1295         (make_pass_sched2): ...new function.
1296         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1297         to a subclass of rtl_opt_pass along with...
1298         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1299         (make_pass_stack_ptr_mod): ...new function.
1300         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1301         to a subclass of rtl_opt_pass along with...
1302         (pass_data_rtl_store_motion): ...new pass_data instance and...
1303         (make_pass_rtl_store_motion): ...new function.
1304         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1305         of gimple_opt_pass along with...
1306         (pass_data_tracer): ...new pass_data instance and...
1307         (make_pass_tracer): ...new function.
1308         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1309         to a subclass of gimple_opt_pass along with...
1310         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1311         (make_pass_diagnose_tm_blocks): ...new function.
1312         (pass_lower_tm): Convert from a global struct to a subclass of
1313         gimple_opt_pass along with...
1314         (pass_data_lower_tm): ...new pass_data instance and...
1315         (make_pass_lower_tm): ...new function.
1316         (pass_tm_init): Convert from a global struct to a subclass of
1317         gimple_opt_pass along with...
1318         (pass_data_tm_init): ...new pass_data instance and...
1319         (make_pass_tm_init): ...new function.
1320         (pass_tm_mark): Convert from a global struct to a subclass of
1321         gimple_opt_pass along with...
1322         (pass_data_tm_mark): ...new pass_data instance and...
1323         (make_pass_tm_mark): ...new function.
1324         (pass_tm_edges): Convert from a global struct to a subclass of
1325         gimple_opt_pass along with...
1326         (pass_data_tm_edges): ...new pass_data instance and...
1327         (make_pass_tm_edges): ...new function.
1328         (pass_tm_memopt): Convert from a global struct to a subclass of
1329         gimple_opt_pass along with...
1330         (pass_data_tm_memopt): ...new pass_data instance and...
1331         (make_pass_tm_memopt): ...new function.
1332         (pass_ipa_tm): Convert from a global struct to a subclass of
1333         simple_ipa_opt_pass along with...
1334         (pass_data_ipa_tm): ...new pass_data instance and...
1335         (make_pass_ipa_tm): ...new function.
1336         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1337         subclass of gimple_opt_pass along with...
1338         (pass_data_call_cdce): ...new pass_data instance and...
1339         (make_pass_call_cdce): ...new function.
1340         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1341         subclass of gimple_opt_pass along with...
1342         (pass_data_build_cfg): ...new pass_data instance and...
1343         (make_pass_build_cfg): ...new function.
1344         (pass_split_crit_edges): Convert from a global struct to a subclass of
1345         gimple_opt_pass along with...
1346         (pass_data_split_crit_edges): ...new pass_data instance and...
1347         (make_pass_split_crit_edges): ...new function.
1348         (pass_warn_function_return): Convert from a global struct to a
1349         subclass of gimple_opt_pass along with...
1350         (pass_data_warn_function_return): ...new pass_data instance and...
1351         (make_pass_warn_function_return): ...new function.
1352         (pass_warn_function_noreturn): Convert from a global struct to a
1353         subclass of gimple_opt_pass along with...
1354         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1355         (make_pass_warn_function_noreturn): ...new function.
1356         (pass_warn_unused_result): Convert from a global struct to a subclass
1357         of gimple_opt_pass along with...
1358         (pass_data_warn_unused_result): ...new pass_data instance and...
1359         (make_pass_warn_unused_result): ...new function.
1360         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1361         a subclass of gimple_opt_pass along with...
1362         (pass_data_merge_phi): ...new pass_data instance and...
1363         (make_pass_merge_phi): ...new function.
1364         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1365         a subclass of gimple_opt_pass along with...
1366         (pass_data_lower_complex): ...new pass_data instance and...
1367         (make_pass_lower_complex): ...new function.
1368         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1369         gimple_opt_pass along with...
1370         (pass_data_lower_complex_O0): ...new pass_data instance and...
1371         (make_pass_lower_complex_O0): ...new function.
1372         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1373         subclass of gimple_opt_pass along with...
1374         (pass_data_lower_eh): ...new pass_data instance and...
1375         (make_pass_lower_eh): ...new function.
1376         (pass_refactor_eh): Convert from a global struct to a subclass of
1377         gimple_opt_pass along with...
1378         (pass_data_refactor_eh): ...new pass_data instance and...
1379         (make_pass_refactor_eh): ...new function.
1380         (pass_lower_resx): Convert from a global struct to a subclass of
1381         gimple_opt_pass along with...
1382         (pass_data_lower_resx): ...new pass_data instance and...
1383         (make_pass_lower_resx): ...new function.
1384         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1385         of gimple_opt_pass along with...
1386         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1387         (make_pass_lower_eh_dispatch): ...new function.
1388         (pass_cleanup_eh): Convert from a global struct to a subclass of
1389         gimple_opt_pass along with...
1390         (pass_data_cleanup_eh): ...new pass_data instance and...
1391         (make_pass_cleanup_eh): ...new function.
1392         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1393         to a subclass of simple_ipa_opt_pass along with...
1394         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1395         (make_pass_ipa_lower_emutls): ...new function.
1396         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1397         a subclass of gimple_opt_pass along with...
1398         (pass_data_if_conversion): ...new pass_data instance and...
1399         (make_pass_if_conversion): ...new function.
1400         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1401         subclass of gimple_opt_pass along with...
1402         (pass_data_build_ssa): ...new pass_data instance and...
1403         (make_pass_build_ssa): ...new function.
1404         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1405         global struct to a subclass of gimple_opt_pass along with...
1406         (pass_data_loop_distribution): ...new pass_data instance and...
1407         (make_pass_loop_distribution): ...new function.
1408         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1409         subclass of gimple_opt_pass along with...
1410         (pass_data_mudflap_1): ...new pass_data instance and...
1411         (make_pass_mudflap_1): ...new function.
1412         (pass_mudflap_2): Convert from a global struct to a subclass of
1413         gimple_opt_pass along with...
1414         (pass_data_mudflap_2): ...new pass_data instance and...
1415         (make_pass_mudflap_2): ...new function.
1416         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1417         subclass of gimple_opt_pass along with...
1418         (pass_data_mudflap_1): ...new pass_data instance and...
1419         (make_pass_mudflap_1): ...new function.
1420         (pass_mudflap_2): Convert from a global struct to a subclass of
1421         gimple_opt_pass along with...
1422         (pass_data_mudflap_2): ...new pass_data instance and...
1423         (make_pass_mudflap_2): ...new function.
1424         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1425         gimple_opt_pass along with...
1426         (pass_data_nrv): ...new pass_data instance and...
1427         (make_pass_nrv): ...new function.
1428         (pass_return_slot): Convert from a global struct to a subclass of
1429         gimple_opt_pass along with...
1430         (pass_data_return_slot): ...new pass_data instance and...
1431         (make_pass_return_slot): ...new function.
1432         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1433         to a subclass of gimple_opt_pass along with...
1434         (pass_data_object_sizes): ...new pass_data instance and...
1435         (make_pass_object_sizes): ...new function.
1436         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1437         global struct to a subclass of gimple_opt_pass along with...
1438         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1439         and...
1440         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1441         (pass_fixup_cfg): Convert from a global struct to a subclass of
1442         gimple_opt_pass along with...
1443         (pass_data_fixup_cfg): ...new pass_data instance and...
1444         (make_pass_fixup_cfg): ...new function.
1445         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1446         (make_pass_mudflap_1): ...new function.
1447         (pass_mudflap_2): Replace declaration with that of...
1448         (make_pass_mudflap_2): ...new function.
1449         (pass_asan): Replace declaration with that of...
1450         (make_pass_asan): ...new function.
1451         (pass_asan_O0): Replace declaration with that of...
1452         (make_pass_asan_O0): ...new function.
1453         (pass_tsan): Replace declaration with that of...
1454         (make_pass_tsan): ...new function.
1455         (pass_tsan_O0): Replace declaration with that of...
1456         (make_pass_tsan_O0): ...new function.
1457         (pass_lower_cf): Replace declaration with that of...
1458         (make_pass_lower_cf): ...new function.
1459         (pass_refactor_eh): Replace declaration with that of...
1460         (make_pass_refactor_eh): ...new function.
1461         (pass_lower_eh): Replace declaration with that of...
1462         (make_pass_lower_eh): ...new function.
1463         (pass_lower_eh_dispatch): Replace declaration with that of...
1464         (make_pass_lower_eh_dispatch): ...new function.
1465         (pass_lower_resx): Replace declaration with that of...
1466         (make_pass_lower_resx): ...new function.
1467         (pass_build_cfg): Replace declaration with that of...
1468         (make_pass_build_cfg): ...new function.
1469         (pass_early_tree_profile): Replace declaration with that of...
1470         (make_pass_early_tree_profile): ...new function.
1471         (pass_cleanup_eh): Replace declaration with that of...
1472         (make_pass_cleanup_eh): ...new function.
1473         (pass_sra): Replace declaration with that of...
1474         (make_pass_sra): ...new function.
1475         (pass_sra_early): Replace declaration with that of...
1476         (make_pass_sra_early): ...new function.
1477         (pass_early_ipa_sra): Replace declaration with that of...
1478         (make_pass_early_ipa_sra): ...new function.
1479         (pass_tail_recursion): Replace declaration with that of...
1480         (make_pass_tail_recursion): ...new function.
1481         (pass_tail_calls): Replace declaration with that of...
1482         (make_pass_tail_calls): ...new function.
1483         (pass_tree_loop): Replace declaration with that of...
1484         (make_pass_tree_loop): ...new function.
1485         (pass_tree_loop_init): Replace declaration with that of...
1486         (make_pass_tree_loop_init): ...new function.
1487         (pass_lim): Replace declaration with that of...
1488         (make_pass_lim): ...new function.
1489         (pass_tree_unswitch): Replace declaration with that of...
1490         (make_pass_tree_unswitch): ...new function.
1491         (pass_predcom): Replace declaration with that of...
1492         (make_pass_predcom): ...new function.
1493         (pass_iv_canon): Replace declaration with that of...
1494         (make_pass_iv_canon): ...new function.
1495         (pass_scev_cprop): Replace declaration with that of...
1496         (make_pass_scev_cprop): ...new function.
1497         (pass_empty_loop): Replace declaration with that of...
1498         (make_pass_empty_loop): ...new function.
1499         (pass_record_bounds): Replace declaration with that of...
1500         (make_pass_record_bounds): ...new function.
1501         (pass_graphite): Replace declaration with that of...
1502         (make_pass_graphite): ...new function.
1503         (pass_graphite_transforms): Replace declaration with that of...
1504         (make_pass_graphite_transforms): ...new function.
1505         (pass_if_conversion): Replace declaration with that of...
1506         (make_pass_if_conversion): ...new function.
1507         (pass_loop_distribution): Replace declaration with that of...
1508         (make_pass_loop_distribution): ...new function.
1509         (pass_vectorize): Replace declaration with that of...
1510         (make_pass_vectorize): ...new function.
1511         (pass_slp_vectorize): Replace declaration with that of...
1512         (make_pass_slp_vectorize): ...new function.
1513         (pass_complete_unroll): Replace declaration with that of...
1514         (make_pass_complete_unroll): ...new function.
1515         (pass_complete_unrolli): Replace declaration with that of...
1516         (make_pass_complete_unrolli): ...new function.
1517         (pass_parallelize_loops): Replace declaration with that of...
1518         (make_pass_parallelize_loops): ...new function.
1519         (pass_loop_prefetch): Replace declaration with that of...
1520         (make_pass_loop_prefetch): ...new function.
1521         (pass_iv_optimize): Replace declaration with that of...
1522         (make_pass_iv_optimize): ...new function.
1523         (pass_tree_loop_done): Replace declaration with that of...
1524         (make_pass_tree_loop_done): ...new function.
1525         (pass_ch): Replace declaration with that of...
1526         (make_pass_ch): ...new function.
1527         (pass_ccp): Replace declaration with that of...
1528         (make_pass_ccp): ...new function.
1529         (pass_phi_only_cprop): Replace declaration with that of...
1530         (make_pass_phi_only_cprop): ...new function.
1531         (pass_build_ssa): Replace declaration with that of...
1532         (make_pass_build_ssa): ...new function.
1533         (pass_build_alias): Replace declaration with that of...
1534         (make_pass_build_alias): ...new function.
1535         (pass_build_ealias): Replace declaration with that of...
1536         (make_pass_build_ealias): ...new function.
1537         (pass_dominator): Replace declaration with that of...
1538         (make_pass_dominator): ...new function.
1539         (pass_dce): Replace declaration with that of...
1540         (make_pass_dce): ...new function.
1541         (pass_dce_loop): Replace declaration with that of...
1542         (make_pass_dce_loop): ...new function.
1543         (pass_cd_dce): Replace declaration with that of...
1544         (make_pass_cd_dce): ...new function.
1545         (pass_call_cdce): Replace declaration with that of...
1546         (make_pass_call_cdce): ...new function.
1547         (pass_merge_phi): Replace declaration with that of...
1548         (make_pass_merge_phi): ...new function.
1549         (pass_split_crit_edges): Replace declaration with that of...
1550         (make_pass_split_crit_edges): ...new function.
1551         (pass_pre): Replace declaration with that of...
1552         (make_pass_pre): ...new function.
1553         (pass_profile): Replace declaration with that of...
1554         (make_pass_profile): ...new function.
1555         (pass_strip_predict_hints): Replace declaration with that of...
1556         (make_pass_strip_predict_hints): ...new function.
1557         (pass_lower_complex_O0): Replace declaration with that of...
1558         (make_pass_lower_complex_O0): ...new function.
1559         (pass_lower_complex): Replace declaration with that of...
1560         (make_pass_lower_complex): ...new function.
1561         (pass_lower_vector): Replace declaration with that of...
1562         (make_pass_lower_vector): ...new function.
1563         (pass_lower_vector_ssa): Replace declaration with that of...
1564         (make_pass_lower_vector_ssa): ...new function.
1565         (pass_lower_omp): Replace declaration with that of...
1566         (make_pass_lower_omp): ...new function.
1567         (pass_diagnose_omp_blocks): Replace declaration with that of...
1568         (make_pass_diagnose_omp_blocks): ...new function.
1569         (pass_expand_omp): Replace declaration with that of...
1570         (make_pass_expand_omp): ...new function.
1571         (pass_expand_omp_ssa): Replace declaration with that of...
1572         (make_pass_expand_omp_ssa): ...new function.
1573         (pass_object_sizes): Replace declaration with that of...
1574         (make_pass_object_sizes): ...new function.
1575         (pass_strlen): Replace declaration with that of...
1576         (make_pass_strlen): ...new function.
1577         (pass_fold_builtins): Replace declaration with that of...
1578         (make_pass_fold_builtins): ...new function.
1579         (pass_stdarg): Replace declaration with that of...
1580         (make_pass_stdarg): ...new function.
1581         (pass_early_warn_uninitialized): Replace declaration with that of...
1582         (make_pass_early_warn_uninitialized): ...new function.
1583         (pass_late_warn_uninitialized): Replace declaration with that of...
1584         (make_pass_late_warn_uninitialized): ...new function.
1585         (pass_cse_reciprocals): Replace declaration with that of...
1586         (make_pass_cse_reciprocals): ...new function.
1587         (pass_cse_sincos): Replace declaration with that of...
1588         (make_pass_cse_sincos): ...new function.
1589         (pass_optimize_bswap): Replace declaration with that of...
1590         (make_pass_optimize_bswap): ...new function.
1591         (pass_optimize_widening_mul): Replace declaration with that of...
1592         (make_pass_optimize_widening_mul): ...new function.
1593         (pass_warn_function_return): Replace declaration with that of...
1594         (make_pass_warn_function_return): ...new function.
1595         (pass_warn_function_noreturn): Replace declaration with that of...
1596         (make_pass_warn_function_noreturn): ...new function.
1597         (pass_cselim): Replace declaration with that of...
1598         (make_pass_cselim): ...new function.
1599         (pass_phiopt): Replace declaration with that of...
1600         (make_pass_phiopt): ...new function.
1601         (pass_forwprop): Replace declaration with that of...
1602         (make_pass_forwprop): ...new function.
1603         (pass_phiprop): Replace declaration with that of...
1604         (make_pass_phiprop): ...new function.
1605         (pass_tree_ifcombine): Replace declaration with that of...
1606         (make_pass_tree_ifcombine): ...new function.
1607         (pass_dse): Replace declaration with that of...
1608         (make_pass_dse): ...new function.
1609         (pass_nrv): Replace declaration with that of...
1610         (make_pass_nrv): ...new function.
1611         (pass_rename_ssa_copies): Replace declaration with that of...
1612         (make_pass_rename_ssa_copies): ...new function.
1613         (pass_sink_code): Replace declaration with that of...
1614         (make_pass_sink_code): ...new function.
1615         (pass_fre): Replace declaration with that of...
1616         (make_pass_fre): ...new function.
1617         (pass_check_data_deps): Replace declaration with that of...
1618         (make_pass_check_data_deps): ...new function.
1619         (pass_copy_prop): Replace declaration with that of...
1620         (make_pass_copy_prop): ...new function.
1621         (pass_vrp): Replace declaration with that of...
1622         (make_pass_vrp): ...new function.
1623         (pass_uncprop): Replace declaration with that of...
1624         (make_pass_uncprop): ...new function.
1625         (pass_return_slot): Replace declaration with that of...
1626         (make_pass_return_slot): ...new function.
1627         (pass_reassoc): Replace declaration with that of...
1628         (make_pass_reassoc): ...new function.
1629         (pass_rebuild_cgraph_edges): Replace declaration with that of...
1630         (make_pass_rebuild_cgraph_edges): ...new function.
1631         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1632         (make_pass_remove_cgraph_callee_edges): ...new function.
1633         (pass_build_cgraph_edges): Replace declaration with that of...
1634         (make_pass_build_cgraph_edges): ...new function.
1635         (pass_local_pure_const): Replace declaration with that of...
1636         (make_pass_local_pure_const): ...new function.
1637         (pass_tracer): Replace declaration with that of...
1638         (make_pass_tracer): ...new function.
1639         (pass_warn_unused_result): Replace declaration with that of...
1640         (make_pass_warn_unused_result): ...new function.
1641         (pass_diagnose_tm_blocks): Replace declaration with that of...
1642         (make_pass_diagnose_tm_blocks): ...new function.
1643         (pass_lower_tm): Replace declaration with that of...
1644         (make_pass_lower_tm): ...new function.
1645         (pass_tm_init): Replace declaration with that of...
1646         (make_pass_tm_init): ...new function.
1647         (pass_tm_mark): Replace declaration with that of...
1648         (make_pass_tm_mark): ...new function.
1649         (pass_tm_memopt): Replace declaration with that of...
1650         (make_pass_tm_memopt): ...new function.
1651         (pass_tm_edges): Replace declaration with that of...
1652         (make_pass_tm_edges): ...new function.
1653         (pass_split_functions): Replace declaration with that of...
1654         (make_pass_split_functions): ...new function.
1655         (pass_feedback_split_functions): Replace declaration with that of...
1656         (make_pass_feedback_split_functions): ...new function.
1657         (pass_strength_reduction): Replace declaration with that of...
1658         (make_pass_strength_reduction): ...new function.
1659         (pass_ipa_lower_emutls): Replace declaration with that of...
1660         (make_pass_ipa_lower_emutls): ...new function.
1661         (pass_ipa_function_and_variable_visibility): Replace declaration with
1662         that of...
1663         (make_pass_ipa_function_and_variable_visibility): ...new function.
1664         (pass_ipa_tree_profile): Replace declaration with that of...
1665         (make_pass_ipa_tree_profile): ...new function.
1666         (pass_early_local_passes): Replace declaration with that of...
1667         (make_pass_early_local_passes): ...new function.
1668         (pass_ipa_whole_program_visibility): Replace declaration with that
1669         of...
1670         (make_pass_ipa_whole_program_visibility): ...new function.
1671         (pass_ipa_lto_gimple_out): Replace declaration with that of...
1672         (make_pass_ipa_lto_gimple_out): ...new function.
1673         (pass_ipa_increase_alignment): Replace declaration with that of...
1674         (make_pass_ipa_increase_alignment): ...new function.
1675         (pass_ipa_inline): Replace declaration with that of...
1676         (make_pass_ipa_inline): ...new function.
1677         (pass_ipa_free_lang_data): Replace declaration with that of...
1678         (make_pass_ipa_free_lang_data): ...new function.
1679         (pass_ipa_free_inline_summary): Replace declaration with that of...
1680         (make_pass_ipa_free_inline_summary): ...new function.
1681         (pass_ipa_cp): Replace declaration with that of...
1682         (make_pass_ipa_cp): ...new function.
1683         (pass_ipa_reference): Replace declaration with that of...
1684         (make_pass_ipa_reference): ...new function.
1685         (pass_ipa_pure_const): Replace declaration with that of...
1686         (make_pass_ipa_pure_const): ...new function.
1687         (pass_ipa_pta): Replace declaration with that of...
1688         (make_pass_ipa_pta): ...new function.
1689         (pass_ipa_lto_finish_out): Replace declaration with that of...
1690         (make_pass_ipa_lto_finish_out): ...new function.
1691         (pass_ipa_tm): Replace declaration with that of...
1692         (make_pass_ipa_tm): ...new function.
1693         (pass_ipa_profile): Replace declaration with that of...
1694         (make_pass_ipa_profile): ...new function.
1695         (pass_ipa_cdtor_merge): Replace declaration with that of...
1696         (make_pass_ipa_cdtor_merge): ...new function.
1697         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1698         of...
1699         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1700         (pass_init_datastructures): Replace declaration with that of...
1701         (make_pass_init_datastructures): ...new function.
1702         (pass_fixup_cfg): Replace declaration with that of...
1703         (make_pass_fixup_cfg): ...new function.
1704         (pass_expand): Replace declaration with that of...
1705         (make_pass_expand): ...new function.
1706         (pass_instantiate_virtual_regs): Replace declaration with that of...
1707         (make_pass_instantiate_virtual_regs): ...new function.
1708         (pass_rtl_fwprop): Replace declaration with that of...
1709         (make_pass_rtl_fwprop): ...new function.
1710         (pass_rtl_fwprop_addr): Replace declaration with that of...
1711         (make_pass_rtl_fwprop_addr): ...new function.
1712         (pass_jump): Replace declaration with that of...
1713         (make_pass_jump): ...new function.
1714         (pass_jump2): Replace declaration with that of...
1715         (make_pass_jump2): ...new function.
1716         (pass_lower_subreg): Replace declaration with that of...
1717         (make_pass_lower_subreg): ...new function.
1718         (pass_cse): Replace declaration with that of...
1719         (make_pass_cse): ...new function.
1720         (pass_fast_rtl_dce): Replace declaration with that of...
1721         (make_pass_fast_rtl_dce): ...new function.
1722         (pass_ud_rtl_dce): Replace declaration with that of...
1723         (make_pass_ud_rtl_dce): ...new function.
1724         (pass_rtl_dce): Replace declaration with that of...
1725         (make_pass_rtl_dce): ...new function.
1726         (pass_rtl_dse1): Replace declaration with that of...
1727         (make_pass_rtl_dse1): ...new function.
1728         (pass_rtl_dse2): Replace declaration with that of...
1729         (make_pass_rtl_dse2): ...new function.
1730         (pass_rtl_dse3): Replace declaration with that of...
1731         (make_pass_rtl_dse3): ...new function.
1732         (pass_rtl_cprop): Replace declaration with that of...
1733         (make_pass_rtl_cprop): ...new function.
1734         (pass_rtl_pre): Replace declaration with that of...
1735         (make_pass_rtl_pre): ...new function.
1736         (pass_rtl_hoist): Replace declaration with that of...
1737         (make_pass_rtl_hoist): ...new function.
1738         (pass_rtl_store_motion): Replace declaration with that of...
1739         (make_pass_rtl_store_motion): ...new function.
1740         (pass_cse_after_global_opts): Replace declaration with that of...
1741         (make_pass_cse_after_global_opts): ...new function.
1742         (pass_rtl_ifcvt): Replace declaration with that of...
1743         (make_pass_rtl_ifcvt): ...new function.
1744         (pass_into_cfg_layout_mode): Replace declaration with that of...
1745         (make_pass_into_cfg_layout_mode): ...new function.
1746         (pass_outof_cfg_layout_mode): Replace declaration with that of...
1747         (make_pass_outof_cfg_layout_mode): ...new function.
1748         (pass_loop2): Replace declaration with that of...
1749         (make_pass_loop2): ...new function.
1750         (pass_rtl_loop_init): Replace declaration with that of...
1751         (make_pass_rtl_loop_init): ...new function.
1752         (pass_rtl_move_loop_invariants): Replace declaration with that of...
1753         (make_pass_rtl_move_loop_invariants): ...new function.
1754         (pass_rtl_unswitch): Replace declaration with that of...
1755         (make_pass_rtl_unswitch): ...new function.
1756         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1757         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1758         (pass_rtl_doloop): Replace declaration with that of...
1759         (make_pass_rtl_doloop): ...new function.
1760         (pass_rtl_loop_done): Replace declaration with that of...
1761         (make_pass_rtl_loop_done): ...new function.
1762         (pass_web): Replace declaration with that of...
1763         (make_pass_web): ...new function.
1764         (pass_cse2): Replace declaration with that of...
1765         (make_pass_cse2): ...new function.
1766         (pass_df_initialize_opt): Replace declaration with that of...
1767         (make_pass_df_initialize_opt): ...new function.
1768         (pass_df_initialize_no_opt): Replace declaration with that of...
1769         (make_pass_df_initialize_no_opt): ...new function.
1770         (pass_reginfo_init): Replace declaration with that of...
1771         (make_pass_reginfo_init): ...new function.
1772         (pass_inc_dec): Replace declaration with that of...
1773         (make_pass_inc_dec): ...new function.
1774         (pass_stack_ptr_mod): Replace declaration with that of...
1775         (make_pass_stack_ptr_mod): ...new function.
1776         (pass_initialize_regs): Replace declaration with that of...
1777         (make_pass_initialize_regs): ...new function.
1778         (pass_combine): Replace declaration with that of...
1779         (make_pass_combine): ...new function.
1780         (pass_if_after_combine): Replace declaration with that of...
1781         (make_pass_if_after_combine): ...new function.
1782         (pass_ree): Replace declaration with that of...
1783         (make_pass_ree): ...new function.
1784         (pass_partition_blocks): Replace declaration with that of...
1785         (make_pass_partition_blocks): ...new function.
1786         (pass_match_asm_constraints): Replace declaration with that of...
1787         (make_pass_match_asm_constraints): ...new function.
1788         (pass_regmove): Replace declaration with that of...
1789         (make_pass_regmove): ...new function.
1790         (pass_split_all_insns): Replace declaration with that of...
1791         (make_pass_split_all_insns): ...new function.
1792         (pass_fast_rtl_byte_dce): Replace declaration with that of...
1793         (make_pass_fast_rtl_byte_dce): ...new function.
1794         (pass_lower_subreg2): Replace declaration with that of...
1795         (make_pass_lower_subreg2): ...new function.
1796         (pass_mode_switching): Replace declaration with that of...
1797         (make_pass_mode_switching): ...new function.
1798         (pass_sms): Replace declaration with that of...
1799         (make_pass_sms): ...new function.
1800         (pass_sched): Replace declaration with that of...
1801         (make_pass_sched): ...new function.
1802         (pass_ira): Replace declaration with that of...
1803         (make_pass_ira): ...new function.
1804         (pass_reload): Replace declaration with that of...
1805         (make_pass_reload): ...new function.
1806         (pass_clean_state): Replace declaration with that of...
1807         (make_pass_clean_state): ...new function.
1808         (pass_branch_prob): Replace declaration with that of...
1809         (make_pass_branch_prob): ...new function.
1810         (pass_value_profile_transformations): Replace declaration with that
1811         of...
1812         (make_pass_value_profile_transformations): ...new function.
1813         (pass_postreload_cse): Replace declaration with that of...
1814         (make_pass_postreload_cse): ...new function.
1815         (pass_gcse2): Replace declaration with that of...
1816         (make_pass_gcse2): ...new function.
1817         (pass_split_after_reload): Replace declaration with that of...
1818         (make_pass_split_after_reload): ...new function.
1819         (pass_branch_target_load_optimize1): Replace declaration with that
1820         of...
1821         (make_pass_branch_target_load_optimize1): ...new function.
1822         (pass_thread_prologue_and_epilogue): Replace declaration with that
1823         of...
1824         (make_pass_thread_prologue_and_epilogue): ...new function.
1825         (pass_stack_adjustments): Replace declaration with that of...
1826         (make_pass_stack_adjustments): ...new function.
1827         (pass_peephole2): Replace declaration with that of...
1828         (make_pass_peephole2): ...new function.
1829         (pass_if_after_reload): Replace declaration with that of...
1830         (make_pass_if_after_reload): ...new function.
1831         (pass_regrename): Replace declaration with that of...
1832         (make_pass_regrename): ...new function.
1833         (pass_cprop_hardreg): Replace declaration with that of...
1834         (make_pass_cprop_hardreg): ...new function.
1835         (pass_reorder_blocks): Replace declaration with that of...
1836         (make_pass_reorder_blocks): ...new function.
1837         (pass_branch_target_load_optimize2): Replace declaration with that
1838         of...
1839         (make_pass_branch_target_load_optimize2): ...new function.
1840         (pass_leaf_regs): Replace declaration with that of...
1841         (make_pass_leaf_regs): ...new function.
1842         (pass_split_before_sched2): Replace declaration with that of...
1843         (make_pass_split_before_sched2): ...new function.
1844         (pass_compare_elim_after_reload): Replace declaration with that of...
1845         (make_pass_compare_elim_after_reload): ...new function.
1846         (pass_sched2): Replace declaration with that of...
1847         (make_pass_sched2): ...new function.
1848         (pass_stack_regs): Replace declaration with that of...
1849         (make_pass_stack_regs): ...new function.
1850         (pass_stack_regs_run): Replace declaration with that of...
1851         (make_pass_stack_regs_run): ...new function.
1852         (pass_df_finish): Replace declaration with that of...
1853         (make_pass_df_finish): ...new function.
1854         (pass_compute_alignments): Replace declaration with that of...
1855         (make_pass_compute_alignments): ...new function.
1856         (pass_duplicate_computed_gotos): Replace declaration with that of...
1857         (make_pass_duplicate_computed_gotos): ...new function.
1858         (pass_variable_tracking): Replace declaration with that of...
1859         (make_pass_variable_tracking): ...new function.
1860         (pass_free_cfg): Replace declaration with that of...
1861         (make_pass_free_cfg): ...new function.
1862         (pass_machine_reorg): Replace declaration with that of...
1863         (make_pass_machine_reorg): ...new function.
1864         (pass_cleanup_barriers): Replace declaration with that of...
1865         (make_pass_cleanup_barriers): ...new function.
1866         (pass_delay_slots): Replace declaration with that of...
1867         (make_pass_delay_slots): ...new function.
1868         (pass_split_for_shorten_branches): Replace declaration with that of...
1869         (make_pass_split_for_shorten_branches): ...new function.
1870         (pass_split_before_regstack): Replace declaration with that of...
1871         (make_pass_split_before_regstack): ...new function.
1872         (pass_convert_to_eh_region_ranges): Replace declaration with that
1873         of...
1874         (make_pass_convert_to_eh_region_ranges): ...new function.
1875         (pass_shorten_branches): Replace declaration with that of...
1876         (make_pass_shorten_branches): ...new function.
1877         (pass_set_nothrow_function_flags): Replace declaration with that of...
1878         (make_pass_set_nothrow_function_flags): ...new function.
1879         (pass_dwarf2_frame): Replace declaration with that of...
1880         (make_pass_dwarf2_frame): ...new function.
1881         (pass_final): Replace declaration with that of...
1882         (make_pass_final): ...new function.
1883         (pass_rtl_seqabstr): Replace declaration with that of...
1884         (make_pass_rtl_seqabstr): ...new function.
1885         (pass_release_ssa_names): Replace declaration with that of...
1886         (make_pass_release_ssa_names): ...new function.
1887         (pass_early_inline): Replace declaration with that of...
1888         (make_pass_early_inline): ...new function.
1889         (pass_inline_parameters): Replace declaration with that of...
1890         (make_pass_inline_parameters): ...new function.
1891         (pass_update_address_taken): Replace declaration with that of...
1892         (make_pass_update_address_taken): ...new function.
1893         (pass_convert_switch): Replace declaration with that of...
1894         (make_pass_convert_switch): ...new function.
1895         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1896         to a subclass of simple_ipa_opt_pass along with...
1897         (pass_data_ipa_tree_profile): ...new pass_data instance and...
1898         (make_pass_ipa_tree_profile): ...new function.
1899         * tree-sra.c (pass_sra_early): Convert from a global struct to a
1900         subclass of gimple_opt_pass along with...
1901         (pass_data_sra_early): ...new pass_data instance and...
1902         (make_pass_sra_early): ...new function.
1903         (pass_sra): Convert from a global struct to a subclass of
1904         gimple_opt_pass along with...
1905         (pass_data_sra): ...new pass_data instance and...
1906         (make_pass_sra): ...new function.
1907         (pass_early_ipa_sra): Convert from a global struct to a subclass of
1908         gimple_opt_pass along with...
1909         (pass_data_early_ipa_sra): ...new pass_data instance and...
1910         (make_pass_early_ipa_sra): ...new function.
1911         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1912         subclass of gimple_opt_pass along with...
1913         (pass_data_ccp): ...new pass_data instance and...
1914         (make_pass_ccp): ...new function.
1915         (pass_fold_builtins): Convert from a global struct to a subclass of
1916         gimple_opt_pass along with...
1917         (pass_data_fold_builtins): ...new pass_data instance and...
1918         (make_pass_fold_builtins): ...new function.
1919         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1920         subclass of gimple_opt_pass along with...
1921         (pass_data_copy_prop): ...new pass_data instance and...
1922         (make_pass_copy_prop): ...new function.
1923         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1924         global struct to a subclass of gimple_opt_pass along with...
1925         (pass_data_rename_ssa_copies): ...new pass_data instance and...
1926         (make_pass_rename_ssa_copies): ...new function.
1927         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1928         subclass of gimple_opt_pass along with...
1929         (pass_data_dce): ...new pass_data instance and...
1930         (make_pass_dce): ...new function.
1931         (pass_dce_loop): Convert from a global struct to a subclass of
1932         gimple_opt_pass along with...
1933         (pass_data_dce_loop): ...new pass_data instance and...
1934         (make_pass_dce_loop): ...new function.
1935         (pass_cd_dce): Convert from a global struct to a subclass of
1936         gimple_opt_pass along with...
1937         (pass_data_cd_dce): ...new pass_data instance and...
1938         (make_pass_cd_dce): ...new function.
1939         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1940         subclass of gimple_opt_pass along with...
1941         (pass_data_dominator): ...new pass_data instance and...
1942         (make_pass_dominator): ...new function.
1943         (pass_phi_only_cprop): Convert from a global struct to a subclass of
1944         gimple_opt_pass along with...
1945         (pass_data_phi_only_cprop): ...new pass_data instance and...
1946         (make_pass_phi_only_cprop): ...new function.
1947         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1948         subclass of gimple_opt_pass along with...
1949         (pass_data_dse): ...new pass_data instance and...
1950         (make_pass_dse): ...new function.
1951         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1952         a subclass of gimple_opt_pass along with...
1953         (pass_data_forwprop): ...new pass_data instance and...
1954         (make_pass_forwprop): ...new function.
1955         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1956         struct to a subclass of gimple_opt_pass along with...
1957         (pass_data_tree_ifcombine): ...new pass_data instance and...
1958         (make_pass_tree_ifcombine): ...new function.
1959         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1960         subclass of gimple_opt_pass along with...
1961         (pass_data_ch): ...new pass_data instance and...
1962         (make_pass_ch): ...new function.
1963         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1964         subclass of gimple_opt_pass along with...
1965         (pass_data_tree_loop): ...new pass_data instance and...
1966         (make_pass_tree_loop): ...new function.
1967         (pass_tree_loop_init): Convert from a global struct to a subclass of
1968         gimple_opt_pass along with...
1969         (pass_data_tree_loop_init): ...new pass_data instance and...
1970         (make_pass_tree_loop_init): ...new function.
1971         (pass_lim): Convert from a global struct to a subclass of
1972         gimple_opt_pass along with...
1973         (pass_data_lim): ...new pass_data instance and...
1974         (make_pass_lim): ...new function.
1975         (pass_tree_unswitch): Convert from a global struct to a subclass of
1976         gimple_opt_pass along with...
1977         (pass_data_tree_unswitch): ...new pass_data instance and...
1978         (make_pass_tree_unswitch): ...new function.
1979         (pass_predcom): Convert from a global struct to a subclass of
1980         gimple_opt_pass along with...
1981         (pass_data_predcom): ...new pass_data instance and...
1982         (make_pass_predcom): ...new function.
1983         (pass_vectorize): Convert from a global struct to a subclass of
1984         gimple_opt_pass along with...
1985         (pass_data_vectorize): ...new pass_data instance and...
1986         (make_pass_vectorize): ...new function.
1987         (pass_graphite): Convert from a global struct to a subclass of
1988         gimple_opt_pass along with...
1989         (pass_data_graphite): ...new pass_data instance and...
1990         (make_pass_graphite): ...new function.
1991         (pass_graphite_transforms): Convert from a global struct to a subclass
1992         of gimple_opt_pass along with...
1993         (pass_data_graphite_transforms): ...new pass_data instance and...
1994         (make_pass_graphite_transforms): ...new function.
1995         (pass_check_data_deps): Convert from a global struct to a subclass of
1996         gimple_opt_pass along with...
1997         (pass_data_check_data_deps): ...new pass_data instance and...
1998         (make_pass_check_data_deps): ...new function.
1999         (pass_iv_canon): Convert from a global struct to a subclass of
2000         gimple_opt_pass along with...
2001         (pass_data_iv_canon): ...new pass_data instance and...
2002         (make_pass_iv_canon): ...new function.
2003         (pass_scev_cprop): Convert from a global struct to a subclass of
2004         gimple_opt_pass along with...
2005         (pass_data_scev_cprop): ...new pass_data instance and...
2006         (make_pass_scev_cprop): ...new function.
2007         (pass_record_bounds): Convert from a global struct to a subclass of
2008         gimple_opt_pass along with...
2009         (pass_data_record_bounds): ...new pass_data instance and...
2010         (make_pass_record_bounds): ...new function.
2011         (pass_complete_unroll): Convert from a global struct to a subclass of
2012         gimple_opt_pass along with...
2013         (pass_data_complete_unroll): ...new pass_data instance and...
2014         (make_pass_complete_unroll): ...new function.
2015         (pass_complete_unrolli): Convert from a global struct to a subclass of
2016         gimple_opt_pass along with...
2017         (pass_data_complete_unrolli): ...new pass_data instance and...
2018         (make_pass_complete_unrolli): ...new function.
2019         (pass_parallelize_loops): Convert from a global struct to a subclass
2020         of gimple_opt_pass along with...
2021         (pass_data_parallelize_loops): ...new pass_data instance and...
2022         (make_pass_parallelize_loops): ...new function.
2023         (pass_loop_prefetch): Convert from a global struct to a subclass of
2024         gimple_opt_pass along with...
2025         (pass_data_loop_prefetch): ...new pass_data instance and...
2026         (make_pass_loop_prefetch): ...new function.
2027         (pass_iv_optimize): Convert from a global struct to a subclass of
2028         gimple_opt_pass along with...
2029         (pass_data_iv_optimize): ...new pass_data instance and...
2030         (make_pass_iv_optimize): ...new function.
2031         (pass_tree_loop_done): Convert from a global struct to a subclass of
2032         gimple_opt_pass along with...
2033         (pass_data_tree_loop_done): ...new pass_data instance and...
2034         (make_pass_tree_loop_done): ...new function.
2035         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2036         struct to a subclass of gimple_opt_pass along with...
2037         (pass_data_cse_reciprocals): ...new pass_data instance and...
2038         (make_pass_cse_reciprocals): ...new function.
2039         (pass_cse_sincos): Convert from a global struct to a subclass of
2040         gimple_opt_pass along with...
2041         (pass_data_cse_sincos): ...new pass_data instance and...
2042         (make_pass_cse_sincos): ...new function.
2043         (pass_optimize_bswap): Convert from a global struct to a subclass of
2044         gimple_opt_pass along with...
2045         (pass_data_optimize_bswap): ...new pass_data instance and...
2046         (make_pass_optimize_bswap): ...new function.
2047         (pass_optimize_widening_mul): Convert from a global struct to a
2048         subclass of gimple_opt_pass along with...
2049         (pass_data_optimize_widening_mul): ...new pass_data instance and...
2050         (make_pass_optimize_widening_mul): ...new function.
2051         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2052         subclass of gimple_opt_pass along with...
2053         (pass_data_phiopt): ...new pass_data instance and...
2054         (make_pass_phiopt): ...new function.
2055         (pass_cselim): Convert from a global struct to a subclass of
2056         gimple_opt_pass along with...
2057         (pass_data_cselim): ...new pass_data instance and...
2058         (make_pass_cselim): ...new function.
2059         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2060         subclass of gimple_opt_pass along with...
2061         (pass_data_phiprop): ...new pass_data instance and...
2062         (make_pass_phiprop): ...new function.
2063         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2064         subclass of gimple_opt_pass along with...
2065         (pass_data_pre): ...new pass_data instance and...
2066         (make_pass_pre): ...new function.
2067         (pass_fre): Convert from a global struct to a subclass of
2068         gimple_opt_pass along with...
2069         (pass_data_fre): ...new pass_data instance and...
2070         (make_pass_fre): ...new function.
2071         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2072         subclass of gimple_opt_pass along with...
2073         (pass_data_reassoc): ...new pass_data instance and...
2074         (make_pass_reassoc): ...new function.
2075         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2076         subclass of gimple_opt_pass along with...
2077         (pass_data_sink_code): ...new pass_data instance and...
2078         (make_pass_sink_code): ...new function.
2079         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2080         subclass of gimple_opt_pass along with...
2081         (pass_data_strlen): ...new pass_data instance and...
2082         (make_pass_strlen): ...new function.
2083         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2084         struct to a subclass of gimple_opt_pass along with...
2085         (pass_data_build_alias): ...new pass_data instance and...
2086         (make_pass_build_alias): ...new function.
2087         (pass_build_ealias): Convert from a global struct to a subclass of
2088         gimple_opt_pass along with...
2089         (pass_data_build_ealias): ...new pass_data instance and...
2090         (make_pass_build_ealias): ...new function.
2091         (pass_ipa_pta): Convert from a global struct to a subclass of
2092         simple_ipa_opt_pass along with...
2093         (pass_data_ipa_pta): ...new pass_data instance and...
2094         (make_pass_ipa_pta): ...new function.
2095         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2096         subclass of gimple_opt_pass along with...
2097         (pass_data_uncprop): ...new pass_data instance and...
2098         (make_pass_uncprop): ...new function.
2099         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2100         global struct to a subclass of gimple_opt_pass along with...
2101         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2102         (make_pass_late_warn_uninitialized): ...new function.
2103         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2104         to a subclass of gimple_opt_pass along with...
2105         (pass_data_init_datastructures): ...new pass_data instance and...
2106         (make_pass_init_datastructures): ...new function.
2107         (pass_early_warn_uninitialized): Convert from a global struct to a
2108         subclass of gimple_opt_pass along with...
2109         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2110         (make_pass_early_warn_uninitialized): ...new function.
2111         (pass_update_address_taken): Convert from a global struct to a
2112         subclass of gimple_opt_pass along with...
2113         (pass_data_update_address_taken): ...new pass_data instance and...
2114         (make_pass_update_address_taken): ...new function.
2115         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2116         struct to a subclass of gimple_opt_pass along with...
2117         (pass_data_release_ssa_names): ...new pass_data instance and...
2118         (make_pass_release_ssa_names): ...new function.
2119         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2120         subclass of gimple_opt_pass along with...
2121         (pass_data_stdarg): ...new pass_data instance and...
2122         (make_pass_stdarg): ...new function.
2123         * tree-switch-conversion.c (pass_convert_switch): Convert from a
2124         global struct to a subclass of gimple_opt_pass along with...
2125         (pass_data_convert_switch): ...new pass_data instance and...
2126         (make_pass_convert_switch): ...new function.
2127         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2128         to a subclass of gimple_opt_pass along with...
2129         (pass_data_tail_recursion): ...new pass_data instance and...
2130         (make_pass_tail_recursion): ...new function.
2131         (pass_tail_calls): Convert from a global struct to a subclass of
2132         gimple_opt_pass along with...
2133         (pass_data_tail_calls): ...new pass_data instance and...
2134         (make_pass_tail_calls): ...new function.
2135         * tree-vect-generic.c (pass_lower_vector): Convert from a global
2136         struct to a subclass of gimple_opt_pass along with...
2137         (pass_data_lower_vector): ...new pass_data instance and...
2138         (make_pass_lower_vector): ...new function.
2139         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2140         gimple_opt_pass along with...
2141         (pass_data_lower_vector_ssa): ...new pass_data instance and...
2142         (make_pass_lower_vector_ssa): ...new function.
2143         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2144         to a subclass of gimple_opt_pass along with...
2145         (pass_data_slp_vectorize): ...new pass_data instance and...
2146         (make_pass_slp_vectorize): ...new function.
2147         (pass_ipa_increase_alignment): Convert from a global struct to a
2148         subclass of simple_ipa_opt_pass along with...
2149         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2150         (make_pass_ipa_increase_alignment): ...new function.
2151         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2152         gimple_opt_pass along with...
2153         (pass_data_vrp): ...new pass_data instance and...
2154         (make_pass_vrp): ...new function.
2155         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2156         subclass of simple_ipa_opt_pass along with...
2157         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2158         (make_pass_ipa_free_lang_data): ...new function.
2159         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2160         gimple_opt_pass along with...
2161         (pass_data_tsan): ...new pass_data instance and...
2162         (make_pass_tsan): ...new function.
2163         (pass_tsan_O0): Convert from a global struct to a subclass of
2164         gimple_opt_pass along with...
2165         (pass_data_tsan_O0): ...new pass_data instance and...
2166         (make_pass_tsan_O0): ...new function.
2167         * var-tracking.c (pass_variable_tracking): Convert from a global
2168         struct to a subclass of rtl_opt_pass along with...
2169         (pass_data_variable_tracking): ...new pass_data instance and...
2170         (make_pass_variable_tracking): ...new function.
2171         * web.c (pass_web): Convert from a global struct to a subclass of
2172         rtl_opt_pass along with...
2173         (pass_data_web): ...new pass_data instance and...
2174         (make_pass_web): ...new function.
2175         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2176         declaration with that of...
2177         (make_pass_mode_switch_use): ...new function.
2178         (pass_resolve_sw_modes): Replace declaration with that of...
2179         (make_pass_resolve_sw_modes): ...new function.
2180         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2181         from a global struct to a subclass of rtl_opt_pass along with...
2182         (pass_data_mode_switch_use): ...new pass_data instance and...
2183         (make_pass_mode_switch_use): ...new function.
2184         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2185         from a global struct to a subclass of rtl_opt_pass along with...
2186         (pass_data_resolve_sw_modes): ...new pass_data instance and...
2187         (make_pass_resolve_sw_modes): ...new function.
2188         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2189         struct to a subclass of rtl_opt_pass along with...
2190         (pass_data_insert_vzeroupper): ...new pass_data instance and...
2191         (make_pass_insert_vzeroupper): ...new function.
2192         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2193         global struct to a subclass of rtl_opt_pass along with...
2194         (pass_data_work_around_errata): ...new pass_data instance and...
2195         (make_pass_work_around_errata): ...new function.
2196         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2197         struct to a subclass of rtl_opt_pass along with...
2198         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2199         (make_pass_mips_machine_reorg2): ...new function.
2201 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2203         * passes.c (pass_manager::operator new): New.
2205 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2207         Handwritten part of conversion of passes to C++ classes.
2209         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2210         (toplev.o): Add dep on PASS_MANAGER_H.
2211         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2212         of early local pases to reflect this moving from a global to a
2213         member of gcc::pass_manager.
2214         (cgraph_add_new_function): Likewise.
2215         * lto-cgraph.c (lto_output_node): Update for conversion of
2216         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2217         * passes.c (opt_pass::clone): New.
2218         (opt_pass::gate): New.
2219         (opt_pass::execute): New.
2220         (opt_pass::opt_pass): New.
2221         (pass_manager::execute_early_local_passes): New.
2222         (pass_manager::execute_pass_mode_switching): new.
2223         (finish_optimization_passes): Convert to...
2224         (pass_manager::finish_optimization_passes): ...this.
2225         (finish_optimization_passes): Update for conversion of passes to
2226         C++ classes.
2227         (register_dump_files_1): Use has_gate since we cannot portably
2228         check a vtable entry against NULL.
2229         (dump_one_pass): Likewise.
2230         (ipa_write_summaries_2): Likewise.
2231         (ipa_write_optimization_summaries_1): Likewise.
2232         (ipa_read_summaries_1): Likewise.
2233         (ipa_read_optimization_summaries_1): Likewise.
2234         (execute_ipa_stmt_fixups): Likewise.
2235         (pass_manager::pass_manager): Rewrite pass-creation, invoking
2236         pass-creation functions rather than wiring up globals, and
2237         storing the results in fields of pass_manager generated using
2238         pass-instances.def.
2239         (pass_manager::dump_profile_report): Update for conversion of
2240         passes to C++ classes.
2241         (pass_manager::execute_ipa_summary_passes): Likewise.
2242         (execute_one_ipa_transform_pass): Likewise.
2243         (execute_one_pass): Use has_gate and has_execute since we cannot
2244         portably check a vtable entry against NULL.
2245         * pass_manager.h (pass_manager::finish_optimization_passes): New.
2246         (pass_manager): Use pass-instances.def to add fields for the
2247         various pass instances.
2248         * toplev.c (finalize): Update for move of
2249         finish_optimization_passes to a method of gcc::pass_manager.
2250         * toplev.h (finish_optimization_passes): Move to method of class
2251         pass_manager.
2252         * tree-pass.h (struct pass_data): New.
2253         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2254         (opt_pass::gate): Convert to virtual function.
2255         (opt_pass::~opt_pass): New.
2256         (opt_pass::clone): New.
2257         (opt_pass::execute): Convert to virtual function.
2258         (opt_pass::opt_pass): New.
2259         (opt_pass::ctxt_): new.
2260         (gimple_opt_pass): Convert to subclass of opt_pass.
2261         (gimple_opt_pass::gimple_opt_pass): New.
2262         (rtl_opt_pass): Convert to subclass of opt_pass.
2263         (rtl_opt_pass::rtl_opt_pass): New.
2264         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2265         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2266         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2267         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2268         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2269         invocation of pass_mode_switching to reflect this moving from a
2270         global to a member of gcc::pass_manager.
2271         (ix86_option_override): Rework how pass_insert_vzeroupper is
2272         added to the pass_manager to reflect autogenerated changes.
2273         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2275 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2277         PR rtl-optimization/57708
2278         * recog.c (peep2_find_free_register): Validate all regs in a
2279         multi-reg mode.
2281 2013-08-05  Jan Hubicka  <jh@suse.cz>
2283         PR lto/57602
2284         * cgraph.c (verify_cgraph_node): Accept local flags from other
2285         partitions.
2286         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2287         (function_and_variable_visibility): Likewise.
2288         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2290 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2292         * graph.c (init_graph_slim_pretty_print): Remove.
2293         (print_graph_cfg): Do not call it.  Use local pretty printer.
2294         (start_graph_dump): Likewise.
2296 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2298         * gimple-pretty-print.c (buffer): Remove.
2299         (initialized): Likewise.
2300         (maybe_init_pretty_print): Likewise.
2301         (print_gimple_stmt): Do not call it.  Use non-static local
2302         pretty_printer variable.
2303         (print_gimple_expr): Likewise.
2304         (print_gimple_seq): Likewise.
2305         (gimple_dump_bb): Likewise.
2307 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2309         * asan.c (asan_pp): Remove.
2310         (asan_pp_initialized): Likewise.
2311         (asan_pp_initialize): Likewise.
2312         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2313         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2314         (asan_add_global): Likewise.
2316 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2318         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2319         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2320         * pretty-print.c (pp_formatted_text_data): Likewise.
2321         (pp_indent): Rename from pp_base_indent.
2322         (pp_format): Rename from pp_base_format.
2323         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2324         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2325         (pp_flush): Rename from pp_base_flush.
2326         (pp_set_line_maximum_length): Rename from
2327         pp_base_set_line_maximum_length.
2328         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2329         (pp_set_prefix): Rename from pp_base_set_prefix.
2330         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2331         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2332         (pp_append_text): Rename from pp_base_append_text.
2333         (pp_formatted_text): Rename from pp_base_formatted_text.
2334         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2335         (pp_remaining_character_count_for_line): Rename from
2336         pp_base_remaining_character_count_for_line.
2337         (pp_newline): Rename from pp_base_newline.
2338         (pp_character): Rename from pp_base_character.
2339         (pp_string): Rename from pp_base_string.
2340         (pp_maybe_space): Rename from pp_base_maybe_space.
2341         * asan.c (asan_pp_string): Adjust.
2342         (asan_emit_stack_protection): Likewise.
2343         (asan_add_global): Likewise.
2344         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2345         * tree-mudflap.c (mf_varname_tree): Likewise.
2346         * tree-pretty-print.c (pp_tree_identifier): Rename from
2347         pp_base_tree_identifier.
2348         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2349         Declare as function.
2351 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2353         * pretty-print.h (pp_bar_bar): New.
2354         (pp_ampersand_ampersand): Likewise.
2355         (pp_less_equal): Likewise.
2356         (pp_greater_equal): Likewise.
2357         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2358         printer functions instead of pp_string or operators and punctuators.
2359         (dump_gimple_call): Likewise.
2360         (dump_gimple_omp_for): Likewise.
2361         (dump_gimple_transaction): Likewise.
2362         (dump_gimple_phi): Likewise.
2363         (pp_gimple_stmt_1): Likewise.
2364         * sched-vis.c (print_insn): Likewise.
2365         * tree-mudflap.c (mf_varname_tree): Likewise.
2366         * tree-pretty-print.c (dump_block_node): Likewise.
2367         (dump_generic_node): Likewise.
2369 2013-08-02  Jan Hubicka  <jh@suse.cz>
2371         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2372         boundaries.
2373         * lto-streamer-out.c (tree_is_indexable): Results decls and
2374         parm decls are not indexable.
2375         (DFS_write_tree_body): Do not follow args and results.
2376         (hash_tree): Likewise.
2377         (output_functions): Rearrange so struct function is needed
2378         only when real body is output; be able to also ouptut abstract
2379         functions; output DECL_ARGUMENTS and DECL_RESULT.
2380         (lto_output): When not in WPA, ale store abstract functions.
2381         (write_symbol): Do not care about RESULT_DECL.
2382         (output_symbol_p): Handle correctly sbtract decls.
2383         * lto-streamer-in.c (input_function): Rearrange so struct
2384         function can be NULL at entry; allow streaming of
2385         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2386         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2387         sanity check during LTO.
2388         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2389         RESULT_DECl and DECL_ARGUMENTS.
2390         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2391         Likewise.
2393 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2395         * pretty-print.h (pp_underscore): New.
2396         (pp_comma): Tidy.
2397         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2398         printer functions instead of pp_character.
2399         (dump_binary_rhs): Likewise.
2400         (dump_ternary_rhs): Likewise.
2401         (dump_gimple_call_args): Likewise.
2402         (pp_points_to_solution): Likewise.
2403         (dump_gimple_call): Likewise.
2404         (dump_gimple_switch): Likewise.
2405         (dump_gimple_cond): Likewise.
2406         (dump_gimple_bind): Likewise.
2407         (dump_gimple_try): Likewise.
2408         (dump_gimple_omp_for): Likewise.
2409         (dump_gimple_omp_continue): Likewise.
2410         (dump_gimple_omp_single): Likewise.
2411         (dump_gimple_omp_sections): Likewise.
2412         (dump_gimple_omp_block): Likewise.
2413         (dump_gimple_omp_critical): Likewise.
2414         (dump_gimple_transaction): Likewise.
2415         (dump_gimple_asm): Likewise.
2416         (dump_gimple_phi): Likewise.
2417         (dump_gimple_omp_parallel): Likewise.
2418         (dump_gimple_omp_task): Likewise.
2419         (dump_gimple_omp_atomic_load): Likewise.
2420         (dump_gimple_omp_atomic_store): Likewise.
2421         (dump_gimple_mem_ops): Likewise.
2422         (pp_gimple_stmt_1): Likewise.
2423         (pp_cfg_jump): Likewise.
2424         (dump_implicit_edges): Likewise.
2425         (gimple_dump_bb_for_graph): Likewise.
2426         * graph.c (draw_cfg_node): Likewise.
2427         * langhooks.c (lhd_print_error_function): Likewise.
2428         * sched-vis.c (print_exp): Likewise.
2429         (print_value): Likewise.
2430         (print_pattern): Likewise.
2431         (print_insn): Likewise.
2432         (rtl_dump_bb_for_graph): Likewise.
2433         * tree-pretty-print.c (dump_function_declaration): Likewise.
2434         (dump_array_domain): Likewise.
2435         (dump_omp_clause): Likewise.
2436         (dump_location): Likewise.
2437         (dump_generic_node): Likewise.
2438         (print_struct_decl): Likewise.
2439         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2441 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2443         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2444         candidate table when replacing a candidate statement.
2445         (replace_rhs_if_not_dup): Likewise.
2446         (replace_one_candidate): Likewise.
2448 2013-08-02  Jan Hubicka  <jh@suse.cz>
2449             Martin Liska  <marxin.liska@gmail.com>
2451         * cgraphunit.c (add_new_function): Fix logic when adding from
2452         late IPA pass.
2453         (assemble_thunk): Rename to ...
2454         (expand_thunk); .. this one; export; get it working with
2455         general functions; make produced gimple valid.
2456         * cgraph.h (expand_thunk): Declare.
2458 2013-08-02  Jan Hubicka  <jh@suse.cz>
2460         * ipa-cp.c (gather_context_independent_values): Use
2461         ipa_get_param_move_cost.
2462         (get_replacement_map): Remove PARAM; move parameter folding
2463         into tree-inline.c
2464         (create_specialized_node): Update.
2465         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2466         assert that we have gimple body; update move_cost.
2467         (count_formal_params): Assert that we have gimple body.
2468         (ipa_dump_param): New function.
2469         (ipa_alloc_node_params): Break out from ...
2470         (ipa_initialize_node_params): ... here.
2471         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2472         (ipa_write_node_info): Stream move costs.
2473         (ipa_read_node_info): Read move costs.
2474         (ipa_update_after_lto_read): Do not recompute node params.
2475         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2476         (ipa_get_param): Check we are not in WPA.
2477         (ipa_get_param_move_cost): New.
2478         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2479         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2480         parm numbers to be present.
2482 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2484         PR rtl-optimization/58048
2485         * lra-constraints.c (process_alt_operands): Don't check asm
2486         operand on register.
2488 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2490         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2491         the implied StoreLoad barrier for atomic operations if before.
2493 2013-08-02  Jan Hubicka  <jh@suse.cz>
2494             Martin Liska  <marxin.liska@gmail.com>
2496         * cgraph.c (cgraph_function_body_availability): Do not check
2497         cgraph flags.
2498         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2499         symtab_node_availability): Declare.
2500         * ipa.c (can_replace_by_local_alias): New.
2501         (function_and_variable_visibility): Use it.
2502         * symtab.c (symtab_for_node_and_aliases,
2503         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2505 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2507         PR rtl-optimization/57963
2508         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2509         (lra_constraints): Use them.
2511 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2513         * config/arm/types.md (define_attr "type"): Add "load_acq"
2514         and "store_rel".
2515         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2516         changes.
2517         (cortex_a53_store1): Likewise.
2519 2013-08-01  Jan Hubicka  <jh@suse.cz>
2521         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2522         partitions are not needed.
2524 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
2526         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2527         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2528         MAYBE_NON_Q_CLASS_P where appropriate.
2530 2013-08-01  Jan Hubicka  <jh@suse.cz>
2532         * cgraph.h (release_function_body): Declare.
2533         * tree.c (free_lang_data_in_decl): Free, parameters and return values
2534         of unused delcarations.
2536 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2538         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2539         RTL form when subtracting a constant.
2541 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2543         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2544         Generate canonical plus rtx with negated immediate instead of minus
2545         where appropriate.
2546         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2548 2013-08-01  Jan Hubicka  <jh@suse.cz>
2550         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2551         (cgraph_release_function_body): Likewise.
2552         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2553         * cgraph.h (cgrpah_node): Rename abstract_and_needed
2554         to used_as_abstract_origin.
2555         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2556         symbols used as abstract origins.
2557         * cgraphunit.c (analyze_functions): Update.
2558         * ipa.c (symtab_remove_unreachable_nodes): Recompute
2559         used_as_abstract_origin.
2560         * tree-inline.c (tree_function_versioning): Update
2561         used_as_abstract_origin; be ready for DECL_RESULT and
2562         DECL_ARGUMENTS to be NULL.
2564         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2565         for abstract functions.
2566         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2567         real symbols.
2569 2013-08-01  Jan Hubicka  <jh@suse.cz>
2571         * profile.c (compute_value_histograms): Fix thinko.
2573 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2575         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
2576         aarch-common-protos.h to extra_headers.
2577         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2578         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2579         * config/aarch64/t-aarch64 (aarch-common.o): Define.
2581 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2583         * config/aarch64/aarch64.md (define_attr "type"): Delete.
2584         Include "../arm/types.md".  Define "type" attribute for all patterns.
2585         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2586         attribute changes.
2588 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
2590         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2591         to support power8 load fusion.
2592         (fusion_gpr_mem_load): Likewise.
2594         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2596         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2597         declarations for power8 load fusion.
2598         (emit_fusion_gpr_load): Likewise.
2600         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2601         tuning for power8, turn on fusion mode by default.  Turn on sign
2602         extending fusion mode if normal fusion mode is on, and we are at
2603         -O2 or -O3.
2604         (fusion_gpr_load_p): New function, return true if we can fuse an
2605         addis instruction with a dependent load to a GPR.
2606         (emit_fusion_gpr_load): Emit the instructions for power8 load
2607         fusion to GPRs.
2609         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2610         (VSX load fusion peepholes): New peepholes to fuse together an
2611         addi instruction with a VSX load instruction.
2613         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2614         peepholes to fuse an addis instruction with a load to a GPR base
2615         register.  If we are supporting sign extending fusions, convert
2616         sign extending loads to zero extending loads and add an explicit
2617         sign extension.
2619 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2621         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
2622         aarch-common-protos.h to extra_headers.
2623         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2624         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2625         (arm_early_store_addr_dep): Likewise.
2626         (arm_no_early_alu_shift_dep): Likewise.
2627         (arm_no_early_alu_shift_value_dep): Likewise.
2628         (arm_no_early_mul_dep): Likewise.
2629         (arm_no_early_store_addr_dep): Likewise.
2630         (arm_mac_accumulator_is_mul_result): Likewise.
2631         (arm_mac_accumulator_is_result): Likewise.
2632         * config/arm/aarch-common.c: ... here.  New file.
2633         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2634         here to ...
2635         (arm_early_store_addr_dep): Likewise.
2636         (arm_no_early_alu_shift_dep): Likewise.
2637         (arm_no_early_alu_shift_value_dep): Likewise.
2638         (arm_no_early_mul_dep): Likewise.
2639         (arm_no_early_store_addr_dep): Likewise.
2640         (arm_mac_accumulator_is_mul_result): Likewise.
2641         (arm_mac_accumulator_is_result): Likewise.
2642         * config/arm/aarch-common-protos.h: ... here.  New file.
2643         * config/arm/t-arm (aarch-common.o): Define.
2645 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2647         * config/arm/arm.md: Include new file "types.md".
2648         (define_attr "type"): Move from here to ...
2649         (define_attr "mul32"): Likewise.
2650         (define_attr "mul64"): Likewise.
2651         * config/arm/types.md: ... here.  New file.
2653 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2655         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2656         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2658 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2660         * gen-pass-instances.awk: Fix offset of substr().
2662 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
2664         * Makefile.in (pass-instances.def): New.
2665         (passes.o): Replace dependency on passes.def with one on
2666         pass-instances.def
2668         * gen-pass-instances.awk: New.
2670         * passes.c (pass_manager::pass_manager): Use pass-instances.def
2671         rather than passes.def, updating local definition of NEXT_PASS
2672         macro to add an extra NUM parameter (currently unused).
2674 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
2676         * Makefile.in (PASS_MANAGER_H): New.
2677         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2678         (passes.o): Likewise.
2679         (statistics.o): Likewise.
2680         (cgraphunit.o): Likewise.
2681         (context.o): Depend on PASS_MANAGER_H.
2683         * pass_manager.h: New.
2685         * cgraphunit.c (cgraph_add_new_function): Update for moves
2686         of globals to fields of pass_manager.
2687         (analyze_function): Likewise.
2688         (expand_function): Likewise.
2689         (ipa_passes): Likewise.
2690         (compile): Likewise.
2692         * context.c (context::context): New.
2693         * context.h  (context::context): New.
2694         (context::get_passes): New.
2695         (context::passes_): New.
2697         * lto-cgraph.c (input_node): Update for moves of globals to
2698         fields of pass_manager.
2700         * passes.c (all_passes): Remove, in favor of a field of the
2701         same name within the new class pass_manager.
2702         (all_small_ipa_passes): Likewise.
2703         (all_lowering_passes): Likewise.
2704         (all_regular_ipa_passes): Likewise.
2705         (all_late_ipa_passes): Likewise.
2706         (all_lto_gen_passes): Likewise.
2707         (passes_by_id): Likewise.
2708         (passes_by_id_size): Likewise.
2709         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2710         the new class pass_manager.
2711         (set_pass_for_id): Convert to...
2712         (pass_manager::set_pass_for_id): ...method.
2713         (get_pass_for_id): Convert to...
2714         (pass_manager::get_pass_for_id): ...method.
2715         (register_one_dump_file): Move body of implementation into...
2716         (pass_manager::register_one_dump_file): ...here.
2717         (register_dump_files_1): Convert to...
2718         (pass_manager::register_dump_files_1): ...method.
2719         (register_dump_files): Convert to...
2720         (pass_manager::register_dump_files): ...method.
2721         (create_pass_tab): Update for moves of globals to fields of
2722         pass_manager.
2723         (dump_passes): Move body of implementation into...
2724         (pass_manager::dump_passes): ...here.
2725         (register_pass): Move body of implementation into...
2726         (pass_manager::register_pass): ...here.
2727         (init_optimization_passes): Convert into...
2728         (pass_manager::pass_manager): ...constructor for new
2729         pass_manager class, and initialize the pass_lists array.
2730         (check_profile_consistency): Update for moves of globals to
2731         fields of pass_manager.
2732         (dump_profile_report): Move body of implementation into...
2733         (pass_manager::dump_profile_report): ...here.
2734         (ipa_write_summaries_1): Update for moves of pass lists from
2735         being globals to fields of pass_manager.
2736         (ipa_write_optimization_summaries): Likewise.
2737         (ipa_read_summaries):  Likewise.
2738         (ipa_read_optimization_summaries): Likewise.
2739         (execute_all_ipa_stmt_fixups): Likewise.
2741         * statistics.c (statistics_fini): Update for moves of globals to
2742         fields of pass_manager.
2744         * toplev.c (general_init): Replace call to
2745         init_optimization_passes with construction of the pass_manager
2746         instance.
2748         * tree-pass.h (all_passes): Remove, in favor of a field of the
2749         same name within the new class pass_manager.
2750         (all_small_ipa_passes): Likewise.
2751         (all_lowering_passes): Likewise.
2752         (all_regular_ipa_passes): Likewise.
2753         (all_lto_gen_passes): Likewise.
2754         (all_late_ipa_passes): Likewise.
2755         (passes_by_id): Likewise.
2756         (passes_by_id_size): Likewise.
2757         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2758         the new class pass_manager.
2759         (get_pass_for_id): Remove.
2761 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2763         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2764         configs.
2766 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2768         * arm.md (mulhi3): New expand pattern.
2770 2013-07-30  Jan Hubicka  <jh@suse.cz>
2771             Martin Liska  <marxin.liska@gmail.com>
2773         * profile.c (compute_value_histograms): Do not ICE when
2774         there is mismatch only on some counters.
2776 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2778         PR rtl-optimization/57637
2779         * function.c (move_insn_for_shrink_wrap): Also check the
2780         GEN set of the LIVE problem for the liveness analysis
2781         if it exists, otherwise give up.
2783 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2785         PR tree-optimization/57993
2786         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2787         replaced statement in the candidate table.
2788         (phi_add_costs): Return infinite cost when the hidden basis does
2789         not dominate all phis on which the candidate is dependent.
2790         (replace_one_candidate): Record replaced statement in the
2791         candidate table.
2793 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
2795         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2796         (ashlv2si3): New expander.
2797         (*ashlv2si3_i): New define_insn_and_split.
2798         * predicates.md (float_operation): Allow patterns with three
2799         basic sub-patterns.
2801         PR rtl-optimization/58021
2802         * mode-switching.c (create_pre_exit): Always split off preceding
2803         insns if we are not at the basic block head.
2805 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
2807         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2808         (UCLIBC_DYNAMIC_LINKER): New macro.
2809         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2810         `-mnan=2008'.
2811         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2812         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
2813         `-mnan=2008'.
2814         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2815         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2816         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2817         for SF and DF modes.  Use ieee_quad_format for TF mode.
2818         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2819         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2820         (mips_option_override): Handle `-mnan=legacy'.
2821         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2822         `-mabs=2008' and `-mnan=2008'.
2823         (OPTION_DEFAULT_SPECS): Add "nan" default.
2824         (ASM_SPEC): Handle `-mnan='.
2825         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2826         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2827         comment accordingly.
2828         (neg<mode>2): Likewise.
2829         * config/mips/mips.opt (mabs, mnan): New options.
2830         * doc/install.texi (Configuration): Document `--with-nan=' option.
2831         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2832         `-mnan=' options.
2833         (MIPS Options): Document them.
2834         * config.gcc <mips*-*-*>: Handle `--with-nan='.
2835         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2836         * configure: Regenerate.
2837         * config.in: Regenerate.
2839 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2841         * config/i386/i386.md (float post-reload splitters): Do not check
2842         for subregs of SSE registers.
2844 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2845             H.J. Lu  <hongjiu.lu@intel.com>
2847         PR target/57954
2848         PR target/57988
2849         * config/i386/i386.md (post-reload splitter
2850         to avoid partial SSE reg dependency stalls): New pattern.
2852 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2854         * config/s390/s390.md ("movcc"): Swap load and store instructions.
2856 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
2858         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2859         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2861 2013-07-26  Cary Coutant  <ccoutant@google.com>
2863         * dwarf2out.c (die_checksum_ordered): Don't include template
2864         instantiations in signature.
2865         (is_template_parameter): New function.
2866         (is_template_instantiation): New function.
2867         (generate_skeleton_bottom_up): Don't include template instantiations
2868         in type unit DIE.
2869         (generate_skeleton): Likewise.
2870         (break_out_comdat_types): Move recursive call to break out nested
2871         types earlier.
2872         (prune_unused_types_mark_generic_parms_dies): Call
2873         is_template_parameter.
2875 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
2877         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2878         uses vector registers.
2879         * config/aarch64/iterators.md: Add attributes rtn and vas.
2881 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2882             Richard Earnshaw  <richard.earnshaw@arm.com>
2884         * combine.c (simplify_comparison): Re-canonicalize operands
2885         where appropriate.
2886         * config/arm/arm.md (movcond_addsi): New splitter.
2888 2013-07-25  Sterling Augustine  <saugustine@google.com>
2890         * dwarf2out.c (size_of_pubnames): Move code to...
2891         (include_pubname_in_output): ...here.  New.
2892         (want_pubnames): Rearrange.
2893         (output_pubnames): Call include_pubname_in_output.  Move assertion.
2895 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2897         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2899 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2901         PR target/38836
2902         * doc/extend.texi: Remove obsolete builtins. Fix
2903         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2905 2013-07-25  Jan Hubicka  <jh@suse.cz>
2907         * cgraph.c (release_function_body): Break out from ...
2908         (cgraph_release_function_body): ... this one; also release DECL_RESULT
2909         and DECL_ARGUMENTS.
2910         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2911         old_tree in the map.
2912         (create_specialized_node): Update.
2913         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2914         into index.
2915         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2916         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2917         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2918         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2919         DECL_RESULT.
2921 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2923         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2924         addsi3_carryin_alt2_<optab>): Correct output template.
2926 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2928         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2929         Adjust for arm_restrict_it.
2930         Remove trailing whitespace.
2932 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
2934         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2935         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2937         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2939 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
2941         PR rtl-optimization/57960
2942         * lra-constraints.c (process_alt_operands): Use the right mode
2943         when checking strict_low.
2945 2013-07-25  Jan Hubicka  <jh@suse.cz>
2947         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2948         * cgraph.c (cgraph_remove_node): Do not release function body
2949         when in cgraph streaming.
2950         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2951         in other partitions are not considered reachable; fix handling of
2952         clones.
2954 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2956         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2958 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2960         PR target/19599
2961         PR target/57731
2962         PR target/57837
2963         * config/arm/arm.md ("*sibcall_insn): Replace use of
2964         Ss with US. Adjust output for v5 and v4t.
2965         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2967         * config/arm/constraints.md ("Ss"): Rename to US.
2969 2013-07-25  Terry Guo  <terry.guo@arm.com>
2971         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2972         shift_add/shift_sub0/shift_sub1 RTXs.
2974 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
2975             Anton Blanchard <anton@au1.ibm.com>
2977         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2978         (altivec_vpks<VI_char>ss): Likewise.
2979         (altivec_vpks<VI_char>us): Likewise.
2980         (altivec_vpku<VI_char>us): Likewise.
2981         (altivec_vpku<VI_char>um): Likewise.
2983 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
2985         Introduce context class.
2987         * Makefile.in (CONTEXT_H): New.
2988         (OBJS): Add context.o.
2989         (toplev.o): Add CONTEXT_H to dependencies.
2990         (context.o): New.
2992         * toplev.c (general_init): Create the singleton gcc::context instance.
2994         * context.c: New.
2996         * context.h: New.
2998 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3000         PR rtl-optimization/57968
3001         * mode-switching.c (create_pre_exit): Allow instructions that
3002         don't set a return register to need a non-exit mode.
3004 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3005             Anton Blanchard <anton@au1.ibm.com>
3007         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3008         operands to vperm for little endian.
3009         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3010         of lvsl to create the control mask for a vperm for little endian.
3012 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3013             Anton Blanchard <anton@au1.ibm.com>
3015         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3016         two operands for little-endian.
3018 2013-07-23  Steve Ellcey  <sellcey@mips.com>
3020         * config/mips/mips.c (mips_case_values_threshold): New.
3021         (TARGET_CASE_VALUES_THRESHOLD): Define.
3023 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3024             Anton Blanchard <anton@au1.ibm.com>
3026         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3027         selection of field for vector splat in little endian mode.
3029 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
3031         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3032         expanders to rs6000.md.
3033         (ior<mode>3): Likewise.
3034         (and<mode>3): Likewise.
3035         (one_cmpl<mode>2): Likewise.
3036         (nor<mode>3): Likewise.
3037         (andc<mode>3): Likewise.
3038         (eqv<mode>3): Likewise.
3039         (nand<mode>3): Likewise.
3040         (orc<mode>3): Likewise.
3042         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3043         declaration.
3045         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3046         to split multi-word logical operations.
3047         (rs6000_split_logical_di): Likewise.
3048         (rs6000_split_logical): Likewise.
3050         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3051         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3052         and allow TImode operations in 32-bit.
3053         (vsx_and<mode>3_64bit): Likewise.
3054         (vsx_ior<mode>3_32bit): Likewise.
3055         (vsx_ior<mode>3_64bit): Likewise.
3056         (vsx_xor<mode>3_32bit): Likewise.
3057         (vsx_xor<mode>3_64bit): Likewise.
3058         (vsx_one_cmpl<mode>2_32bit): Likewise.
3059         (vsx_one_cmpl<mode>2_64bit): Likewise.
3060         (vsx_nor<mode>3_32bit): Likewise.
3061         (vsx_nor<mode>3_64bit): Likewise.
3062         (vsx_andc<mode>3_32bit): Likewise.
3063         (vsx_andc<mode>3_64bit): Likewise.
3064         (vsx_eqv<mode>3_32bit): Likewise.
3065         (vsx_eqv<mode>3_64bit): Likewise.
3066         (vsx_nand<mode>3_32bit): Likewise.
3067         (vsx_nand<mode>3_64bit): Likewise.
3068         (vsx_orc<mode>3_32bit): Likewise.
3069         (vsx_orc<mode>3_64bit): Likewise.
3071         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3072         logical types in GPRs.
3074         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3075         logical insns to rs6000.md, and allow TImode operations in
3076         32-bit.
3077         (altivec_ior<mode>3): Likewise.
3078         (altivec_xor<mode>3): Likewise.
3079         (altivec_one_cmpl<mode>2): Likewise.
3080         (altivec_nor<mode>3): Likewise.
3081         (altivec_andc<mode>3): Likewise.
3083         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3084         attributes for moving the 128-bit logical operations into
3085         rs6000.md.
3086         (BOOL_REGS_OUTPUT): Likewise.
3087         (BOOL_REGS_OP1): Likewise.
3088         (BOOL_REGS_OP2): Likewise.
3089         (BOOL_REGS_UNARY): Likewise.
3090         (BOOL_REGS_AND_CR0): Likewise.
3091         (one_cmpl<mode>2): Add support for DI logical operations on
3092         32-bit, splitting the operations to 32-bit.
3093         (anddi3): Likewise.
3094         (iordi3): Likewise.
3095         (xordi3): Likewise.
3096         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3097         changes to combine the 32/64-bit code, allow logical operations on
3098         TI mode in 32-bit, and to use similar match_operator patterns like
3099         scalar mode uses.  Combine the Altivec and VSX code for logical
3100         operations, and move it here.
3101         (ior<mode>3, 128-bit types): Likewise.
3102         (xor<mode>3, 128-bit types): Likewise.
3103         (one_cmpl<mode>3, 128-bit types): Likewise.
3104         (nor<mode>3, 128-bit types): Likewise.
3105         (andc<mode>3, 128-bit types): Likewise.
3106         (eqv<mode>3, 128-bit types): Likewise.
3107         (nand<mode>3, 128-bit types): Likewise.
3108         (orc<mode>3, 128-bit types): Likewise.
3109         (and<mode>3_internal): Likewise.
3110         (bool<mode>3_internal): Likewise.
3111         (boolc<mode>3_internal1): Likewise.
3112         (boolc<mode>3_internal2): Likewise.
3113         (boolcc<mode>3_internal1): Likewise.
3114         (boolcc<mode>3_internal2): Likewise.
3115         (eqv<mode>3_internal1): Likewise.
3116         (eqv<mode>3_internal2): Likewise.
3117         (one_cmpl1<mode>3_internal): Likewise.
3119 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3121         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3122         Rename flag_stack_usage to flag_stack_usage_info.
3124 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3126         * config/microblaze/sync.md: New file.
3127         * config/microblaze/microblaze.md: Include sync.md
3128         * config/microblaze/microblaze.c: Add print_operand 'y'.
3129         * config/microblaze/constraints.md: Add memory_contraint
3130         'Q' which is a single register.
3132 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
3134         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3136 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3138         * reload.c (find_reloads): Exit loop once we find this operand
3139         cannot be reloaded somehow for this alternative.
3141         * reload.c (find_reloads): Exit loop once we find a hard register.
3143         * rtlanal.c (computed_jump_p): Exit loop once we find label
3144         reference is used.
3146         * i386.c (ix86_pad_returns): Exit loop after setting replace.
3148         * cfgloopmanip.c (remove_path): Exit loop after setting
3149         irred_invalidated.
3151         * gensupport.c (subst_dup): Avoid loop if code is not
3152         MATCH_DUP nor MATCH_OP_DUP.
3154 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
3156         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3158 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3160         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3161         true for SP_REGNUM if mode == ptr_mode.
3162         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3163         with value R0_REGNUM + 31.
3165 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3167         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3168         pad pointer-typed argument downward.
3170 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3172         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3173         and __ILP32__ when the ILP32 model is in use.
3175 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3177         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3178         (aarch64_load_symref_appropriately): In the case of
3179         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3180         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3181         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3182         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3183         if 'mode' doesn't equal to 'ptr_mode'.
3184         (aarch64_output_mi_thunk): Add an assertion on the alignment of
3185         'vcall_offset'; change to call aarch64_emit_move differently depending
3186         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3187         to calculate the upper bound of 'vcall_offset'.
3188         (aarch64_cannot_force_const_mem): Change to also return true if
3189         mode != ptr_mode.
3190         (aarch64_legitimize_reload_address): In the case of large
3191         displacements, add new local variable 'xmode' and an assertion
3192         based on it; change to use 'xmode' to generate the new rtx and
3193         reload.
3194         (aarch64_asm_trampoline_template): Change to generate the template
3195         differently depending on TARGET_ILP32 or not; change to use
3196         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3197         (aarch64_trampoline_size): Removed.
3198         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3199         and replace immediate literals with it.  Change to use 'ptr_mode'
3200         instead of 'DImode' and call convert_memory_address if the mode
3201         of 'fnaddr' doesn't equal to 'ptr_mode'.
3202         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3203         to output symbol.
3204         (aarch64_elf_asm_destructor): Likewise.
3205         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3206         on TARGET_ILP32 instead of aarch64_trampoline_size.
3207         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3208         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3209         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3210         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
3211         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3212         (storewb_pair<GPI:mode>_<P:mode>): ... this.
3213         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3214         depending on the value of 'mode'.
3215         (add_losym_<mode>): New.
3216         (ldr_got_small_<mode>): New, based on ldr_got_small.
3217         (ldr_got_small): Remove.
3218         (ldr_got_small_sidi): New.
3219         * config/aarch64/iterators.md (P): New.
3220         (PTR): Change to 'ptr_mode' in the condition.
3222 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3224         * config.gcc (aarch64*-*-*): Support --with-abi.
3225         (aarch64*-*-elf): Support --with-multilib-list.
3226         (aarch64*-*-linux*): Likewise.
3227         (supported_defaults): Add abi to aarch64*-*-*.
3228         * configure.ac: Mention AArch64 for --with-multilib-list.
3229         * configure: Re-generated.
3230         * config/aarch64/biarchilp32.h: New file.
3231         * config/aarch64/biarchlp64.h: New file.
3232         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3233         (ABI_SPEC): Ditto.
3234         (MULTILIB_DEFAULTS): Ditto.
3235         (DRIVER_SELF_SPECS): Ditto.
3236         (ASM_SPEC): Update to also substitute -mabi.
3237         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3238         file whose name depends on -mabi= and -mbig-endian.
3239         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3240         TARGET_ILP32.
3241         (POINTER_SIZE): New define.
3242         (POINTERS_EXTEND_UNSIGNED): Ditto.
3243         (enum aarch64_abi_type): New enumeration tag.
3244         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3245         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3246         (TARGET_ILP32): New define.
3247         * config/aarch64/aarch64.opt (mabi): New.
3248         (aarch64_abi): New.
3249         (ilp32, lp64): New values for -mabi.
3250         * config/aarch64/t-aarch64 (comma): New define.
3251         (MULTILIB_OPTIONS): Ditto.
3252         (MULTILIB_DIRNAMES): Ditto.
3253         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3254         * doc/invoke.texi: Document -mabi for AArch64.
3256 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
3258         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3260 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3261             Anton Blanchard <anton@au1.ibm.com>
3263         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3264         endianness when selecting field to splat.
3266 2013-07-22  Eric Christopher  <echristo@gmail.com>
3268         * dwarf2out.c (die_odr_checksum): New function to use
3269         CHECKSUM_ macros and ULEB128 for DIE tag.
3270         (generate_type_signature): Use.
3272 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3274         * config.gcc (sparc*-*-*): Accept leon3 processor.
3275         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3276         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3277         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3278         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3279         (mfix-ut699): Adjust comment.
3280         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3281         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3282         (CPP_CPU_SPEC): Likewise.
3283         (ASM_CPU_SPEC): Likewise.
3284         * config/sparc/sparc.c (leon3_cost): New constant.
3285         (sparc_option_override): Add leon3 support.
3286         (mem_ref): New function.
3287         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3288         (sparc_do_work_around_errata): Look into the instruction in the delay
3289         slot and adjust accordingly.  Add fix for the data cache nullify issues
3290         of the UT699.  Change insertion position for the NOP.
3291         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3292         (leon3_load): New reservation.
3293         (leon_store): Bump latency to 2.
3294         (grfpu): New automaton.
3295         (grfpu_alu): New unit.
3296         (grfpu_ds): Likewise.
3297         (leon_fp_alu): Adjust.
3298         (leon_fp_mult): Delete.
3299         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3300         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3301         * config/sparc/sparc.md (cpu): Add leon3.
3302         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3303         (swapsi): Likewise.
3304         (atomic_test_and_set): Likewise.
3305         (ldstub): Likewise.
3307 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3309         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3310         default for R5900 targets.
3311         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3312         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3313         * config/mips/mips.c (mips_option_override): Report an error for
3314         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3315         for -march=r5900 -mhard-float.
3317 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3319         * df-problems.c (can_move_insns_across): Exit loop once we
3320         find a non-fixed, non-global register.
3322         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3323         setting can_throw.
3325         * omega.c (omega_eliminate_red): Break after setting red_found.
3326         (omega_problem_has_red_equations): Similarly after setting found.
3327         (omega_query_variable): Similarly after setting coupled.
3329 2013-07-22  Marek Polacek  <polacek@redhat.com>
3331         * gimplify.c: Don't include gimple.h twice.
3333 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3335         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3336         instead of TARGET_THUMB1.
3337         (Pz): New constraint.
3338         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3339         encodings.
3340         (compare_negsi_si): Likewise.
3341         (compare_addsi2_op0): Likewise.
3342         (compare_addsi2_op1): Likewise.
3343         (addsi3_carryin_<optab>): Likewise.
3344         (addsi3_carryin_alt2_<optab>): Likewise.
3345         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3346         for arm_restrict_it.
3347         (subsi3_carryin): Likewise.
3348         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3349         (minmax_arithsi): Disable for arm_restrict_it.
3350         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3351         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3352         (satsi_<SAT:code>_shift): Likewise.
3353         (arm_shiftsi3): Add alternative for 16-bit encoding.
3354         (arm32_movhf): Disable for arm_restrict_it.
3355         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3356         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3358 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3360         * config/arm/arm.md (attribute "insn"): Delete.
3361         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3362         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3363         (not_shiftsi): Update for attribute change.
3364         (not_shiftsi_compare0): Likewise.
3365         (not_shiftsi_compare0_scratch): Likewise.
3366         (arm_one_cmplsi2): Likewise.
3367         (thumb1_one_cmplsi2): Likewise.
3368         (notsi_compare0): Likewise.
3369         (notsi_compare0_scratch): Likewise.
3370         (thumb1_movdi_insn): Likewise.
3371         (arm_movsi_insn): Likewise.
3372         (movhi_insn_arch4): Likewise.
3373         (movhi_bytes): Likewise.
3374         (arm_movqi_insn): Likewise.
3375         (thumb1_movqi_insn): Likewise.
3376         (arm32_movhf): Likewise.
3377         (thumb1_movhf): Likewise.
3378         (arm_movsf_soft_insn): Likewise.
3379         (thumb1_movsf_insn): Likewise.
3380         (thumb_movdf_insn): Likewise.
3381         (movsicc_insn): Likewise.
3382         (movsfcc_soft_insn): Likewise.
3383         (and_scc): Likewise.
3384         (cond_move): Likewise.
3385         (if_move_not): Likewise.
3386         (if_not_move): Likewise.
3387         (if_shift_move): Likewise.
3388         (if_move_shift): Likewise.
3389         (if_shift_shift): Likewise.
3390         (if_not_arith): Likewise.
3391         (if_arith_not): Likewise.
3392         (cond_move_not): Likewise.
3393         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3394         (neon_mov<mode>): Likewise.
3395         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3396         (thumb2_movsi_vfp): Likewise.
3397         (movsf_vfp): Likewise.
3398         (thumb2_movsf_vfp): Likewise.
3399         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3400         change.
3401         (cortexa7_older_only): Likewise.
3402         (cortexa7_younger): Likewise.
3403         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3404         (1020alu_shift_op): Likewise.
3405         (1020alu_shift_reg_op): Likewise.
3406         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3407         (alu_shift_op): Likewise.
3408         (alu_shift_reg_op): Likewise.
3409         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3410         (11_alu_shift_op): Likewise.
3411         (11_alu_shift_reg_op): Likewise.
3412         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3413         (9_alu_shift_reg_op): Likewise.
3414         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3415         change.
3416         (cortex_a15_alu_shift): Likewise.
3417         (cortex_a15_alu_shift_reg): Likewise.
3418         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3419         (cortex_a5_alu_shift): Likewise.
3420         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3421         change.
3422         (cortex_a53_alu_shift): Likewise.
3423         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3424         change.
3425         (cortex_a7_alu_reg): Likewise.
3426         (cortex_a7_alu_shift): Likewise.
3427         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3428         (cortex_a8_alu_shift): Likewise.
3429         (cortex_a8_alu_shift_reg): Likewise.
3430         (cortex_a8_mov): Likewise.
3431         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3432         (cortex_a9_dp_shift): Likewise.
3433         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3434         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3435         (cortex_r4_mov): Likewise.
3436         (cortex_r4_alu_shift): Likewise.
3437         (cortex_r4_alu_shift_reg): Likewise.
3438         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3439         (526_alu_shift_op): Likewise.
3440         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3441         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3442         (626te_alu_shift_op): Likewise.
3443         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3444         (726te_alu_op): Likewise.
3445         (726te_alu_shift_op): Likewise.
3446         (726te_alu_shift_reg_op): Likewise.
3447         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3448         (mp626_alu_shift_op): Likewise.
3449         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3450         (pj4_alu_e1_conds): Likewise.
3451         (pj4_alu): Likewise.
3452         (pj4_alu_conds): Likewise.
3453         (pj4_shift): Likewise.
3454         (pj4_shift_conds): Likewise.
3455         (pj4_alu_shift): Likewise.
3456         (pj4_alu_shift_conds): Likewise.
3458 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3460         * config/arm/predicates.md (shiftable_operator_strict_it):
3461         New predicate.
3462         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3463         Disable cond_exec version for arm_restrict_it.
3464         (thumb2_smaxsi3): Convert to generate cond_exec.
3465         (thumb2_sminsi3): Likewise.
3466         (thumb32_umaxsi3): Likewise.
3467         (thumb2_uminsi3): Likewise.
3468         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3469         (thumb2_neg_abssi2): Likewise.
3470         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3471         (thumb2_movsicc_insn): Adjust alternatives.
3472         (thumb2_mov_negscc): Disable for arm_restrict_it.
3473         (thumb2_mov_negscc_strict_it): New pattern.
3474         (thumb2_mov_notscc_strict_it): New pattern.
3475         (thumb2_mov_notscc): Disable for arm_restrict_it.
3476         (thumb2_ior_scc): Likewise.
3477         (thumb2_ior_scc_strict_it): New pattern.
3478         (thumb2_cond_move): Adjust for arm_restrict_it.
3479         (thumb2_cond_arith): Disable for arm_restrict_it.
3480         (thumb2_cond_arith_strict_it): New pattern.
3481         (thumb2_cond_sub): Adjust for arm_restrict_it.
3482         (thumb2_movcond): Likewise.
3483         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3484         (thumb2_zero_extendhisi2_v6): Likewise.
3485         (thumb2_zero_extendqisi2_v6): Likewise.
3486         (orsi_notsi_si): Likewise.
3487         (orsi_not_shiftsi_si): Likewise.
3489 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3491         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3492         instruction sequence is 1 byte shorter.
3494 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3496         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3497         it is not needed after split.
3499 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3501         PR target/51784
3502         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3503         second label for nonlocal goto receivers. Don't output pic base labels
3504         unless we're producing PIC; mark that action unreachable().
3505         (ix86_save_reg): If the function contains a nonlocal label, save the
3506         PIC base reg.
3507         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3508         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3509         (update_pic_label_number_if_needed): New.
3510         (machopic_output_function_base_name): Adjust for nonlocal receiver
3511         case.
3512         (machopic_should_output_picbase_label): New.
3513         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3514         (nonlocal_goto_receiver): New insn and split.
3516 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3518         * config/aarch64/aarch64-builtins.c
3519         (aarch64_fold_builtin): Fold abs in all modes.
3520         * config/aarch64/aarch64-simd-builtins.def
3521         (abs): Enable for all modes.
3522         * config/aarch64/arm_neon.h
3523         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3524         (vabs_f64): Add missing intrinsic.
3526 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
3528         * config/aarch64/arm_neon.h (vabs_s64): New function
3530 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
3532         PR target/57516
3533         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3534         * config/avr/avr.md (adjust_len): Add `round'.
3535         * config/avr/avr-protos.h (avr_out_round): New prototype.
3536         (avr_out_plus): Add `out_label' argument.
3537         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3538         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3539         Handle the case where `insn' is just a pattern.
3540         (avr_out_bitop): Handle the case where `insn' is just a pattern.
3541         (avr_out_round): New function.
3542         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3544 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
3546         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3547         Add check for flag_stack_usage to handle -fstack-usage support
3549 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
3551         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3552         interaction for new Power8 flags and VSX.
3554 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
3556         PR middle-end/57698
3557         * tree-inline.c (expand_call_inline): Emit errors during
3558         early_inlining only if optimization is not turned on.
3560 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3562         * passes.def: New.
3564         * passes.c (init_optimization_passes): Move the construction of
3565         the pass hierarchy into a new passes.def file.
3567         * Makefile.in (passes.o): Add dependency on passes.def.
3569 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3571         * passes.c (init_optimization_passes): Introduce macros for
3572         constructing the tree of passes (INSERT_PASSES_AFTER,
3573         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3574         TERMINATE_PASS_LIST).
3576 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
3577             Wei Mi  <wmi@google.com>
3579         PR rtl-optimization/57878
3580         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3581         top.
3582         (reload_pseudo_compare_func): Check nregs first for reload
3583         pseudos.
3585 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3587         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3589 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
3591         * read-rtl.c (validate_const_int): Once an invalid character is
3592         seen, quit the loop.
3594         * gengtype.c (write_roots): Similarly once we find the "deletable"
3595         or "if_marked" option.
3597 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3599         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3600         "xtab" and "sat".  Move value "clz" from here to ...
3601         (attriubte "type"): ... here.
3602         (satsi_<SAT:code>): Delete "insn" attribute.
3603         (satsi_<SAT:code>_shift): Likewise.
3604         (arm_zero_extendqisi2addsi): Likewise.
3605         (arm_extendqisi2addsi): Likewise.
3606         (clzsi2): Update for attribute changes.
3607         (rbitsi2): Likewise.
3608         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3609         attribute.
3610         (arm_usatsihi): Likewise.
3611         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3613 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3615         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3616         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
3617         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
3618         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
3619         in alphabetical order.
3620         (attribute "core_cycles"): Update for attribute changes.
3621         (arm_addsi3): Likewise.
3622         (addsi3_compare0): Likewise.
3623         (addsi3_compare0_scratch): Likewise.
3624         (addsi3_compare_op1): Likewise.
3625         (addsi3_compare_op2): Likewise.
3626         (compare_addsi2_op0): Likewise.
3627         (compare_addsi2_op1): Likewise.
3628         (addsi3_carryin_shift_<optab>): Likewise.
3629         (subsi3_carryin_shift): Likewise.
3630         (rsbsi3_carryin_shift): Likewise.
3631         (arm_subsi3_insn): Likewise.
3632         (subsi3_compare0): Likewise.
3633         (subsi3_compare): Likewise.
3634         (arm_andsi3_insn): Likewise.
3635         (thumb1_andsi3_insn): Likewise.
3636         (andsi3_compare0): Likewise.
3637         (andsi3_compare0_scratch): Likewise.
3638         (zeroextractsi_compare0_scratch
3639         (andsi_not_shiftsi_si): Likewise.
3640         (iorsi3_insn): Likewise.
3641         (iorsi3_compare0): Likewise.
3642         (iorsi3_compare0_scratch): Likewise.
3643         (arm_xorsi3): Likewise.
3644         (thumb1_xorsi3_insn): Likewise.
3645         (xorsi3_compare0): Likewise.
3646         (xorsi3_compare0_scratch): Likewise.
3647         (satsi_<SAT:code>_shift): Likewise.
3648         (rrx): Likewise.
3649         (arm_shiftsi3): Likewise.
3650         (shiftsi3_compare0): Likewise.
3651         (not_shiftsi): Likewise.
3652         (not_shiftsi_compare0): Likewise.
3653         (not_shiftsi_compare0_scratch): Likewise.
3654         (arm_one_cmplsi2): Likewise.
3655         (thumb_one_complsi2): Likewise.
3656         (notsi_compare0): Likewise.
3657         (notsi_compare0_scratch): Likewise.
3658         (thumb1_zero_extendhisi2): Likewise.
3659         (arm_zero_extendhisi2): Likewise.
3660         (arm_zero_extendhisi2_v6): Likewise.
3661         (arm_zero_extendhisi2addsi): Likewise.
3662         (thumb1_zero_extendqisi2): Likewise.
3663         (thumb1_zero_extendqisi2_v6): Likewise.
3664         (arm_zero_extendqisi2): Likewise.
3665         (arm_zero_extendqisi2_v6): Likewise.
3666         (arm_zero_extendqisi2addsi): Likewise.
3667         (thumb1_extendhisi2): Likewise.
3668         (arm_extendhisi2): Likewise.
3669         (arm_extendhisi2_v6): Likewise.
3670         (arm_extendqisi): Likewise.
3671         (arm_extendqisi_v6): Likewise.
3672         (arm_extendqisi2addsi): Likewise.
3673         (thumb1_extendqisi2): Likewise.
3674         (thumb1_movdi_insn): Likewise.
3675         (arm_movsi_insn): Likewise.
3676         (movsi_compare0): Likewise.
3677         (movhi_insn_arch4): Likewise.
3678         (movhi_bytes): Likewise.
3679         (arm_movqi_insn): Likewise.
3680         (thumb1_movqi_insn): Likewise.
3681         (arm32_movhf): Likewise.
3682         (thumb1_movhf): Likewise.
3683         (arm_movsf_soft_insn): Likewise.
3684         (thumb1_movsf_insn): Likewise.
3685         (movdf_soft_insn): Likewise.
3686         (thumb_movdf_insn): Likewise.
3687         (arm_cmpsi_insn): Likewise.
3688         (cmpsi_shiftsi): Likewise.
3689         (cmpsi_shiftsi_swp): Likewise.
3690         (arm_cmpsi_negshiftsi_si): Likewise.
3691         (movsicc_insn): Likewise.
3692         (movsfcc_soft_insn): Likewise.
3693         (arith_shiftsi): Likewise.
3694         (arith_shiftsi_compare0
3695         (arith_shiftsi_compare0_scratch
3696         (sub_shiftsi): Likewise.
3697         (sub_shiftsi_compare0
3698         (sub_shiftsi_compare0_scratch
3699         (and_scc): Likewise.
3700         (cond_move): Likewise.
3701         (if_plus_move): Likewise.
3702         (if_move_plus): Likewise.
3703         (if_move_not): Likewise.
3704         (if_not_move): Likewise.
3705         (if_shift_move): Likewise.
3706         (if_move_shift): Likewise.
3707         (if_shift_shift): Likewise.
3708         (if_not_arith): Likewise.
3709         (if_arith_not): Likewise.
3710         (cond_move_not): Likewise.
3711         (thumb1_ashlsi3): Set type attribute.
3712         (thumb1_ashrsi3): Likewise.
3713         (thumb1_lshrsi3): Likewise.
3714         (thumb1_rotrsi3): Likewise.
3715         (shiftsi3_compare0_scratch): Likewise.
3716         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3717         (neon_mov<mode>): Likewise.
3718         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3719         attribute changes.
3720         (thumb2_movsi_insn): Likewise.
3721         (thumb2_cmpsi_neg_shiftsi): Likewise.
3722         (thumb2_extendqisi_v6): Likewise.
3723         (thumb2_zero_extendhisi2_v6): Likewise.
3724         (thumb2_zero_extendqisi2_v6): Likewise.
3725         (thumb2_shiftsi3_short): Likewise.
3726         (thumb2_addsi3_compare0_scratch): Likewise.
3727         (orsi_not_shiftsi_si): Likewise.
3728         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3729         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3730         changes.
3731         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3732         (1020alu_shift_op): Likewise.
3733         (1020alu_shift_reg_op): Likewise.
3734         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3735         (alu_shift_op): Likewise.
3736         (alu_shift_reg_op): Likewise.
3737         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3738         (11_alu_shift_op): Likewise.
3739         (11_alu_shift_reg_op): Likewise.
3740         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3741         (9_alu_shift_reg_op): Likewise.
3742         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3743         changes.
3744         (cortex_a15_alu_shift): Likewise.
3745         (cortex_a15_alu_shift_reg): Likewise.
3746         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3747         changes.
3748         (cortex_a5_alu_shift): Likewise.
3749         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3750         changes.
3751         (cortex_a53_alu_shift): Likewise.
3752         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3753         changes.
3754         (cortex_a7_alu_reg): Likewise.
3755         (cortex_a7_alu_shift): Likewise.
3756         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3757         changes.
3758         (cortex_a8_alu_shift): Likewise.
3759         (cortex_a8_alu_shift_reg): Likewise.
3760         (cortex_a8_mov): Likewise.
3761         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3762         (cortex_a9_dp_shift): Likewise.
3763         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3764         changes.
3765         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3766         changes.
3767         (cortex_r4_mov): Likewise.
3768         (cortex_r4_alu_shift): Likewise.
3769         (cortex_r4_alu_shift_reg): Likewise.
3770         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3771         (526_alu_shift_op): Likewise.
3772         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3773         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3774         (626te_alu_shift_op): Likewise.
3775         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3776         (726te_alu_op): Likewise.
3777         (726te_alu_shift_op): Likewise.
3778         (726te_alu_shift_reg_op): Likewise.
3779         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3780         (mp626_alu_shift_op): Likewise.
3781         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3782         (pj4_alu_e1_conds): Likewise.
3783         (pj4_alu): Likewise.
3784         (pj4_alu_conds): Likewise.
3785         (pj4_shift): Likewise.
3786         (pj4_shift_conds): Likewise.
3787         (pj4_alu_shift): Likewise.
3788         (pj4_alu_shift_conds): Likewise.
3789         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3790         changes.
3791         (cortexa7_older_only): Likewise.
3792         (cortexa7_younger): Likewise.
3794 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3796         * ipa-pure-const.c (generate_summary): Rename to...
3797         (pure_const_generate_summary): ... this.
3799 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
3801         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3803 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
3805         PR target/57909
3806         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3807         usage in HI mode.
3809 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3811         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3812         enabled without -march=zEC12.
3813         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3814         flags to be set.
3816 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3818         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3819         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3820         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3821         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3822         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3823         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3824         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3825         ISA_HAS_FP4.
3826         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3827         and ISA_HAS_NMADD3_NMSUB3.
3828         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3829         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3830         (nmsub4<mode>, nmsub3<mode>): Likewise.
3831         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3833 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3835         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3836         TARGET_MIPS5400 checking.
3838 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
3839             Peter Bergner  <bergner@vnet.ibm.com>
3841         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3842         registers in the comment.
3843         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3844         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3845         rather than FIRST_PSEUDO_REGISTERS.
3847 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
3849         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3850         enable extra ISA flags with TARGET_HTM.
3852 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3854         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3855         Fix comment typos.
3857 2013-07-15  Cong Hou  <congh@google.com>
3859         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3860         in compare function for sorting.
3862 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
3864         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3865         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3866         * config/rs6000/rs6000.opt: Add -mhtm option.
3867         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3868         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3869         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3870         __HTM__ if the HTM instructions are available.
3871         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3872         htm_spr_reg_operand): New define_predicates.
3873         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3874         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3875         Include htm.md.
3876         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3877         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3878         HTM builtin functions.
3879         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3880         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3881         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3882         (rs6000_builtin_mask_calculate): Likewise.
3883         (rs6000_option_override_internal): Likewise.
3884         (bdesc_htm): Add new HTM builtin support.
3885         (htm_spr_num): New function.
3886         (htm_spr_regno): Likewise.
3887         (rs6000_htm_spr_icode): Likewise.
3888         (htm_expand_builtin): Likewise.
3889         (htm_init_builtins): Likewise.
3890         (rs6000_expand_builtin): Add support for HTM builtin functions.
3891         (rs6000_init_builtins): Likewise.
3892         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3893         option.
3894         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3895         (TARGET_HTM, MASK_HTM): Define macros.
3896         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3897         (FIXED_REGISTERS): Likewise.
3898         (CALL_USED_REGISTERS): Likewise.
3899         (CALL_REALLY_USED_REGISTERS): Likewise.
3900         (REG_ALLOC_ORDER): Likewise.
3901         (enum reg_class): Likewise.
3902         (REG_CLASS_NAMES): Likewise.
3903         (REG_CLASS_CONTENTS): Likewise.
3904         (REGISTER_NAMES): Likewise.
3905         (ADDITIONAL_REGISTER_NAMES): Likewise.
3906         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3907         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3908         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3909         * config/rs6000/htm.md: New file.
3910         * config/rs6000/htmintrin.h: New file.
3911         * config/rs6000/htmxlintrin.h: New file.
3913 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3915         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3916         Define SYMBOL_TINY_GOT, update comment.
3917         * config/aarch64/aarch64.c
3918         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3919         (aarch64_expand_mov_immediate): Likewise.
3920         (aarch64_print_operand): Likewise.
3921         (aarch64_classify_symbol): Likewise.
3922         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3923         (ldr_got_tiny): Define.
3925 2013-07-13  Tobias Grosser  <tobias@grosser.es>
3927         PR tree-optimization/54094
3928         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3929           scheduling dimension for the parallelism check from the polyhedral
3930           information in the AST.
3931         * graphite-dependences.c (carries_deps): Do not assume the schedule is
3932           in 2D + 1 form.
3934 2013-07-13  Jason Merrill  <jason@redhat.com>
3936         * print-tree.c (debug_vec_tree): Use debug_raw.
3937         (debug_raw (vec<tree, va_gc> &)): New.
3938         (debug_raw (vec<tree, va_gc> *)): New.
3939         * tree.h: Declare them.
3941 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
3943         * ifcvt.c (ifcvt_after_combine): New static variable.
3944         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3945         for size.
3946         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
3947         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3948         rest_of_handle_if_after_reload): Pass new argument for if_convert.
3950 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
3952         * config/mips/mips.c (mips_expand_call): Remove empty statement.
3954 2013-07-12  Michael Matz  <matz@suse.de>
3956         PR middle-end/55771
3957         * convert.c (convert_to_real): Reject non-float inner types.
3959 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
3961         * config/aarch64/aarch64-protos.h
3962         (aarch64_simd_immediate_valid_for_move): Remove.
3963         * config/aarch64/aarch64.c (simd_immediate_info): New member.
3964         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3965         cases.
3966         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3968 2013-07-11  Steve Ellcey  <sellcey@mips.com>
3970         * config/mips/mips.c (mips_conditional_register_usage): Do not
3971         use t[0-7] registers in MIPS16 mode when optimizing for size.
3973 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
3975         * config/i386/i386.c (dispatch_function_versions): Fix array
3976         indexing of function_version_info to match actual_versions.
3978 2013-07-11  Teresa Johnson  <tejohnson@google.com>
3980         * vec.h (struct va_gc): Move release out-of-line.
3981         (va_gc::release): Call ggc_free on released vec.
3983 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3985         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3986         Require GOT register as additional operand in UNSPEC.
3987         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3988         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3989         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3990         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3991         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3992         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3993         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3995 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
3997         PR target/57631
3998         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3999         name seen by assembler/linker rather if available.
4001 2013-07-11  Andreas Schwab  <schwab@suse.de>
4003         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4005 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4007         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4009 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4011         * doc/tm.texi.in: Move hook documentation to ....
4012         * target.def: ... here.
4014         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4015         text on @hook line.
4016         * doc/tm.texi: Regenerate.
4018 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
4020         PR c++/57869
4021         * doc/invoke.texi: Document -Wconditionally-supported.
4023 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4025         PR target/57844
4026         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4027         of my_fp.
4029 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4031         PR target/57506
4032         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4033         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4034         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4035         Remove duplicate devices.
4036         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4037         * config/avr/t-multilib: Regenerate.
4038         * config/avr/avr-tables.opt: Regenerate.
4039         * doc/avr-mmcu.texi: Regenerate.
4041 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4043         PR target/56987
4044         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4046 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
4048         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4049         the cost of MULT when optimizing for size.
4051 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4053         * config/cr16/cr16-protos.h: Don't include target.h.
4055 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
4057         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4058         adjust register size for TDmode and TFmode for VSX registers.
4060 2013-07-08  Kai Tietz  <ktietz@redhat.com>
4062         PR target/56892
4063         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4064         hook_bool_const_tree_true.
4066 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4068         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4069         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4070         * config/s390/s390.md: Define FPR*_REGNUM constants.
4071         Fix FPR2_REGNUM constant (18 -> 17).
4072         ("*trunc<BFP:mode><DFP_ALL:mode>2")
4073         ("*trunc<DFP_ALL:mode><BFP:mode>2")
4074         ("trunc<BFP:mode><DFP_ALL:mode>2")
4075         ("trunc<DFP_ALL:mode><BFP:mode>2")
4076         ("*extend<BFP:mode><DFP_ALL:mode>2")
4077         ("*extend<DFP_ALL:mode><BFP:mode>2")
4078         ("extend<BFP:mode><DFP_ALL:mode>2")
4079         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4080         FPR4_REGNUM.
4082 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
4084         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4086 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4088         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4089         and cfun_fpr_bit_p to cfun_fpr_save_p.
4090         (s390_frame_area, s390_register_info, s390_frame_info)
4091         (s390_emit_prologue, s390_emit_epilogue)
4092         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4093         register numbers.
4094         * config/s390/s390.h: Define *_REGNUM macros for floating point
4095         register numbers.
4097 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
4099         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4101 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
4103         PR rtl-optimization/57786
4104         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4105         and break out of the loop when it is set to false.
4107 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
4109         PR target/57819
4110         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4111         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4112         (const_int 63)) 0)).
4113         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4114         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4115         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4117         PR rtl-optimization/57829
4118         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4119         mask bits outside of mode are just sign-extension from mode to HWI.
4121 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
4123         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4124         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4125         adjust_address instead of change_address to keep info about alignment.
4126         (emit_strmov): Remove.
4127         (emit_memmov): New function.
4128         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4129         (expand_movmem_epilogue): Likewise and return updated rtx for
4130         destination.
4131         (expand_constant_movmem_prologue): Likewise and return updated rtx for
4132         destination and source.
4133         (decide_alignment): Refactor, handle vector_loop.
4134         (ix86_expand_movmem): Likewise.
4135         (ix86_expand_setmem): Likewise.
4136         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4138 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
4140         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4141         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4143 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4145         * config/i386/sse.md (sse_movlhps): Change alternative 3
4146         of operand 2 to "m".
4148 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4150         PR target/57807
4151         * config/i386/sse.md (iptr): New mode attribute.
4152         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4153         (<sse>_vm<plusminus_insn><mode>3): Ditto.
4154         (<sse>_vmmul<mode>3): Ditto.
4155         (<sse>_vmdiv<mode>3): Ditto.
4156         (sse_vmrcpv4sf2): Ditto.
4157         (<sse>_vmsqrt<mode>2): Ditto.
4158         (sse_vmrsqrtv4sf2): Ditto.
4159         (<sse>_vm<code><mode>3): Ditto.
4160         (avx_vmcmp<mode>3): Ditto.
4161         (<sse>_vmmaskcmp<mode>3): Ditto.
4162         (<sse>_comi): Ditto.
4163         (<sse>_ucomi): Ditto.
4164         (*xop_vmfrcz_<mode>): Ditto.
4165         (*fmai_fmadd_<mode>): Ditto.
4166         (*fmai_fmsub_<mode>): Ditto.
4167         (*fmai_fnmadd_<mode>): Ditto.
4168         (*fmai_fnmsub_<mode>): Ditto.
4169         (*fma4i_vmfmadd_<mode>): Ditto.
4170         (*fma4i_vmfmsub_<mode>): Ditto.
4171         (*fma4i_vmfnmadd_<mode>): Ditto.
4172         (*fma4i_vmfnmsub_<mode>): Ditto.
4173         (*xop_vmfrcz_<mode>): Ditto.
4174         (sse_cvtps2pi): Ditto.
4175         (sse_cvttps2pi): Ditto.
4176         (sse_cvtss2si): Ditto.
4177         (sse_cvtss2si_2): Ditto.
4178         (sse_cvtss2siq_2): Ditto.
4179         (sse_cvttss2si): Ditto.
4180         (sse_cvttss2siq): Ditto.
4181         (sse_cvtsd2si): Ditto.
4182         (sse_cvtsd2si_2): Ditto.
4183         (sse_cvtsd2siq_2): Ditto.
4184         (sse_cvttsd2si): Ditto.
4185         (sse_cvttsd2siq): Ditto.
4186         (sse_cvtsd2ss): Ditto.
4187         (sse_cvtss2sd): Ditto.
4188         (avx2_pbroadcast<mode>): Ditto.
4189         (avx2_pbroadcast<mode>_1): Ditto.
4190         (*avx_vperm_broadcast_v4sf): Ditto.
4192         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4193         (sse_movlhps): Ditto.
4194         (sse_storehps): Ditto.
4195         (sse_loadhps): Ditto.
4196         (sse_storelps): Ditto.
4197         (sse_loadlps): Ditto.
4198         (*vec_concatv4sf): Ditto.
4199         (*vec_interleave_highv2df): Ditto.
4200         (*vec_interleave_lowv2df): Ditto.
4201         (*vec_extractv2df_1_sse): Ditto.
4202         (*vec_extractv2df_0_sse): Ditto.
4203         (sse2_storelpd): Ditto.
4204         (sse2_loadlpd): Ditto.
4205         (sse2_movsd): Ditto.
4206         (*vec_concatv4si): Ditto.
4207         (vec_concatv2di): Ditto.
4209         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4210         for Intel asm dialect.
4211         (mmx_punpcklwd): Ditto.
4212         (mmx_punpckldq): Ditto.
4214         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4215         for intel assembler dialect.
4217 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
4219         PR target/29776
4220         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4221         for BUILT_IN_C{LZ,LRSB}*.
4222         * tree.h (CASE_INT_FN): Add FN##IMAX case.
4223         * tree-vrp.c (extract_range_basic): Handle
4224         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
4225         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4226         fall thru to code calling set_value*.
4227         * builtins.c (expand_builtin): Remove *IMAX cases.
4228         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4229         if width is bigger than 2*HWI.
4231 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
4233         PR rtl-optimization/55342
4234         * lra-int.h (lra_subreg_reload_pseudos): New.
4235         * lra.c: Add undoing optional reloads to the block diagram.
4236         (lra_subreg_reload_pseudos): New.
4237         (lra_optional_reload_pseudos): Change comments.
4238         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
4239         lra_optional_reload_pseudos after undo transformations.
4240         * lra-assigns.c (pseudo_prefix_title): New.
4241         (lra_setup_reg_renumber): Use it.
4242         (spill_for): Ditto.  Check subreg reload pseudos too.
4243         (assign_by_spills): Consider subreg reload pseudos too.
4244         * lra-constraints.c (simplify_operand_subreg): Use
4245         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4246         (curr_insn_transform): Recognize and do optional reloads.
4247         (undo_optional_reloads): New.
4248         (lra_undo_inheritance): Call undo_optional_reloads.
4250 2013-07-05  Thomas Quinot  <quinot@adacore.com>
4252         * tree-complex.c (expand_complex_operations_1): Fix typo.
4254 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
4256         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4257         (tune_params): New member 'const vec_costs'.
4258         * config/aarch64/aarch64.c (generic_vector_cost): New.
4259         (generic_tunings): New member 'generic_vector_cost'.
4260         (aarch64_builtin_vectorization_cost): New.
4261         (aarch64_add_stmt_cost): New.
4262         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4263         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4265 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4267         PR target/57777
4268         * config/i386/predicates.md (vsib_address_operand): Disallow
4269         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4271 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4273         PR middle-end/55030
4274         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4275         expand_builtin_setjmp_receiver.
4276         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4277         with NULL for the label parameter.
4278         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4279         the frame-pointer.  Adjust comments.
4280         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4281         only if LABEL is non-NULL.
4283 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4285         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4286         (ARM_ABI_AAPCS64): Ditto.
4287         (arm_abi): Ditto.
4288         (ARM_DEFAULT_ABI): Ditto.
4290 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4292         * config/aarch64/aarch64-builtins.c
4293         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4294         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4295         (st1): Likewise.
4296         * config/aarch64/aarch64-simd.md
4297         (aarch64_ld1<VALL:mode>): New.
4298         (aarch64_st1<VALL:mode>): Likewise.
4299         * config/aarch64/arm_neon.h
4300         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4302 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4304         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4305         target ISA is AVX.
4306         (ix86_option_override_internal):Turn on all -mavx target flags by
4307         default as they are dependent on AVX anyway.
4309 2013-07-02  Cary Coutant  <ccoutant@google.com>
4311         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4312         deterministic hash.
4313         (loc_checksum_ordered): Likewise.
4314         (hash_loc_operands): Remove inline keyword.
4316 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4318         PR tree-optimization/57741
4319         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4320         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4321         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4322         Allow REAL_CST step_exprs if flag_associative_math.
4323         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4325 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4327         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4329 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4331         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4333 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4335         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4336         encoding.
4337         (iorsi3_insn): Likewise.
4338         (arm_xorsi3): Likewise.
4340 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4342         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4343         to ...
4344         (attribute "type"): ... here, and prefix with "wmmx_".
4345         (attribute "core_cycles"): Update for attribute changes.
4346         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4347         (tbcstv4hi): Likewise.
4348         (tbcstv2si): Likewise.
4349         (iwmmxt_iordi3): Likewise.
4350         (iwmmxt_xordi3): Likewise.
4351         (iwmmxt_anddi3): Likewise.
4352         (iwmmxt_nanddi3): Likewise.
4353         (iwmmxt_arm_movdi): Likewise.
4354         (iwmmxt_movsi_insn): Likewise.
4355         (mov<mode>_internal): Likewise.
4356         (and<mode>3_iwmmxt): Likewise.
4357         (ior<mode>3_iwmmxt): Likewise.
4358         (xor<mode>3_iwmmxt): Likewise.
4359         (add<mode>3_iwmmxt): Likewise.
4360         (ssaddv8qi3): Likewise.
4361         (ssaddv4hi3): Likewise.
4362         (ssaddv2si3): Likewise.
4363         (usaddv8qi3): Likewise.
4364         (usaddv4hi3): Likewise.
4365         (usaddv2si3): Likewise.
4366         (sub<mode>3_iwmmxt): Likewise.
4367         (sssubv8qi3): Likewise.
4368         (sssubv4hi3): Likewise.
4369         (sssubv2si3): Likewise.
4370         (ussubv8qi3): Likewise.
4371         (ussubv4hi3): Likewise.
4372         (ussubv2si3): Likewise.
4373         (mulv4hi3_iwmmxt): Likewise.
4374         (smulv4hi3_highpart): Likewise.
4375         (umulv4hi3_highpart): Likewise.
4376         (iwmmxt_wmacs): Likewise.
4377         (iwmmxt_wmacsz): Likewise.
4378         (iwmmxt_wmacu): Likewise.
4379         (iwmmxt_wmacuz): Likewise.
4380         (iwmmxt_clrdi): Likewise.
4381         (iwmmxt_clrv8qi): Likewise.
4382         (iwmmxt_clr4hi): Likewise.
4383         (iwmmxt_clr2si): Likewise.
4384         (iwmmxt_uavgrndv8qi3): Likewise.
4385         (iwmmxt_uavgrndv4hi3): Likewise.
4386         (iwmmxt_uavgv8qi3): Likewise.
4387         (iwmmxt_uavgv4hi3): Likewise.
4388         (iwmmxt_tinsrb): Likewise.
4389         (iwmmxt_tinsrh): Likewise.
4390         (iwmmxt_tinsrw): Likewise.
4391         (iwmmxt_textrmub): Likewise.
4392         (iwmmxt_textrmsb): Likewise.
4393         (iwmmxt_textrmuh): Likewise.
4394         (iwmmxt_textrmsh): Likewise.
4395         (iwmmxt_textrmw): Likewise.
4396         (iwmxxt_wshufh): Likewise.
4397         (eqv8qi3): Likewise.
4398         (eqv4hi3): Likewise.
4399         (eqv2si3): Likewise.
4400         (gtuv8qi3): Likewise.
4401         (gtuv4hi3): Likewise.
4402         (gtuv2si3): Likewise.
4403         (gtv8qi3): Likewise.
4404         (gtv4hi3): Likewise.
4405         (gtv2si3): Likewise.
4406         (smax<mode>3_iwmmxt): Likewise.
4407         (umax<mode>3_iwmmxt): Likewise.
4408         (smin<mode>3_iwmmxt): Likewise.
4409         (umin<mode>3_iwmmxt): Likewise.
4410         (iwmmxt_wpackhss): Likewise.
4411         (iwmmxt_wpackwss): Likewise.
4412         (iwmmxt_wpackdss): Likewise.
4413         (iwmmxt_wpackhus): Likewise.
4414         (iwmmxt_wpackwus): Likewise.
4415         (iwmmxt_wpackdus): Likewise.
4416         (iwmmxt_wunpckihb): Likewise.
4417         (iwmmxt_wunpckihh): Likewise.
4418         (iwmmxt_wunpckihw): Likewise.
4419         (iwmmxt_wunpckilb): Likewise.
4420         (iwmmxt_wunpckilh): Likewise.
4421         (iwmmxt_wunpckilw): Likewise.
4422         (iwmmxt_wunpckehub): Likewise.
4423         (iwmmxt_wunpckehuh): Likewise.
4424         (iwmmxt_wunpckehuw): Likewise.
4425         (iwmmxt_wunpckehsb): Likewise.
4426         (iwmmxt_wunpckehsh): Likewise.
4427         (iwmmxt_wunpckehsw): Likewise.
4428         (iwmmxt_wunpckelub): Likewise.
4429         (iwmmxt_wunpckeluh): Likewise.
4430         (iwmmxt_wunpckeluw): Likewise.
4431         (iwmmxt_wunpckelsb): Likewise.
4432         (iwmmxt_wunpckelsh): Likewise.
4433         (iwmmxt_wunpckelsw): Likewise.
4434         (ror<mode>3): Likewise.
4435         (ashr<mode>3_iwmmxt): Likewise.
4436         (lshr<mode>3_iwmmxt): Likewise.
4437         (ashl<mode>3_iwmmxt): Likewise.
4438         (ror<mode>3_di): Likewise.
4439         (ashr<mode>3_di): Likewise.
4440         (lshr<mode>3_di): Likewise.
4441         (ashl<mode>3_di): Likewise.
4442         (iwmmxt_wmadds): Likewise.
4443         (iwmmxt_wmaddu): Likewise.
4444         (iwmmxt_tmia): Likewise.
4445         (iwmmxt_tmiaph): Likewise.
4446         (iwmmxt_tmiabb): Likewise.
4447         (iwmmxt_tmiatb): Likewise.
4448         (iwmmxt_tmiabt): Likewise.
4449         (iwmmxt_tmiatt): Likewise.
4450         (iwmmxt_tmovmskb): Likewise.
4451         (iwmmxt_tmovmskh): Likewise.
4452         (iwmmxt_tmovmskw): Likewise.
4453         (iwmmxt_waccb): Likewise.
4454         (iwmmxt_wacch): Likewise.
4455         (iwmmxt_waccw): Likewise.
4456         (iwmmxt_waligni): Likewise.
4457         (iwmmxt_walignr): Likewise.
4458         (iwmmxt_walignr0): Likewise.
4459         (iwmmxt_walignr1): Likewise.
4460         (iwmmxt_walignr2): Likewise.
4461         (iwmmxt_walignr3): Likewise.
4462         (iwmmxt_wsadb): Likewise.
4463         (iwmmxt_wsadh): Likewise.
4464         (iwmmxt_wsadbz): Likewise.
4465         (iwmmxt_wsadhz): Likewise.
4466         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4467         (iwmmxt_wabsdiffb): Likewise.
4468         (iwmmxt_wabsdiffh): Likewise.
4469         (iwmmxt_wabsdiffw): Likewise.
4470         (iwmmxt_waddsubhx): Likewise
4471         (iwmmxt_wsubaddhx): Likewise.
4472         (addc<mode>3): Likewise.
4473         (iwmmxt_avg4): Likewise.
4474         (iwmmxt_avg4r): Likewise.
4475         (iwmmxt_wmaddsx): Likewise.
4476         (iwmmxt_wmaddux): Likewise.
4477         (iwmmxt_wmaddsn): Likewise.
4478         (iwmmxt_wmaddun): Likewise.
4479         (iwmmxt_wmulwsm): Likewise.
4480         (iwmmxt_wmulwum): Likewise.
4481         (iwmmxt_wmulsmr): Likewise.
4482         (iwmmxt_wmulumr): Likewise.
4483         (iwmmxt_wmulwsmr): Likewise.
4484         (iwmmxt_wmulwumr): Likewise.
4485         (iwmmxt_wmulwl): Likewise.
4486         (iwmmxt_wqmulm): Likewise.
4487         (iwmmxt_wqmulwm): Likewise.
4488         (iwmmxt_wqmulmr): Likewise.
4489         (iwmmxt_wqmulwmr): Likewise.
4490         (iwmmxt_waddbhusm): Likewise.
4491         (iwmmxt_waddbhusl): Likewise.
4492         (iwmmxt_wqmiabb): Likewise.
4493         (iwmmxt_wqmiabt): Likewise.
4494         (iwmmxt_wqmiatb): Likewise.
4495         (iwmmxt_wqmiatt): Likewise.
4496         (iwmmxt_wqmiabbn): Likewise.
4497         (iwmmxt_wqmiabtn): Likewise.
4498         (iwmmxt_wqmiatbn): Likewise.
4499         (iwmmxt_wqmiattn): Likewise.
4500         (iwmmxt_wmiabb): Likewise.
4501         (iwmmxt_wmiabt): Likewise.
4502         (iwmmxt_wmiatb): Likewise.
4503         (iwmmxt_wmiatt): Likewise.
4504         (iwmmxt_wmiabbn): Likewise.
4505         (iwmmxt_wmiabtn): Likewise.
4506         (iwmmxt_wmiatbn): Likewise.
4507         (iwmmxt_wmiattn): Likewise.
4508         (iwmmxt_wmiawbb): Likewise.
4509         (iwmmxt_wmiawbt): Likewise.
4510         (iwmmxt_wmiawtb): Likewise.
4511         (iwmmxt_wmiawtt): Likewise.
4512         (iwmmxt_wmiawbbn): Likewise.
4513         (iwmmxt_wmiawbtn): Likewise.
4514         (iwmmxt_wmiawtbn): Likewise.
4515         (iwmmxt_wmiawttn): Likewise.
4516         (iwmmxt_wmerge): Likewise.
4517         (iwmmxt_tandc<mode>3): Likewise.
4518         (iwmmxt_torc<mode>3): Likewise.
4519         (iwmmxt_torvsc<mode>3): Likewise.
4520         (iwmmxt_textrc<mode>3): Likewise.
4521         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4522         (wmmxt_pack): Likewise.
4523         (wmmxt_mult_c1): Likewise.
4524         (wmmxt_mult_c2): Likewise.
4525         (wmmxt_alu_c1): Likewise.
4526         (wmmxt_alu_c2): Likewise.
4527         (wmmxt_alu_c3): Likewise.
4528         (wmmxt_transfer_c1): Likewise.
4529         (wmmxt_transfer_c2): Likewise.
4530         (wmmxt_transfer_c3): Likewise.
4531         (marvell_f_iwmmxt_wstr): Likewise.
4532         (marvell_f_iwmmxt_wldr): Likewise.
4534 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
4536         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4538 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4540         Revert:
4541         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4542         * lra-constraints.c (need_for_split_p): Check call used hard regs
4543         living through calls.
4545         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4546         call used regs for call insn.
4548 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
4550         PR target/57736
4551         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4552         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4554 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
4556         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4558 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4560         * lra-constraints.c (need_for_split_p): Check call used hard regs
4561         living through calls.
4563 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
4565         PR target/57744
4566         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4567         to tie with any other modes.  Eliminate Altivec vector mode tests,
4568         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
4569         code, to return 0 if testing MODE2 for a condition, if we've
4570         already tested MODE1 for the same condition.
4572 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4574         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4575         layout.
4577 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4579         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4580         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4582 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4584         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4585         Define.
4586         (aarch64_symbolic_constant_p): Remove.
4587         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4588         static.  Fix line length and white space.
4589         (aarch64_symbolic_constant_p): Remove.
4590         * config/aarch64/predicates.md (aarch64_valid_symref):
4591         Use aarch64_classify_symbol_expression.
4593 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4595         * config/arm/constraints.md (Ts): New constraint.
4596         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4597         16-bit encodings.
4598         (compare_scc): Use "Ts" constraint for operand 0.
4599         (ior_scc_scc): Likewise.
4600         (and_scc_scc): Likewise.
4601         (and_scc_scc_nodom): Likewise.
4602         (ior_scc_scc_cmp): Likewise for operand 7.
4603         (and_scc_scc_cmp): Likewise.
4604         * config/arm/thumb2.md (thumb2_movsi_insn):
4605         Add alternatives for 16-bit encodings.
4606         (thumb2_movhi_insn): Likewise.
4607         (thumb2_movsicc_insn): Likewise.
4608         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
4609         (thumb2_negscc): Use "Ts" constraint.
4610         Move mvn instruction outside cond_exec block.
4611         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4612         for 16-bit encodings.
4614 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4616         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4617         encoding.
4618         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4619         (mulsi3subsi): Likewise.
4620         (mulsidi3adddi): Likewise.
4621         (mulsidi3_v6): Likewise.
4622         (umulsidi3_v6): Likewise.
4623         (umulsidi3adddi_v6): Likewise.
4624         (smulsi3_highpart_v6): Likewise.
4625         (umulsi3_highpart_v6): Likewise.
4626         (mulhisi3tb): Likewise.
4627         (mulhisi3bt): Likewise.
4628         (mulhisi3tt): Likewise.
4629         (maddhisi4): Likewise.
4630         (maddhisi4tb): Likewise.
4631         (maddhisi4tt): Likewise.
4632         (maddhidi4): Likewise.
4633         (maddhidi4tb): Likewise.
4634         (maddhidi4tt): Likewise.
4635         (zeroextractsi_compare0_scratch): Likewise.
4636         (insv_zero): Likewise.
4637         (insv_t2): Likewise.
4638         (anddi_notzesidi_di): Likewise.
4639         (anddi_notsesidi_di): Likewise.
4640         (andsi_notsi_si): Likewise.
4641         (iordi_zesidi_di): Likewise.
4642         (xordi_zesidi_di): Likewise.
4643         (andsi_iorsi3_notsi): Likewise.
4644         (smax_0): Likewise.
4645         (smax_m1): Likewise.
4646         (smin_0): Likewise.
4647         (not_shiftsi): Likewise.
4648         (unaligned_loadsi): Likewise.
4649         (unaligned_loadhis): Likewise.
4650         (unaligned_loadhiu): Likewise.
4651         (unaligned_storesi): Likewise.
4652         (unaligned_storehi): Likewise.
4653         (extv_reg): Likewise.
4654         (extzv_t2): Likewise.
4655         (divsi3): Likewise.
4656         (udivsi3): Likewise.
4657         (arm_zero_extendhisi2addsi): Likewise.
4658         (arm_zero_extendqisi2addsi): Likewise.
4659         (compareqi_eq0): Likewise.
4660         (arm_extendhisi2_v6): Likewise.
4661         (arm_extendqisi2addsi): Likewise.
4662         (arm_movt): Likewise.
4663         (thumb2_ldrd): Likewise.
4664         (thumb2_ldrd_base): Likewise.
4665         (thumb2_ldrd_base_neg): Likewise.
4666         (thumb2_strd): Likewise.
4667         (thumb2_strd_base): Likewise.
4668         (thumb2_strd_base_neg): Likewise.
4669         (arm_negsi2): Add alternative for 16-bit encoding.
4670         (arm_one_cmplsi2): Likewise.
4672 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4674         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4675         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4676         (movdfcc): Likewise.
4677         * config/arm/vfp.md (*thumb2_movsf_vfp):
4678         Disable predication for arm_restrict_it.
4679         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4680         (*thumb2_movdfcc_vfp): Likewise.
4681         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4682         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4683         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4684         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4685         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4686         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4687         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4688         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4689         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4690         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4691         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4692         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4693         Disable predication for arm_restrict_it.
4695 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
4697         * config/i386/bmiintrin.h (_bextr_u32): New.
4698         (_bextr_u64): Ditto.
4700 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
4702         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4703         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4704         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4705         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4706         * config/mips/n32-elf.h: ...this new file.
4708 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
4710         PR target/57224
4711         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4712         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4714 2013-06-27  Catherine Moore <clm@codesourcery.com>
4716         * config/mips/mips-tables.opt: Regenerate.
4717         * config/mips/mips-cpus.def: Add m14ke and m14kec.
4718         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4719         * doc/invoke.texi: Add -m14kc.
4721 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4723         PR target/57623
4724         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4725         constraints of operand 1 and 2.
4727         PR target/57623
4728         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4729         to match RTL canonicalization.  Swap predicates and
4730         constraints of operand 1 and 2.
4732 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
4734         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4735         Process OP_INOUT regs for splitting too.
4737 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4739         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4740         decl before the loop, initialize to NULL.
4741         (vectorizable_load): Initialize ptr_incr to NULL.
4743 2013-06-27  Martin Jambor  <mjambor@suse.cz>
4745         PR lto/57208
4746         * ipa-ref.h (ipa_maybe_record_reference): Declare.
4747         * ipa-ref.c (ipa_maybe_record_reference): New function.
4748         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4749         * ipa-cp.c (create_specialized_node): Record potential references from
4750         aggvals.
4751         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4753 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4755         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4756         parameter 'mode' of type 'enum machine_mode mode'; change to pass
4757         'mode' to force_reg.
4758         (aarch64_add_offset): Update calls to aarch64_force_temporary.
4759         (aarch64_expand_mov_immediate): Likewise.
4761 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4763         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4764         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4766 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4768         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4769         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4770         (struct machine_function): Add tbegin_p.
4771         (s390_canonicalize_comparison): Fold CC mode compares to
4772         conditional jump if possible.
4773         (s390_emit_jump): Return the emitted jump.
4774         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4775         Handle CCRAWmode compares.
4776         (s390_option_override): Default to -mhtm if available.
4777         (s390_reg_clobbered_rtx): Handle floating point regs as well.
4778         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4779         FPRs instead of df_regs_ever_live_p.
4780         (s390_optimize_nonescaping_tx): New function.
4781         (s390_init_frame_layout): Extend clobbered_regs array to cover
4782         FPRs as well.
4783         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4784         (s390_expand_tbegin): New function.
4785         (enum s390_builtin): New enum definition.
4786         (code_for_builtin): New array definition.
4787         (s390_init_builtins): New function.
4788         (s390_expand_builtin): New function.
4789         (TARGET_INIT_BUILTINS): Define.
4790         (TARGET_EXPAND_BUILTIN): Define.
4791         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4792         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4793         (s390_alc_comparison): Likewise.
4794         * config/s390/s390-modes.def: Add CCRAWmode.
4795         * config/s390/s390.h (processor_flags): Add PF_TX.
4796         (TARGET_CPU_HTM): Define macro.
4797         (TARGET_HTM): Define macro.
4798         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4799         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4800         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4801         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4802         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4803         (TBEGIN_MASK, TBEGINC_MASK): New constants.
4804         ("*cc_to_int"): Move up.
4805         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4806         constants other than 0.
4807         ("*ccraw_to_int"): New insn and splitter definition.
4808         ("tbegin", "tbegin_nofloat", "tbegin_retry")
4809         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4810         ("tx_assist"): New expander.
4811         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4812         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4813         * config/s390/s390.opt: Add -mhtm option.
4814         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4815         * config/s390/htmxlintrin.h: New file.
4816         * config/s390/htmintrin.h: New file.
4817         * config/s390/s390intrin.h: New file.
4818         * doc/extend.texi: Document htm builtins.
4819         * config.gcc: Add the new header files to extra_headers.
4821 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
4823         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4824         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4826 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
4827             Pat Haugen <pthaugen@us.ibm.com>
4828             Peter Bergner <bergner@vnet.ibm.com>
4830         * config/rs6000/power8.md: New.
4831         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4832         setting for power8 entry.
4833         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4834         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4835         test for Power4/Power5 only.
4836         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4837         support.
4838         (force_new_group): Adjust comment.
4839         * config/rs6000/rs6000.md: Include power8.md.
4841 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
4843         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4844         * config/arm/arm-protos.h (arm_max_conditional_execute): New
4845         declaration.
4846         (tune_params): Update comment.
4847         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4848         (arm_max_conditional_execute): New function.
4849         (thumb2_final_prescan_insn): Use max_insn_skipped and
4850         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4852 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
4854         PR tree-optimization/57705
4855         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4856         SSA_NAME step, provided that it is not defined inside the loop.
4857         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4858         (get_initial_def_for_induction): Handle SSA_NAME IV step.
4860 2013-06-25  Martin Jambor  <mjambor@suse.cz>
4862         PR middle-end/57670
4863         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4864         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4865         calls in the dump.
4866         (ipa_note_param_call): Initialize member_ptr flag.
4867         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4868         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4869         (ipa_write_indirect_edge_info): Stream member_ptr flag.
4870         (ipa_read_indirect_edge_info): Likewise.
4872 2013-06-25  Richard Biener  <rguenther@suse.de>
4874         PR middle-end/56977
4875         * passes.c (init_optimization_passes): Move pass_fold_builtins
4876         and pass_dce earlier with -Og.
4878 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
4880         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4881         <BIT_FIELD_REF>: Remove trailing TAB.
4882         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4883         remove blank line.
4885 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4887         PR tree-optimization/57358
4888         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4889         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4890         (ipa_analyze_params_uses): Generate pessimistic info when true.
4892 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4894         PR tree-optimization/57539
4895         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4896         global.inlined_to of the new node to it.  All callers changed.
4897         * ipa-inline-transform.c (clone_inlined_nodes): New variable
4898         inlining_into, pass it to cgraph_clone_node.
4899         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4900         ipa_free_edge_args_substructures.
4901         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4902         rdesc linked list.  Do not assert rdesc edges have inlined caller.
4903         Assert we have found an rdesc in the rdesc list.
4905 2013-06-24  Richard Biener  <rguenther@suse.de>
4907         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4908         (pointer_set_lookup): Declare.
4909         (class pointer_map): New template class implementing a
4910         generic pointer to T map.
4911         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4912         pointer_map<T>::contains, pointer_map<T>::insert,
4913         pointer_map<T>::traverse): New functions.
4914         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4915         (pointer_set_lookup): New function.
4916         (pointer_set_contains): Use pointer_set_lookup.
4917         (pointer_set_insert): Likewise.
4918         (insert_aux): Remove.
4919         (struct pointer_map_t): Embed a pointer_set_t.
4920         (pointer_map_create): Adjust.
4921         (pointer_map_destroy): Likewise.
4922         (pointer_map_contains): Likewise.
4923         (pointer_map_insert): Likewise.
4924         (pointer_map_traverse): Likewise.
4925         * tree-streamer.h (struct streamer_tree_cache_d): Use a
4926         pointer_map<unsigned> instead of a pointer_map_t.
4927         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4928         (streamer_tree_cache_lookup): Likewise.
4929         (streamer_tree_cache_create): Likewise.
4930         (streamer_tree_cache_delete): Likewise.
4931         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4932         pointer_map<unsigned> instead of a pointer_map_t.
4933         (lto_init_tree_ref_encoder): Adjust.
4934         (lto_destroy_tree_ref_encoder): Likewise.
4935         * lto-section-out.c (lto_output_decl_index): Likewise.
4936         (lto_record_function_out_decl_state): Likewise.
4937         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4939 2013-06-24  Richard Biener  <rguenther@suse.de>
4941         PR tree-optimization/57488
4942         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4944 2013-06-24  Alan Modra  <amodra@gmail.com>
4946         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4947         (gen_easy_altivec_constant): Likewise.
4948         * config/rs6000/predicates.md (easy_vector_constant_add_self,
4949         easy_vector_constant_msb): Likewise.
4951 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
4953         PR target/57688
4954         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4955         add missing return true.
4957 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
4959         PR target/52483
4960         * config/sh/predicates.md (general_extend_operand): Invoke
4961         general_movsrc_operand for memory operands.
4962         (general_movsrc_operand): Allow reg+reg addressing, do not use
4963         general_operand for memory operands.
4965 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
4967         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
4968         when current target options does not apply.
4969         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4970         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4971         * config/i386/bmiintrin.h: Pass appropriate target
4972         attributes to header.
4973         * config/i386/mmintrin.h: Ditto.
4974         * config/i386/nmmintrin.h: Ditto.
4975         * config/i386/avx2intrin.h: Ditto.
4976         * config/i386/fxsrintrin.h: Ditto.
4977         * config/i386/tbmintrin.h: Ditto.
4978         * config/i386/xsaveintrin.h: Ditto.
4979         * config/i386/f16cintrin.h: Ditto.
4980         * config/i386/xtestintrin.h: Ditto.
4981         * config/i386/xsaveoptintrin.h: Ditto.
4982         * config/i386/bmi2intrin.h: Ditto.
4983         * config/i386/lzcntintrin.h: Ditto.
4984         * config/i386/smmintrin.h: Ditto.
4985         * config/i386/wmmintrin.h: Ditto.
4986         * config/i386/x86intrin.h: Remove all header include guards.
4987         * config/i386/prfchwintrin.h: Ditto.
4988         * config/i386/pmmintrin.h: Ditto.
4989         * config/i386/tmmintrin.h: Ditto.
4990         * config/i386/xmmintrin.h: Ditto.
4991         * config/i386/popcntintrin.h: Ditto.
4992         * config/i386/rdseedintrin.h: Ditto.
4993         * config/i386/ammintrin.h: Ditto.
4994         * config/i386/emmintrin.h: Ditto.
4995         * config/i386/immintrin.h: Remove all header include guards.
4996         * config/i386/fma4intrin.h: Ditto.
4997         * config/i386/lwpintrin.h: Ditto.
4998         * config/i386/xopintrin.h: Ditto.
4999         * config/i386/ia32intrin.h: Ditto.
5000         * config/i386/avxintrin.h: Ditto.
5001         * config/i386/rtmintrin.h: Ditto.
5002         * config/i386/fmaintrin.h: Ditto.
5003         * config/i386/mm3dnow.h: Ditto.
5005 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5007         * common/config/i386/i386-common.c: Handle LZCNT.
5009 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5011         * doc/extend.texi: Use __atomic_store_n instead of
5012         __atomic_store in HLE example.
5014 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
5016         * config/sh/sh.c: Remove <cstdlib> workaround.
5018 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5020         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
5022 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5024         * doc/extend.texi: Document that __atomic_clear and
5025           __atomic_test_and_set should only be used with bool.
5027 2013-06-20  Jan Hubicka  <jh@suse.cz>
5029         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5030         types_same_for_odr.
5031         * tree.c (decls_same_for_odr): New function.
5032         (same_for_edr): New function.
5033         (types_same_for_odr): New function.
5034         (get_binfo_at_offset): Use it.
5035         * tree.h (types_same_for_odr): Declare.
5037 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
5038             Jason Merrill  <jason@redhat.com>
5040         * system.h: Include <cstdlib> as well as <stdlib.h>.
5042 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
5044         PR target/57655
5045         * config/i386/i386.c (construct_container): Report error if
5046         long double is used with disabled x87 float returns.
5048 2013-06-20  Jan Hubicka  <jh@suse.cz>
5050         * lto-cgraph.c (input_symtab): Do not set cgraph state.
5052 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
5054         PR rtl-optimization/57425
5055         PR rtl-optimization/57569
5056         * alias.c (write_dependence_p): Remove parameters mem_mode and
5057         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
5058         Changed all callers.
5059         (canon_anti_dependence): Get comments and semantics in sync.
5060         Add parameter mem_canonicalized.  Changed all callers.
5061         * rtl.h (canon_anti_dependence): Update prototype.
5063 2013-06-20  Richard Biener  <rguenther@suse.de>
5065         * data-streamer-in.c (streamer_read_uhwi): Optimize single
5066         byte case, inline streamer_read_uchar and defer section
5067         overrun check.
5069 2013-06-20  Richard Biener  <rguenther@suse.de>
5071         PR tree-optimization/57584
5072         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5073         SSA names into the expanded expression that take part in
5074         abnormal coalescing.
5076 2013-06-19  Sharad Singhai  <singhai@google.com>
5078         * gcov.c (print_usage): Handle new option.
5079         (process_args): Ditto.
5080         (get_gcov_intermediate_filename): New function.
5081         (output_intermediate_file): New function.
5082         (output_gcov_file): New function
5083         (generate_results): Handle new option.
5084         (release_function): Relase demangled name.
5085         (read_graph_file): Handle demangled name.
5086         (output_lines): Ditto.
5087         * doc/gcov.texi: Document gcov intermediate format.
5089 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
5091         PR bootstrap/57604
5092         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5093         (lra_emit_add): Use the functions.  Add comment about Y as an
5094         address segment.
5096 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
5098         PR driver/57652
5099         * collect2.c (collect_atexit): New.
5100         (collect_exit): Delete.
5101         (main): Register collect_atexit with atexit.
5102         (collect_wait): Change collect_exit to exit.
5103         (do_wait): Same.
5104         * collect2.h (collect_exit): Delete.
5105         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5107 2013-06-19  Wei Mi  <wmi@google.com>
5109         PR rtl-optimization/57518
5110         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5111         if regno is used in paradoxical subreg.
5112         (update_equiv_regs): Check pdx_subregs[regno] before
5113         set a reg to be equivalent with a mem.
5115 2013-06-19  Matthias Klose  <doko@ubuntu.com>
5117         PR driver/57651
5118         * file-find.h (find_a_file): Add a mode parameter.
5119         * file-find.c (find_a_file): Likewise.
5120         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5121         with X_OK for the executables.
5122         * collect2.c (main): Call find_a_file with X_OK.
5124 2013-06-19  Steve Ellcey  <sellcey@mips.com>
5126         PR target/56942
5127         * config/mips/mips.md (casesi_internal_mips16_<mode>):
5128         Use NEXT_INSN instead of next_real_insn.
5130 2013-06-19  Jan Hubicka  <jh@suse.cz>
5132         * cgraph.h (const_value_known_p): Replace by ...
5133         (ctor_for_folding): .. this one.
5134         * cgraphunit.c (process_function_and_variable_attributes): Use it.
5135         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5136         * expr.c (expand_expr_real_1): Likewise.
5137         (string_constant): Likewise.
5138         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5139         * ipa.c (process_references): Likewise.
5140         (symtab_remove_unreachable_nodes): Likewise.
5141         * ipa-inline-analysis.c (param_change_prob): Likewise.
5142         * gimple-fold.c (canonicalize_constructor_val): Likewise.
5143         (get_base_constructor): Likwise.
5144         * varpool.c (varpool_remove_node): Likewise.
5145         (varpool_remove_initializer): LIkewise.
5146         (dump_varpool_node): LIkwise.
5147         (const_value_known_p): Rewrite to ...
5148         (ctor_for_folding): ... this one.
5150 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
5152         PR driver/57651
5153         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5154         PERSONALITY in $PATH derived prefixes.
5156 2013-06-19  Jeff Law  <law@redhat.com>
5158         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5159         in comment.
5161         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5162         (simplify_bitwise_binary): Use it to simpify certain binary ops on
5163         booleans.
5165 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
5167         * config/arm/vfp.md: Move VFP instruction classification documentation
5168         to ...
5169         * config/arm/arm.md: ... here.  Update instruction classification
5170         documentation.
5172 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
5174         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5175         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5176         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
5178 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
5180         PR target/57609
5181         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5182         with NEXT_INSN.  Use tablejump_p to check for jump table data
5183         insns.
5185 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
5187         PR c++/56544
5188         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5189         that now in C++ the value is correct per the C++ standards.
5191 2013-06-19  Richard Biener  <rguenther@suse.de>
5193         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5194         for global context.
5196 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5198         Revert:
5199         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5201         PR target/57609
5202         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5203         with next_active_insn.
5205 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
5207         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5208         functions are inlined during failures to flag an error.
5209         * tree-inline.c (expand_call_inline): Allow the error to be flagged
5210         in early inline pass.
5212 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
5214         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5215         in comments.
5217 2013-06-18  Julian Brown  <julian@codesourcery.com>
5219         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5220         Permit virtual register pre-reload if !strict.
5221         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5222         change.
5223         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5224         prototype.
5225         * config/arm/neon.md (movmisalign<mode>): Use
5226         neon_perm_struct_or_reg_operand instead of
5227         neon_struct_or_register_operand.
5228         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5229         neon_permissive_struct_operand instead of neon_struct_operand.
5230         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5231         neon_vector_mem_operand.
5232         * config/arm/predicates.md (neon_struct_operand): Adjust call to
5233         neon_vector_mem_operand.
5234         (neon_permissive_struct_operand): New.
5235         (neon_struct_or_register_operand): Rename to...
5236         (neon_perm_struct_or_reg_operand): This. Adjust call to
5237         neon_vector_mem_operand.
5239 2013-06-18  Richard Biener  <rguenther@suse.de>
5241         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5242         * lto-streamer.h: Include pointer-set.h.
5243         (struct lto_decl_slot): Remove.
5244         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5245         Remove next_index entry.
5246         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5247         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5248         (lto_init_tree_ref_encoder): Adjust.
5249         (lto_destroy_tree_ref_encoder): Likewise.
5250         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5251         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5252         (lto_output_decl_index): Adjust.
5253         (lto_new_out_decl_state): Likewise.
5254         (lto_record_function_out_decl_state): Likewise.
5255         * lto-streamer-out.c (copy_function): Likewise.
5257 2013-06-18  Richard Biener  <rguenther@suse.de>
5259         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5260         * cgraphunit.c: Include cfgloop.h.
5261         (init_lowered_empty_function): Initialize the loop tree.
5262         (assemble_thunk): Insert new BBs into loops.
5264 2013-06-18  Richard Biener  <rguenther@suse.de>
5266         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5267         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5268         the map from cache entry to cache index optional.
5269         (streamer_tree_cache_replace_tree): Adjust accordingly.
5270         (streamer_tree_cache_append): Likewise.
5271         (streamer_tree_cache_delete): Likewise.
5272         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5273         streamer cache map from cache entry to cache index.
5274         * lto-streamer-out.c (create_output_block): Adjust.
5276 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5278         * config/arm/arm.md (attribute "insn"): Move multiplication and
5279         division attributes to...
5280         (attribute "type"): ... here.  Remove mult.
5281         (attribute "mul32"): New attribute.
5282         (attribute "mul64"): Add umaal.
5283         (*arm_mulsi3): Update attributes.
5284         (*arm_mulsi3_v6): Likewise.
5285         (*thumb_mulsi3): Likewise.
5286         (*thumb_mulsi3_v6): Likewise.
5287         (*mulsi3_compare0): Likewise.
5288         (*mulsi3_compare0_v6): Likewise.
5289         (*mulsi_compare0_scratch): Likewise.
5290         (*mulsi_compare0_scratch_v6): Likewise.
5291         (*mulsi3addsi): Likewise.
5292         (*mulsi3addsi_v6): Likewise.
5293         (*mulsi3addsi_compare0): Likewise.
5294         (*mulsi3addsi_compare0_v6): Likewise.
5295         (*mulsi3addsi_compare0_scratch): Likewise.
5296         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5297         (*mulsi3subsi): Likewise.
5298         (*mulsidi3adddi): Likewise.
5299         (*mulsi3addsi_v6): Likewise.
5300         (*mulsidi3adddi_v6): Likewise.
5301         (*mulsidi3_nov6): Likewise.
5302         (*mulsidi3_v6): Likewise.
5303         (*umulsidi3_nov6): Likewise.
5304         (*umulsidi3_v6): Likewise.
5305         (*umulsidi3adddi): Likewise.
5306         (*umulsidi3adddi_v6): Likewise.
5307         (*smulsi3_highpart_nov6): Likewise.
5308         (*smulsi3_highpart_v6): Likewise.
5309         (*umulsi3_highpart_nov6): Likewise.
5310         (*umulsi3_highpart_v6): Likewise.
5311         (mulhisi3): Likewise.
5312         (*mulhisi3tb): Likewise.
5313         (*mulhisi3bt): Likewise.
5314         (*mulhisi3tt): Likewise.
5315         (maddhisi4): Likewise.
5316         (*maddhisi4tb): Likewise.
5317         (*maddhisi4tt): Likewise.
5318         (maddhidi4): Likewise.
5319         (*maddhidi4tb): Likewise.
5320         (*maddhidi4tt): Likewise.
5321         (divsi3): Likewise.
5322         (udivsi3): Likewise.
5323         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5324         (thumb2_mulsi_short_compare0): Likewise.
5325         (thumb2_mulsi_short_compare0_scratch): Likewise.
5326         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5327         (1020mult2): Likewise.
5328         (1020mult3): Likewise.
5329         (1020mult4): Likewise.
5330         (1020mult5): Likewise.
5331         (1020mult6): Likewise.
5332         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5333         change.
5334         (cortex_a15_mult64): Likewise.
5335         (cortex_a15_sdiv): Likewise.
5336         (cortex_a15_udiv): Likewise.
5337         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5338         (mult2): Likewise.
5339         (mult3): Likewise.
5340         (mult4): Likewise.
5341         (mult5): Likewise.
5342         (mult6): Likewise.
5343         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5344         (pj4_ir_div): Likewise.
5345         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5346         (11_mult2): Likewise.
5347         (11_mult3): Likewise.
5348         (11_mult4): Likewise.
5349         (11_mult5): Likewise.
5350         (11_mult6): Likewise.
5351         (11_mult7): Likewise.
5352         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5353         (cortex_a8_mla): Likewise.
5354         (cortex_a8_mull): Likewise.
5355         (cortex_a8_smulwy): Likewise.
5356         (cortex_a8_smlald): Likewise.
5357         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5358         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5359         (cortex_r4_mul_3): Likewise.
5360         (cortex_r4_mla_4): Likewise.
5361         (cortex_r4_mla_3): Likewise.
5362         (cortex_r4_smlald): Likewise.
5363         (cortex_r4_mull): Likewise.
5364         (cortex_r4_sdiv): Likewise.
5365         (cortex_r4_udiv): Likewise.
5366         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5367         (cortex_a7_idiv): Likewise.
5368         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5369         (9_mult2): Likewise.
5370         (9_mult3): Likewise.
5371         (9_mult4): Likewise.
5372         (9_mult5): Likewise.
5373         (9_mult6): Likewise.
5374         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5375         (cortex_a53_sdiv): Likewise.
5376         (cortex_a53_udiv): Likewise.
5377         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5378         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5379         (mp626_mult2): Likewise.
5380         (mp626_mult3): Likewise.
5381         (mp626_mult4): Likewise.
5382         * config/arm/fa526.md (526_mult1): Update attribute change.
5383         (526_mult2): Likewise.
5384         * config/arm/arm-generic.md (mult): Update attribute change.
5385         (mult_ldsched_strongarm): Likewise.
5386         (mult_ldsched): Likewise.
5387         (multi_cycle): Likewise.
5388         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5389         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5390         (606te_mult2): Likewise.
5391         (606te_mult3): Likewise.
5392         (606te_mult4): Likewise.
5393         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5394         (cortex_a9_mac16): Likewise.
5395         (cortex_a9_multiply): Likewise.
5396         (cortex_a9_mac): Likewise.
5397         (cortex_a9_multiply_long): Likewise.
5398         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5399         (626te_mult2): Likewise.
5400         (626te_mult3): Likewise.
5401         (626te_mult4): Likewise.
5403 2013-06-18  Richard Biener  <rguenther@suse.de>
5405         PR lto/57334
5406         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5408 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5410         PR target/57609
5411         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5412         with next_active_insn.
5414 2013-06-18  Alan Modra  <amodra@gmail.com>
5416         * config/rs6000/rs6000.h (enum data_align): New.
5417         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5418         (DATA_ABI_ALIGNMENT): Define.
5419         (CONSTANT_ALIGNMENT): Correct comment.
5420         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5421         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5423 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5425         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5426         ATTRIBUTE_UNUSED marking.
5428 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5430         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5431         alternative and update.
5432         (aarch64_dup_lanedi): Delete.
5433         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5434         * config/aarch64/aarch64-simd-builtins.def: Update.
5436 2013-06-17  Richard Biener  <rguenther@suse.de>
5438         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5439         (lto_input_scc): Declare.
5440         (lto_input_tree_1): Likewise.
5441         (struct lto_stats_d): Add num_tree_bodies_output and
5442         num_pickle_refs_output.
5443         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5444         (lto_read_tree_1): Split out from ...
5445         (lto_read_tree): ... this.
5446         (lto_input_scc): New function.
5447         (lto_input_tree_1): Split out from ...
5448         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5449         (lto_data_in_create): Create the streamer cache without hashes.
5450         * lto-streamer-out.c (create_output_block): Create the streamer
5451         cache with hashes when not doing WPA.
5452         (lto_write_tree_1): Split out from ...
5453         (lto_write_tree): ... this.
5454         (get_symbol_initial_value): New function.
5455         (lto_output_tree_1): Split out from ...
5456         (lto_output_tree): ... this.  Write trees as series of SCCs
5457         using a DFS walk via DFS_write_tree.
5458         (struct sccs, struct scc_entry): New types.
5459         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5460         (DFS_write_tree_body): New function.
5461         (DFS_write_tree): Likewise.
5462         (hash_tree): Likewise.
5463         (scc_entry_compare): Likewise.
5464         (hash_scc): Likewise.
5465         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5466         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5467         TREE_CHAIN as regular reference.
5468         (streamer_read_integer_cst): Remove.
5469         (streamer_get_pickled_tree): Adjust.
5470         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5471         of DECL_EXTERNALs in BLOCK_VARS for now.
5472         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5473         reference.
5474         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5475         Add hash value argument and record that if hashes are recorded
5476         in the cache.
5477         (streamer_tree_cache_insert_1): Adjust.
5478         (streamer_tree_cache_insert): Likewise.
5479         (streamer_tree_cache_insert_at): Rename to ...
5480         (streamer_tree_cache_replace_tree): ... this and adjust.
5481         (streamer_tree_cache_append): Adjust.
5482         (record_common_node): Likewise.
5483         (streamer_tree_cache_create): Add argument whether to
5484         record hash values together with trees.
5485         (streamer_tree_cache_delete): Adjust.
5486         * tree-streamer.h (struct streamer_tree_cache_d): Add
5487         vector of hashes.
5488         (streamer_read_integer_cst): Remove.
5489         (streamer_tree_cache_insert): Adjust.
5490         (streamer_tree_cache_append): Likewise.
5491         (streamer_tree_cache_insert_at): Rename to ...
5492         (streamer_tree_cache_replace_tree): ... this and adjust.
5493         (streamer_tree_cache_create): Add argument whether to record hashes.
5494         (streamer_tree_cache_get): Rename to ...
5495         (streamer_tree_cache_get_tree): ... this.
5496         (streamer_tree_cache_get_hash): New function.
5497         * tree.c (cache_integer_cst): New function.
5498         * tree.h (cache_integer_cst): Declare.
5499         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5500         * lto-symtab.c (lto_varpool_replace_node): Only release
5501         DECL_INITIAL of non-prevailing decls.
5502         * varpool.c (varpool_remove_initializer): Do not release
5503         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5505 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5507         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5508         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5509         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5510         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5511         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5512         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5513         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5514         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5515         instead of TARGET_64BIT.
5516         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5517         Require ISA_HAS_<D>DIV.
5519 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
5521         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5522         (mips*-*-linux*): Move default with_llsc setting to where other
5523         defaults are set.
5524         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5525         with_arch block.
5526         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5527         Likewise.  Remove default with_tune setting.  Move default float
5528         setting to its own block.  Handle with_llsc in the same block as above.
5530 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
5532         PR rtl-optimization/57425
5533         PR rtl-optimization/57569
5534         * alias.c (write_dependence_p): Add new parameters mem_mode,
5535         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
5536         Changed all callers.
5537         (canon_anti_dependence): New function.
5538         * cse.c (check_dependence): Use canon_anti_dependence.
5539         * cselib.c (cselib_invalidate_mem): Likewise.
5540         * rtl.h (canon_anti_dependence): Declare.
5542 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5544         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5545         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5546         ".set mips3" for 64-bit targets.
5548 2013-06-15  Dehao Chen  <dehao@google.com>
5550         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5551         * gimple-low.c (gimple_check_call_matching_types): Likewise.
5552         (gimple_check_call_args): Likewise.
5553         * value-prof.c (check_ic_target): Likewise.
5554         * ipa-inline.c (early_inliner): Likewise.
5555         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5556         * cgraph.c (cgraph_create_edge_1): Likewise.
5557         (cgraph_make_edge_direct): Likewise.
5559 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
5561         PR target/57615
5562         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5563         rs6000_output_move_128bit to handle emitting quad memory
5564         operations.  Set attribute length to 8 bytes.
5566 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
5568         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5569         New pattern.
5570         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5571         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5572         (aarch64_<su>mlsl<mode>): Likewise.
5574 2013-06-14  Mike Stump  <mikestump@comcast.net>
5576         * Makefile.in (TARGET_H): Add insn-codes.h.
5578 2013-06-14  Alan Modra  <amodra@gmail.com>
5580         PR middle-end/57134
5581         PR middle-end/57586
5582         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5583         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
5584         bitfield expansion when EXPAND_MEMORY.
5585         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5587 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
5589         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5590         test for clearing quad memory on 32-bit later.
5592 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5594         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5595         (fold_negate_expr): Likewise.
5596         (fold_real_zero_addition_p): Handle vectors.
5597         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5599 2013-06-14  Alan Modra  <amodra@gmail.com>
5601         * varasm.c (force_const_mem): Revert 2013-06-07 change.
5603 2013-06-13  Jan Hubicka  <jh@suse.cz>
5605         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5606         Local comdats are not externally visible.
5607         * symtab.c (dump_symtab_base): Dump externally visible.
5608         (verify_symtab_base): Verify back links in the symtab hash.
5610 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5612         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5613         CONVERT_EXPR as equal nodes.
5615 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5617         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5619 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5621         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5622         Generalize to complex and vector.
5623         * tree.c (build_all_ones_cst): New function.
5624         * tree.h (build_all_ones_cst): Declare it.
5626 2013-06-13  Alan Modra  <amodra@gmail.com>
5628         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5629         * config/rs6000/rs6000.md (signbittf2): New insn.
5630         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5631         (abstf2_internal, cmptf_internal2): Likewise.
5632         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5634 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5635             Pat Haugen <pthaugen@us.ibm.com>
5636             Peter Bergner <bergner@vnet.ibm.com>
5638         * config/rs6000/rs6000.c (emit_load_locked): Add support for
5639         power8 byte, half-word, and quad-word atomic instructions.
5640         (emit_store_conditional): Likewise.
5641         (rs6000_expand_atomic_compare_and_swap): Likewise.
5642         (rs6000_expand_atomic_op): Likewise.
5644         * config/rs6000/sync.md (larx): Add new modes for power8.
5645         (stcx): Likewise.
5646         (AINT): New mode iterator to include TImode as well as normal
5647         integer modes on power8.
5648         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5649         that VSX registers are not considered.  Use AINT mode iterator
5650         instead of INT1 to allow inclusion of quad word atomic operations
5651         on power8.
5652         (load_locked<mode>): Likewise.
5653         (store_conditional<mode>): Likewise.
5654         (atomic_compare_and_swap<mode>): Likewise.
5655         (atomic_exchange<mode>): Likewise.
5656         (atomic_nand<mode>): Likewise.
5657         (atomic_fetch_<fetchop_name><mode>): Likewise.
5658         (atomic_nand_fetch<mode>): Likewise.
5659         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5660         each type.
5661         (ATOMIC): On power8, add QImode, HImode modes.
5662         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5663         modes that promote to SImode.
5664         (load_lockedti): Convert TImode arguments to PTImode, so that we
5665         get a guaranteed even/odd register pair.
5666         (load_lockedpti): Likewise.
5667         (store_conditionalti): Likewise.
5668         (store_conditionalpti): Likewise.
5670         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5671         atomic load/store instructions.
5672         (HSI): Likewise.
5674 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
5676         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5677         loads.
5678         (insn_count): New attribute, with most cases extracted from...
5679         (length): ...here.  Redefine most cases in terms of insn_count.
5680         (single_insn): Delete.
5681         (can_delay): Use insn_count to check for single instructions.
5682         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5683         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5684         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5685         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5686         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5687         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5688         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5689         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5690         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5691         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5692         rather than "length".
5693         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
5694         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5695         Use "insn_count" rather than "length".
5696         * config/mips/mips-dsp.md
5697         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5698         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5699         length attributes.
5701 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
5703         PR tree-optimization/57361
5704         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5706 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
5708         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5709         to split.
5710         (aarch64_simd_combine<mode>): New instruction expansion.
5711         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5712         function prototype.
5713         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5714         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5716 2013-06-12  Jan Hubicka  <jh@suse.cz>
5718         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5719         decl has when in streaming stage.
5720         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5721         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5723 2013-06-12  Roland Stigge <stigge@antcom.de>
5725         PR target/57578
5726         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5728 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5730         PR tree-optimization/57537
5731         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5732         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5734 2013-06-12  Richard Biener  <rguenther@suse.de>
5736         * data-streamer.h (streamer_write_char_stream): CSE
5737         obs->current_pointer.
5738         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5739         streamer_write_char_stream manually and optimize the resulting loop.
5740         (streamer_write_hwi_stream): Likewise.
5742 2013-06-12  Jan Hubicka  <jh@suse.cz>
5744         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5745         * cgraph.h (varpool_create_empty_node): Declare.
5746         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5747         duplicated nodes.
5748         * symtab.c (symtab_unregister_node): Be lax about missin entries
5749         in node hash.
5750         (symtab_get_node): Update comment.
5751         * varpool.c (varpool_create_empty_node): Break out from ...
5752         (varpool_node_for_decl): ... here.
5753         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5755 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
5757         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5758         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5759         part.  Use straight-line flow at the end.
5760         <COMPONENT_REF>: Remove superfluous else.
5761         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5763 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5765         PR target/56564
5766         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5767         target hook even for !TREE_PUBLIC decls.  If no resolution info
5768         is available, return false for common and external decls.
5770 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
5772         * config/rl78/constraints.md (U): New constraint.
5773         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5774         valloc attribute.
5776 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5778         PR target/57589
5779         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5780         to allow returning address to AT_PLATFORM name.
5782 2013-06-11  Jan Hubicka  <jh@suse.cz>
5784         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5785         * cgraph.h (symtab_node_base): Add weakref flag.
5786         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5787         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5788         (output_weakrefs): Use weakref flag.
5789         * fold-const.c (simple_operand_p): Handle WEAK.
5790         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5791         * ipa.c (varpool_externally_visible_p): Drop weakref.
5792         (function_and_variable_visibility): Update comment; fix weakref
5793         sanity checks; do not clear DECL_WEAK on them.
5794         * lto-cgraph.c (lto_output_node): update.
5795         (lto_output_varpool_node): Update.
5796         (input_overwrite_node): Update.
5797         (input_node): Update.
5798         (input_varpool_node): Update.
5799         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5800         (lto_symtab_merge_symbols): Add sanity check.
5801         (lto_symtab_prevailing_decl): Do not special case weakrefs.
5802         * passes.c (rest_of_decl_compilation): Set static flag, too.
5803         * symtab.c (dump_symtab_base): Dump weakref.
5804         (verify_symtab_base): Sanity check weakrefs.
5805         (symtab_make_decl_local): Remove duplicated code.
5806         (symtab_alias_ultimate_target): Simplify.
5807         * varpool.c (varpool_create_variable_alias): Set weakref flag.
5809 2013-06-11  DJ Delorie  <dj@redhat.com>
5811         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5812         (rl78_unwind_word_mode): New.
5814 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5816         * final.c (debug_prefix_maps): Make static.
5818 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5820         * function.c (initial_trampoline): Remove stray copy.
5822 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
5824         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5826 2013-06-11  Martin Jambor  <mjambor@suse.cz>
5828         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5829         within bounds at the beginning of the function.
5831 2013-06-11  Alan Modra  <amodra@gmail.com>
5833         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
5834         reporting.
5835         (get_named_section): Don't NULL !DECL_P decl.
5837 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
5839         * doc/invoke.texi (core-avx2): Document.
5840         (slm): Likewise.
5841         (atom): Updated with MOVBE.
5843 2013-06-11  Richard Biener  <rguenther@suse.de>
5845         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5847 2013-06-11  Anton Blanchard  <anton@samba.org>
5849         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5850         correct shift value in little-endian mode.
5852 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
5854         PR target/56564
5855         * varasm.c (get_variable_align): Move #endif to the right place.
5857 2013-06-10  Cary Coutant  <ccoutant@google.com>
5859         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5860         for hash so that hash table traversal order is deterministic.
5862 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
5863             Pat Haugen <pthaugen@us.ibm.com>
5864             Peter Bergner <bergner@vnet.ibm.com>
5866         * config/rs6000/vector.md (GPR move splitter): Do not split moves
5867         of vectors in GPRS if they are direct moves or quad word load or
5868         store moves.
5870         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5871         declaration.
5872         (direct_move_p): Likewise.
5873         (quad_load_store_p): Likewise.
5875         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5876         classes into bins based on the physical register type.
5877         (reg_class_to_reg_type): Likewise.
5878         (IS_STD_REG_TYPE): Likewise.
5879         (IS_FP_VECT_REG_TYPE): Likewise.
5880         (reload_fpr_gpr): Arrays to determine what insn to use if we can
5881         use direct move instructions.
5882         (reload_gpr_vsx): Likewise.
5883         (reload_vsx_gpr): Likewise.
5884         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5885         information that is a simplification of register classes.  Also
5886         precalculate direct move reload helpers.
5887         (direct_move_p): New function to return true if the operation can
5888         be done as a direct move instruciton.
5889         (quad_load_store_p): New function to return true if the operation
5890         is a quad memory operation.
5891         (rs6000_legitimize_address): If quad memory, only allow register
5892         indirect for TImode addresses.
5893         (rs6000_legitimate_address_p): Likewise.
5894         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5895         (rs6000_reload_register_type): Likewise.
5896         (register_to_reg_type): Return register type.
5897         (rs6000_secondary_reload_simple_move): New helper function for
5898         secondary reload and secondary memory needed to identify anything
5899         that is a simple move, and does not need reloading.
5900         (rs6000_secondary_reload_direct_move): New helper function for
5901         secondary reload to identify cases that can be done with several
5902         instructions via the direct move instructions.
5903         (rs6000_secondary_reload_move): New helper function for secondary
5904         reload to identify moves between register types that can be done.
5905         (rs6000_secondary_reload): Add support for quad memory operations
5906         and for direct move.
5907         (rs6000_secondary_memory_needed): Likewise.
5908         (rs6000_debug_secondary_memory_needed): Change argument names.
5909         (rs6000_output_move_128bit): New function to return the move to
5910         use for 128-bit moves, including knowing about the various
5911         limitations of quad memory operations.
5913         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5914         memory operations.  call rs6000_output_move_128bit for the actual
5915         instruciton(s) to generate.
5916         (vsx_movti_64bit): Likewise.
5918         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5919         (UNSPEC_P8V_MTVSRWZ): Likewise.
5920         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5921         (UNSPEC_P8V_MTVSRD): Likewise.
5922         (UNSPEC_P8V_XXPERMDI): Likewise.
5923         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5924         (UNSPEC_FUSION_GPR): Likewise.
5925         (FMOVE128_GPR): New iterator for direct move.
5926         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5927         (f32_sv): Likewise.
5928         (f32_dm): Likewise.
5929         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5930         loads and direct move instructions.
5931         (zero_extendsidi2_lfiwzx): Likewise.
5932         (extendsidi2_lfiwax): Likewise.
5933         (extendsidi2_nocell): Likewise.
5934         (floatsi<mode>2_lfiwax): Likewise.
5935         (lfiwax): Likewise.
5936         (floatunssi<mode>2_lfiwzx): Likewise.
5937         (lfiwzx): Likewise.
5938         (fix_trunc<mode>_stfiwx): Likewise.
5939         (fixuns_trunc<mode>_stfiwx): Likewise.
5940         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5941         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5942         (parity<mode>2_cmpb): Set length/type attr.
5943         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5944         for 'mr.' to fast_compare.
5945         (bpermd_<mode>): Change type attr to popcnt.
5946         (p8_fmrgow_<mode>): New insns for power8 direct move support.
5947         (p8_mtvsrwz_1): Likewise.
5948         (p8_mtvsrwz_2): Likewise.
5949         (reload_fpr_from_gpr<mode>): Likewise.
5950         (p8_mtvsrd_1): Likewise.
5951         (p8_mtvsrd_2): Likewise.
5952         (p8_xxpermdi_<mode>): Likewise.
5953         (reload_vsx_from_gpr<mode>): Likewise.
5954         (reload_vsx_from_gprsf): Likewise.
5955         (p8_mfvsrd_3_<mode>): LIkewise.
5956         (reload_gpr_from_vsx<mode>): Likewise.
5957         (reload_gpr_from_vsxsf): Likewise.
5958         (p8_mfvsrd_4_disf): Likewise.
5959         (multi-word GPR splits): Do not split direct moves or quad memory
5960         operations.
5962 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
5964         * tree-into-ssa.c (interesting_blocks): Make static.
5966 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
5968         PR target/56564
5969         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5970         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5971         Use DATA_ABI_ALIGNMENT for that case instead if defined.
5972         (get_variable_align): New function.
5973         (get_variable_section, emit_bss, emit_common,
5974         assemble_variable_contents, place_block_symbol): Use
5975         get_variable_align instead of DECL_ALIGN.
5976         (assemble_noswitch_variable): Add align argument, use it
5977         instead of DECL_ALIGN.
5978         (assemble_variable): Adjust caller.  Use get_variable_align
5979         instead of DECL_ALIGN.
5980         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5981         caller.
5982         (DATA_ABI_ALIGNMENT): Define.
5983         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5984         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
5985         opt is false, only return the psABI mandated alignment increase.
5986         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5987         (DATA_ABI_ALIGNMENT): ... this.
5988         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5989         (DATA_ABI_ALIGNMENT): ... this.
5990         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5991         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5992         (DATA_ABI_ALIGNMENT): ... this.
5993         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5994         * doc/tm.texi: Regenerated.
5996 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
5998         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5999         cmp_code to construct REG_EQUAL note.
6001 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6003         PR target/57568
6004         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6005         that operands[2] doesn't overlap with operands[0].
6007 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6008             Jan Hubicka  <jh@suse.cz>
6010         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6011         hack to mark symbols as used.
6013 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6015         PR rtl-optimization/57559
6016         * lra-constraints.c (process_alt_operands): Don't discourage
6017         memory with known offset for offsetable memory constraint.
6018         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6020 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
6022         * varasm.c (struct oc_local_state): Reorder fields.
6023         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6024         and adjust accordingly.
6025         (output_constructor): Reorder initialization code and adjust call to
6026         output_constructor_bitfield.
6028 2013-06-07  Jan Hubicka  <jh@suse.cz>
6030         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6032 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
6034         * tree-object-size.c (unknown): Make const.
6036 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6038         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6039         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6040         for last alternative in the cpu_facility attribute.
6042 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6044         PR target/56315
6045         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6046         (xordi3): Change operand 2 constraint to arm_xordi_operand.
6047         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6048         * config/arm/constraints.md (Dg): New constraint.
6049         * config/arm/neon.md (xordi3_neon): Remove.
6050         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6051         * config/arm/predicates.md (arm_xordi_operand): New predicate.
6053 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6055         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6056         Clean up alternatives.
6058 2013-06-07  Alan Modra  <amodra@gmail.com>
6060         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6061         va_list_gpr_size.
6063 2013-06-07  Alan Modra  <amodra@gmail.com>
6065         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6067 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6069         * config/arm/constraints.md (Df): New constraint.
6070         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6071         Correct length attribute for last two alternatives.
6073 2013-06-07  Alan Modra  <amodra@gmail.com>
6075         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6076         override user -mfp-in-toc.
6077         (offsettable_ok_by_alignment): Consider just the current access
6078         rather than the whole object, unless BLKmode.  Handle
6079         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6080         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6081         for -mcmodel=medium.
6082         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6083         override user -mfp-in-toc or -msum-in-toc.  Default to
6084         -mno-fp-in-toc for -mcmodel=medium.
6086 2013-06-06  DJ Delorie  <dj@redhat.com>
6088         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6089         TARGET_VALID_POINTER_MODE.
6091 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
6092             Pat Haugen <pthaugen@us.ibm.com>
6093             Peter Bergner <bergner@vnet.ibm.com>
6095         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6096         Document new power8 builtins.
6098         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6099         condition code register, to allow 128-bit logical operations to be
6100         done in the VSX or GPR registers.
6101         (nor<mode>3): Use the canonical form for nor.
6102         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6103         vclz*, and vpopcnt* vector instructions.
6104         (nand<mode>3): Likewise.
6105         (orc<mode>3): Likewise.
6106         (clz<mode>2): LIkewise.
6107         (popcount<mode>2): Likewise.
6109         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6110         that only the GPRs are recognized.
6112         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6113         support for new power8 builtins.
6115         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6116         builtin functions.
6117         (xscvdpspn): Likewise.
6118         (vclz): Likewise.
6119         (vclzb): Likewise.
6120         (vclzh): Likewise.
6121         (vclzw): Likewise.
6122         (vclzd): Likewise.
6123         (vpopcnt): Likewise.
6124         (vpopcntb): Likewise.
6125         (vpopcnth): Likewise.
6126         (vpopcntw): Likewise.
6127         (vpopcntd): Likewise.
6128         (vgbbd): Likewise.
6129         (vmrgew): Likewise.
6130         (vmrgow): Likewise.
6131         (eqv): Likewise.
6132         (eqv_v16qi3): Likewise.
6133         (eqv_v8hi3): Likewise.
6134         (eqv_v4si3): Likewise.
6135         (eqv_v2di3): Likewise.
6136         (eqv_v4sf3): Likewise.
6137         (eqv_v2df3): Likewise.
6138         (nand): Likewise.
6139         (nand_v16qi3): Likewise.
6140         (nand_v8hi3): Likewise.
6141         (nand_v4si3): Likewise.
6142         (nand_v2di3): Likewise.
6143         (nand_v4sf3): Likewise.
6144         (nand_v2df3): Likewise.
6145         (orc): Likewise.
6146         (orc_v16qi3): Likewise.
6147         (orc_v8hi3): Likewise.
6148         (orc_v4si3): Likewise.
6149         (orc_v2di3): Likewise.
6150         (orc_v4sf3): Likewise.
6151         (orc_v2df3): Likewise.
6153         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6154         allow power8 quad mode in 64-bit.
6155         (rs6000_builtin_vectorized_function): Add support to vectorize
6156         ISA 2.07 count leading zeros, population count builtins.
6157         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6158         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6159         (builtin_function_type): Add vgbbd builtin function which takes an
6160         unsigned argument.
6161         (altivec_expand_vec_perm_const): Add support for new power8 merge
6162         instructions.
6164         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6165         that does not include TImdoe for use with 32-bit.
6166         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6167         instructions.
6168         (UNSPEC_VSX_CVDPSPN): Likewise.
6169         (vsx_xscvdpspn): Likewise.
6170         (vsx_xscvspdpn): Likewise.
6171         (vsx_xscvdpspn_scalar): Likewise.
6172         (vsx_xscvspdpn_directmove): Likewise.
6173         (vsx_and<mode>3): Split logical operations into 32-bit and
6174         64-bit. Add support to do logical operations on TImode as well as
6175         VSX vector types.  Allow logical operations to be done in either
6176         VSX registers or in general purpose registers in 64-bit mode.  Add
6177         splitters if GPRs were used. For AND, add clobber of CCmode to
6178         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
6179         encoding.
6180         (vsx_and<mode>3_32bit): Likewise.
6181         (vsx_and<mode>3_64bit): Likewise.
6182         (vsx_ior<mode>3): Likewise.
6183         (vsx_ior<mode>3_32bit): Likewise.
6184         (vsx_ior<mode>3_64bit): Likewise.
6185         (vsx_xor<mode>3): Likewise.
6186         (vsx_xor<mode>3_32bit): Likewise.
6187         (vsx_xor<mode>3_64bit): Likewise.
6188         (vsx_one_cmpl<mode>2): Likewise.
6189         (vsx_one_cmpl<mode>2_32bit): Likewise.
6190         (vsx_one_cmpl<mode>2_64bit): Likewise.
6191         (vsx_nor<mode>3): Likewise.
6192         (vsx_nor<mode>3_32bit): Likewise.
6193         (vsx_nor<mode>3_64bit): Likewise.
6194         (vsx_andc<mode>3): Likewise.
6195         (vsx_andc<mode>3_32bit): Likewise.
6196         (vsx_andc<mode>3_64bit): Likewise.
6197         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6198         and xxlorc instructions.
6199         (vsx_eqv<mode>3_64bit): Likewise.
6200         (vsx_nand<mode>3_32bit): Likewise.
6201         (vsx_nand<mode>3_64bit): Likewise.
6202         (vsx_orc<mode>3_32bit): Likewise.
6203         (vsx_orc<mode>3_64bit): Likewise.
6205         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6207         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6208         instruction.
6209         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6210         (p8_vmrgow): Likewise.
6211         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6212         GPRs to be split under VSX.
6213         (p8v_clz<mode>2): Add power8 count leading zero support.
6214         (p8v_popcount<mode>2): Add power8 population count support.
6215         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6216         support.
6218         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6219         instruction.
6221         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6222         builtin functions.
6223         (vec_nand): Likewise.
6224         (vec_vclz): Likewise.
6225         (vec_vclzb): Likewise.
6226         (vec_vclzd): Likewise.
6227         (vec_vclzh): Likewise.
6228         (vec_vclzw): Likewise.
6229         (vec_vgbbd): Likewise.
6230         (vec_vmrgew): Likewise.
6231         (vec_vmrgow): Likewise.
6232         (vec_vpopcnt): Likewise.
6233         (vec_vpopcntb): Likewise.
6234         (vec_vpopcntd): Likewise.
6235         (vec_vpopcnth): Likewise.
6236         (vec_vpopcntw): Likewise.
6238 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6240         PR rtl-optimization/57468
6241         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6242         spilled pseudos.
6244 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6246         PR rtl-optimization/57459
6247         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6248         type when setting live regs.
6250 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6252         * config/s390/s390.opt (mlra): New option.
6253         * config/s390/s390.c (s390_decompose_address): Check displacement
6254         for all registers for LRA.
6255         (s390_secondary_reload): Don't used secondary reloads for LRA.
6256         (s390_lra_p): New function.
6257         (TARGET_LRA_P): Define.
6258         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6259         of attribute cpu_facility to zarch for the last alternative.
6260         (*cmpmem_short): Ditto.
6262 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6264         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6265         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6266         (arm_expand_prologue): Likewise.
6268 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6270         PR c++/53743
6271         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6272         as this is now done by redirect_edge_and_branch_force.
6273         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6274         barriers, and fix interaction with splitting.
6275         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6276         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6277         reflect changes made in the routine.
6278         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6279         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6280         since this is called in cfglayout mode, and replace partition fixup
6281         with assert as that is now done by force_nonfallthru_and_redirect.
6282         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6283         already be marked with region crossing note.
6284         (insert_section_boundary_note): Make non-static, gate on flag
6285         has_bb_partition, rewrite to also check for multiple partitions.
6286         (rest_of_handle_reorder_blocks): Remove call to
6287         insert_section_boundary_note, now done later during free_cfg.
6288         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6289         * bb-reorder.h (insert_section_boundary_note): Declare.
6290         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6291         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6292         invoke insert_section_boundary_note.
6293         (try_redirect_by_replacing_jump): Remove unnecessary
6294         check for region crossing note.
6295         (fixup_partition_crossing): New function.
6296         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6297         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6298         in non-cfglayout mode.
6299         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6300         remove old code that tried to do this. Emit barrier correctly
6301         when we are in cfglayout mode.
6302         (last_bb_in_partition): New function.
6303         (rtl_split_edge): Correctly fixup partition boundaries.
6304         (commit_one_edge_insertion): Remove old code that tried to
6305         fixup region crossing edge since this is now handled in
6306         split_block, and set up insertion point correctly since
6307         block may now end in a jump.
6308         (verify_hot_cold_block_grouping): Guard against checking when not in
6309         linearized RTL mode.
6310         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6311         notes.
6312         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6313         rtl_verify_flow_info, so not called in cfglayout mode.
6314         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6315         (fixup_reorder_chain): Remove old code that attempted to fixup region
6316         crossing note as this is now handled in force_nonfallthru_and_redirect.
6317         (duplicate_insn_chain): Don't duplicate switch section notes.
6318         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6319         note.
6320         * basic-block.h (emit_barrier_after_bb): Declare.
6322 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6324         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6325         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6326         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6328 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6330         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6331         where appropriate.
6332         * config/arm/ldmstm.md: Regenerate.
6334 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6336         * config/arm/sync.md (atomic_loaddi_1):
6337         Disable predication for arm_restrict_it.
6338         (arm_load_exclusive<mode>): Likewise.
6339         (arm_load_exclusivesi): Likewise.
6340         (arm_load_exclusivedi): Likewise.
6341         (arm_load_acquire_exclusive<mode>): Likewise.
6342         (arm_load_acquire_exclusivesi): Likewise.
6343         (arm_load_acquire_exclusivedi): Likewise.
6344         (arm_store_exclusive<mode>): Likewise.
6345         (arm_store_exclusive<mode>): Likewise.
6346         (arm_store_release_exclusivedi): Likewise.
6347         (arm_store_release_exclusive<mode>): Likewise.
6349 2013-06-06  Richard Biener  <rguenther@suse.de>
6351         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6352         after LTO_null.
6353         (lto_tag_is_tree_code_p): Adjust.
6354         (lto_tag_is_gimple_code_p): Likewise.
6355         (lto_gimple_code_to_tag): Likewise.
6356         (lto_tag_to_gimple_code): Likewise.
6357         (lto_tree_code_to_tag): Likewise.
6358         (lto_tag_to_tree_code): Likewise.
6359         * data-streamer.h (streamer_write_hwi_in_range): Use
6360         uhwi streaming to stream the normalized range.
6361         (streamer_read_hwi_in_range): Likewise.
6363 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6365         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6366         (predicable_short_it): Likewise.
6367         (predicated): Likewise.
6368         (enabled): Handle above.
6369         (define_cond_exec): Set predicated attribute to yes.
6371 2013-06-05  Mike Stump  <mikestump@comcast.net>
6373         * gdbinit.in (__FUNCTION__): Add.
6375 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6377         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6378         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6380 2013-06-05  Jan Hubicka  <jh@suse.cz>
6382         * varasm.c (mark_decl_referenced): Revert the removal until targets
6383         are fixed.
6385 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6387         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6388         instead of mark_decl_referenced.
6390 2013-06-05  Jan Hubicka  <jh@suse.cz>
6392         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6393         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6394         and symtab_used_from_object_file_p.
6395         (cgraph_make_node_local_1): Clear forced_by_abi.
6396         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6397         * cgraph.h (symtab_node_base): Add forced_by_abi.
6398         (decide_is_variable_needed): Remove.
6399         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6400         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6401         (decide_is_symbol_needed): ... this one; handle symbols in general;
6402         always analyze virtuals; honnor forced_by_abi.
6403         (cgraph_finalize_function): Update.
6404         (varpool_finalize_decl): Update.
6405         (symbol_defined_and_needed): Remove.
6406         (analyze_functions): Update.
6407         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6408         output_refs, input_overwrite_node): Handle forced_by_abi.
6409         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6410         (address_taken_from_non_vtable_p): ... this one.
6411         (comdat_can_be_unshared_p_1): New function.
6412         (cgraph_comdat_can_be_unshared_p): Rename to ...
6413         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6414         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6415         (function_and_variable_visibility): Clear forced_by_abi as needed.
6416         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6417         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6418         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6419         * varpool.c (decide_is_variable_needed): Remove.
6421 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6423         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6424         (arm_option_override): Override arm_restrict_it where appropriate.
6425         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6426         * config/arm/arm.opt (mrestrict-it): New command-line option.
6427         * doc/invoke.texi: Document -mrestrict-it.
6429 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6431         * tsan.c (tsan_atomic_table): Make const.
6433 2013-06-05  Richard Biener  <rguenther@suse.de>
6435         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6436         index associated with the tree we are supposed to replace.
6437         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6438         TREE_ASM_WRITTEN as zero for everything but SSA names.
6440 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6442         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6444 2013-06-04  Jan Hubicka  <jh@suse.cz>
6446         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6447         (input_node, input_varpool_node): Handle correctly external same
6448         body aliases.
6449         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6450         nodes at ltrans stage.
6452 2013-06-04  Jan Hubicka  <jh@suse.cz>
6454         * ipa-inline.c (update_caller_keys): Fix availability test.
6455         (update_callee_keys): Likewise.
6456         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6457         to follow ELF standard.
6459 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6461         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6462         (mips64r5900el-*-elf*): New configurations.
6463         * config/mips/mips-cpus.def (r5900): New processor.
6464         * config/mips/mips-tables.opt: Regenerate.
6465         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6466         (mips_issue_rate): Handle PROCESSOR_R5900.
6467         (mips_reorg_process_insns): Force reorder mode for the R5900.
6468         * config/mips/mips.h (TARGET_MIPS5900): Define.
6469         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6470         TARGET_MIPS5900.
6471         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6472         TARGET_MIPS5900.
6473         * config/mips/mips.md (processor): Add r5900.
6474         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6476 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6478         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6479         into function to generate MOVI instruction.
6480         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6481         (aarch64_preferred_simd_mode): Turn into wrapper.
6482         (aarch64_output_scalar_simd_mov_immediate): New function.
6483         * config/aarch64/aarch64-protos.h: Add prototype for above.
6485 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6487         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6488         element_char member.
6489         (sizetochar): Return signed char.
6490         (aarch64_simd_valid_immediate): Remove elchar and other
6491         unnecessary variables.
6492         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6493         Calculate element_char as required.
6494         * config/aarch64/aarch64-protos.h: Update and move prototype
6495         for aarch64_output_simd_mov_immediate.
6496         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6497         Update arguments.
6499 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6501         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6502         information completed by aarch64_simd_valid_immediate.
6503         (aarch64_legitimate_constant_p): Update arguments.
6504         (aarch64_simd_valid_immediate): Work with struct rather than many
6505         pointers.
6506         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6507         (aarch64_simd_make_constant): Update arguments.
6508         (aarch64_output_simd_mov_immediate): Work with struct rather than
6509         many pointers.  Output immediate directly rather than as operand.
6510         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6511         Update prototype.
6512         * config/aarch64/constraints.md (Dn): Update arguments.
6514 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6516         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6517         longer static.
6518         (aarch64_simd_immediate_valid_for_move): Remove.
6519         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6520         (aarch64_simd_make_constant): Update call.
6521         (aarch64_output_simd_mov_immediate): Update call.
6522         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6523         Add prototype.
6524         * config/aarch64/constraints.md (Dn): Update call.
6526 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6528         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6529         return type to bool for prototype.
6530         (aarch64_legitimate_constant_p): Check for true instead of not -1.
6531         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6532         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6534 2013-06-04  Catherine Moore  <clm@codesourcery.com>
6536         * config/mips/mips.opt (meva): New.
6537         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6538         (ASM_SPEC): Handle -meva.
6539         * doc/invoke.texi (meva):  Document.
6541 2013-06-04  Alan Modra  <amodra@gmail.com>
6543         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6544         constant output.
6546 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6548         * rtl.def: Add extra fourth optional field to define_cond_exec.
6549         * gensupport.c (process_one_cond_exec): Process attributes from
6550         define_cond_exec.
6551         * doc/md.texi: Document fourth field in define_cond_exec.
6553 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
6555         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6556         out the processing order as in store_bit_field_1.
6558 2013-06-04  Jan Hubicka  <jh@suse.cz>
6560         PR middle-end/57500
6561         * cgraphunit.c (cgraph_process_same_body_aliases): Create
6562         non-VAR_DECL node if it does not exist yet.
6564 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
6566         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6567         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6568         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6569         target_cpu_default setting.
6571 2013-06-03  Teresa Johnson  <tejohnson@google.com>
6573         * dumpfile.c (opt_info_switch_p): Change -fopt-info
6574         default to -fopt-info=optimized instead of all.
6575         * doc/invoke.texi: Ditto.
6576         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6577         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6578         (execute_vect_slp): Emit BB vectorization success under
6579         MSG_OPTIMIZED_LOCATIONS.
6580         * tree-vect-slp.c (vect_slp_transform_bb): Change
6581         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6582         * tree-vect-loop.c (vect_transform_loop): Ditto.
6584 2013-06-03  Jason Merrill  <jason@redhat.com>
6586         PR c++/57415
6587         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6588         Use TARGET_EXPR for C++.
6590 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
6592         PR rtl-optimization/57268
6593         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6594         if DEBUG_INSN_P (insn).
6596         Reapply
6597         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6599         PR rtl-optimization/57268
6600         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6601         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6603 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6605         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6606         (ix86_avoid_lea_for_addr): Likewise.
6607         (exact_dependency_1): Likewise.
6608         (ix86_adjust_cost): Likewise.
6609         (swap_top_of_ready_list): Fix formatting and !reload_completed check
6610         removed.
6611         (do_reorder_for_imul): Fix typo, formatting and
6612         !reload_completed check removed.
6613         (ix86_sched_reorder): Fix typo and formatting.
6614         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6615         list.
6617 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
6619         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6621 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
6623         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6624         <STRING_CST>: Likewise.
6625         <VECTOR_CST>: Likewise.
6627 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
6628             Mikael Morin  <mikael@gcc.gnu.org>
6630         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6631         * config.in: Regenerated.
6632         * configure: Regenerated.
6634 2013-06-01  Jan Hubicka  <jh@suse.cz>
6636         PR middle-end/57366
6637         * cgraphunit.c (compile): When weakref is not supported,
6638         set up transparent aliases before final output pass.
6639         * varasm.c (assemble_alias): Do not try to do it here.
6641 2013-06-01  Jan Hubicka  <jh@suse.cz>
6643         PR middle-end/57467
6644         * passes.c (for_per_function): Skip unanalyzed functions.
6646 2013-06-01  Jan Hubicka  <jh@suse.cz>
6648         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6649         (lto_symtab_merge_symbols_1): ... this one.
6650         (lto_symtab_merge_cgraph_nodes): Rename to ...
6651         (lto_symtab_merge_symbols): ... this one; simplify.
6652         * cgraph.c (same_body_aliases_done): Rename to ...
6653         (cpp_implicit_aliases_done): ... this one.
6654         (cgraph_create_function_alias): Update.
6655         (cgraph_same_body_alias): Update.
6656         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6657         (verify_edge_corresponds_to_fndecl): Simplify.
6658         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6659         (cgraph_node): Remove same_body_alias.
6660         (varpool_node): Remove alias_of and extra_name_alias.
6661         (same_body_aliases_done): Rename to ..
6662         (cpp_implicit_aliases_done): ... this one.
6663         (symtab_alias_ultimate_target): Add default parameter.
6664         (symtab_resolve_alias): New function.
6665         (fixup_same_cpp_alias_visibility): Declare.
6666         (cgraph_function_node): Add default parameter.
6667         (cgraph_node_asm_name): Likewise.
6668         (cgraph_function_or_thunk_node): Add default parameter; do
6669         not ICE when it is NULL.
6670         (varpool_variable_node): Likewise.
6671         * tree-emutls.c (create_emultls_var): Update.
6672         (ipa_lower_emutls): Update.
6673         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6674         (cgraph_reset_node): Reset alias info.
6675         (cgraph_finalize_function): Update.
6676         (fixup_same_cpp_alias_visibility): Move to symtab.c.
6677         (analyze_function): Simplify.
6678         (cgraph_process_same_body_aliases): Simplify.
6679         (analyze_functions): Fixup same body aliases.
6680         (handle_alias_pairs): Simplify.
6681         (assemble_thunk): Update.
6682         (assemble_thunks_and_aliases): Update.
6683         (output_weakrefs): Rewrite.
6684         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6685         (lto_output_varpool_node): Likewise.
6686         (compute_ltrans_boundary): Remve assert.
6687         (get_alias_symbol): New functoin.
6688         (input_node): Rewrite alias handling.
6689         (input_varpool_node): Likewise.
6690         * ipa-pure-const.c (propagate_pure_const): Fix formating.
6691         * ipa.c (process_references): Handle weakrefs correctly.
6692         (symtab_remove_unreachable_nodes): Likewise.
6693         * trans-mem.c (get_cg_data): Update.
6694         (ipa_tm_create_version_alias): Update.
6695         (ipa_tm_execute): Update.
6696         * symtab.c (dump_symtab_base): Dump aliases.
6697         (verify_symtab_base): Verify aliases.
6698         (symtab_node_availability): New function.
6699         (symtab_alias_ultimate_target): Simplify.
6700         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6701         handle all the fixup cases.
6702         (symtab_resolve_alias): New function.
6703         * passes.c (ipa_write_summaries): Handle weakrefs.
6704         * varpool.c (varpool_analyze_node): Simplify.
6705         (assemble_aliases): Update.
6706         (varpool_create_variable_alias): Simplify.
6707         (varpool_extra_name_alias): Simplify.
6708         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6709         (lto_symtab_merge_symbols): ... this one.
6711 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
6713         Revert
6714         PR rtl-optimization/57268
6715         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6716         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6718 2013-06-01  Tobias Burnus  <burnus@net-b.de>
6720         Partially reverted:
6721         2013-05-31  Tobias Burnus  <burnus@net-b.de>
6723         PR middle-end/57073
6724         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6725         further up.
6727 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6729         PR rtl-optimization/57268
6730         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6731         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6733 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6735         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6736         unordered comparison operators when -fno-trapping-math is in effect
6737         on the e500.
6738         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6739         and implement unordered comparison operators properly on the e500.
6741 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6743         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6744         for constant scalar integers.
6745         (simplify_relational_operation_1): Likewise.
6747 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
6749         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6750         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
6751         Fix comment.
6753 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6754             Igor Zamyatin  <igor.zamyatin@intel.com>
6756         Silvermont (SLM) architecture performance tuning.
6757         * config/i386/i386.h (enum ix86_tune_indices): Add
6758         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6759         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6761         * config/i386/i386.c (initial_ix86_tune_features)
6762         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6763         (ix86_lea_outperforms): Handle Silvermont tuning.
6764         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6765         call.
6766         (ix86_use_lea_for_mov): Likewise.
6767         (ix86_avoid_lea_for_addr): Likewise.
6768         (ix86_lea_for_add_ok): Likewise.
6769         (exact_dependency_1): New function.
6770         (exact_store_load_dependency): Likewise.
6771         (ix86_adjust_cost): Handle Silvermont tuning.
6772         (do_reoder_for_imul): Likewise.
6773         (swap_top_of_ready_list): New function.
6774         (ix86_sched_reorder): Changed to handle Silvermont tuning.
6776         * config/i386/i386.md (peepholes that split memory operand in fp
6777         converts): New.
6779 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6781         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6782         Remove un-necessary braces.
6784 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6786         * config/aarch64/aarch64.c (aarch64_classify_symbol):
6787         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6789 2013-05-31  Tobias Burnus  <burnus@net-b.de>
6791         PR middle-end/57073
6792         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6794 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6796         PR target/56315
6797         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6798         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6799         * config/arm/neon.md (iordi3_neon): Remove.
6800         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6801         * config/arm/predicates.md (imm_for_neon_logic_operand):
6802         Move to earlier in the file.
6803         (neon_logic_op2): Likewise.
6804         (arm_iordi_operand_neon): New predicate.
6806 2013-05-31  Richard Biener  <rguenther@suse.de>
6808         PR tree-optimization/57478
6809         PR tree-optimization/57453
6810         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6811         are life as well.
6813 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6815         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6816         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6818 2013-05-30  Tobias Burnus  <burnus@net-b.de>
6819             Thomas Koenig  <tkoenig@gcc.gnu.org>
6821         PR middle-end/57073
6822         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6823         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6825 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
6827         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6829 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
6831         * target.def (register_usage_leveling_p): New hook.
6832         * targhooks.c (default_register_usage_leveling_p): New.
6833         * targhooks.h (default_register_usage_leveling_p): New prototype.
6834         * lra-assigns.c (register_usage_leveling_p): Use the hook.
6835         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6836         * doc/tm.texi: Update.
6837         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6839 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
6841         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6842         (*insv_reg<mode>): New define_insn.
6844 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
6846         PR rtl-optimization/57439
6847         * postreload.c (move2add_valid_value_p): Check that we have
6848         a zero subreg_regno_offset when accessing the register in
6849         the requested mode.
6851 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6852             Igor Zamyatin  <igor.zamyatin@intel.com>
6854         Silvermont (SLM) architecture pipeline model, tuning and
6855         insn selection.
6856         * config.gcc: Add slm config options and target.
6858         * config/i386/slm.md: New.
6860         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6862         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
6863         PROCESSOR_SLM.
6864         (ix86_target_macros_internal): Likewise.
6866         * gcc/config/i386/i386.c (slm_cost): New cost.
6867         (m_SLM): New macro flag.
6868         (initial_ix86_tune_features): Set m_SLM.
6869         (x86_accumulate_outgoing_args): Likewise.
6870         (x86_arch_always_fancy_math_387): Likewise.
6871         (processor_target_table): Add slm cost.
6872         (cpu_names): Add slm cpu name.
6873         (x86_option_override_internal): Set SLM ISA.
6874         (ix86_issue_rate): New case PROCESSOR_SLM.
6875         (ia32_multipass_dfa_lookahead): Likewise.
6876         (fold_builtin_cpu): Add slm.
6878         * config/i386/i386.h (TARGET_SLM): New target macro.
6879         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6880         (processor_type): Add PROCESSOR_SLM.
6882         * config/i386/i386.md (cpu): Add new value "slm".
6883         (slm.md): Include slm.md.
6885 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
6886             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6888         * config/arm/arm-protos.h: Add and update function protos.
6889         * config/arm/arm.c (use_simple_return_p): New added.
6890         (thumb2_expand_return): Check simple_return flag.
6891         * config/arm/arm.md: Add simple_return and conditional simple_return.
6892         * config/arm/iterators.md: Add iterator for return and simple_return.
6894 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6896         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6897         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6898         (arm_emit_vfp_multi_reg_pop): Likewise.
6899         (thumb2_emit_ldrd_pop): Likewise.
6900         (arm_expand_epilogue): Add misc REG_CFA notes.
6901         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6903 2013-05-29  Lawrence Crowl  <crowl@google.com>
6905         * config/arm/t-arm: Update for below.
6907         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6908         Change type to hash_table.  Update dependent calls and types.
6910         * config/i386/t-cygming: Update for below.
6912         * config/i386/t-interix: Update for below.
6914         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6915         Change type to hash_table.  Update dependent calls and types.
6916         (i386_find_on_wrapper_list::wrappers): Likewise.
6918         * config/ia64/t-ia64: Update for below.
6920         * config/ia64/ia64.c (bundle_state_table):
6921         Change type to hash_table.  Update dependent calls and types.
6923         * config/mips/mips.c (mips_reorg_process_insns::htab):
6924         Change type to hash_table.  Update dependent calls and types.
6926         * config/sol2.c (solaris_comdat_htab):
6927         Change type to hash_table.  Update dependent calls and types.
6929         * config/t-sol2: Update for above.
6931 2013-05-29  Teresa Johnson  <tejohnson@google.com>
6933         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6934         functions are not yet marked as defined.
6936 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
6937             Pat Haugen <pthaugen@us.ibm.com>
6938             Peter Bergner <bergner@vnet.ibm.com>
6940         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6941         instructions.
6942         (VEC_A): Likewise.
6943         (VEC_C): Likewise.
6944         (vrotl<mode>3): Likewise.
6945         (vashl<mode>3): Likewise.
6946         (vlshr<mode>3): Likewise.
6947         (vashr<mode>3): Likewise.
6949         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6950         support for power8 V2DI builtins.
6952         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6953         power8 V2DI builtins.
6954         (vupkhsw): Likewise.
6955         (vupklsw): Likewise.
6956         (vaddudm): Likewise.
6957         (vminsd): Likewise.
6958         (vmaxsd): Likewise.
6959         (vminud): Likewise.
6960         (vmaxud): Likewise.
6961         (vpkudum): Likewise.
6962         (vpksdss): Likewise.
6963         (vpkudus): Likewise.
6964         (vpksdus): Likewise.
6965         (vrld): Likewise.
6966         (vsld): Likewise.
6967         (vsrd): Likewise.
6968         (vsrad): Likewise.
6969         (vsubudm): Likewise.
6970         (vcmpequd): Likewise.
6971         (vcmpgtsd): Likewise.
6972         (vcmpgtud): Likewise.
6973         (vcmpequd_p): Likewise.
6974         (vcmpgtsd_p): Likewise.
6975         (vcmpgtud_p): Likewise.
6976         (vupkhsw): Likewise.
6977         (vupklsw): Likewise.
6978         (vaddudm): Likewise.
6979         (vmaxsd): Likewise.
6980         (vmaxud): Likewise.
6981         (vminsd): Likewise.
6982         (vminud): Likewise.
6983         (vpksdss): Likewise.
6984         (vpksdus): Likewise.
6985         (vpkudum): Likewise.
6986         (vpkudus): Likewise.
6987         (vrld): Likewise.
6988         (vsld): Likewise.
6989         (vsrad): Likewise.
6990         (vsrd): Likewise.
6991         (vsubudm): Likewise.
6993         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6994         support for power8 V2DI instructions.
6996         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6997         power8 V2DI instructions.  Combine pack and unpack insns to use an
6998         iterator for each mode.  Check whether a particular mode supports
6999         Altivec instructions instead of just checking TARGET_ALTIVEC.
7000         (UNSPEC_VPKUWUM): Likewise.
7001         (UNSPEC_VPKSHSS): Likewise.
7002         (UNSPEC_VPKSWSS): Likewise.
7003         (UNSPEC_VPKUHUS): Likewise.
7004         (UNSPEC_VPKSHUS): Likewise.
7005         (UNSPEC_VPKUWUS): Likewise.
7006         (UNSPEC_VPKSWUS): Likewise.
7007         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7008         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7009         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7010         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7011         (UNSPEC_VUPKHSB): Likewise.
7012         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7013         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7014         (UNSPEC_VUPKHSH): Likewise.
7015         (UNSPEC_VUPKLSB): Likewise.
7016         (UNSPEC_VUPKLSH): Likewise.
7017         (VI2): Likewise.
7018         (VI_char): Likewise.
7019         (VI_scalar): Likewise.
7020         (VI_unit): Likewise.
7021         (VP): Likewise.
7022         (VP_small): Likewise.
7023         (VP_small_lc): Likewise.
7024         (VU_char): Likewise.
7025         (add<mode>3): Likewise.
7026         (altivec_vaddcuw): Likewise.
7027         (altivec_vaddu<VI_char>s): Likewise.
7028         (altivec_vadds<VI_char>s): Likewise.
7029         (sub<mode>3): Likewise.
7030         (altivec_vsubcuw): Likewise.
7031         (altivec_vsubu<VI_char>s): Likewise.
7032         (altivec_vsubs<VI_char>s): Likewise.
7033         (altivec_vavgs<VI_char>): Likewise.
7034         (altivec_vcmpbfp): Likewise.
7035         (altivec_eq<mode>): Likewise.
7036         (altivec_gt<mode>): Likewise.
7037         (altivec_gtu<mode>): Likewise.
7038         (umax<mode>3): Likewise.
7039         (smax<mode>3): Likewise.
7040         (umin<mode>3): Likewise.
7041         (smin<mode>3): Likewise.
7042         (altivec_vpkuhum): Likewise.
7043         (altivec_vpkuwum): Likewise.
7044         (altivec_vpkshss): Likewise.
7045         (altivec_vpkswss): Likewise.
7046         (altivec_vpkuhus): Likewise.
7047         (altivec_vpkshus): Likewise.
7048         (altivec_vpkuwus): Likewise.
7049         (altivec_vpkswus): Likewise.
7050         (altivec_vpks<VI_char>ss): Likewise.
7051         (altivec_vpks<VI_char>us): Likewise.
7052         (altivec_vpku<VI_char>us): Likewise.
7053         (altivec_vpku<VI_char>um): Likewise.
7054         (altivec_vrl<VI_char>): Likewise.
7055         (altivec_vsl<VI_char>): Likewise.
7056         (altivec_vsr<VI_char>): Likewise.
7057         (altivec_vsra<VI_char>): Likewise.
7058         (altivec_vsldoi_<mode>): Likewise.
7059         (altivec_vupkhsb): Likewise.
7060         (altivec_vupkhs<VU_char>): Likewise.
7061         (altivec_vupkls<VU_char>): Likewise.
7062         (altivec_vupkhsh): Likewise.
7063         (altivec_vupklsb): Likewise.
7064         (altivec_vupklsh): Likewise.
7065         (altivec_vcmpequ<VI_char>_p): Likewise.
7066         (altivec_vcmpgts<VI_char>_p): Likewise.
7067         (altivec_vcmpgtu<VI_char>_p): Likewise.
7068         (abs<mode>2): Likewise.
7069         (vec_unpacks_hi_v16qi): Likewise.
7070         (vec_unpacks_hi_v8hi): Likewise.
7071         (vec_unpacks_lo_v16qi): Likewise.
7072         (vec_unpacks_hi_<VP_small_lc>): Likewise.
7073         (vec_unpacks_lo_v8hi): Likewise.
7074         (vec_unpacks_lo_<VP_small_lc>): Likewise.
7075         (vec_pack_trunc_v8h): Likewise.
7076         (vec_pack_trunc_v4si): Likewise.
7077         (vec_pack_trunc_<mode>): Likewise.
7079         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7080         V2DI builtins.
7081         (vec_vmaxsd): Likewise.
7082         (vec_vmaxud): Likewise.
7083         (vec_vminsd): Likewise.
7084         (vec_vminud): Likewise.
7085         (vec_vpksdss): Likewise.
7086         (vec_vpksdus): Likewise.
7087         (vec_vpkudum): Likewise.
7088         (vec_vpkudus): Likewise.
7089         (vec_vrld): Likewise.
7090         (vec_vsld): Likewise.
7091         (vec_vsrad): Likewise.
7092         (vec_vsrd): Likewise.
7093         (vec_vsubudm): Likewise.
7094         (vec_vupkhsw): Likewise.
7095         (vec_vupklsw): Likewise.
7097 2013-05-29  Jan Hubicka  <jh@suse.cz>
7099         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7100         flags; reorder rest of fields in more consistent way.
7101         (varpool_node): Remove analyzed, finalized and alias.
7102         (cgraph_ndoe): Likewise.
7103         (symtab_alias_ultimate_target): New function.
7104         (cgraph_function_node): Move offline.
7105         (cgraph_reset_node): Declare.
7106         (cgraph_comdat_can_be_unshared_p): Remove.
7107         (varpool_remove_initializer): Declare.
7108         (varpool_first_defined_variable, varpool_next_defined_variable
7109         cgraph_first_defined_function, cgraph_next_defined_function): Update.
7110         (cgraph_function_with_gimple_body_p): Update.
7111         (varpool_all_refs_explicit_p): Update.
7112         (symtab_alias_target): New function.
7113         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7114         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7115         (cgraph_function_or_thunk_node): Simplify using
7116         symtab_alias_ultimate_target.
7117         (varpool_variable_node): Likewise.
7118         * cgraph.c (cgraph_create_function_alias): Update.
7119         (cgraph_add_thunk): Update.
7120         (cgraph_remove_node): Update.
7121         (dump_cgraph_node): Do not dump removed flags.
7122         (cgraph_function_body_availability): Update.
7123         (cgraph_propagate_frequency): Update.
7124         (verify_cgraph_node): Check sanity of local flag.
7125         (cgraph_function_node): Move here from cgraph.h; revamp for
7126         cgraph_function_or_thunk_node.
7127         * lto-symtab.c (lto_varpool_replace_node): Update.
7128         (lto_symtab_resolve_can_prevail_p): Update.
7129         (lto_symtab_merge_cgraph_nodes): Update.
7130         * ipa-cp.c (determine_versionability, initialize_node_lattices,
7131         propagate_constants_accross_call, devirtualization_time_bonus,
7132         ipcp_propagate_stage): Update.
7133         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7134         * ipa-inline-transform.c (clone_inlined_nodes,
7135         preserve_function_body_p): Update.
7136         * ipa-reference.c (propagate): Update.
7137         (write_node_summary_p): Update.
7138         * toplev.c (wrapup_global_declaration_2): Update.
7139         * cgraphunit.c (cgraph_analyze_function): Rename to ...
7140         (analyze_function) ... this one.
7141         (cgraph_process_new_functions): Update.
7142         (cgraph_reset_node): Export.
7143         (cgraph_finalize_function): Update.
7144         (cgraph_add_new_function): Update.
7145         (process_function_and_variable_attributes): Update.
7146         (varpool_finalize_decl): Update.
7147         (symbol_finalized): Remove.
7148         (symbol_finalized_and_needed): Rename to ...
7149         (symbol_defined_and_needed): ... update.
7150         (cgraph_analyze_functions): Update.
7151         (handle_alias_pairs): Update.
7152         (mark_functions_to_output): Update.
7153         (assemble_thunk): Update.
7154         (output_in_order): Update.
7155         (output_weakrefs): Update.
7156         (finalize_compilation_unit): Update.
7157         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7158         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7159         input_node, input_varpool_node): Update.
7160         * dbxout.c (dbxout_expand_expr): Update.
7161         * cgraphclones.c (cgraph_clone_node): Update.
7162         (cgraph_copy_node_for_versioning): Update.
7163         (cgraph_materialize_clone): Update.
7164         (cgraph_materialize_all_clones): Update.
7165         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7166         propagate_pure_const, propagate_nothrow): Update.
7167         * lto-streamer-out.c (lto_output, write_symbol): Update.
7168         * ipa-utils.c (ipa_reverse_postorder): Update.
7169         * ipa-inline.c (can_inline_edge_p): Update.
7170         (update_caller_keys, ipa_inline): Update.
7171         * dwarf2out.c (reference_to_unused,
7172         premark_types_used_by_global_vars_helper): Update.
7173         * tree-eh.c (tree_could_trap_p): Update.
7174         * ipa-split.c (consider_split, execute_split_functions): Update.
7175         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7176         has_addr_references_p): Update; move ahead in file for better
7177         readability.
7178         (process_references): Simplify.
7179         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7180         bodies are removed.
7181         (cgraph_comdat_can_be_unshared_p): Make static.
7182         (cgraph_externally_visible_p): Update.
7183         (varpool_externally_visible_p): Update.
7184         (function_and_variable_visibility): Update.
7185         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7186         ipa_tm_mark_force_output_node): Update.
7187         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7188         estimate_edge_devirt_benefit, inline_generate_summary,
7189         inline_write_summary): Update.
7190         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7191         * ipa-prop.c (ipa_compute_jump_functions): Update.
7192         (ipa_print_node_params, ipa_prop_read_section,
7193         ipa_update_after_lto_read, read_replacements_section): Update.
7194         * varasm.c (mark_decl_referenced): Update.
7195         (assemble_alias, dump_tm_clone_pairs): Update.
7196         * tree-inline.c (copy_bb): Update.
7197         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7198         Update.
7199         * symtab.c (dump_symtab_base): Print new flags.
7200         (verify_symtab_base): Verify new flags.
7201         (symtab_alias_ultimate_target): New function.
7202         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7203         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7204         Update.
7205         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7206         Update.
7207         * i386.c (ix86_get_function_versions_dispatcher,
7208         ix86_generate_version_dispatcher_body): Update.
7209         (fold_builtin_cpu): Use varpool_add_new_variable.
7210         * varpool.c (varpool_remove_initializer): Break out from ...
7211         (varpool_remove_node): ... this one.
7212         (dump_varpool_node, varpool_node_for_asm,
7213         cgraph_variable_initializer_availability, varpool_analyze_node,
7214         varpool_assemble_decl, varpool_remove_unreferenced_decls,
7215         varpool_finalize_named_section_flags, varpool_create_variable_alias):
7216         Update.
7218 2013-05-29  Jan Hubicka  <jh@suse.cz>
7220         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7222 2013-05-29  Easwaran Raman  <eraman@google.com>
7224         PR tree-optimization/57442
7225         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7226         when control exits the main loop.
7228 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
7230         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7231         and RX600.
7232         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7233         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7234         * rx/t-rx: Add rx100 under multi library matches option for nofpu
7235         option.
7237 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7239         PR tree-optimization/57441
7240         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7241         Don't limit size of incr_vec to number of candidates.
7243 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
7245         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7246         and mips16 directories.
7247         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7248         (MULTILIB_DIRNAMES): Ditto.
7249         (MULTILIB_EXCEPTIONS): Add new exceptions.
7250         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7251         (MULTILIB_DIRNAMES): Ditto.
7252         (MULTILIB_EXCEPTIONS): Add new exceptions.
7254 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7255             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7257         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7258         SYMBOL_TINY_ABSOLUTE.
7259         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7260         SYMBOL_TINY_ABSOLUTE.
7261         (aarch64_expand_mov_immediate): Likewise.
7262         (aarch64_classify_symbol): Likewise.
7263         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7264         Permit SYMBOL_TINY_ABSOLUTE.
7265         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7267 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7268             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7270         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7271         Refactor if/switch.  Replace gcc_assert with if.
7273 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7275         * config/i386/i386.c (initial_ix86_tune_features): Enable
7276         FP Reassociation for AMD bdver1 and bdver2.
7278 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7280         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7281         and IMAGPART_EXPR do not occur within other handled_components.
7283 2013-05-29  Richard Biener  <rguenther@suse.de>
7285         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7286         access on whether the use is in the BB we currently try to
7287         vectorize.
7288         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7289         vectorize to vect_bb_slp_scalar_cost.
7291 2013-05-29  Richard Biener  <rguenther@suse.de>
7293         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7294         computing scalar cost offsetted by stmts that are kept live
7295         by scalar uses.
7296         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7297         for computation of scalar cost.
7299 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7301         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7303 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7305         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7306         array notation built-in reduction functions.
7307         * doc/passes.texi (Passes): Added documentation about changes done
7308         for Cilk Plus.
7309         * doc/invoke.texi (C Dialect Options): Added documentation about
7310         the -fcilkplus flag.
7311         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7312         (BUILTINS_DEF): Depend on cilkplus.def.
7313         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7314         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7315         * cilkplus.def: New file.
7317 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7319         PR rtl-optimization/57439
7320         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7322 2013-05-28  Easwaran Raman  <eraman@google.com>
7324         PR tree-optimization/57337
7325         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7326         (find_insert_point): Correctly identify the insertion point
7327         when two statements with the same UID is compared.
7329 2013-05-28  Richard Biener  <rguenther@suse.de>
7331         PR tree-optimization/56787
7332         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7333         from the list of data references.
7334         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7335         clobbers.
7336         (vect_analyze_loop_operations): Likewise.
7337         (vect_transform_loop): Remove clobbers.
7339 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7341         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7342         and REALPART_EXPRs have scalar type.
7344 2013-05-28  Richard Biener  <rguenther@suse.de>
7346         PR tree-optimization/57411
7347         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7348         virtual operands.
7349         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7350         virtual operand propagation.
7352 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7354         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7355         destination register for bmasksi_vis.
7356         (vector_init_bshuffle): Likewise.
7357         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7359 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7361         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7362         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7363         mode if the instruction isn't available in the original mode.
7364         * config/sparc/sparc.opt (mfix-ut699): New option.
7365         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7366         (divdf3): Turn into expander.
7367         (divdf3_nofix): New insn.
7368         (divdf3_fix): Likewise.
7369         (divsf3): Disable if -mfix-ut699.
7370         (sqrtdf2): Turn into expander.
7371         (sqrtdf2_nofix): New insn.
7372         (sqrtdf2_fix): Likewise.
7373         (sqrtsf2): Disable if -mfix-ut699.
7375 2013-05-27  Richard Biener  <rguenther@suse.de>
7377         PR middle-end/57412
7378         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7379         block for the new loop.
7381 2013-05-27  Richard Biener  <rguenther@suse.de>
7383         PR tree-optimization/57343
7384         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7385         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7386         (number_of_iterations_cond): Do not build the folded tree.
7388 2013-05-27  Richard Biener  <rguenther@suse.de>
7390         Revert
7391         PR middle-end/57381
7392         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7393         OEP_CONSTANT_ADDRESS_OF retained.
7395         PR tree-optimization/57417
7396         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7397         for unchanged base.
7398         (set_ssa_val_to): Compare addresses using
7399         get_addr_base_and_unit_offset.
7401 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7403         PR rtl-optimization/56833
7404         * postreload.c (move2add_record_mode): New function.
7405         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7406         (move2add_use_add2_insn): Use move2add_record_sym_value.
7407         (move2add_use_add3_insn): Likewise.
7408         (reload_cse_move2add): Use move2add_valid_value_p and
7409         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7410         affected regs by setting reg_mode to VOIDmode.
7411         (move2add_note_store): Don't pretend the inside of a SUBREG is
7412         the actual destination.  Invalidate single/leading registers by
7413         setting reg_mode to VOIDmode.
7414         Use move2add_record_sym_value, move2add_valid_value_p and
7415         move2add_record_mode.
7417 2013-05-27  Richard Biener  <rguenther@suse.de>
7419         PR tree-optimization/57396
7420         * tree-affine.c (double_int_constant_multiple_p): Properly
7421         return false for val == 0 and div != 0.
7423 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7425         * config/mips/mips.h: Use #elif in preprocessor conditions.
7427 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7429         PR target/53916
7430         * config/mips/constraints.md (kl): New constraint.
7431         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7432         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7433         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7434         from LO for MIPS16.
7435         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7437 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7439         PR target/55777
7440         * config/mips/mips.c (mips_can_inline_p): New function.
7441         (TARGET_CAN_INLINE_P): Define.
7443 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7445         * sched-int.h (ds_t, dw_t): Make unsigned int.
7446         Fix documentation that describes how all the ds_t bits are used.
7447         Reserve the last bit for delayed-branch scheduling.
7448         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7449         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7450         (gen_dep_weak_1): Remove prototype.
7451         * sched-deps.c (get_dep_weak_1): Make static.
7452         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7453         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7454         * doc/tm.texi: Regenerate.
7455         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7457 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7459         PR debug/56950
7460         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7462 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7463             Sandra Loosemore  <sandra@codesourcery.com>
7465         * config.gcc (powerpc-*): Allow native for with-cpu.
7467 2013-05-24  Jeff Law  <law@redhat.com>
7469         PR tree-optimization/57124
7470         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7471         conversion feeding a condition if the range has an overflow
7472         if -fstrict-overflow.  Add warnings for when we do make the
7473         transformation.
7475 2013-05-24  Dehao Chen  <dehao@google.com>
7477         * tree-cfg.c (locus_discrim_map): Fix the typo.
7478         (locus_discrim_hasher): Likewise.
7479         (locus_discrim_hasher::hash): Likewise.
7480         (locus_discrim_hasher::equal): Likewise.
7482 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7484         PR tree-optimization/57294
7485         * cgraph.h (ipa_record_stmt_references): Declare.
7486         * cgraphbuild.c (ipa_record_stmt_references): New function.
7487         (build_cgraph_edges): Use ipa_record_stmt_references.
7488         (rebuild_cgraph_edges): Likewise.
7489         (cgraph_rebuild_references): Likewise.
7490         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7491         associated with the old statement and build references from the
7492         newly built statements.
7493         * ipa-ref.c (ipa_remove_stmt_references): New function.
7494         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7496 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7498         * lra-constraints.c (emit_spill_move): Use smaller mode for
7499         mem-mem moves.
7500         (check_and_process_move): Consider mem-reg moves for secondary
7501         too.
7502         (curr_insn_transform): Don't lose insns emitted before for
7503         secondary memory moves.
7504         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7505         reg set up in the current insn.
7507 2013-05-24  Dehao Chen  <dehao@google.com>
7509         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7510         hash function.
7511         (locus_descrim_hasher::equal): Likewise.
7512         (build_gimple_cfg): New discriminator assignment algorithm.
7513         (make_edges): Likewise.
7514         (next_discriminator_for_locus): Likewise.
7515         (same_line_p): Likewise.
7516         (assign_discriminators): Likewise.
7517         (make_cond_expr_edges): Likewise.
7518         (make_gimple_switch_edges): Likewise.
7519         (make_goto_expr_edges): Likewise.
7520         (make_gimple_asm_edges): Likewise.
7522 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
7524         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7525         X format specifier to only display bottom 16 bits.
7526         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7527         immediate to match for operand 2, since it will be masked.
7529 2013-05-24  Richard Biener  <rguenther@suse.de>
7531         PR tree-optimization/57287
7532         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7533         all SSA names that occur in abnormal PHIs.
7535 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7537         PR tree-ssa/57385
7538         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7539         that index is not negative.
7541 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
7543         PR rtl-optimization/55177
7544         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7545         (simplify_byte_swapping_operation): New.
7546         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7547         (simplify_relational_operation_1): Deal with BSWAP.
7549 2013-05-23  Richard Henderson  <rth@redhat.com>
7551         PR target/56742
7552         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7553         (ix86_reorg): Call it.
7555 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
7557         PR target/57379
7558         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7559         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7560         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7562 2013-05-23  Christian Bruel  <christian.bruel@st.com>
7564         PR debug/57351
7565         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7567 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7568             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7570         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7571         * config/aarch64/constraints.md (Usa): Remove.
7572         * doc/md.texi (AArch64 Usa): Remove.
7574 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7575             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7577         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7578         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7579         * config/aarch64/predicates.md (aarch64_const_address): Remove.
7580         (aarch64_mov_operand): Use aarch64_mov_operand_p.
7582 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
7584         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7585         instruction (AdvSIMD).
7586         * config/aarch64/aarch64-builtins.c
7587         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7588         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7590 2013-05-23  Martin Jambor  <mjambor@suse.cz>
7592         PR middle-end/57347
7593         * tree.h (contains_bitfld_component_ref_p): Declare.
7594         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7595         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
7596         caller.
7597         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7598         not access a bit-field.  Assert all final offsets are byte-aligned.
7600 2013-05-23  Richard Biener  <rguenther@suse.de>
7602         PR tree-optimization/57380
7603         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7604         least one invariant or re-used load.
7605         * passes.c (init_optimization_passes): Move pass_phiprop before
7606         pass_forwprop.
7608 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
7610         * config/aarch64/aarch64-simd.md
7611         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7613 2013-05-23  Richard Biener  <rguenther@suse.de>
7615         PR middle-end/57381
7616         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7617         OEP_CONSTANT_ADDRESS_OF retained.
7619 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
7621         PR middle-end/57344
7622         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7623         don't lower unit.  Handle unit not being always BITS_PER_WORD.
7625 2013-05-23  Richard Biener  <rguenther@suse.de>
7627         PR rtl-optimization/57341
7628         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7629         instead of true_dependence.
7631 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
7633         * bb-reorder.c (branch_threshold): Make const.
7634         (exec_threshold): Ditto.
7636 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7637             Pat Haugen <pthaugen@us.ibm.com>
7638             Peter Bergner <bergner@vnet.ibm.com>
7640         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7641         documentation for the power8 crypto builtins.
7643         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7645         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7646         macros for defining power8 builtin functions.
7647         (BU_P8V_AV_2): Likewise.
7648         (BU_P8V_AV_P): Likewise.
7649         (BU_P8V_VSX_1): Likewise.
7650         (BU_P8V_OVERLOAD_1): Likewise.
7651         (BU_P8V_OVERLOAD_2): Likewise.
7652         (BU_CRYPTO_1): Likewise.
7653         (BU_CRYPTO_2): Likewise.
7654         (BU_CRYPTO_3): Likewise.
7655         (BU_CRYPTO_OVERLOAD_1): Likewise.
7656         (BU_CRYPTO_OVERLOAD_2): Likewise.
7657         (XSCVSPDP): Fix typo, point to the correct instruction.
7658         (VCIPHER): Add power8 crypto builtins.
7659         (VCIPHERLAST): Likewise.
7660         (VNCIPHER): Likewise.
7661         (VNCIPHERLAST): Likewise.
7662         (VPMSUMB): Likewise.
7663         (VPMSUMH): Likewise.
7664         (VPMSUMW): Likewise.
7665         (VPERMXOR_V2DI): Likewise.
7666         (VPERMXOR_V4SI: Likewise.
7667         (VPERMXOR_V8HI: Likewise.
7668         (VPERMXOR_V16QI: Likewise.
7669         (VSHASIGMAW): Likewise.
7670         (VSHASIGMAD): Likewise.
7671         (VPMSUM): Likewise.
7672         (VPERMXOR): Likewise.
7673         (VSHASIGMA): Likewise.
7675         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7676         __CRYPTO__ if the crypto instructions are available.
7677         (altivec_overloaded_builtins): Add support for overloaded power8
7678         builtins.
7680         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7681         support for power8 crypto builtins.
7682         (builtin_function_type): Likewise.
7683         (altivec_init_builtins): Add support for builtins that take vector
7684         long long (V2DI) arguments.
7686         * config/rs6000/crypto.md: New file, define power8 crypto
7687         instructions.
7689 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7690             Pat Haugen <pthaugen@us.ibm.com>
7691             Peter Bergner <bergner@vnet.ibm.com>
7693         * doc/invoke.texi (Option Summary): Add power8 options.
7694         (RS/6000 and PowerPC Options): Likewise.
7696         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7697         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
7698         wm, wn, wr documentation.
7700         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7701         registers if direct move instructions are enabled.
7702         (wn): New constraint for no registers.
7703         (wq): New constraint for quad word even GPR registers.
7704         (wr): New constraint if 64-bit instructions are enabled.
7705         (wv): New constraint if power8 vector instructions are enabled.
7706         (wQ): New constraint for quad word memory locations.
7708         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7709         constraint for 0..15 for crypto instructions.
7710         (gpc_reg_operand): If VSX allow registers in VSX registers as well
7711         as GPR and floating point registers.
7712         (int_reg_operand): New predicate to match only GPR registers.
7713         (base_reg_operand): New predicate to match base registers.
7714         (quad_int_reg_operand): New predicate to match even GPR registers
7715         for quad memory operations.
7716         (vsx_reg_or_cint_operand): New predicate to allow vector logical
7717         operations in both GPR and VSX registers.
7718         (quad_memory_operand): New predicate for quad memory operations.
7719         (reg_or_indexed_operand): New predicate for direct move support.
7721         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7722         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7723         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7724         (POWERPC_MASKS): Add power8 options.
7725         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7726         various options.
7728         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7729         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7731         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7732         (-mpower8-fusion): New power8 options.
7733         (-mpower8-fusion-sign): Likewise.
7734         (-mpower8-vector): Likewise.
7735         (-mcrypto): Likewise.
7736         (-mdirect-move): Likewise.
7737         (-mquad-memory): Likewise.
7739         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7740         power8.
7741         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7742         registers.
7743         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7744         (rs6000_debug_vector_unit): Add p8_vector.
7745         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7746         definitions.  Also print fusion state.
7747         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7748         (rs6000_builtin_mask_calculate): Add power8 builtin support.
7749         (rs6000_option_override_internal): Add support for power8.
7750         (rs6000_common_init_builtins): Add debugging for skipped builtins
7751         if -mdebug=builtin.
7752         (rs6000_adjust_cost): Add power8 support.
7753         (rs6000_issue_rate): Likewise.
7754         (insn_must_be_first_in_group): Likewise.
7755         (insn_must_be_last_in_group): Likewise.
7756         (force_new_group): Likewise.
7757         (rs6000_register_move_cost): Likewise.
7758         (rs6000_opt_masks): Likewise.
7760         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7761         power8 capable assembler, default to power7 options.
7762         (TARGET_DIRECT_MOVE): Likewise.
7763         (TARGET_CRYPTO): Likewise.
7764         (TARGET_P8_VECTOR): Likewise.
7765         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7766         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7767         (VECTOR_MEM_P8_VECTOR_P): Likewise.
7768         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7769         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7770         (TARGET_XSCVDPSPN): Likewise.
7771         (TARGET_XSCVSPDPN): Likewsie.
7772         (TARGET_SYNC_HI_QI): Likewise.
7773         (TARGET_SYNC_TI): Likewise.
7774         (MASK_CRYPTO): Likewise.
7775         (MASK_DIRECT_MOVE): Likewise.
7776         (MASK_P8_FUSION): Likewise.
7777         (MASK_P8_VECTOR): Likewise.
7778         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7779         temporary used by some of the direct move instructions to get two FP
7780         temporary registers does not force creation of a stack frame.
7781         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7782         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7783         that any VSX registers are tieable, even if they are also an
7784         Altivec vector mode.
7785         (r6000_reg_class_enum): Add wm, wr, wv constraints.
7786         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7787         (RS6000_BTM_CRYPTO): Likewise.
7788         (RS6000_BTM_COMMON): Likewise.
7790         * config/rs6000/rs6000.md (cpu attribute): Add power8.
7791         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7792         (enum rs6000_vector): Add power8 vector support.
7794 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7796         PR target/19599
7797         PR target/57340
7798         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7799         (any_sibcall_could_use_r3): this and handle indirect calls.
7800         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7802 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7804         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7806 2013-05-22  Richard Biener  <rguenther@suse.de>
7808         PR middle-end/57349
7809         * profile.c (branch_prob): Do not split blocks that are
7810         abnormally receiving from ECF_RETURNS_TWICE functions.
7812 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7814         * recog.c (offsettable_address_addr_space_p): Fix calculation of
7815         address mode.  Move pointer mode initialization to the same place.
7817 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7819         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7820         while it has any effect.
7822 2013-05-21  Easwaran Raman  <eraman@google.com>
7824         PR tree-optimization/57322
7825         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7826         UID of the statement added to the BB to be 1.
7828 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
7830         PR tree-optimization/57331
7831         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7832         of conversion from pointer type to integral type with integer.
7834 2013-05-21  Martin Jambor  <mjambor@suse.cz>
7836         PR lto/57289
7837         * ipa-prop.c (ipa_read_node_info): Process param_used and
7838         controlled_uses in the same order as when writing.
7840 2013-05-21  Magnus Granberg  <baldrick@free.fr>
7842         PR plugins/56754
7843         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7845 2013-05-21  Richard Biener  <rguenther@suse.de>
7847         PR tree-optimization/57318
7848         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7849         estimate stmts with side-effects as likely eliminated.
7851 2013-05-21  Richard Biener  <rguenther@suse.de>
7853         PR tree-optimization/57330
7854         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7855         preserve the call stmts fntype.
7857 2013-05-21  Richard Biener  <rguenther@suse.de>
7859         PR tree-optimization/57303
7860         * tree-ssa-sink.c (statement_sink_location): Improve killing
7861         stmt detection and properly handle self-assignments.
7863 2013-05-21  Christian Bruel  <christian.bruel@st.com>
7865         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7866         spanning registers. LEAF_REG_REMAP is supported only for contiguous
7867         registers. Set register size out of the PARALLEL loop.
7869 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
7871         PR target/56547
7872         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7873         (*fmasf4, *fmasf4_media): New insns.
7875 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7877         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7878         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7879         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7880         (mips_idiv_insns): Update the comments to say that the returned
7881         instruction counts are in units of BASE_INSN_LENGTH.
7882         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7883         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
7884         using 2 rather than 4 as the length of indirect MIPS16 and
7885         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
7886         length of a NOP.  Don't divide MIPS16 lengths by 2.
7887         (mips16_split_long_branches): Assume a branch is long if the
7888         length is greater than 4 rather than 8.
7889         * config/mips/mips.md (length): Give MIPS16 lengths directly,
7890         rather than multiplying them by 2.  Multiply instruction counts
7891         by BASE_INSN_LENGTH rather than 4.
7892         (*jump_mips16, tls_get_tp_mips16_<mode>)
7893         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7895 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7897         * config/mips/mips.md (extended_mips16): Remove branch case.
7898         (length): Remove duplicated extended_mips16 test.
7900 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7902         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7904 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
7906         * recog.h (Recog_data): Rename to...
7907         (recog_data_d): ...this.
7908         (recog_data): Update accordingly.
7909         * recog.c (recog_data): Likewise.
7910         * reload.c (save_recog_data): Likewise.
7911         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7912         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7914 2013-05-17  Julian Brown  <julian@codesourcery.com>
7916         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7917         found in a REG_EQUAL note, invalidate it.
7919 2013-05-17  Easwaran Raman  <eraman@google.com>
7921         * tree-ssa-reassoc.c (find_insert_point): New function.
7922         (insert_stmt_after): Likewise.
7923         (get_def_stmt): Likewise.
7924         (ensure_ops_are_available): Likewise.
7925         (not_dominated_by): Likewise.
7926         (rewrite_expr_tree): Do not move statements beyond what is
7927         necessary. Remove call to swap_ops_for_binary_stmt...
7928         (reassociate_bb): ... and move it here.
7929         (build_and_add_sum): Assign UIDs for new statements.
7930         (linearize_expr): Likewise.
7931         (do_reassoc): Renumber gimple statement UIDs.
7933 2013-05-17  Jan Hubicka  <jh@suse.cz>
7935         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7936         weakrefs.
7937         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7938         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7939         weakrefs.
7940         (output_weakrefs): Update.
7942 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
7943             Martin Jambor  <mjambor@suse.cz>
7945         PR middle-end/57276
7946         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7947         value that corresponds to the given aggval is found in values vector.
7949 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
7951         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7952         sse, sse2, sse3, ssse3 and sse4a flags to options.
7954 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
7956         * gengtype-state.c: (s_expr_writer): New class, to handle
7957         prettifying of output layout of s-expressions.
7958         (state_writer): New class, to write out gtype.state.
7959         (state_written_type_count): Move this variable into member data of
7960         state_writer.
7961         (s_expr_writer::s_expr_writer): New code: constructor for new class
7962         (state_writer::state_writer(): ditto
7963         (s_expr_writer::write_new_line): New function
7964         (s_expr_writer::write_any_indent): ditto
7965         (s_expr_writer::begin_s_expr): ditto
7966         (s_expr_writer::end_s_expr): ditto
7967         (write_state_fileloc): convert to method of state_writer...
7968         (state_writer:: write_state_fileloc): ...and use methods of
7969         s_expr_writer to write indentation into the gtype.state output file
7970         to visually represent the hierarchical structure of the list
7971         structures
7972         (write_state_fields): ditto, renaming to...
7973         (state_writer::write_state_fields)
7974         (write_state_a_string): ditto, renaming to...
7975         (state_writer::write_state_a_string)
7976         (write_state_string_option): ditto, renaming to...
7977         (state_writer::write_state_string_option)
7978         (write_state_type_option): ditto, renaming to...
7979         (state_writer::write_state_type_option)
7980         (write_state_nested_option): ditto, renaming to...
7981         (state_writer::write_state_nested_option)
7982         (write_state_option): ditto, renaming to...
7983         (state_writer::write_state_option)
7984         (write_state_options): ditto, renaming to...
7985         (state_writer::write_state_options)
7986         (write_state_lang_bitmap): ditto, renaming to...
7987         (state_writer::write_state_lang_bitmap)
7988         (write_state_version): ditto, renaming to...
7989         (state_writer::write_state_version)
7990         (write_state_scalar_type): ditto, renaming to...
7991         (state_writer::write_state_scalar_type)
7992         (write_state_string_type): ditto, renaming to...
7993         (state_writer::write_state_string_type)
7994         (write_state_undefined_type): ditto, renaming to...
7995         (state_writer::write_state_undefined_type)
7996         (write_state_struct_union_type): ditto, renaming to...
7997         (state_writer::write_state_struct_union_type)
7998         (write_state_struct_type): ditto, renaming to...
7999         (state_writer::write_state_struct_type)
8000         (write_state_user_struct_type): ditto, renaming to...
8001         (state_writer::write_state_user_struct_type)
8002         (write_state_lang_struct_type): ditto, renaming to...
8003         (state_writer::write_state_lang_struct_type)
8004         (write_state_param_struct_type): ditto, renaming to...
8005         (state_writer::write_state_param_struct_type)
8006         (write_state_pointer_type): ditto, renaming to...
8007         (state_writer::write_state_pointer_type)
8008         (write_state_array_type): ditto, renaming to...
8009         (state_writer::write_state_array_type)
8010         (write_state_gc_used): ditto, renaming to...
8011         (state_writer::write_state_gc_used)
8012         (write_state_common_type_content): ditto, renaming to...
8013         (state_writer::write_state_common_type_content)
8014         (write_state_type): ditto, renaming to...
8015         (state_writer::write_state_type)
8016         (write_state_pair_list): ditto, renaming to...
8017         (state_writer::write_state_pair_list)
8018         (write_state_pair): ditto, renaming to...
8019         (state_writer::write_state_pair)
8020         (write_state_typedefs): ditto, renaming to...
8021         (state_writer::write_state_typedefs)
8022         (write_state_structures): ditto, renaming to...
8023         (state_writer::write_state_structures)
8024         (write_state_param_structs): ditto, renaming to...
8025         (state_writer::write_state_param_structs)
8026         (write_state_variables): ditto, renaming to...
8027         (state_writer::write_state_variables)
8028         (write_state_srcdir): ditto, renaming to...
8029         (state_writer::write_state_srcdir)
8030         (write_state_files_list): ditto, renaming to...
8031         (state_writer::write_state_files_list)
8032         (write_state_languages): ditto, renaming to...
8033         (state_writer::write_state_languages)
8034         (write_state): create a state_writer instance and use it when
8035         writing out the state file
8037 2013-05-17  Mike Stump  <mikestump@comcast.net>
8039         PR rtl-optimization/57304
8040         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8041         accessing DF_REF_REAL_LOC.
8043 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8045         PR rtl-optimization/57281
8046         PR rtl-optimization/57300
8047         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8048         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8049         what the other splitter did if the registers are dead.
8051 2013-05-17  Richard Biener  <rguenther@suse.de>
8053         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8054         MEM_REF offsets.
8056 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8058         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8059         linking.
8061 2013-05-17  Marek Polacek  <polacek@redhat.com>
8063         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8064         length when doing non-zero store of storing '\0' to '\0'.
8066 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8068         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8069         vect_external_def oprnd1 with loop_vinfo, try to emit
8070         optional cast, negation and and stmts on the loop preheader
8071         edge instead of into the pattern def seq.
8073         PR tree-optimization/57051
8074         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8075         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8077 2013-05-16  Nick Clifton  <nickc@redhat.com>
8079         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8080         (rl78_is_naked_func): New function.
8081         (rl78_expand_prologue): Skip prologue generation for naked functions.
8082         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8083         * doc/extend.texi (naked): Add RL78 to the list of processors
8084         that supports this attribute.
8086 2013-05-16  Jeff Law  <law@redhat.com>
8088         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8090 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8092         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8093         cache parameters using detect_caches_amd also for CYRIX,
8094         NSC and TM2 signatures.
8096 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8097             Dzianis Kahanovich  <mahatma@eu.by>
8099         PR target/45359
8100         PR target/46396
8101         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8102         VIA/Centaur processors and determine their cache parameters
8103         using detect_caches_amd.
8105 2013-05-16  Teresa Johnson  <tejohnson@google.com>
8107         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8108         (rtl_verify_edges): New function.
8109         (rtl_verify_bb_insns): Ditto.
8110         (rtl_verify_bb_pointers): Ditto.
8111         (rtl_verify_bb_insn_chain): Ditto.
8112         (rtl_verify_fallthru): Ditto.
8113         (rtl_verify_bb_layout): Ditto.
8114         (rtl_verify_flow_info_1): Outline checks into new functions.
8115         (rtl_verify_flow_info): Ditto.
8117 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
8119         * cfghooks.c (copy_bbs): Add update_dominance argument.
8120         * cfghooks.h (copy_bbs): Update prototype.
8121         * tree-cfg.c (gimple_duplicate_sese_region):
8122         Add update_dominance argument.
8123         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8124         * tree-ssa-loop-ch.c (copy_loop_headers): Update
8125         gimple_duplicate_sese_region call.
8126         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8127         Update copy_bbs call.
8128         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8129         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8131 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8133         * tree-vectorizer.h (NUM_PATTERNS): Increment.
8134         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8135         vect_recog_rotate_pattern.
8136         (vect_recog_rotate_pattern): New function.
8138 2013-05-16  Jason Merrill  <jason@redhat.com>
8140         * Makefile.in (LLINKER): New variable.
8141         (mostlyclean): Remove link mutex.
8142         * configure.ac: Handle --enable-link-mutex.
8143         * lock-and-run.sh: New script.
8145 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8147         PR target/19599
8148         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8149         for NULL decl.
8151 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8153         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8155 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
8157         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8158         * config/arm/arm.c (next_consecutive_mem): New function.
8159         (gen_movmem_ldrd_strd): Likewise.
8160         * config/arm/arm.md (movmemqi): Update condition and code.
8161         (unaligned_loaddi, unaligned_storedi): New patterns.
8163 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8165         * config.gcc: Obsolete *-*-solaris2.9*.
8166         * doc/install.texi (Specific, *-*-solaris2*): Document it.
8168 2013-05-16  Richard Biener  <rguenther@suse.de>
8170         * passes.c (init_optimization_passes): Move pass_parallelize_loops
8171         earlier, after GRAPHITE transforms and IV canonicalization.
8173 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8175         * omp-low.c (extract_omp_for_data): For collapsed loops,
8176         if at least one of the loops is known at compile time to
8177         iterate zero times, set count to 0.
8178         (expand_omp_regimplify_p): New function.
8179         (expand_omp_for_generic): For collapsed loops, if at least
8180         one of the loops isn't known to iterate at least once,
8181         add runtime check with setting count to 0.
8182         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8183         For unsigned types if it isn't known at compile time that
8184         the loop will iterate at least once, add runtime check to bypass
8185         the whole loop if initial condition isn't true.
8187 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
8189         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8191 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
8193         PR middle-end/57286
8194         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8195         transformations to avoid an infinite loop.
8197 2013-05-16  Marek Polacek  <polacek@redhat.com>
8199         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8201 2013-05-15  Leif Ekblad  <leif@rdos.net>
8203         * config/i386/i386.c (ix86_decompose_address): Use
8204         DEFAULT_TLS_SEG_REG to access TLS segment register.
8205         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8206         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8207         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8209 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
8211         PR target/57260
8212         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8213         sibling calls to functions that would normally be lazily bound,
8214         unless $gp is call-clobbered.
8216 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
8218         * config/i386/i386.c (ix86_option_override_internal): Update
8219         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
8220         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8221         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8222         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8223         of TARGET_3DNOW.
8224         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8226 2013-05-15  Andreas Schwab  <schwab@suse.de>
8228         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8229         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
8230         third operand.
8232 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8234         * loop-unroll.c (report_unroll_peel): Check decision before
8235         emitting unroll/peel message.
8237 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8239         * function.h (has_bb_partition): New rtl_data flag.
8240         (bb_reorder_complete): Ditto.
8241         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8242         instead of flag_reorder_blocks_and_partition.
8243         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8244         with some enhancements.
8245         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8246         * bb-reorder.c (connect_traces): Check for has_bb_partition
8247         instead of flag_reorder_blocks_and_partition.
8248         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8249         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8250         verify_hot_cold_block_grouping.
8251         (partition_hot_cold_basic_blocks): Set has_bb_partition.
8253 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8255         PR target/19599
8256         * config/arm/predicates.md (call_insn_operand): New predicate.
8257         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
8258         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8259         if insn is not a tail call.
8260         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8261         registers.
8262         * config/arm/arm.h (enum reg_class): New caller save register class.
8263         (REG_CLASS_NAMES): Likewise.
8264         (REG_CLASS_CONTENTS): Likewise.
8265         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8266         without decls.
8268 2013-05-15  Richard Biener  <rguenther@suse.de>
8270         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8271         of MSG_OPTIMIZED_LOCATIONS.
8272         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8273         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8274         message.
8275         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8276         of MSG_OPTIMIZED_LOCATIONS.
8277         (execute_vect_slp): Likewise.
8278         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8279         (vect_create_cond_for_alias_checks): Likewise.
8280         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8281         (vect_recog_widen_mult_pattern): Likewise.
8282         (vect_recog_widen_sum_pattern): Likewise.
8283         (vect_recog_over_widening_pattern): Likewise.
8284         (vect_recog_widen_shift_pattern): Likewise.
8285         (vect_recog_vector_vector_shift_pattern): Likewise.
8286         (vect_recog_divmod_pattern): Likewise.
8287         (vect_recog_mixed_size_cond_pattern): Likewise.
8288         (vect_recog_bool_pattern): Likewise.
8289         (vect_pattern_recog_1): Likewise.
8291 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8293         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8294         non-functions to builtin_unreachable.
8295         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8296         correct when new direct edges were discovered.
8298 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8300         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8301         header, print symbol order instead of node uid, print more information
8302         about indirect edge targets.
8303         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8304         uids.
8305         (ipa_make_edge_direct_to_target): Likewise.
8306         (remove_described_reference): Likewise.
8307         (propagate_controlled_uses): Likewise.
8308         (ipa_print_node_params): Also print symbol order.
8309         (ipcp_transform_function): Print symbol order instead of node uids.
8310         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8311         (cgraph_get_create_real_symbol_node): Likewise.
8312         * ipa-cp.c (print_lattice): Likewise.
8313         (print_all_lattices): Likewise.
8314         (determine_versionability): Likewise.
8315         (initialize_node_lattices): Likewise.
8316         (estimate_local_effects): Likewise.
8317         (update_profiling_info): Likewise.
8318         (create_specialized_node): Likewise.
8319         (perhaps_add_new_callers): Likewise.
8320         (decide_about_value): Likewise.
8321         (decide_whether_version_node): Likewise.
8322         (identify_dead_nodes): Likewise.
8323         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8324         (dump_inline_summary): Likewise.
8325         (estimate_node_size_and_time): Likewise.
8326         (inline_analyze_function): Likewise.
8327         * ipa-inline.c (report_inline_failed_reason): Likewise.
8328         (want_early_inline_function_p): Likewise.
8329         (edge_badness): Likewise.
8330         (update_edge_key): Likewise.
8331         (inline_small_functions): Likewise.  Add dumping of order to two other
8332         dumps.
8333         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8334         instead of node uids.
8335         (propagate_pure_const): Likewise.
8336         (propagate_pure_const): Likewise.
8337         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8338         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8339         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8340         of node uids.
8341         * tree-pretty-print.c (dump_function_header): Likewise.
8342         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8343         Print symbol order instead of node uids.
8345 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8347         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8348         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8350 2013-05-15  Richard Biener  <rguenther@suse.de>
8352         PR tree-optimization/57275
8353         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8354         return value for fail to do runtime alias checks for gather loads.
8356 2013-05-15  Jan Hubicka  <jh@suse.cz>
8358         PR lto/57038
8359         PR lto/47375
8360         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8361         weakrefs are not external.
8362         (lto_symtab_merge_decls): Fix thinko when dealing with
8363         non-lto_symtab decls.
8364         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8365         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8366         * varpool.c (dump_varpool_node): Dump more flags.
8368 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8370         * config/i386/i386.c (processor_alias_table): Add instruction
8371         FSGSBASE for AMD bdver3 architecture.
8373 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8375         * tree.c (warn_deprecated_use): Print file:line using locus color.
8376         * diagnostic.c (diagnostic_report_current_module): Print file:line
8377         and file:line:column using locus color.
8379 2013-05-14  Mike Stump  <mikestump@comcast.net>
8381         * gdbinit.in: Add __null.
8383 2013-05-14  Mike Stump  <mikestump@comcast.net>
8385         * recog.h: Rename struct recog_data to Recog_data.
8386         * recog.c: Likewise.
8387         * reload.c (can_reload_into): Likewise.
8388         * config/picochip/picochip.c: Likewise.
8390 2013-05-14  Mike Stump  <mikestump@comcast.net>
8392         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8394 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8396         * resource.h (struct resources): Remove unch_memory member.
8397         (CLEAR_RESOURCE): Don't clear unch_memory.
8398         * resource.c (mark_referenced_resources): Don't set it.
8399         (mark_set_resources): Likewise.
8400         (mark_target_live_regs): Don't clear it.
8401         (init_resource_info): Likewise.
8402         * reorg.c (resource_conflicts_p): Don't compare it.
8403         (redundant_insn): Don't set it.
8405         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8406         Remove prototypes.
8407         * emit-rtl.c (next_label): Remove unused function.
8408         (skip_consecutive_labels, link_cc0_insns): Move to ...
8409         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8410         only place where these functions are used, and make them static.
8412 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8414         * fold-const.c (fold_negate_expr): Handle vectors.
8415         (fold_truth_not_expr): Make it static.
8416         (fold_invert_truthvalue): New static function.
8417         (invert_truthvalue_loc): Handle vectors. Do not call
8418         fold_truth_not_expr directly.
8419         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8420         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8421         (fold_comparison): Handle vector constants.
8422         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8423         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8424         * tree.h (fold_truth_not_expr): Remove declaration.
8426 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8428         * config/aarch64/aarch64-simd.md
8429         (aarch64_vcond_internal<mode>): Rename to...
8430         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8431         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8432         float modes. Clarify all iterator modes.
8433         (vcond<mode><mode>): Use new name for vcond expanders.
8434         (vcond<v_cmp_result><mode>): Likewise.
8435         (vcondu<mode><mode>: Likewise.
8436         * config/aarch64/iterators.md (VDQF_COND): New.
8438 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8440         PR bootstrap/57266
8441         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8442         variable for the shift amount. Check that we shift by non-negative
8443         amounts.
8445 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8447         PR target/42017
8448         * config/arm/arm.h (EPILOGUE_USES): Only return true
8449         for LR_REGNUM after epilogue_completed.
8451 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8453         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8454         is error_mark_node.
8456 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8458         PR target/57261
8459         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8460         and Solaris 11+/x86 with gld.
8461         * configure: Regenerate.
8463 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8465         * expmed.c (expand_shift_1): Canonicalize rotates by
8466         constant bitsize / 2 to bitsize - 1.
8467         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8468         case ROTATERT>: Likewise.
8470         Revert:
8471         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8473         * config/i386/i386.md (rotateinv): New code attr.
8474         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8475         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8476         roll $31, %eax, etc.
8478 2013-05-14  Richard Biener  <rguenther@suse.de>
8480         PR middle-end/57235
8481         * tree-eh.c (sink_clobbers): Give up for successors with
8482         multiple predecessors and no virtual uses.
8484 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8486         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8487         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8489 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8491         PR middle-end/57251
8492         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8493         the case when both op0 and op1 have VOIDmode.
8495 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8497         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8498         in multiply-accumulate mode.
8500 2013-05-13  Guozhi Wei  <carrot@google.com>
8502         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8504 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8506         PR target/56975
8507         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8508         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8509         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8510         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8511         (ix86_expand_prologue): Likewise.
8512         (ix86_expand_split_stack_prologue): Likewise.
8513         (legitimate_pic_address_disp_p): Likewise.
8514         (legitimize_pic_address): Likewise.
8515         (legitimize_tls_address): Likewise.
8516         (legitimize_pe_coff_symbol): Likewise.
8517         (output_pic_addr_const): Likewise.
8518         (construct_plt_address): Likewise.
8519         (ix86_expand_call): Likewise.
8520         (x86_output_mi_thunk): Likewise.
8521         (x86_function_profiler): Likewise.
8523 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
8525         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8526         similar switch cases.
8527         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8528         (aarch64_simd_mov_to_<mode>low): Delete.
8529         (aarch64_simd_mov_to_<mode>high): Delete.
8530         (move_lo_quad_<mode>): Add w<-r alternative.
8531         (aarch64_simd_move_hi_quad_<mode>): Likewise.
8532         (aarch64_simd_mov_from_*): Update type attribute.
8533         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8534         statement.
8536 2013-05-13  Jan Hubicka  <jh@suse.cz>
8538         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8539         * config/i386/i386.c (ix86_compute_frame_layout,
8540         ix86_expand_epilogue, emit_i387_cw_initialization,
8541         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8542         ix86_local_alignment): Fix use of size/speed predicates.
8544 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8546         PR tree-optimization/45216
8547         PR tree-optimization/57157
8548         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8549         the (-Y) & (B - 1) variant if OP is |.
8550         * expmed.c (expand_shift_1): For rotations by const0_rtx just
8551         return shifted.  Use (-op1) & (prec - 1) as other_amount
8552         instead of prec - op1.
8554 2013-05-13  Martin Jambor  <mjambor@suse.cz>
8556         PR middle-end/42371
8557         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8558         (ipa_constant_data): New type.
8559         (ipa_jump_func): Use ipa_constant_data to hold information about
8560         constant jump functions.
8561         (ipa_get_jf_constant): Adjust to jump function type changes.
8562         (ipa_get_jf_constant_rdesc): New function.
8563         (ipa_param_descriptor): New field controlled_uses.
8564         (ipa_get_controlled_uses): New function.
8565         (ipa_set_controlled_uses): Likewise.
8566         * ipa-ref.h (ipa_find_reference): Declare.
8567         * ipa-prop.c (ipa_cst_ref_desc): New type.
8568         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8569         changes.
8570         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
8571         New parameter cs.  Adjust all callers.
8572         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8573         (remove_described_reference): New function.
8574         (jfunc_rdesc_usable): Likewise.
8575         (try_make_edge_direct_simple_call): Decrement controlled use count,
8576         attempt to remove reference if it hits zero.
8577         (combine_controlled_uses_counters): New function.
8578         (propagate_controlled_uses): Likewise.
8579         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8580         (ipa_edge_duplication_hook): Duplicate reference descriptions.
8581         (ipa_print_node_params): Print described use counter.
8582         (ipa_write_jump_function): Adjust to jump function type changes.
8583         (ipa_read_jump_function): New parameter CS, pass it to
8584         ipa_set_jf_constant.  Adjust caller.
8585         (ipa_write_node_info): Stream controlled use count
8586         (ipa_read_node_info): Likewise.
8587         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8588         asserting.
8589         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8590         count.  Remove cloning-added reference if it reaches zero.
8591         * ipa-ref.c (ipa_find_reference): New function.
8593 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8595         * config/i386/i386.c (processor_target_table): Modified default
8596         alignment values for AMD BD and BT architectures.
8598 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
8600         * tree-vect-generic.c (uniform_vector_p): Move ...
8601         * tree.c (uniform_vector_p): ... here.
8602         * tree.h (uniform_vector_p): Declare it.
8603         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8604         into a scalar.
8606 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8608         PR tree-optimization/57230
8609         * tree-ssa-strlen.c (handle_char_store): Record length for
8610         array store from STRING_CST.
8612         PR tree-optimization/57230
8613         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8614         check.
8616 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
8618         * config/epiphany/epiphany.c (epiphany_init): Check size of
8619         NUM_MODES_FOR_MODE_SWITCHING.
8620         (epiphany_expand_prologue):
8621         Remove CONFIG_REGNUM initial value handling code.
8622         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8623         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8624         (emit_set_fp_mode, epiphany_mode_after): Likewise.
8625         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8626         Don't return 1 for FP_MODE_NONE.
8627         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8628         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8629         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8630         * config/epiphany/epiphany.md (save_config): New pattern.
8632 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
8634         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8636 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
8638         * config/i386/i386.md (memory): Handle sseishft1.
8639         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8640         (*vec_extractv2di_1): Ditto.
8642 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
8644         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8645         saved registers.
8647 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8649         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8650         Add mthumb/march=armv7-a multilib.
8651         Add mthumb/march=armv7-r multilib.
8652         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8654 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
8656         * config/v850/t-rtems: Add more multilibs.
8658 2013-05-10  Richard Biener  <rguenther@suse.de>
8660         PR tree-optimization/57214
8661         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8662         not propagate from SSA names that occur in abnormal PHI nodes.
8664 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
8666         * stor-layout.c (element_precision): New function.
8667         * machmode.h (element_precision): Declare it.
8668         * tree.c (build_minus_one_cst): New function.
8669         (element_precision): Likewise.
8670         * tree.h (build_minus_one_cst): Declare new function.
8671         (element_precision): Likewise.
8672         * fold-const.c (operand_equal_p): Use element_precision.
8673         (fold_binary_loc): Handle vector types.
8674         * convert.c (convert_to_integer): Use element_precision.
8675         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8676         separately.
8678 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
8680         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8681         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8682         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8683         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8684         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8685         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8686         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8687         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8688         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8689         (Uuw8): New constraints.
8690         (Usb4): Move into alphabetical order.
8691         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8692         (sd8_operand, ub8_operand, uw8_operand): New predicates.
8693         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8694         previously unnamed patterns.
8695         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8696         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8697         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8698         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8699         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
8700         of specific lengths.
8702 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8704         * config/i386/i386.md (rotateinv): New code attr.
8705         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8706         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8707         roll $31, %eax, etc.
8709         PR tree-optimization/45216
8710         PR tree-optimization/57157
8711         * tree-ssa-forwprop.c (simplify_rotate): New function.
8712         (ssa_forward_propagate_and_combine): Call it.
8714 2013-05-10  Richard Biener  <rguenther@suse.de>
8716         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8717         disable peeling when we version for aliasing.
8718         (vector_alignment_reachable_p): Honor explicit user alignment.
8719         (vect_supportable_dr_alignment): Likewise.
8720         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8721         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8722         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8723         then peeling to arrange for the cost-model check to come first.
8725 2013-05-10  Alan Modra  <amodra@gmail.com>
8727         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8728         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8729         * configure: Regenerate.
8731 2013-05-10  Alan Modra  <amodra@gmail.com>
8733         PR target/55033
8734         * varasm.c (default_elf_select_section): Move !DECL_P check..
8735         (get_named_section): ..to here before calling get_section_name.
8736         Adjust assertion.
8737         (default_section_type_flags): Add DECL_P check.
8738         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8739         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8741 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
8743         * config/epiphany/epiphany.c (epiphany_expand_prologue):
8744         When using gen_stack_adjust_str with a register offset, add a
8745         REG_FRAME_RELATED_EXPR note.
8747 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
8749         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8750         (*vec_extractv4si_zext_mem): Ditto.
8751         (*vec_extractv2di): Add 0->x and x->x alternatives.
8752         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8753         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8755 2013-05-09  Jason Merrill  <jason@redhat.com>
8757         N3639 C++1y VLA support
8758         * gimplify.c (gimplify_vla_decl): Don't touch an existing
8759         DECL_VALUE_EXPR.
8761         * tree.c (build_constructor_va): New.
8762         * tree.h: Declare it.
8764 2013-05-09  Martin Jambor  <mjambor@suse.cz>
8766         PR lto/57084
8767         * gimple-fold.c (canonicalize_constructor_val): Call
8768         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8770 2013-05-09  Jan Hubicka  <jh@suse.cz>
8771             Richard Biener  <rguenther@suse.de>
8773         PR lto/54095
8774         * symtab.c (symtab_make_decl_local): Do not add private names.
8776 2013-05-09  Jan Hubicka  <jh@suse.cz>
8778         PR lto/54095
8779         * symtab.c (insert_to_assembler_name_hash): Handle clones.
8780         (unlink_from_assembler_name_hash): Likewise.
8781         (symtab_prevail_in_asm_name_hash, symtab_register_node,
8782         symtab_unregister_node, symtab_initialize_asm_name_hash,
8783         change_decl_assembler_name): Update.
8785 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
8787         * config/aarch64/aarch64.md: New movtf split.
8788         (*movtf_aarch64): Update.
8789         (aarch64_movdi_tilow): Handle TF modes and rename to
8790         aarch64_movdi_<mode>low.
8791         (aarch64_movdi_tihigh): Handle TF modes and rename to
8792         aarch64_movdi_<mode>high
8793         (aarch64_movtihigh_di): Handle TF modes and rename to
8794         aarch64_mov<mode>high_di
8795         (aarch64_movtilow_di): Handle TF modes and rename to
8796         aarch64_mov<mode>low_di
8797         (aarch64_movtilow_tilow): Remove spurious whitespace.
8798         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8799         splits.
8800         (aarch64_print_operand): Update.
8802 2013-05-09  Alan Modra  <amodra@gmail.com>
8804         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8805         powerpc64le.
8806         * configure: Regenerate.
8808 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8810         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8811         splitter preparation statements.
8812         * config/i386/sse.md (*vec_extract* splitters): Ditto.
8813         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8814         adjust_address_nv.
8816 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8818         * gimple-ssa-strength-reduction.c (count_candidates): Change
8819         return value to int.
8820         (analyze_candidates_and_replace): Change type of length to int.
8822 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8824         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8825         (*vec_extract<mode>): Use VI12_128 mode iterator.
8826         (*vec_extract<mode>_mem): Ditto.
8827         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8828         attribute.
8830 2013-05-08  Diego Novillo  <dnovillo@google.com>
8832         PR bootstrap/54659
8834         Revert:
8835         2012-08-17  Diego Novillo  <dnovillo@google.com>
8837         PR bootstrap/54281
8838         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8839         * config.in: Regenerate.
8840         * configure: Regenerate.
8841         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8843 2013-05-08  Jan Hubicka  <jh@suse.cz>
8845         PR lto/54095
8846         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8847         * cgraph.h (symtab_node_base): Add unique_name.
8848         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8849         input_overwrite_node, input_varpool_node): Stream unique_name.
8850         * cgraphclones.c (cgraph_create_virtual_clone,
8851         cgraph_function_versioning): Set unique_name.
8852         * ipa.c (function_and_variable_visibility): Set unique_name.
8854 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8856         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8857         (alloc_cand_and_find_basis): Restrict conditional candidate
8858         processing to CAND_MULTs.
8860 2013-05-08  Jan Hubicka  <jh@suse.cz>
8862         PR lto/54095
8863         lto-symtab.c (lto_symtab_symbol_p): New function.
8864         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8865         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8866         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8867         Skip static symbols.
8869 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
8871         PR tree-optimization/57200
8872         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8873         Only call inform if the preceding warning_at returns true.
8875 2013-05-07  Han Shen  <shenhan@google.com>
8877         * cfgexpand.c (record_or_union_type_has_array_p): New function.
8878         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8879         * common.opt (fstack-protector-strong): New option.
8880         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8881         * doc/invoke.texi (Optimization Options): Document
8882         "-fstack-protector-strong".
8883         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8885 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8887         * config/mips/mips.c (mips_machine_reorg2): Return 0.
8889 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
8891         * ira.c (update_equiv_regs): Add insn having equiv memory even if
8892         it is not lhs of the insn.
8893         (setup_reg_equiv): Remove insn having equiv memory which it is not
8894         lhs of the insn.
8895         * lra-constraints.c (process_address): Try to improve generation
8896         code for address base + disp.
8897         (lra_constraints): Make correct the code for checking insn setting
8898         up backward equivalence.  Remove insn only if it is in the init
8899         insn list.
8900         * lra-eliminations.c (update_reg_eliminate): Change return value.
8901         (lra_eliminate): Use the result.
8903 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
8905         * config/i386/sse.md (ssescalarnummask): New mode attribute.
8906         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8907         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8908         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8909         register target operands.
8910         (*vec_extractv8hi_sse2): New pattern.
8911         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8912         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8913         (*vec_extract<mode>_mem): New insn and split pattern.
8915 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
8917         * config/arm/arm.c (arm_asan_shadow_offset): New function.
8918         (TARGET_ASAN_SHADOW_OFFSET): Define.
8919         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8920         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8922 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8924         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8925         (incr_vec_index): Return -1 if increment not found.
8926         (create_add_on_incoming_edge): Assert if increment not found.
8927         (record_increment): Limit number of increments recorded.
8928         (all_phi_incrs_profitable): Return false if an increment not found.
8929         (replace_profitable_candidates): Don't process increments that were
8930         not recorded.
8931         (analyze_candidates_and_replace): Limit size of incr_vec.
8933 2013-05-07  Richard Biener  <rguenther@suse.de>
8935         * calls.c (special_function_p): setjmp-like functions are leaf.
8936         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8937         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8939 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
8941         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8942         (aarch64_simd_mov<mode>): New expander.
8943         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8944         (aarch64_simd_mov_to_<mode>high): Likewise.
8945         (aarch64_simd_mov_from_<mode>low): Likewise.
8946         (aarch64_simd_mov_from_<mode>high): Likewise.
8947         (aarch64_dup_lane<mode>): Update.
8948         (aarch64_dup_lanedi): New instruction pattern.
8949         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8950         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8952 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8954         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8955         (replace_mult_candidate): Remove unnecessary argument; remove
8956         unnecessary parameter from call to introduce_cast_before_cand.
8957         (replace_unconditional_candidate): Remove unnecessary parameter
8958         from call to replace_mult_candidate.
8959         (replace_conditional_candidate): Likewise.
8960         (insert_initializers): Use make_temp_ssa_name.
8961         (introduce_cast_before_cand): Remove unnecessary argument; use
8962         make_temp_ssa_name.
8963         (replace_one_candidate): Remove unnecessary argument; remove
8964         unnecessary parameter from calls to introduce_cast_before_cand.
8965         (replace_profitable_candidates): Remove unnecessary parameters
8966         from calls to replace_one_candidate.
8968 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8970         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8971         phi def as possibly hiding a basis for a CAND_ADD whose operands
8972         have been commuted in the analysis.
8973         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8975 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
8977         * config/aarch64/aarch64.md
8978         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8979         shift value between 0-4.
8981 2013-05-07  Richard Biener  <rguenther@suse.de>
8983         * double-int.h (rshift): New overload.
8984         * double-int.c (rshift): New function.
8985         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8986         (create_reference_ops_from_ref): Remove.
8987         (vn_reference_insert): Use shared ops for constructing the
8988         reference and copy it.
8990 2013-05-07  Richard Biener  <rguenther@suse.de>
8992         PR middle-end/57190
8993         * tree-eh.c (sink_clobbers): Properly propagate
8994         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8996 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
8998         PR tree-optimization/57149
8999         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9000         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9001         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9002         uninit_undefined_value_p instead of ssa_undefined_value_p.
9004         PR debug/57184
9005         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9006         for modifier == EXPAND_INITIALIZER.
9008 2013-05-07  Anton Blanchard  <anton@samba.org>
9010         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9011         for powerpc64 little endian.
9012         * configure: Regenerate.
9014 2013-05-06  Graham Stott  <grahams@btinternet.com>
9016         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9017         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9018         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9019         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9021 2013-05-06  Graham Stott  <grahams@btinternet.com>
9023         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9024         codes which allow non-lvalues.
9026 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
9028         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9029         components are all 1s.
9030         (integer_minus_onep): New function.
9031         * tree.h (integer_minus_onep): Declare it.
9032         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9033         integer_minus_onep instead of integer_all_onesp.
9035 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9037         PR target/52933
9038         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9039         variations of these patterns.
9041 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9043         * config/i386/i386.md (isa): Add x64_sse4 member.
9044         (enabled): Handle x64_sse4.
9045         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9046         instruction for 64bit SSE4_1 targets.  Update insn attributes.
9047         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9048         instruction for SSE4_1 targets.  Update insn attributes.
9049         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9050         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9051         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9052         const_1 selector.
9053         (*vec_extractv4si): Rename from *sse4_1_pextrd.
9054         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9055         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9057 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9059         PR target/57108
9060         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9062 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
9064         * final.c (do_assembler_dialects): Don't handle curly braces and
9065         vertical bar escaped by % as dialect delimiters.
9066         (output_asm_insn): Print curly braces and vertical bar if escaped
9067         by % and ASSEMBLER_DIALECT defined.
9068         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9069         * doc/tm.texi: Regenerated.
9071 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9073         * config/mips/mips.c: Include tree-pass.h.
9074         (mips_reorg): Split in pre- and post-dbr_schedule parts.
9075         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9076         (pass_mips_machine_reorg2): New machine specific pass.
9077         (insert_pass_mips_machine_reorg2): New pass plugin definition.
9078         (mips_option_override): Register the new pass.
9079         * rtl.h (cleanup_barriers): Remove prototype.
9080         (dbr_schedule): Likewise.
9081         * jump.c (cleanup_barriers): Make static.
9082         * reorg.c (dbr_schedule): Likewise.
9084 2013-05-06  Richard Biener  <rguenther@suse.de>
9086         PR tree-optimization/57185
9087         * tree-parloops.c (add_field_for_reduction): Handle anonymous
9088         SSA names properly.
9090 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9092         PR target/57106
9093         * config/i386/i386.c (add_parameter_dependencies): Add dependence
9094         between "first_arg" and "insn", not "last" and "insn".
9096 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9098         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9099         (find_candidates_in_block): Re-enable slsr_process_phi.
9100         (create_phi_basis): Fix double counting of candidate adjustment.
9102 2013-05-06  Richard Biener  <rguenther@suse.de>
9104         PR middle-end/57147
9105         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9106         the edge is also fallthru, preserve it and just clear the
9107         abnormal flag.
9108         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9109         also complex, preserve that and just clear the fallthru flag.
9110         * tree-inline.c (update_ssa_across_abnormal_edges): Also
9111         update virtual operands.
9113 2013-05-06  Alan Modra  <amodra@gmail.com>
9115         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9116         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9117         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9118         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9119         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9120         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9122 2013-05-06  Alan Modra  <amodra@gmail.com>
9124         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9125         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
9126         (DEFAULT_ASM_ENDIAN): Define.
9127         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9128         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9129         Update -K PIC clause from sysv4.h.
9130         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9131         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9133 2013-05-06  Alan Modra  <amodra@gmail.com>
9135         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9136         twice for little-endian.
9137         (ashrdi3_no_power, ashrdi3): Support little-endian.
9139 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9141         PR target/55303
9142         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9143         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9144         related expanders.
9145         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9146         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9147         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9148         New predicates.
9150 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
9151             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
9153         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9154         * config/pa/pa.opt: Make mbig-switch a no-op.
9155         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9156         (CASE_VECTOR_MODE): Always return SImode.
9157         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9158         for the !TARGET_BIG_SWITCH case.
9159         * config/pa/pa-linux.h: Likewise.
9160         * config/pa/pa-openbsd.h: Likewise.
9161         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9162         * config/pa/pa.md (short_jump): Remove define_insn.
9163         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9164         (casesi0): Remove define_insn.
9165         (type): Remove btable_branch.
9166         (pa_combine_type): Likewise.
9167         (in_nullified_branch_delay): Likewise.
9168         (in_call_delay): Likewise.
9169         (define_delay): Likewise.
9170         (define_insn_reservation "Z3"): Likewise.
9171         (define_insn_reservation "Z4"): Likewise.
9172         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9173         (pa_adjust_insn_length): Remove adjustment for btable branches.
9174         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9175         and mno-big-switch
9177 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
9179         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9180         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9181         Add m->r,x alternatives.
9182         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9183         splitters using SWI48x mode iterator.
9184         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
9185         TARGET_64BIT.  Add m->x alternative.
9186         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9187         Add o->x alternative.  Enable for TARGET_SSE.
9188         (sse_storeq): Remove expander.
9189         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9190         with memory input operand.
9191         (*vec_extractv2di_1 splitter): New.
9192         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9193         * config/i386/i386.md (ssevecmodelower): New mode attribute.
9195 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9197         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
9198         (INT_LOWPART): Delete.
9199         (extract_MB): Adjust.
9200         (extract_ME): Adjust.
9201         (print_operand): Adjust.
9203 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9205         * config/rs6000/predicates.md (reg_or_add_cint_operand,
9206         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9207         (reg_or_logical_cint_operand, easy_fp_constant,
9208         logical_const_operand): Delete "CONST_DOUBLE" case.
9209         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9210         "HOST_BITS_PER_WIDE_INT == 64" test.
9211         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
9212         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9213         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9214         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9215         test.
9216         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9217         CONST_DOUBLE DImode/VOIDmode case.
9218         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9219         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9220         CONST_DOUBLE VOIDmode case.
9221         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9222         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9223         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9224         Delete CONST_DOUBLE case.
9225         (splitters for mov FMOVE64 const_double): Delete
9226         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9227         "HOST_BITS_PER_WIDE_INT >= 64" test.
9228         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9229         case.
9230         (mov DI const_double): Delete.
9232 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
9234         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9235         on op shows all bits zero in mode of a lowpart subreg, return zero.
9237 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
9239         PR target/57150
9240         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9241         to save TFmode registers and DImode to save TImode registers for
9242         caller save operations.
9243         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9244         mark being partially clobbered since they only use the first
9245         double word.
9247         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9248         and TDmode only use the upper 64-bits of each VSX register.
9250 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9252         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9253         (find_candidates_in_block): Disable slsr_process_phi.
9255 2013-05-03  Guozhi Wei  <carrot@google.com>
9257         * coverage.c (coverage_obj_init): Move the construction of gcov
9258         constructor to ...
9259         (build_init_ctor): ... here.
9261 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9263         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9264         (slsr_cand_d): Redefine def_phi.
9265         (stride_status, phi_adjust_status, count_phis_status): New enums.
9266         (find_phi_def): New.
9267         (find_basis_for_base_expr): New.
9268         (find_basis_for_candidate): Handle hidden bases.
9269         (alloc_cand_and_find_basis): Handle phi candidates.
9270         (slsr_process_phi): New.
9271         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9272         (create_mul_imm_cand): Likewise.
9273         (create_add_ssa_cand): Exclude phi base candidates.
9274         (create_add_imm_cand): Likewise.
9275         (slsr_process_cast): Likewise.
9276         (slsr_process_copy): Likewise.
9277         (find_candidates_in_block): Handle phi candidates.
9278         (dump_candidate): Likewise.
9279         (unconditional_cands): Delete.
9280         (unconditional_cands_with_known_stride_p): Delete.
9281         (phi_dependent_cand_p): New.
9282         (cand_increment): Handle phi-dependent candidates.
9283         (replace_dependent): Delete.
9284         (replace_mult_candidate): New.
9285         (replace_unconditional_candidate): New.
9286         (incr_vec_index): Move to avoid forward reference.
9287         (create_add_on_incoming_edge): New.
9288         (create_phi_basis): New.
9289         (replace_dependents): Delete.
9290         (replace_conditional_candidate): New.
9291         (phi_add_costs): New.
9292         (replace_uncond_cands_and_profitable_phis): New.
9293         (record_increment): Handle phi adjustments.
9294         (record_phi_increments): New.
9295         (record_increments): Handle phi adjustments.
9296         (phi_incr_cost): New.
9297         (lowest_cost_path): Handle phis.
9298         (total_savings): Likewise.
9299         (analyze_increments): Likewise.
9300         (ncd_with_phi): New.
9301         (ncd_of_cand_and_phis): New.
9302         (nearest_common_dominator_for_cands): Handle phi increments.
9303         (all_phi_incrs_profitable): New.
9304         (replace_profitable_candidates): Handle phi-dependent candidates.
9305         (analyze_candidates_and_replace): Likewise.
9307 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9309         PR bootstrap/57154
9310         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9311         do not exceed REG_BR_PROB_BASE.
9313 2013-05-03  Jeff Law  <law@redhat.com>
9315         PR tree-optimization/57144
9316         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9317         operand of the condition will bit into the new type when eliminating
9318         a cast feeding a condition.
9320 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9322         PR rtl-optimization/57130
9323         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9324         of COMPARE as in_code to the recursive call if needed.
9326 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9328         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9329         (enabled): Handle new members.
9330         * config/i386/sse.md (*vec_concatv2si): Merge from
9331         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9332         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9334 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9336         PR tree-optimization/57027
9337         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9338         for fnms opportunity, check we got the prerequisite kind
9339         of tree / gimple before using accessor functions.
9341 2013-05-03  Richard Biener  <rguenther@suse.de>
9343         * double-int.h (lshift): New overload without precision
9344         and arith argument.
9345         (operator *=, operator +=, operator -=): Move ...
9346         * double-int.c (operator *=, operator +=, operator -=): ... here
9347         and implement more efficiently.
9348         (mul_double_with_sign): Remove.
9349         (lshift_double): Adjust to take unsinged shift argument, push
9350         dispatching code to callers.
9351         (mul_double_wide_with_sign): Add early out for callers that
9352         are not interested in high parts or overflow.
9353         (lshift): New function.
9354         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9355         dispatch code here.
9356         (lrotate, rrotate): Use logical shifts.
9357         * expr.c (get_inner_reference): Use lshift.
9358         * fixed-value.c (do_fixed_divide): Likewise.
9359         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9360         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9361         (indirect_refs_may_alias_p): Likewise.
9362         (stmt_kills_ref_p_1): Likewise.
9364 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9366         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9368 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9370         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9371         scalar form of FABD instruction.
9373 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9375         * lra-constraints.c (process_alt_operands): Add checking alt
9376         number to choose the best alternative.
9378 2013-05-02  Richard Biener  <rguenther@suse.de>
9380         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9381         bitmap and its handling.
9382         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9384 2013-05-02  Richard Biener  <rguenther@suse.de>
9386         PR middle-end/57140
9387         * tree-inline.c (copy_loops): Properly handle removed loops.
9388         (copy_cfg_body): Mark destination loops for fixup if source
9389         loops needed fixup.
9391 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9393         PR target/56732
9394         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9395         generating simple_return for naked functions.
9397 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9399         PR middle-end/56988
9400         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9401         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9402         flags match.
9403         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9404         ipa_agg_replacement_value structures.
9405         (known_aggs_to_agg_replacement_list): Likewise.
9406         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9407         (read_agg_replacement_chain): Likewise.
9408         (ipcp_transform_function): Also check that by_ref flags match.
9410 2013-05-02  Richard Biener  <rguenther@suse.de>
9412         * graphds.h (struct graph): Add obstack member.
9413         * graphds.c (new_graph): Initialize obstack and allocate
9414         vertices from it.
9415         (add_edge): Allocate edge from the obstack.
9416         (free_graph): Free the obstack instead of all edges and vertices.
9418 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9420         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9421         divides.
9422         * cfg.c (update_bb_profile_for_threading): Ditto.
9423         * tree-inline.c (copy_bb): Ditto.
9424         (copy_edges_for_bb): Ditto.
9425         (initialize_cfun): Ditto.
9426         (copy_cfg_body): Ditto.
9427         (expand_call_inline): Ditto.
9428         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9429         (estimate_node_size_and_time): Ditto.
9430         (inline_merge_summary): Ditto.
9431         * cgraphclones.c (cgraph_clone_edge): Ditto.
9432         (cgraph_clone_node): Ditto.
9433         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9434         (compute_trg_info): Ditto.
9436 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9438         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9439         S reg when fp attribute set.
9440         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9442 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9444         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9445         New pattern.
9446         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9447         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9448         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9450 2013-05-02  Richard Biener  <rguenther@suse.de>
9452         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9453         (struct instantiate_cache_entry): New type.
9454         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9455         (struct instantiate_cache_type): New type.
9456         (set_instantiated_value, get_instantiated_value): Remove.
9457         (get_instantiated_value_entry): New function.
9458         (instantiate_scev_name): Use the new cache and adjust.
9459         (instantiate_scev_poly): Adjust.
9460         (instantiate_scev_binary): Likewise.
9461         (instantiate_array_ref): Likewise.
9462         (instantiate_scev_convert): Likewise.
9463         (instantiate_scev_not): Likewise.
9464         (instantiate_scev_3): Likewise.
9465         (instantiate_scev_2): Likewise.
9466         (instantiate_scev_r): Likewise.
9467         (instantiate_scev): Likewise.
9468         (resolve_mixers): Likewise.
9470 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9472         PR target/57091
9473         * lra-constraints.c (best_small_class_operands_num): Remove.
9474         (process_alt_operands): Remove small_class_operands_num.  Take
9475         small classes operands into losers and only if the operand is not
9476         matched.  Modify debugging output.
9477         (curr_insn_transform): Remove best_small_class_operands_num.
9478         Print insn name.
9480 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9482         * config/aarch64/aarch64-builtins.c
9483         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9484         * config/aarch64/aarch64-simd-builtins.def
9485         (reduc_splus_): Add new modes.
9486         (reduc_uplus_): New.
9487         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9488         (reduc_uplus_v4sf): Likewise.
9489         (reduc_splus_v4sf): Likewise.
9490         (aarch64_addv<mode>): Likewise.
9491         (reduc_uplus_<mode>): Likewise.
9492         (reduc_splus_<mode>): Likewise.
9493         (aarch64_addvv2di): Likewise.
9494         (reduc_uplus_v2di): Likewise.
9495         (reduc_splus_v2di): Likewise.
9496         (aarch64_addvv2si): Likewise.
9497         (reduc_uplus_v2si): Likewise.
9498         (reduc_splus_v2si): Likewise.
9499         (reduc_<sur>plus_<mode>): New.
9500         (reduc_<sur>plus_v2di): Likewise.
9501         (reduc_<sur>plus_v2si): Likewise.
9502         (reduc_<sur>plus_v4sf): Likewise.
9503         (aarch64_addpv4sf): Likewise.
9504         * config/aarch64/arm_neon.h
9505         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9506         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9507         add UNSPEC_SADDV, UNSPEC_UADDV.
9508         (SUADDV): New.
9509         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9511 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9513         * config/aarch64/arm_neon.h
9514         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9516 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9518         * config/aarch64/aarch64-builtins
9519         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9521 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9523         * config/aarch64/aarch64-simd-builtins.def
9524         (reduc_smax_): New.
9525         (reduc_smin_): Likewise.
9526         (reduc_umax_): Likewise.
9527         (reduc_umin_): Likewise.
9528         (reduc_smax_nan_): Likewise.
9529         (reduc_smin_nan_): Likewise.
9530         (fmax): Remove.
9531         (fmin): Likewise.
9532         (smax): Update for V2SF, V4SF and V2DF modes.
9533         (smin): Likewise.
9534         (smax_nan): New.
9535         (smin_nan): Likewise.
9536         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9537         (<su><maxmin><mode>3): ...This, refactor.
9538         (s<maxmin><mode>3): New.
9539         (<maxmin_uns><mode>3): Likewise.
9540         (reduc_<maxmin_uns>_<mode>): Refactor.
9541         (reduc_<maxmin_uns>_v4sf): Likewise.
9542         (reduc_<maxmin_uns>_v2si): Likewise.
9543         (aarch64_<fmaxmin><mode>: Remove.
9544         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9545         new builtin names.
9546         (vmin<q>_f<32,64>): Likewise.
9547         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9548         (FMAXMIN): New.
9549         (su): Add mappings for smax, smin, umax, umin.
9550         (maxmin): New.
9551         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9552         (FMAXMIN): Rename as...
9553         (FMAXMIN_UNS): ...This.
9554         (maxminv): Remove.
9555         (fmaxminv): Likewise.
9556         (fmaxmin): Likewise.
9557         (maxmin_uns): New.
9558         (maxmin_uns_op): Likewise.
9560 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9562         * config/aarch64/arm_neon.h
9563         (vac<ge, gt><sd>_f<32, 64>): Rename to...
9564         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9565         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9567 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9569         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9570         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9572 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9574         * config/aarch64/aarch64-simd.md
9575         (vcond<mode>_internal): Handle special cases for constant masks.
9576         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9577         (vcondu<mode><mode>): Likewise.
9578         (vcond<v_cmp_result><mode>): New.
9580 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9582         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9583         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9584         * config/aarch64/aarch64-simd-builtins.def
9585         (cmeq): Update to BUILTIN_VALLDI.
9586         (cmgt): Likewise.
9587         (cmge): Likewise.
9588         (cmle): Likewise.
9589         (cmlt): Likewise.
9590         * config/aarch64/arm_neon.h
9591         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9592         to builtins or C as appropriate.
9594 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9596         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9597         (cmgeu): ...This.
9598         (cmhi): Rename to...
9599         (cmgtu): ...This.
9600         * config/aarch64/aarch64-simd.md
9601         (simd_mode): Add SF.
9602         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9603         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9604         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9605         (cstore<mode>_neg): ...This.
9606         * config/aarch64/iterators.md
9607         (VALLF): new.
9608         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9609         (COMPARISONS): New.
9610         (UCOMPARISONS): Likewise.
9611         (optab): Add missing comparisons.
9612         (n_optab): New.
9613         (cmp_1): Likewise.
9614         (cmp_2): Likewise.
9615         (CMP): Likewise.
9616         (cmp): Remove.
9617         (VCMP_S): Likewise.
9618         (VCMP_U): Likewise.
9619         (V_cmp_result): Add DF, SF modes.
9620         (v_cmp_result): Likewise.
9621         (v): Likewise.
9622         (vmtype): Likewise.
9623         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9625 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
9627         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9628         define_insn to define_insn_and_split.
9629         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9630         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9631         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9632         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9633         (thumb2_negscc): Likewise.
9635 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9637         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9639 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9641         * config/arm/thumb2.md: Remove trailing whitespaces.
9643 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9645         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9646         Use gen_int_mode rather than GEN_INT.
9648 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
9650         * value-prof.c (stream_in_histogram_value): Remove the strayed
9651         debug_gimple_stmt.
9653 2013-04-30  Richard Biener  <rguenther@suse.de>
9655         PR middle-end/57122
9656         * cfghooks.c (split_edge): Properly check for the loop latch edge.
9658 2013-04-30  Richard Biener  <rguenther@suse.de>
9660         PR middle-end/57107
9661         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9663 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
9665         PR rtl-optimization/56957
9666         PR rtl-optimization/57105
9667         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9668         variable.  Use just INSN_UID for determining whether an insn
9669         should be only disconnected from the insn stream.
9670         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9672 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
9674         PR tree-optimization/57104
9675         * tsan.c (instrument_expr): Don't instrument accesses to
9676         DECL_HARD_REGISTER VAR_DECLs.
9678 2013-04-30  Richard Biener  <rguenther@suse.de>
9680         * function.h (loops_for_fn): New inline function.
9681         (set_loops_for_fn): Likewise.
9682         * cfgloop.h (place_new_loop): Add struct function parameter.
9683         (get_loop): Likewise.
9684         (get_loops): Likewise.
9685         (number_of_loops): Likewise.
9686         (fel_next): Adjust.
9687         (fel_init): Likewise.
9688         * cfg.c (get_loop_copy): Adjust.
9689         * cfgloop.c (flow_loops_dump): Likewise.
9690         (record_loop_exits): Likewise.
9691         (verify_loop_structure): Likewise.
9692         * cfgloopanal.c (mark_irreducible_loops): Likewise.
9693         (estimate_reg_pressure_cost): Likewise.
9694         (mark_loop_exit_edges): Likewise.
9695         * cfgloopmanip.c (place_new_loop): Likewise.
9696         (add_loop): Likewise.
9697         (duplicate_loop): Likewise.
9698         * graph.c (draw_cfg_nodes): Likewise.
9699         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9700         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9701         (extract_affine_chrec): Likewise.
9702         (build_scop_iteration_domain): Likewise.
9703         * graphite.c (graphite_initialize): Likewise.
9704         * ira-build.c (create_loop_tree_nodes): Likewise.
9705         (more_one_region_p): Likewise.
9706         (rebuild_regno_allocno_maps): Likewise.
9707         (mark_loops_for_removal): Likewise.
9708         (mark_all_loops_for_removal): Likewise.
9709         (remove_unnecessary_regions): Likewise.
9710         (ira_build): Likewise.
9711         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9712         * loop-init.c (fix_loop_structure): Likewise.
9713         (gate_rtl_move_loop_invariants): Likewise.
9714         (gate_rtl_unswitch): Likewise.
9715         (gate_rtl_unroll_and_peel_loops): Likewise.
9716         (rtl_doloop): Likewise.
9717         * lto-streamer-in.c (input_cfg): Likewise.
9718         * lto-streamer-out.c (output_cfg): Likewise.
9719         * modulo-sched.c (sms_schedule): Likewise.
9720         * predict.c (tree_estimate_probability): Likewise.
9721         (tree_estimate_probability_driver): Likewise.
9722         (estimate_loops): Likewise.
9723         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9724         (move_sese_region_to_fn): Likewise.
9725         (debug_loop_num): Likewise.
9726         * tree-chrec.c (chrec_evaluate): Likewise.
9727         (hide_evolution_in_other_loops_than_loop): Likewise.
9728         (chrec_component_in_loop_num): Likewise.
9729         (reset_evolution_in_loop): Likewise.
9730         (evolution_function_is_invariant_rec_p): Likewise.
9731         * tree-if-conv.c (main_tree_if_conversion): Likewise.
9732         * tree-inline.c (copy_loops): Likewise.
9733         (copy_cfg_body): Likewise.
9734         (tree_function_versioning): Likewise.
9735         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9736         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9737         Likewise.
9738         (add_to_evolution_1): Likewise.
9739         (scev_const_prop): Likewise.
9740         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9741         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9742         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9743         (tree_ssa_lim_initialize): Likewise.
9744         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9745         (verify_loop_closed_ssa): Likewise.
9746         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
9747         (tree_ssa_loop_im): Likewise.
9748         (tree_ssa_loop_unswitch): Likewise.
9749         (tree_vectorize): Likewise.
9750         (check_data_deps): Likewise.
9751         (tree_ssa_loop_ivcanon): Likewise.
9752         (tree_ssa_loop_bounds): Likewise.
9753         (tree_complete_unroll): Likewise.
9754         (tree_complete_unroll_inner): Likewise.
9755         (tree_parallelize_loops): Likewise.
9756         (tree_ssa_loop_prefetch): Likewise.
9757         (tree_ssa_loop_ivopts): Likewise.
9758         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9759         * tree-vectorizer.c (vectorize_loops): Likewise.
9761 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
9763         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9764         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9765         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9766         with EABI_LINK_SPEC.
9768 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9770         PR target/44578
9771         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9772         alternative.
9774 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
9776         PR target/57097
9777         * lra-constraints.c (process_alt_operands): Discourage a bit more
9778         using memory for pseudos.  Print cost dump for alternatives.
9779         Modify cost values for conflicts with early clobbers.
9780         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9782 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9784         PR target/57098
9785         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9787 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9789         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9790         from/to S register.
9791         (movdi_aarch64): Support LDR/STR from/to D register.
9793 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9795         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9796         or higher by default.
9798 2013-04-29  Richard Biener  <rguenther@suse.de>
9800         PR middle-end/57075
9801         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9802         even if not adding abnormal edges for calls that can make
9803         abnormal gotos.
9805 2013-04-29  Richard Biener  <rguenther@suse.de>
9807         PR middle-end/57103
9808         * tree-cfg.c (move_stmt_op): Fix condition under which to update
9809         TREE_BLOCK.
9810         (move_stmt_r): Remove redundant checking.
9812 2013-04-29  Teresa Johnson  <tejohnson@google.com>
9814         PR bootstrap/57077
9815         * basic-block.h (apply_scale): New function.
9816         (apply_probability): Use apply_scale.
9817         * gimple-streamer-in.c (input_bb): Ditto.
9818         * lto-streamer-in.c (input_cfg): Ditto.
9819         * lto-cgraph.c (merge_profile_summaries): Ditto.
9820         * tree-optimize.c (execute_fixup_cfg): Ditto.
9821         * tree-inline.c (copy_bb): Update comment to use apply_scale.
9822         (copy_edges_for_bb): Ditto.
9823         (copy_cfg_body): Ditto.
9825 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9827         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9828         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9829         (tail_merge_optimize): Handle current_loops == NULL.
9831 2013-04-26  Jeff Law  <law@redhat.com>
9833         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9834         (simplify_cond_using_ranges): Generalize code to simplify
9835         COND_EXPRs where one argument is a constant and the other
9836         is an SSA_NAME created by an integral type conversion.
9838 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9840         * config/arm/arm.md (store_minmaxsi): Use only when
9841         optimize_insn_for_size_p.
9843 2013-04-29  Christian Bruel  <christian.bruel@st.com>
9845         PR target/57108
9846         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9848 2013-04-29  Richard Biener  <rguenther@suse.de>
9850         PR middle-end/57089
9851         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9852         loop tree make sure to schedule a fixup for the child as well.
9853         (expand_omp_for_generic): Properly add loops.
9854         (expand_omp_for_static_nochunk): Likewise.
9855         (expand_omp_for_static_chunk): Likewise.
9856         (expand_omp_for): For the degenerate case fixup loops.
9857         (expand_omp_sections): Fix default bb placement in loops.
9858         (expand_omp_atomic_pipeline): Properly add loops.
9860 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9862         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9864 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9866         * tree-ssa-tail-merge.c: Update header comment.
9868 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9870         * config/aarch64/arm_neon.h
9871         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9872         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9873         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9874         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9875         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9876         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9877         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9878         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9880 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9882         * config/aarch64/aarch64-simd.md
9883         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9884         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9885         fix_trunc, fixuns_trunc.
9886         (ftrunc<VDQF:mode>2): New.
9887         * config/aarch64/iterators.md (optab): Add fix, fixuns.
9888         (fix_trunc_optab): New.
9890 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9892         * config/aarch64/aarch64-builtins.c
9893         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9894         iceilf, lround, iroundf.
9896 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9898         PR target/54349
9899         * config/i386/i386.h (enum ix86_tune_indices)
9900         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9901         New, split from X86_TUNE_INTER_UNIT_MOVES.
9902         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9903         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9904         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9905         (TARGET_INTER_UNIT_MOVES): Remove.
9906         * config/i386/i386.c (initial_ix86_tune_features): Update.
9907         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9908         (ix86_expand_convert_uns_didf_sse): Use
9909         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9910         (ix86_expand_vector_init_one_nonzero): Ditto.
9911         (ix86_expand_vector_init_interleave): Ditto.
9912         (inline_secondary_memory_needed): Return true for moves from SSE class
9913         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9914         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9915         * config/i386/constraints.md (Yi, Ym): Depend on
9916         TARGET_INTER_UNIT_MOVES_TO_VEC.
9917         (Yj, Yn): New constraints.
9918         * config/i386/i386.md (*movdi_internal): Change constraints of
9919         operand 1 from Yi to Yj and from Ym to Yn.
9920         (*movsi_internal): Ditto.
9921         (*movdf_internal): Ditto.
9922         (*movsf_internal): Ditto.
9923         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9924         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9925         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9926         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9927         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9928         * config/i386/sse.md (movdi_to_sse): Ditto.
9929         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9930         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9931         TARGET_INTER_UNIT_MOVES.
9932         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9933         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9934         instead of TARGET_INTER_UNIT_MOVES.
9935         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9936         operand 1 from Yi to Yj and from Ym to Yn.
9938 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9940         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9941         (float_truncate_hi_): Likewise.
9942         (float_extend_lo_): Likewise.
9943         (float_truncate_lo_): Likewise.
9944         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9945         (aarch64_float_extend_lo_v2df): Likewise.
9946         (vec_unpacks_hi_v4sf): Likewise.
9947         (aarch64_float_truncate_lo_v2sf): Likewise.
9948         (aarch64_float_truncate_hi_v4sf): Likewise.
9949         (vec_pack_trunc_v2df): Likewise.
9950         (vec_pack_trunc_df): Likewise.
9952 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9954         * config/aarch64/aarch64-builtins.c
9955         (aarch64_fold_builtin): Fold float conversions.
9956         * config/aarch64/aarch64-simd-builtins.def
9957         (floatv2si, floatv4si, floatv2di): New.
9958         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9959         * config/aarch64/aarch64-simd.md
9960         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9961         * config/aarch64/iterators.md (FLOATUORS): New.
9962         (optab): Add float, floatuns.
9963         (su_optab): Likewise.
9965 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9967         * config/aarch64/aarch64-builtins.c
9968         (aarch64_builtin_vectorized_function): Use new names for
9969         fcvt builtins.
9970         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9971         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9972         (fcvtzu): Split as...
9973         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9974         (fcvtas): Split as...
9975         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9976         (fcvtau): Split as...
9977         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9978         (fcvtps): Split as...
9979         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9980         (fcvtpu): Split as...
9981         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9982         (fcvtms): Split as...
9983         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9984         (fcvtmu): Split as...
9985         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9986         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9987         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9988         (lfrintnusf, lfrintnudf): Likewise.
9989         * config/aarch64/aarch64-simd.md
9990         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9991         define_insn.
9992         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9993         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9994         (fcvt_pattern): Likewise.
9996 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9998         * config/aarch64/aarch64-simd.md
9999         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10000         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10002 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10004         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10005         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10006         (vrnd<a,m,n,p>_f32): Implement using builtins.
10007         (vrnd<i,x><q>_f<32, 64>): New.
10009 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10011         * config/aarch64/aarch64-builtins.c
10012         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10013         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10014         (frintz): Rename to...
10015         (btrunc): ...this.
10016         (frintp): Rename to...
10017         (ceil): ...this.
10018         (frintm): Rename to...
10019         (floor): ...this.
10020         (frinti): Rename to...
10021         (nearbyint): ...this.
10022         (frintx): Rename to...
10023         (rint): ...this.
10024         (frinta): Rename to...
10025         (round): ...this.
10026         * config/aarch64/aarch64-simd.md
10027         (aarch64_frint<frint_suffix><mode>): Delete.
10028         (<frint_pattern><mode>2): Convert to insn.
10029         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10030         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10031         (frint_pattern): Likewise.
10032         (frint_suffix): Likewise.
10034 2013-04-29  Richard Biener  <rguenther@suse.de>
10036         PR tree-optimization/57081
10037         * loop-init.c: Include tree-flow.h.
10038         (loop_optimizer_finalize): Free number of iteration estimates.
10039         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10041 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
10043         PR tree-optimization/57083
10044         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10045         non-singleton shift count range, zero extend low_bound for uns case.
10047         * config/i386/predicates.md (general_vector_operand): New predicate.
10048         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10049         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10050         if they aren't nonimmediate operands.  If their original values
10051         satisfy const_vector_equal_evenodd_p, don't shift them.
10052         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10053         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
10054         if not nonimmediate_operand.
10055         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10056         instead of register_operand.
10057         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10059 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
10061         * stor-layout.c (finalize_size_functions): Allocate a structure and
10062         reset cfun before dumping the functions.
10064 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
10066         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10068         PR target/56866
10069         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10070         use xop_pmacsdqh if uns_p.
10071         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10072         the immediate rotate count.
10074 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
10076         * rtl.h (struct rtx_def): Add comment for field jump.
10077         (LRA_SUBREG_P): New macro.
10078         * recog.c (register_operand): Check LRA_SUBREG_P.
10079         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10080         stack frame.
10081         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10082         (lra_final_code_change): Skip subreg change for operators.
10083         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10084         if there are no operand changes.
10085         * lra-constraints.c (curr_insn_set): New.
10086         (match_reload): Set LRA_SUBREG_P.
10087         (emit_spill_move): Ditto.
10088         (check_and_process_move): Use curr_insn_set. Process only single
10089         set insns.  Don't initialize sec_mem_p and change_p.
10090         (simplify_operand_subreg): Use LRA_SUBREG_P.
10091         (reg_in_class_p): New function.
10092         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10093         of #ifdef.  Add code to remove cycling.
10094         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10095         non-null disp.  Reload inner instead of disp when base and index
10096         are null.  Try to put lo_sum into register.
10097         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10098         (check_and_process_move): Move code for move cost check to
10099         simple_move_p.  Remove equiv_substitution.
10100         (simple_move_p): New function.
10101         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10102         curr_insn_set.  Call check_and_process_move only for single set
10103         insns.  Use the new function.  Move call of check_and_process_move
10104         after operand equiv substitution and address process.
10106 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10108         PR go/57045
10109         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10110         with nonlocal goto receivers or returns twice calls, ignore
10111         unininitialized values from abnormal edges to nl goto receiver
10112         or returns twice call.
10114 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10116         PR tree-optimization/57051
10117         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10118         and VEC_RSHIFT_EXPR if shift count is a multiple of element
10119         bitsize.
10121 2013-04-26  Richard Biener  <rguenther@suse.de>
10123         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10124         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
10125         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10126         (fixup_loop_arrays_after_move): New function.
10127         (move_sese_region_to_fn): Properly outline the loop tree parts
10128         of the SESE region.
10130 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
10132         * config/i386/i386.md (type, unit): Fix long lines.
10134 2013-04-26  Richard Biener  <rguenther@suse.de>
10136         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10137         (lto-streamer-out.o): Likewise.
10138         * cfgloop.c (init_loops_structure): Export, add struct function
10139         argument and adjust.
10140         (flow_loops_find): Adjust.
10141         * cfgloop.h (enum loop_estimation): Add EST_LAST.
10142         (init_loops_structure): Declare.
10143         * lto-streamer-in.c: Include cfgloop.h.
10144         (input_cfg): Input the loop tree.
10145         * lto-streamer-out.c: Include cfgloop.h.
10146         (output_cfg): Output the loop tree.
10147         (output_struct_function_base): Do not drop PROP_loops.
10149 2013-03-26  Richard Biener  <rguenther@suse.de>
10151         * tree-cfg.c (execute_build_cfg): Build the loop tree.
10152         (pass_build_cfg): Provide PROP_loops.
10153         (move_sese_region_to_fn): Remove loops that are outlined into fn
10154         for now.
10155         * tree-inline.c: Include cfgloop.h.
10156         (initialize_cfun): Do not drop PROP_loops.
10157         (copy_loops): New function.
10158         (copy_cfg_body): Copy loop structure.
10159         (tree_function_versioning): Initialize destination loop tree.
10160         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10161         (pass_parallelize_loops): Do IL verification.
10162         * loop-init.c (loop_optimizer_init): Fixup loops if required.
10163         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10164         the CFG make sure we fixup loops as well.
10165         * tree-ssa-tail-merge.c: Include cfgloop.h.
10166         (replace_block_by): When merging loop latches mark loops for fixup.
10167         * lto-streamer-out.c (output_struct_function_base): Drop
10168         PROP_loops for now.
10169         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10170         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10171         * ipa-split.c: Include cfgloop.h.
10172         (split_function): Add the new return block to the loop tree root.
10173         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10174         whether we have removed the forwarder block.
10175         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10176         * cfgloop.h (place_new_loop): Declare.
10177         * cfgloopmanip.c (place_new_loop): Export.
10178         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10179         (tree-switch-conversion.o): Likewise.
10180         (tree-complex.o): Likewise.
10181         (tree-inline.o): Likewise.
10182         (tree-ssa-tailmerge.o): Likewise.
10183         (ipa-split.o): Likewise.
10184         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10185         (tree-ssa-copy.o): Likewise.
10186         * tree-switch-conversion.c: Include cfgloop.h
10187         (process_switch): If we emit a bit-test cascade, schedule loops
10188         for fixup.
10189         * tree-complex.c: Include cfgloop.h.
10190         (expand_complex_div_wide): Properly add new basic-blocks to loops.
10191         * asan.c: Include cfgloop.h.
10192         (create_cond_insert_point): Properly add new basic-blocks to
10193         loops, schedule loop fixup.
10194         * cfgloop.c (verify_loop_structure): Check that looks are not
10195         marked for fixup.
10196         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10197         to loops.
10198         (expand_omp_for_generic): Likewise.
10199         (expand_omp_sections): Likewise.
10200         (expand_omp_atomic_pipeline): Schedule loops for fixup.
10201         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10202         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10203         is initialized, not when loops are present.
10204         * tree-parloops.c (parallelize_loops): Remove checking here.
10205         * passes.c (init_optimization_passes): Schedule a copy-propagation
10206         pass before complete unrolling of inner loops.
10208 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10210         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10211         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10212         (colorize_init): Add argument to _WIN32 version.
10213         * toplev.c: Include diagnostic-color.h.
10214         (process_options): Default to -fdiagnostics-color=auto if
10215         GCC_COLORS env var is in the environment.
10216         * common.opt (fdiagnostics-color=): Add Var and Init.
10217         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10218         env var is in the environment, the default is auto rather than never.
10220         * diagnostic.h (file_name_as_prefix): Add context argument.
10221         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
10222         the string as locus.
10223         * langhooks.c (lhd_print_error_function): Adjust caller.
10225 2013-04-25  Lawrence Crowl  <crowl@google.com>
10227         * var-tracking.c (shared_hash_def::htab):
10228         Change type to hash_table.  Update dependent calls and types.
10230 2013-04-25  Lawrence Crowl  <crowl@google.com>
10232         * Makefile.in: Update as needed below.
10234         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10235         Move declaration to after the type's method definitons.
10237         * attribs.c (htab_t scoped_attributes::attribute_hash):
10238         Change type to hash_table.  Update dependent calls and types.
10240         * bitmap.c (htab_t bitmap_desc_hash):
10241         Change type to hash_table.  Update dependent calls and types.
10243         * cselib.c (htab_t cselib_hash_table):
10244         Change type to hash_table.  Update dependent calls and types.
10246         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10247         (hash_string_slot_node): Move implementation into lto-streamer.h
10248         struct string_slot_hasher.
10249         (eq_string_slot_node): Likewise.
10251         * data-streamer-out.c: Update output_block::string_hash_table
10252         dependent calls and types.
10254         * dwarf2cfi.c (htab_t trace_index):
10255         Change type to hash_table.  Update dependent calls and types.
10257         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10258         Change type to hash_table.  Update dependent calls and types.
10259         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10260         (htab_t optimize_external_refs::map): Likewise.
10261         (htab_t output_comp_unit::extern_map): Likewise.
10262         (htab_t output_comdat_type_unit::extern_map): Likewise.
10263         (htab_t output_macinfo::macinfo_htab): Likewise.
10264         (htab_t optimize_location_lists::htab): Likewise.
10265         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10267         * except.c (htab_t ehspec_hash_type):
10268         Change type to hash_table.  Update dependent calls and types.
10269         (assign_filter_values::ttypes): Likewise.
10270         (assign_filter_values::ehspec): Likewise.
10271         (sjlj_assign_call_site_values::ar_hash): Likewise.
10272         (convert_to_eh_region_ranges::ar_hash): Likewise.
10274         * gcse.c (htab_t pre_ldst_table):
10275         Change type to hash_table.  Update dependent calls and types.
10277         * ggc-common.c (htab_t saving_htab):
10278         Change type to hash_table.  Update dependent calls and types.
10279         (htab_t loc_hash): Likewise.
10280         (htab_t ptr_hash): Likewise.
10281         (call_count): Rename ggc_call_count.
10282         (call_alloc): Rename ggc_call_alloc.
10283         (loc_descriptor): Rename make_loc_descriptor.
10284         (add_statistics): Rename ggc_add_statistics.
10286         * ggc-common.c (saving_htab):
10287         Change type to hash_table.  Update dependent calls and types.
10289         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10290         (push_gimplify_context): Likewise.
10291         (pop_gimplify_context): Likewise.
10292         (struct gimple_temp_hash_elt): Added.
10293         (struct gimplify_hasher): Likewise.
10294         (struct gimplify_ctx.temp_htab):
10295         Change type to hash_table.  Update dependent calls and types.
10297         * gimple-fold.c: Include gimplify-ctx.h.
10299         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10300         Change type to hash_table.  Update dependent calls and types.
10301         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10302         avoid potential global name collision.
10304         * gimplify.c: Include gimplify-ctx.h.
10305         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10306         (htab_t gimplify_ctx::temp_htab):
10307         Update dependent calls and types for new type hash_table.
10308         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10309         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10311         * gimplify-ctx.h: New.
10312         (struct gimple_temp_hash_elt): Move from gimplify.c.
10313         (class gimplify_hasher): New.
10314         (struct gimplify_ctx): Move from gimple.h.
10315         (htab_t gimplify_ctx::temp_htab):
10316         Change type to hash_table.  Update dependent calls and types.
10318         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10319         (htab_t ivs_params::newivs_index):
10320         Change type to hash_table.  Update dependent calls and types.
10321         (htab_t ivs_params::params_index): Likewise.
10322         (htab_t print_generated_program::params_index): Likewise.
10323         (htab_t gloog::newivs_index): Likewise.
10324         (htab_t gloog::params_index): Likewise.
10326         * graphite.c: Include graphite-htab.h.
10327         4htab_t graphite_transform_loops::bb_pbb_mapping):
10328         Change type to hash_table.  Update dependent calls and types.
10330         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10331         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10332         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10334         * graphite-dependences.c: Include graphite-htab.h.
10335         (loop_is_parallel_p): Change hash table type of parameter.
10337         * graphite-htab.h: New.
10338         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10339         (extern find_pbb_via_hash): Move from graphite-poly.h.
10340         (extern loop_is_parallel_p): Move from graphite-poly.h.
10341         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10343         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10344         (extern loop_is_parallel_p): Move to graphite-htab.h.
10345         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10347         * haifa-sched.c (htab_t delay_htab):
10348         Change type to hash_table.  Update dependent calls and types.
10349         (htab_t delay_htab_i2): Likewise.
10351         * ira-color.c (htab_t allocno_hard_regs_htab):
10352         Change type to hash_table.  Update dependent calls and types.
10354         * ira-costs.c (htab_t cost_classes_htab):
10355         Change type to hash_table.  Update dependent calls and types.
10357         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10358         Change type to hash_table.  Update dependent calls and types.
10360         * loop-iv.c (htab_t bivs):
10361         Change type to hash_table.  Update dependent calls and types.
10363         * loop-unroll.c (htab_t opt_info::insns_to_split):
10364         Change type to hash_table.  Update dependent calls and types.
10365         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10367         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10368         (struct string_slot_hasher): New.
10369         (htab_t output_block::string_hash_table):
10370         Change type to hash_table.  Update dependent calls and types.
10372         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10373         (htab_t file_name_hash_table):
10374         Change type to hash_table.  Update dependent calls and types.
10376         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10377         calls and types.
10379         * lto-streamer.c (htab_t tree_htab):
10380         Change type to hash_table.  Update dependent calls and types.
10382         * omp-low.c: Include gimplify-ctx.h.
10384         * passes.c (htab_t name_to_pass_map):
10385         Change type to hash_table.  Update dependent calls and types.
10386         (pass_traverse): Rename to passes_pass_traverse.
10388         * plugin.c (htab_t event_tab):
10389         Change type to hash_table.  Update dependent calls and types.
10391         * postreload-gcse.c (htab_t expr_table):
10392         Change type to hash_table.  Update dependent calls and types.
10393         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10395         * sese.c (debug_rename_map_1): Make extern.
10396         (htab_t copy_bb_and_scalar_dependences::rename_map):
10397         Change type to hash_table.  Update dependent calls and types.
10399         * sese.h (extern debug_rename_map): Move to .c file.
10401         * store-motion.c (htab_t store_motion_mems_table):
10402         Change type to hash_table.  Update dependent calls and types.
10404         * trans-mem.c (htab_t tm_new_mem_hash):
10405         Change type to hash_table.  Update dependent calls and types.
10407         * tree-browser.c (htab_t TB_up_ht):
10408         Change type to hash_table.  Update dependent calls and types.
10410         * tree-cfg.c (htab_t discriminator_per_locus):
10411         Change type to hash_table.  Update dependent calls and types.
10413         * tree-complex.c: Include tree-hasher.h
10414         (htab_t complex_variable_components):
10415         Change type to hash_table.  Update dependent calls and types.
10417         * tree-eh.c (htab_t finally_tree):
10418         Change type to hash_table.  Update dependent calls and types.
10420         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10421         struct int_tree_hasher.
10422         (extern int_tree_map_eq): Likewise.
10423         (uid_decl_map_hash): Removed.
10424         (extern decl_tree_map_eq): Likewise.
10426         * tree-hasher.h: New.
10427         (struct int_tree_hasher): New.
10428         (typedef int_tree_htab_type): New.
10430         * tree-inline.c: Include gimplify-ctx.h.
10432         * tree-mudflap.c: Include gimplify-ctx.h.
10434         * tree-parloops.c: Include tree-hasher.h.
10435         (htab_t eliminate_local_variables_stmt::decl_address):
10436         Change type to hash_table.  Update dependent calls and types.
10437         (htab_t separate_decls_in_region::decl_copies): Likewise.
10439         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10440         Change type to hash_table.  Update dependent calls and types.
10442         * tree-sra.c (candidates):
10443         Change type to hash_table.  Update dependent calls and types.
10445         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10446         in tree-flow.h.
10447         (int_tree_map_hash): Likewise.
10449         * tree-ssa-dom.c (htab_t avail_exprs):
10450         Change type to hash_table.  Update dependent calls and types.
10452         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10453         Change type to hash_table.  Update dependent calls and types.
10455         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10456         Change type to hash_table.  Update dependent calls and types.
10458         * tree-ssa-phiopt.c (seen_ssa_names):
10459         Change type to hash_table.  Update dependent calls and types.
10461         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10462         Change type to hash_table.  Update dependent calls and types.
10464         * tree-ssa-uncprop.c (equiv):
10465         Change type to hash_table.  Update dependent calls and types.
10467 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10469         PR rtl-optimization/57003
10470         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10471         call note_stores with kill_clobbered_value callback again after
10472         killing regs_invalidated_by_call.
10474 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10476         * config/aarch64/aarch64-simd.md
10477         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10478         (aarch64_simd_bsl<mode>): Likewise.
10479         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10481 2013-04-25  Marek Polacek  <polacek@redhat.com>
10483         PR tree-optimization/57066
10484         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10486 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10488         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10490 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10492         * config/aarch64/aarch64-builtins.c
10493         (aarch64_fold_builtin): New.
10494         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10495         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10496         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10497         * config/aarch64/arm_neon.h
10498         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10500 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10501             Tejas Belagod  <tejas.belagod@arm.com>
10503         * config/aarch64/aarch64-builtins.c
10504         (aarch64_gimple_fold_builtin): New.
10505         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10506         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10507         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10508         (addvv4sf): Update.
10509         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10511 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10513         * config/aarch64/aarch64.md
10514         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10516 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10518         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10519         (*ngcsi_uxtw): New pattern.
10521 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10522             Julian Brown  <julian@codesourcery.com>
10524         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10525         (TB_DREG): Add T_V4HF.
10526         (v4hf_UP): New macro.
10527         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10528         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10529         Handle initialisation of V4HF. Adjust initialisation of reinterpret
10530         built-ins.
10531         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10532         (arm_vector_mode_supported_p): Handle V4HF.
10533         (arm_mangle_map): Handle V4HFmode.
10534         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10535         * config/arm/arm_neon_builtins.def: Add entries for
10536         vcvtv4hfv4sf, vcvtv4sfv4hf.
10537         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10538         (neon_vcvtv4hfv4sf): Likewise.
10539         * config/arm/neon-gen.ml: Handle half-precision floating point
10540         features.
10541         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10542         * config/arm/arm_neon.h: Regenerate.
10543         * config/arm/neon.ml (type elts): Add F16.
10544         (type vectype): Add T_float16x4, T_floatHF.
10545         (type vecmode): Add V4HF.
10546         (type features): Add Requires_FP_bit feature.
10547         (elt_width): Handle F16.
10548         (elt_class): Likewise.
10549         (elt_of_class_width): Likewise.
10550         (mode_of_elt): Refactor.
10551         (type_for_elt): Handle F16, fix error messages.
10552         (vectype_size): Handle T_float16x4.
10553         (vcvt_sh): New function.
10554         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10555         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10556         (string_of_mode): Handle V4HF.
10557         * doc/arm-neon-intrinsics.texi: Regenerate.
10559 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10561         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10562         format specifier in 'X' case.
10564 2013-04-25  Alan Modra  <amodra@gmail.com>
10566         PR target/57052
10567         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10568         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10569         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10570         Repeat for many other rotate/shift and mask patterns using subregs.
10571         Name lshiftrt insns.
10572         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10573         on WORDS_BIG_ENDIAN.
10575 2013-04-25  Alan Modra  <amodra@gmail.com>
10577         * config.gcc: Support little-endian powerpc-linux targets.
10578         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10579         (LINK_OS_LINUX_SPEC): Define.
10580         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10581         Preserve MASK_LITTLE_ENDIAN.
10582         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10583         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10584         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10585         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10586         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10587         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10588         Correct fp word order for little-endian.  Don't shift toc entries
10589         smaller than a word for little-endian.
10590         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10591         (bswapdi2 splits): Correct low-part subreg for little-endian.
10592         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10593         low/high where such is correct only for be.
10594         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10595         little-endian for -mcall-aixdesc.
10597 2013-04-25  Alan Modra  <amodra@gmail.com>
10599         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10600         replace_equiv_address_nv.
10602 2013-04-25  Alan Modra  <amodra@gmail.com>
10604         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10606 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10608         Revert:
10609         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10610         * rtl.h (struct rtx_def): ...
10612 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10614         PR rtl-optimizations/57046
10615         * lra-constraints (split_reg): Set up lra_risky_transformations_p
10616         for multi-reg splits.
10618 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
10620         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10622 2013-04-24  Sterling Augustine  <saugustine@google.com>
10624         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10625         (comp_dir_string, debug_str_dwo_section): New.
10626         (DEBUG_STR_DWO_SECTION): Rename to ...
10627         (DEBUG_DWO_STR_SECTION): ... this.
10628         (DEBUG_NORM_STR_SECTION): Delete.
10629         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10630         (DEBUG_STR_DWO_SECTION_FLAGS): New.
10631         (find_AT_string): Move most logic to ...
10632         (find_AT_string_in_table): ... here.  New.
10633         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10634         add_skeleton_AT_string.  Delete logic.
10635         (output_skeleton_debug_sections): Remove call to
10636         add_top_level_skeleton_die_attrs.
10637         (add_comp_dir_attribute): Move logic to comp_dir_string.
10638         (dwarf2out_init): Initialize debug_str_dwo_section.
10639         (output_indirect_string): Call find_string_form.
10640         (output_indirect_strings): Rewrite.
10641         (prune_unused_types): Empty skeleton_debug_str_hash.
10642         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10643         (dwarf2out_finish):  Call output_indirect_strings.
10645 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10647         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10649 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10651         * rtl.h (struct rtx_def): Add comment for field jump.
10652         (LRA_SUBREG_P): New macro.
10653         * recog.c (register_operand): Check LRA_SUBREG_P.
10654         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10655         stack frame.
10656         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10657         (lra_final_code_change): Skip subreg change for operators.
10658         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10659         if there are no operand changes.
10660         * lra-constraints.c (curr_insn_set): New.
10661         (match_reload): Set LRA_SUBREG_P.
10662         (emit_spill_move): Ditto.
10663         (check_and_process_move): Use curr_insn_set. Process only single
10664         set insns.  Don't initialize sec_mem_p and change_p.
10665         (simplify_operand_subreg): Use LRA_SUBREG_P.
10666         (reg_in_class_p): New function.
10667         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10668         of #ifdef.  Add code to remove cycling.
10669         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10670         non-null disp.  Reload inner instead of disp when base and index
10671         are null.  Try to put lo_sum into register.
10672         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10673         (check_and_process_move): Move code for move cost check to
10674         simple_move_p.  Remove equiv_substitution.
10675         (simple_move_p): New function.
10676         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10677         curr_insn_set.  Call check_and_process_move only for single set
10678         insns.  Use the new function.  Move call of check_and_process_move
10679         after operand equiv substitution and address process.
10681 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
10683         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10684         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10685         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10687 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10689         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10691 2013-04-24  Marek Polacek  <polacek@redhat.com>
10693         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10694         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10695         (select_loops_exit_conditions): Likewise.
10696         (number_of_iterations_for_all_loops): Likewise.
10697         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10698         (scev_analysis): Likewise.
10700 2013-04-02  Catherine Moore  <clm@codesourcery.com>
10701             Chao-ying Fu <fu@mips.com>
10703         * config/mips/micromips.md (jraddiusp): New pattern.
10704         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10705         instruction if possible.
10707 2013-04-24  Alan Modra  <amodra@gmail.com>
10709         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10711 2013-04-24  Julian Brown  <julian@codesourcery.com>
10712             Chung-Lin Tang  <cltang@codesourcery.com>
10714         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10715         dependency behavior in enumeration type DIE generation. Add TODO note
10716         to comments about future DW_FORM_sdata/udata re-work of related code.
10718 2013-04-23  Lawrence Crowl  <crowl@google.com>
10720         * Makefile.in: Update as needed below.
10722         * hash-table.h (class hash_table):
10723         Correct many methods with parameter types compare_type to the correct
10724         value_type.  (Correct code was unlikely to notice the change.)
10725         (hash_table::elements_with_deleted) New.
10726         (class hashtable::iterator): New.
10727         (hashtable::begin()): New.
10728         (hashtable::end()): New.
10729         (FOR_EACH_HASH_TABLE_ELEMENT): New.
10731         * statistics.c (statistics_hashes):
10732         Change type to hash_table.  Update dependent calls and types.
10734         * tree-into-ssa.c (var_infos):
10735         Change type to hash_table.  Update dependent calls and types.
10737         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10738         Change type to hash_table.  Update dependent calls and types.
10740         * tree-ssa-loop-im.c (struct mem_ref.refs):
10741         Change type to hash_table.  Update dependent calls and types.
10743         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10744         Change type to hash_table.  Update dependent calls and types.
10746         * tree-ssa-sccvn.c (vn_tables_s::nary):
10747         Change type to hash_table.  Update dependent calls and types.
10748         (vn_tables_s::phis): Likewise.
10749         (vn_tables_s::references): Likewise.
10751         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10752         (vn_reference_eq): Update parameter and return types.
10754         * tree-ssa-structalias.c (pointer_equiv_class_table):
10755         Change type to hash_table.  Update dependent calls and types.
10756         (location_equiv_class_table): Likewise.
10758         * tree-vect-data-refs.c: Consequential changes for making
10759         peeling a hash_table.
10761         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10762         (destroy_loop_vec_info): Dependent hash_table update.
10764         * tree-vectorizer.h (peeling_htab):
10765         Change type to hash_table.  Update dependent calls and types.
10767 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
10769         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10770         to check the register content is equal or not.
10771         * lra-constraints.c (match_reload): Use lra_assign_reg_val
10772         to assign register content record.
10773         * lra-eliminations.c (update_reg_eliminate): Use
10774         lra_update_reg_val_offset to update register content offset.
10775         * lra-int.h (struct lra_reg): Add offset member.
10776         (lra_reg_val_equal_p): New static inline function.
10777         (lra_update_reg_val_offset): New static inline function.
10778         (lra_assign_reg_val): New static inline function.
10779         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10780         to assign register content record.
10781         (initialize_lra_reg_info_element): Initial offset to zero.
10783 2013-04-23  Catherine Moore  <clm@codesourcery.com>
10785         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10786         operands.  Record compression.
10788 2013-04-23  Xinliang David Li  <davidxl@google.com>
10790         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10792 2013-04-23  Richard Biener  <rguenther@suse.de>
10794         PR middle-end/57036
10795         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10796         parameter, only add abnormal goto edges from the copied body
10797         if the call could perform abnormal gotos.
10798         (copy_cfg_body): Adjust.
10800 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
10802         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10804 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
10806         * coretypes.h (gimple_stmt_iterator): Add struct to make
10807         compatible with C.
10809 2013-04-23  Richard Biener  <rguenther@suse.de>
10811         PR tree-optimization/57026
10812         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10813         from SSA names occuring in abnormal PHI nodes.
10815 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10817         * lto/lto.c (print_lto_report_1): Fix LTO report names.
10819 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10821         * lto/lto.c (print_lto_report_1): Declare early.
10822         (read_cgraph_and_symbols): Call print_lto_report_1 early.
10824 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10826         * common.opt (-flto-report-wpa): Add.
10827         * doc/invoke.texi (-flto-report-wpa): Add.
10828         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10829         (lto_main): dito.
10831 2013-04-22  Xinliang David Li  <davidxl@google.com>
10833         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10834         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10835         * Makefile.in: New dependency
10837             David Daney <ddaney.cavm@gmail.com>
10839         * configure.ac (gcc_cv_as_micromips_support):  Use the
10840         --fatal-warnings option.
10841         * configure: Regenerate.
10843 2013-04-22  Marek Polacek  <polacek@redhat.com>
10845         PR sanitizer/56990
10846         * tsan.c (instrument_expr): Don't instrument expression
10847         in case its size is zero.
10849 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
10851         PR target/57032
10852         Revert:
10853         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
10855         * config/alpha/alpha.c (TARGET_LRA_P): New define.
10857 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10859         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10860         (gimple_stmt_iterator): New typedef.
10861         * gimple.h (gimple_stmt_iterator): Rename to...
10862         (gimple_stmt_iterator_d): ... This.
10863         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10864         trees be valid for GIMPLE and GENERIC.
10865         (TARGET_GIMPLE_FOLD_BUILTIN): New.
10866         * gimple-fold.c (gimple_fold_call): Call target hook
10867         gimple_fold_builtin.
10868         * hooks.c (hook_bool_gsiptr_false): New.
10869         * hooks.h (hook_bool_gsiptr_false): New.
10870         * target.def (fold_stmt): New.
10871         * doc/tm.texi: Regenerate.
10873 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
10875         PR target/57018
10876         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10877         a set sp if no stack realignment.
10879 2013-04-22  Nick Clifton  <nickc@redhat.com>
10881         * config.gcc (tilegx-linux): Extend extra_objs rather than
10882         overwriting it.
10883         (tilepro-linux): Likewise.
10885 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10887         * config/aarch64/aarch64-builtins.c
10888         (CF): Remove.
10889         (CF0, CF1, CF2, CF3, CF4, CF10): New.
10890         (VAR<1-12>): Add MAP parameter.
10891         (BUILTIN_*): Likewise.
10892         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10893         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10894         (aarch64_ushl_n<mode>): Likewise.
10895         (aarch64_sshr_n<mode>): Likewise.
10896         (aarch64_ushr_n<mode>): Likewise.
10897         (aarch64_<maxmin><mode>): Likewise.
10898         (aarch64_sqrt<mode>): Likewise.
10899         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10900         (vshr<q>_n_*): Likewise.
10902 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10904         * config/aarch64/aarch64-builtins.c
10905         (aarch64_simd_builtin_type_mode): Handle SF types.
10906         (sf_UP): Define.
10907         (BUILTIN_GPF): Define.
10908         (aarch64_init_simd_builtins): Handle SF types.
10909         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10910         (frecps): Likewise.
10911         (frecpx): Likewise.
10912         * config/aarch64/aarch64-simd.md
10913         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10914         (aarch64_frecpe<mode>): New.
10915         (aarch64_frecps<mode>): Likewise.
10916         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10917         (v8type): Add frecp<esx>.
10918         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10919         (aarch64_frecps<mode>): Likewise.
10920         * config/aarch64/iterators.md (FRECP): New.
10921         (frecp_suffix): Likewise.
10922         * config/aarch64/arm_neon.h
10923         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10925 2013-04-22  Christian Bruel  <christian.bruel@st.com>
10927         PR target/56995
10928         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10929         (REG_CLASS_NAMES): Idem.
10930         (REG_CLASS_CONTENTS): Idem.
10931         (REGCLASS_HAS_FP_REG): Idem.
10932         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10933         (sh_conditional_register_usage): Idem.
10935 2013-04-21  Jeff Law  <law@redhat.com>
10937         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10938         (ssa_forward_propagate_and_combine): Use it.
10940 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10942         * lra.c: Update the flow chart diagram.
10944 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10946         PR rtl-optimization/56847
10947         * lra-constraints.c (process_alt_operands): Discourage alternative
10948         with non-matche doffsettable memory constraint fro memory with
10949         known offset.
10951 2013-04-19  Richard Biener  <rguenther@suse.de>
10953         PR tree-optimization/56982
10954         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10955         function.
10956         * gimplify.c (gimplify_call_expr): Notice special calls.
10957         (gimplify_modify_expr): Likewise.
10958         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10959         abnormal control flow receivers.
10960         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10961         in the same way as cfun->has_nonlocal_labels.
10962         (gimple_purge_dead_abnormal_call_edges): Likewise.
10963         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10964         receivers start a basic-block.
10966 2013-04-19  Richard Biener  <rguenther@suse.de>
10968         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10969         member ...
10970         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
10971         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10972         (SLP_TREE_LOAD_PERMUTATION): Add.
10973         (vect_transform_slp_perm_load): Adjust prototype.
10974         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10975         (vect_free_slp_instance): Likewise.
10976         (vect_create_new_slp_node): Likewise.
10977         (vect_supported_slp_permutation_p): Remove.
10978         (vect_slp_rearrange_stmts): Adjust.
10979         (vect_supported_load_permutation_p): Likewise.  Inline
10980         vect_supported_slp_permutation_p here.
10981         (vect_analyze_slp_instance): Compute load permutations per
10982         slp node instead of per instance.
10983         (vect_get_slp_defs): Adjust.
10984         (vect_transform_slp_perm_load): Likewise.
10985         (vect_schedule_slp_instance): Remove redundant code.
10986         (vect_schedule_slp): Remove hack for PR56270, add it ...
10987         * tree-vect-stmts.c (vectorizable_load): ... here, do not
10988         CSE loads for SLP.  Adjust.
10990 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10992         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10993         spelling in two comments.
10995 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10997         PR target/56797
10998         * config/arm/arm.c (load_multiple_sequence): Require SP
10999         as base register for loads if SP is in the register list.
11001 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11003         PR tree-optimization/56718
11004         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11005         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11006         and made public.  Adjusted all callers.
11007         (ipa_intraprocedural_devirtualization): New function.
11008         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11009         (ipa_intraprocedural_devirtualization): Likewise.
11010         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11012 2013-04-19  Richard Biener  <rguenther@suse.de>
11014         PR tree-optimization/57000
11015         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11017 2013-04-19  Terry Guo  <terry.guo@arm.com>
11019         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11020         Replace with ...
11021         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
11022         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11023         (cortex_m4_fmacs): Use new reservations.
11024         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11026 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
11028         PR rtl-optimization/56999
11029         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11030         related code.
11031         (lra_coalesce): Remove split_origin_bitmap and related code.
11032         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
11033         ranges if necessary.
11035 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
11037         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11038         New array.
11039         (ix86_expand_call): Remove clobbered_registers array and use
11040         x86_64_ms_sysv_extra_clobbered_registers instead.
11041         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11042         Declare here.
11043         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11044         predicate.
11045         * config/i386/i386.md (*call_rex64_ms_sysv): Use
11046         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
11047         (*call_value_rex64_ms_sysv): Ditto.
11049 2013-04-18  Cary Coutant  <ccoutant@google.com>
11051         * dwarf2out.c (output_pubnames): Check die_perennial_p of
11052         parent instead of die_mark.
11054 2013-04-18  Diego Novillo  <dnovillo@google.com>
11056         * gimple.c (create_gimple_tmp): New.
11057         (get_expr_type): New.
11058         (build_assign): New.
11059         (build_type_cast): New.
11060         * gimple.h (enum ssa_mode): Define.
11061         (gimple_seq_set_location): New.
11062         * asan.c (build_check_stmt): Change some gimple_build_* calls
11063         to use build_assign and build_type_cast.
11065 2013-04-18  Richard Biener  <rguenther@suse.de>
11067         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11068         handle negative step.  Remove redundant checks.
11069         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11070         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11071         for negative step and grouped loads fail to vectorize.
11073 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11075         * emit-rtl.c (reset_insn_used_flags): New function.
11076         (reset_all_used_flags): Use it.
11077         (verify_insn_sharing): New function.
11078         (verify_rtl_sharing): Fix verification for SEQUENCEs.
11080 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
11082         PR tree-optimization/56984
11083         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11084         and (x >> M) >= N don't register any assertion if N << M is the
11085         minimum value.
11087 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11089         * lower-subreg.c (resolve_simple_move): If called self-recursive,
11090         do not delete_insn insns that have not yet been emitted, only
11091         unlink them with remove_insn.
11092         * df-scan.c (df_insn_delete): Revert r197492.
11094 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
11096         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11097         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11099 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11101         * config/arm/arm.md (movsicc_insn): Convert define_insn into
11102         define_insn_and_split.
11103         (and_scc,ior_scc,negscc): Likewise.
11104         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11106 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11108         * config/arm/arm.c (use_return_insn): Return 0 for targets that
11109         can benefit from using a sequence of LDRD instructions in epilogue
11110         instead of a single LDM instruction.
11112 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11114         PR 45688
11115         * doc/extend.texi: Fix typo.
11117 2013-04-17  Richard Biener  <rguenther@suse.de>
11119         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11120         (vect_build_slp_tree): ... here.
11121         (vect_build_slp_tree_1): Compute which stmts of the SLP group
11122         match.  Remove special-casing of mismatched complex loads.
11123         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11124         re-try the match with swapped commutative operands.
11125         (vect_supported_load_permutation_p): Remove special-casing of
11126         mismatched complex loads.
11127         (vect_analyze_slp_instance): Adjust.
11129 2013-04-17  Richard Biener  <rguenther@suse.de>
11131         PR rtl-optimization/56921
11132         * cfgloop.h (struct loop): Add simple_loop_desc member.
11133         (struct niter_desc): Mark with GTY(()).
11134         (simple_loop_desc): Do not use aux field but simple_loop_desc.
11135         * loop-iv.c (get_simple_loop_desc): Likewise.
11136         (free_simple_loop_desc): Likewise.
11138         Revert
11139         2013-04-16  Richard Biener  <rguenther@suse.de>
11141         PR rtl-optimization/56921
11142         * loop-init.c (pass_rtl_move_loop_invariants): Add
11143         TODO_do_not_ggc_collect to todo_flags_finish.
11144         (pass_rtl_unswitch): Same.
11145         (pass_rtl_unroll_and_peel_loops): Same.
11146         (pass_rtl_doloop): Same.
11148 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
11150         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11151         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11152         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11153         references.
11154         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11155         * tree-streamer.c (record_common_node): Adjust reference in comment.
11157 2013-04-17  Terry Guo  <terry.guo@arm.com>
11159         * config/arm/cortex-m4.md: Add a new bypass.
11161 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11163         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11164         New pattern.
11165         (*subs_<optab><mode>_multp2): New pattern.
11166         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11167         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11169 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11171         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11172         (*subs_mul_imm_<mode>): New pattern.
11174 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
11176         PR target/56948
11177         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11178         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
11179         (vsx_movti_32bit): Change j->wa to O->wa.
11181 2013-04-16  Richard Biener  <rguenther@suse.de>
11183         PR rtl-optimization/56921
11184         * loop-init.c (pass_rtl_move_loop_invariants): Add
11185         TODO_do_not_ggc_collect to todo_flags_finish.
11186         (pass_rtl_unswitch): Same.
11187         (pass_rtl_unroll_and_peel_loops): Same.
11188         (pass_rtl_doloop): Same.
11190 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11192         * config/arm/arm.c (emit_multi_reg_push): New declaration
11193         for an existing function.
11194         (arm_emit_strd_push): New function.
11195         (arm_expand_prologue): Used here.
11196         (arm_emit_ldrd_pop): New function.
11197         (arm_expand_epilogue): Used here.
11198         (arm_get_frame_offsets): Update condition.
11199         (arm_emit_multi_reg_pop): Add a special case for load of a single
11200         register with writeback.
11202 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
11204         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11205         description.
11207 2013-04-16  Richard Biener  <rguenther@suse.de>
11209         PR tree-optimization/56756
11210         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11211         (first_mem_ref_loc): New.
11212         (execute_sm): Place the load temporarily before a previous
11213         access instead of in the latch edge to ensure its SSA dependencies
11214         are defined at points dominating the load.
11216 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
11218         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11219         correct fix by moving header and footer insn to the footer of
11220         the merged basic block.  Clear BB_END of the merged-away block.
11222         PR middle-end/43631
11223         * emit-rtl.c (make_note_raw): New function.
11224         (link_insn_into_chain): New static inline function.
11225         (add_insn): Use it.
11226         (add_insn_before, add_insn_after): Factor insn chain linking code...
11227         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11228         using link_insn_into_chain.
11229         (note_outside_basic_block_p): New helper function for emit_note_after
11230         and emit_note_before.
11231         (emit_note_after): Use nobb variant of add_insn_after if the note
11232         should not be contained in a basic block.
11233         (emit_note_before): Use nobb variant of add_insn_before if the note
11234         should not be contained in a basic block.
11235         (emit_note_copy): Use make_note_raw.
11236         (emit_note): Likewise.
11237         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11238         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11239         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11240         the moved barrier the tail of the basic block it follows.
11241         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11243 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
11245         PR tree-optimization/56962
11246         * gimple-ssa-strength-reduction.c (record_increment): Only set
11247         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11248         either rhs1 or rhs2 is equal to c->base_expr.
11250 2013-04-15  Richard Biener  <rguenther@suse.de>
11252         PR tree-optimization/56933
11253         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11254         member.
11255         (GROUP_READ_WRITE_DEPENDENCE): Remove.
11256         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11257         * tree-vect-data-refs.c (vect_analyze_group_access): Move
11258         dependence check ...
11259         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11260         ... here.
11261         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11262         GROUP_READ_WRITE_DEPENDENCE.
11264 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11266         * emit-rtl.c (reset_all_used_flags): New function.
11267         (verify_rtl_sharing): Call reset_all_used_flags before and after
11268         performing the checks.
11270 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11272         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11273         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11274         * config/arm/constraints.md (De): New constraint.
11275         * config/arm/neon.md (anddi3_neon): Delete.
11276         (neon_vand<mode>): Expand to standard anddi3 pattern.
11277         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11278         Move earlier in the file.
11279         (neon_inv_logic_op2): Likewise.
11280         (arm_anddi_operand_neon): New predicate.
11282 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11284         * configure.ac (gcc_cv_ld_as_needed): Set
11285         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11286         Use -z ignore, -z record on *-*-solaris2*.
11287         (HAVE_LD_AS_NEEDED): Update comment.
11288         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11289         * configure: Regenerate.
11290         * config.in: Regenerate.
11291         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11292         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11293         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11294         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11295         equivalents.  Fix markup.
11296         * doc/tm.texi: Regenerate.
11298 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11300         * config/i386/i386.opt: New option mstack-protector-guard=.
11301         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11302         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11303         TARGET_SSP_TLS_GUARD.
11304         * config/i386/i386.c (ix86_option_override_internal): Set
11305         ix86_stack_protector_guard.
11306         * config/i386/i386.md (stack_protect_set): Enable for
11307         TARGET_SSP_TLS_GUARD only.
11308         (stack_protect_set_<mode>): Ditto.
11309         (stack_protect_test): Ditto.
11310         (stack_protect_test_<mode>): Ditto.
11311         * doc/invoke.texi (i386 Option): Document.
11313 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11315         PR target/56890
11316         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11317         (S_MODES): Set H_MODE bit.
11318         (SF_MODES): Set only S_MODE and SF_MODE bits.
11319         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11320         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11321         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11322         <MODE_FLOAT>: Likewise.
11324 2013-04-15  Joey Ye  <joey.ye@arm.com>
11326         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11328 2013-04-15  Joey Ye  <joey.ye@arm.com>
11330         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11331         for real far jump.
11332         (thumb_far_jump_used_p): Count instruction size and set
11333         far_jump_used.
11335 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11337         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11338         * resource.c (mark_target_live_regs): Reformat conditional block.
11340 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11342         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11343         notes, they are emitted only just before final.
11344         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11346 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11348         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11349         * cfgrtl.c (delete_insn): Call it here instead.
11350         * lra-spills.c (lra_final_code_change): Use delete_insn.
11351         * haifa-sched.c (sched_remove_insn): Likewise.
11352         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11353         returning to the nop pool.
11354         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11355         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11357 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11359         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11360         * doc/tm.texi: Regenerated.
11362 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11364         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11365         QImode checks.
11367 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11369         * df-core.c (df_find_def): Compare register numbers.
11370         (df_find_use): Likewise.
11372 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11374         PR target/56903
11375         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11376         lra_in_progress for return.
11378 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11380         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11381         define_insn into define_insn_and_split and emit movsicc patterns.
11383 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11385         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11387 2013-04-12  Richard Biener  <rguenther@suse.de>
11389         * tree-pass.h (TODO_do_not_ggc_collect): New.
11390         * passes.c (execute_one_ipa_transform_pass): Honor
11391         TODO_do_not_ggc_collect.
11392         (execute_one_pass): Likewise.
11394         Revert
11395         2013-04-10  Richard Biener  <rguenther@suse.de>
11397         * passes.c (init_optimization_passes): Remove reload pass.
11398         * ira.c (do_reload): Merge into ...
11399         (ira): ... this.
11400         (rest_of_handle_reload): Remove.
11401         (pass_reload): Likewise.
11402         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11403         of reload for vzeroupper pass placement.
11405 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11407         PR tree-optimization/56918
11408         PR tree-optimization/56920
11409         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11410         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11411         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11412         use wide_mul_with_sign method.
11414 2013-04-12  Richard Biener  <rguenther@suse.de>
11416         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11417         not be considered a gimple constant.
11419 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11421         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11422         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11423         operations.
11425 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11426             Jakub Jelinek  <jakub@redhat.com>
11428         * opts.c: Include diagnostic-color.h.
11429         (common_handle_option): Handle OPT_fdiagnostics_color_.
11430         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11431         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11432         (diagnostic-color.o): New.
11433         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11434         (diagnostic_color_rule): New enum.
11435         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11436         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11437         the location string.
11438         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11439         either NULL, or color kind.
11440         * diagnostic-color.c: New file.
11441         * diagnostic-color.h: New file.
11442         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11443         arguments.
11444         * doc/invoke.texi (-fdiagnostics-color): Document.
11445         * pretty-print.h (pp_show_color): Define.
11446         (struct pretty_print_info): Add show_color field.
11447         * diagnostic.c: Include diagnostic-color.h.
11448         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11449         macros.  Colorize error:, warning: etc. strings and also the location
11450         string.
11451         (diagnostic_show_locus): Colorize the caret line.
11452         * pretty-print.c: Include diagnostic-color.h.
11453         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11454         inside of %< %> quotes or quoted through q format modifier.
11456 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11458         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11460 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11462         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11463         code in CC_NZ mode.
11464         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11465         pattern.
11467 2013-04-11  Marek Polacek  <polacek@redhat.com>
11469         PR tree-optimization/48184
11470         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11472 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11474         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11475         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11476         (skip_simple_arithmetic): Tidy up.
11477         * tree.h (skip_simple_constant_arithmetic): Declare.
11479 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11481         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11483 2013-04-11  Richard Biener  <rguenther@suse.de>
11485         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11486         generate vector constants.
11488 2013-04-11  Richard Biener  <rguenther@suse.de>
11490         PR tree-optimization/56878
11491         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11492         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11493         New function.
11494         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11495         Prefer to align the DR with the most invariant base address.
11497 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11499         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11500         comment.
11502 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11504         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11505         floating-point vector comparisons against 0.
11507 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11509         PR tree-optimization/56899
11510         * fold-const.c (extract_muldiv_1): Apply distributive law
11511         only if TYPE_OVERFLOW_WRAPS (ctype).
11513 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11515         PR target/56124
11516         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11517         loading has side effect.
11519 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
11521         * config/sparc/sparc.c: Include tree-pass.h.
11522         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11523         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
11524         head of file.  Change return type.  Split off gate function.
11525         (sparc_gate_work_around_errata): New function.
11526         (pass_work_around_errata): New pass definition.
11527         (insert_pass_work_around_errata) New pass insert definition to
11528         insert pass_work_around_errata just after delayed-branch scheduling.
11529         (sparc_option_override): Insert the pass.
11530         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11532 2013-04-10  David S. Miller  <davem@davemloft.net>
11534         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11535         or -mcpu=hypersparc.
11537         * target.def (cstore_mode): New hook.
11538         * target.h: Include insn-codes.h
11539         * targhooks.c: Likewise.
11540         (default_cstore_mode): New function.
11541         * targhooks.h: Declare it.
11542         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11543         * doc/tm.texi: Rebuild.
11544         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11545         target hook, rather than inspecting the insn_data.
11546         * config/sparc/sparc.c (sparc_cstore_mode): New function.
11547         (TARGET_CSTORE_MODE): Redefine.
11548         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11549         result patterns.
11550         * config/sparc/predicates.md (cstore_result_operand): New special
11551         predicate.
11552         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11553         Use it for operand 0.
11554         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11555         (*snesi_special): Likewise.
11556         (*snesi_zero): Likewise.
11557         (*seqsi_zero): Likewise.
11558         (*sltu_insn): Likewise.
11559         (*sgeu_insn): Likewise.
11560         (*seqdi_special): Make operand 0 and comparison operation be of
11561         DImode.
11562         (*snedi_special): Likewise.
11563         (*snedi_special_vis3): Likewise.
11564         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11565         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11566         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11567         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11568         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11569         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11570         (*sltu_extend_sp64): Likewise.
11571         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11572         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11573         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11574         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11575         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11577 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
11579         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11580         (aarch64_start_file): Use the new function.
11582 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11583             Jason Merrill  <jason@redhat.com>
11585         * common.opt: Add -gdwarf.
11586         * opts.c (common_handle_option): Handle it.
11587         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11589 2013-04-10  Richard Biener  <rguenther@suse.de>
11591         * passes.c (execute_todo): Do not call ggc_collect conditional here.
11592         (execute_one_ipa_transform_pass): But unconditionally here.
11593         (execute_one_pass): And here.
11594         (init_optimization_passes): Remove reload pass.
11595         * tree-pass.h (TODO_ggc_collect): Remove.
11596         (pass_reload): Likewise.
11597         * ira.c (do_reload): Merge into ...
11598         (ira): ... this.
11599         (rest_of_handle_reload): Remove.
11600         (pass_reload): Likewise.
11601         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11602         of reload for vzeroupper pass placement.
11603         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11604         and todo_flags_finish of all passes.
11606 2013-04-10  Richard Biener  <rguenther@suse.de>
11608         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11609         first_const_oprnd field, rename first_def_type to first_op_type.
11610         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11611         (vect_get_and_check_slp_defs): Always use the type of the
11612         operand.  Allow mixed vect_external_def, vect_constant_def types.
11613         (vect_get_constant_vectors): Handle mixed vect_external_def,
11614         vect_constant_def types.
11616 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
11618         PR tree-optimization/55524
11619         * tree-ssa-math-opts.c
11620         (convert_mult_to_fma): Don't use an fms construct
11621         when we don't have an fms operation, but fnma, and it looks
11622         likely that we'll be able to use the latter.
11624 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11626         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11627         function.
11628         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11629         inline fail caused by overwritable functions.
11631 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
11633         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11634         unnecessary bits in the constant power of two case.
11636 2013-04-10  Richard Biener  <rguenther@suse.de>
11638         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11639         broken code swapping operands.
11640         (vect_build_slp_tree): Do not compute load permutations here.
11641         (vect_analyze_slp_instance): Compute load permutations here,
11642         after building the SLP tree.
11644 2013-04-09  Christian Bruel  <christian.bruel@st.com>
11646         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11647         of next/prev_real_insn.
11649 2013-04-09  Jan Hubicka  <jh@suse.cz>
11651         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11652         Drop aliased parameter.
11653         (function_and_variable_visibility): Do not handle alias pairs.
11654         * cgraph.c (varpool_externally_visible_p): Update prototype.
11655         * varpool.c (varpool_add_new_variable): Update.
11657 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11659         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11661 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
11663         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11665         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11667 2013-04-09  Marek Polacek  <polacek@redhat.com>
11669         PR tree-optimization/48762
11670         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11672 2013-04-09  Richard Biener  <rguenther@suse.de>
11674         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11675         dealing with cost.
11676         (vect_build_slp_tree): Likewise.
11677         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11678         calculating the cost of a SLP instance.
11679         (vect_analyze_slp_instance): Use it from here, after building
11680         the SLP tree.
11682 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11684         PR middle-end/56883
11685         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11686         expand_omp_for_static_chunk): Use simple_p = true in
11687         force_gimple_operand_gsi calls when assigning to addressable decls.
11689 2013-04-09  Jeff Law  <law@redhat.com>
11691         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11692         when the boolean was created by converting a wider object which
11693         had a boolean range.
11695 2013-04-09  Richard Biener  <rguenther@suse.de>
11697         * tree-vectorizer.h (slp_void_p): Remove.
11698         (slp_tree): Typedef before _slp_tree declaration.
11699         (struct _slp_tree): Use a vector of slp_tree as children.
11700         (vect_get_place_in_interleaving_chain): Remove.
11701         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11702         Move ...
11703         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11704         and make static.
11705         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11706         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11707         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11708         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11709         Use slp_node instead of slp_void_p and adjust.
11711 2013-04-09  Richard Biener  <rguenther@suse.de>
11713         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11714         work that is not necessary.
11716 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11718         PR tree-optimization/56854
11719         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11720         forward into clobber stmts if it would change MEM_REF lhs into
11721         non-MEM_REF.
11723 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
11725         * tree.c (type_hash_lookup, type_hash_add): Make static.
11726         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11728 2013-04-09  Richard Biener  <rguenther@suse.de>
11730         * tree.h (unsave_expr_now): Remove.
11731         * tree-inline.c (mark_local_for_remap_r): Remove.
11732         (unsave_expr_1): Likewise.
11733         (unsave_r): Likewise.
11734         (unsave_expr_now): Likewise.
11735         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11736         (propagate_tree_value): Likewise.
11738 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
11740         * doc/rtl.texi (sequence): Rewrite documentation to match the
11741         current use of SEQUENCE rtl objects.
11742         * rtl.def (SEQUENCE): Likewise.
11744         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11745         Update documentation.
11746         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11747         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11749         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11751 2013-04-08  Teresa Johnson  <tejohnson@google.com>
11753         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11754         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11755         methods.
11756         (estimate_edge_size_and_time): Add comment to suggest using rounding
11757         methods.
11758         (estimate_node_size_and_time): Ditto.
11759         (remap_edge_change_prob): Use helper rounding divide methods.
11760         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11761         (gimple_mod_pow2_value_transform): Ditto.
11762         (gimple_mod_subtract_transform): Ditto.
11763         (gimple_ic_transform): Ditto.
11764         (gimple_stringops_transform): Ditto.
11765         * stmt.c (conditional_probability): Ditto.
11766         (emit_case_dispatch_table): Ditto.
11767         * lto-cgraph.c (merge_profile_summaries): Ditto.
11768         * tree-optimize.c (execute_fixup_cfg): Ditto.
11769         * cfgcleanup.c (try_forward_edges): Ditto.
11770         * cfgloopmanip.c (scale_loop_profile): Ditto.
11771         (loopify): Ditto.
11772         (duplicate_loop_to_header_edge): Ditto.
11773         (lv_adjust_loop_entry_edge): Ditto.
11774         * tree-vect-loop.c (vect_transform_loop): Ditto.
11775         * profile.c (compute_branch_probabilities): Ditto.
11776         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11777         * lto-streamer-in.c (input_cfg): Ditto.
11778         * gimple-streamer-in.c (input_bb): Ditto.
11779         * ipa-cp.c (update_profiling_info): Ditto.
11780         (update_specialized_profile): Ditto.
11781         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11782         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11783         rounding methods.
11784         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11785         (compute_trg_info): Ditto.
11786         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11787         (purge_dead_edges): Ditto.
11788         * loop-unswitch.c (unswitch_loop): Ditto.
11789         * cgraphclones.c (cgraph_clone_edge): Ditto.
11790         (cgraph_clone_node): Ditto.
11791         * tree-inline.c (copy_bb): Ditto.
11792         (copy_edges_for_bb): Ditto.
11793         (initialize_cfun): Ditto.
11794         (copy_cfg_body): Ditto.
11795         (expand_call_inline): Ditto.
11797 2013-04-08  Kai Tietz  <ktietz@redhat.com>
11799         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11800         TARGET_CYGWIN64 by TARGET_64BIT.
11802 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11804         * config/epiphany/epiphany.md (GPR_1): New constant.
11805         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11806         * config/epiphany/epiphany.c (gen_compare_reg):
11807         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11808         is already in place.
11809         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11810         Don't require being called during rtl expansion; If y operlaps r0,
11811         return 0.
11812         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11813         (epiphany_expand_epilogue): Likewise.
11815         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11816         Don't use CC_FPmode for ORDERED / UNORDERED.
11817         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11819         * config/epiphany/constraints.md (CnL): New constraint.
11820         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11821         * config/epiphany/predicates.md (add_operand): Allow 1024.
11823         * config/epiphany/epiphany.md (logical_op): New code iterator.
11824         (op_mnc): New code attribute.
11825         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11826         (mov_f+1, mov_f+2): New peephole2 patterns.
11828         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11829         (cstoresi4): Also allow re-use of zero result when doing a NE
11830         comparison to a non-zero operand.
11831         Use (clobber (scratch)) for first insn if the gpr output is not needed.
11833         * config/epiphany/epiphany.md (<insn_opname>v2si3):
11834         Use gen_addsi3_i / gen_subsi3_i.
11836 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11838         PR c++/34949
11839         PR c++/50243
11840         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11841         contain anything but clobbers, at most one __builtin_stack_restore,
11842         optionally debug stmts and final resx, and if it has at least one
11843         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
11844         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11845         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11846         which isn't defaut definition, remove them.
11847         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11848         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11849         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11850         with MEM_REF LHS with SSA_NAME address.
11852 2013-04-08  Jeff Law  <law@redhat.com>
11854         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11856 2013-04-08  Richard Biener  <rguenther@suse.de>
11858         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11859         extra newline.
11860         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11861         determined vector type.
11862         (vect_analyze_data_refs): Likewise.
11863         (vect_get_new_vect_var): Adjust.
11864         (vect_create_destination_var): Preserve SSA name versions.
11865         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11866         not dump anything here.
11868 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11870         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11871         Add member lr_slot_known.
11872         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11873         if necessary.
11874         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11875         Remove code that sets lr_slot_offset according to what a previous
11876         version of epiphany_emit_save_restore used to do.
11877         (epiphany_emit_save_restore): When doing an lr save or restore,
11878         set/verify lr_slot_known and lr_slot_offset.
11880 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
11882         PR target/54338
11883         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11884         in ALL_REGS.
11886 2013-04-08  Richard Biener  <rguenther@suse.de>
11888         * alias.c (find_base_term): Fix thinko in previous change.
11890 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11892         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11893         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11894         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
11895         if possible to compute val.
11896         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11897         For QImode integers don't require anything about precision.  Use
11898         const_with_all_bytes_same to find out if the constant doesn't have
11899         repeated bytes in it.
11901 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11903         * config/s390/s390.c (s390_expand_insv): Only accept insertions
11904         within mode size.
11906 2013-04-08  Marek Polacek  <polacek@redhat.com>
11908         PR rtl-optimization/48182
11909         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11910         value to 1.
11912 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11914         PR target/55487
11915         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11916         nuses, make sure we have a label.
11918 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11920         PR target/56843
11921         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11922         (rs6000_emit_swdiv_low_precision): Remove.
11923         (rs6000_emit_swdiv): Rewrite to handle between one and four
11924         iterations of Newton-Raphson generally; modify required number of
11925         iterations for some cases.
11926         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11928 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
11930         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11931         set-but-unused variable.
11933         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11934         basic blocks of released function bodies garbage-collectable.
11936         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11937         (struct rtl_opt_pass): Add TODO_df_finish.
11939         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11941 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11943         * config/arm/constraints.md (q): New constraint.
11944         * config/arm/ldrdstrd.md: New file.
11945         * config/arm/arm.md (ldrdstrd.md) New include.
11946         (arm_movdi): Use "q" instead of "r" constraint
11947         for double-word memory access.
11948         (movdf_soft_insn): Likewise.
11949         * config/arm/vfp.md (movdi_vfp): Likewise.
11950         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11951         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11952         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11953         (mem_ok_for_ldrd_strd): Likewise.
11954         (output_move_double): Update assertion.
11956 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11958         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11960 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11962         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11963         define_insn_and_split.
11964         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11966 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11968         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11969         define_insn_and_split.
11970         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11971         (shiftsi3_compare): New pattern.
11972         (rrx): New pattern.
11973         * config/arm/unspecs.md (UNSPEC_RRX): New.
11975 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11977         * config/arm/arm.md (negdi_extendsidi): New pattern.
11978         (negdi_zero_extendsidi): Likewise.
11980 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11982         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11983         define_insn_and_split.
11984         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11985         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11987 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11989         * config/arm/arm.md (arm_subdi3): Convert define_insn into
11990         define_insn_and_split.
11991         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11992         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11994 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11996         * config/arm/arm.md (subsi3_carryin): New pattern.
11997         (subsi3_carryin_const): Likewise.
11998         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11999         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12001 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12003         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12005 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12007         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12008         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12010 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12012         * config/arm/arm.c (arm_expand_builtin): Change fcode
12013         type to unsigned int.
12015 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12017         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12019 2013-04-04  Ian Lance Taylor  <iant@google.com>
12021         * doc/standards.texi (Standards): The Go frontend supports the Go 1
12022         language standard.
12024 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
12026         PR middle-end/56729
12027         * df-scan.c (df_insn_delete): Disable failing assert.
12029 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12031         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12032         New function prototype.
12033         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12034         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12035         (arm_builtin_vectorized_function): New function.
12037 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12039         * config/arm/arm_neon_builtins.def: New file.
12040         * config/arm/arm.c (neon_builtin_data): Move contents to
12041         arm_neon_builtins.def.
12042         (enum arm_builtins): Include neon builtin definitions.
12043         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12044         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12046 2013-04-04  Marek Polacek  <polacek@redhat.com>
12048         PR tree-optimization/48186
12049         * predict.c (maybe_hot_frequency_p): Return false if
12050         HOT_BB_FREQUENCY_FRACTION is 0.
12051         (cgraph_maybe_hot_edge_p): Likewise.
12053 2013-04-04  Richard Biener  <rguenther@suse.de>
12055         PR tree-optimization/56826
12056         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12057         more accurately.
12059 2013-04-04  Richard Biener  <rguenther@suse.de>
12061         PR tree-optimization/56213
12062         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12063         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12065 2013-04-04  Richard Biener  <rguenther@suse.de>
12067         PR tree-optimization/56837
12068         * tree-loop-distribution.c (classify_partition): For non-zero
12069         values require that the value has the same precision as its
12070         mode to be useful as memset value.
12072 2013-04-03  Nick Clifton  <nickc@redhat.com>
12074         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12075         (fmssf4): Use fmsf.s on E3V5 architectures.
12076         (fnmasf4): Use fnmaf.s on E3V5 architectures.
12077         (fnmssf4): Use fnmsf.s on E3V5 architectures.
12079 2013-04-03  Jeff Law  <law@redhat.com>
12081         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12082         (lra-eliminations.o): Likewise.
12084 2013-04-03  Teresa Johnson  <tejohnson@google.com>
12086         * gcov-io.c (compute_working_sets): Moved most of body of old
12087         compute_working_sets here from profile.c.
12088         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12089         (gcov_working_set_t): Moved typedef here from basic-block.h
12090         (compute_working_set): Declare.
12091         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12092         (get_working_sets): Renamed from compute_working_set,
12093         replace most of body with call to new compute_working_sets.
12094         (get_exec_counts): Replace call to compute_working_sets
12095         to get_working_sets.
12096         * profile.h (get_working_sets): Renamed from compute_working_set.
12097         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12098         to get_working_sets.
12099         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12100         * gcov-dump.c (dump_working_sets): New function.
12102 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
12104         * hwint.c (sext_hwi, zext_hwi): New functions.
12105         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12106         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12107         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12108         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12109         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12110         (sext_hwi, zext_hwi): New functions.
12112 2013-04-03  Jeff Law  <law@redhat.com>
12114         PR tree-optimization/56799
12115         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12116         back test for widening conversion erroneously dropped in prior change.
12118 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12120         PR target/56809
12121         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12122         instead of next_real_insn.
12124 2013-04-03  Marek Polacek  <polacek@redhat.com>
12126         PR sanitizer/55702
12127         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12129 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12131         PR target/56809
12132         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12133         next_real_insn.
12134         (thumb1_output_casesi): Likewise.
12135         (thumb2_output_casesi): Likewise.
12137 2013-04-03  Richard Biener  <rguenther@suse.de>
12139         PR tree-optimization/56817
12140         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12141         Split out ...
12142         (tree_unroll_loops_completely_1): ... new function to manually
12143         walk the loop tree, properly defering outer loops of unrolled
12144         loops to later iterations.
12146 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12148         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12149         (vectorizable_load): Likewise.
12150         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12151         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12153 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12155         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12156         BIT_FIELD_REF.
12158 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12160         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12162 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
12164         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12166 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12168         PR tree-optimization/56790
12169         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12170         folding.
12172 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12174         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12175         Handle VEC_MERGE.
12176         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12177         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12178         equal arguments.
12180 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
12182         PR c/19449
12183         * tree.h (force_folding_builtin_constant_p): New decl.
12184         * builtins.c (force_folding_builtin_constant_p): New variable.
12185         (fold_builtin_constant_p): Fold immediately also if
12186         force_folding_builtin_constant_p.
12188 2013-04-03  Richard Biener  <rguenther@suse.de>
12190         PR tree-optimization/56812
12191         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12192         DRs of the same interleaving chain are independent.
12194 2013-04-02  Jason Merrill  <jason@redhat.com>
12196         * gdbinit.in (pbb): Use debug fn.
12198 2013-04-02  Lawrence Crowl  <crowl@google.com>
12200         * sese.h (struct ivtype_map_elt_s): Remove unused.
12201         (extern debug_ivtype_map): Remove unused.
12202         (extern eq_ivtype_map_elts): Remove unused.
12203         * sese.c (debug_ivtype_map): Removed unused.
12204         (debug_ivtype_map_1): Removed unused.
12205         (debug_ivtype_elt): Remove unused.
12206         (eq_ivtype_map_elts): Remove unused.
12209 2013-04-02  Kai Tietz  <ktietz@redhat.com>
12211         PR target/52790
12212         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12213         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12214         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12215         function.
12216         (legitimize_pe_coff_symbol): Likewise.
12217         (is_imported_p): New helper-function.
12218         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12219         for Windows x64 targets.
12220         (ix86_expand_prologue): Optimize for pe-coff targets.
12221         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12222         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12223         medium/large code-model.
12224         (legitimize_pic_address): Likewise.
12225         (legitimize_tls_address): Likewise.
12226         (ix86_expand_call): Likewise.
12227         (x86_output_mi_thunk): Likewise.
12228         (get_dllimport_decl): Add new beimport argument.
12229         (construct_plt_address): Don't assert for x64 pe-coff targets.
12230         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12231         targets.
12232         (SYMBOL_FLAG_STUBVAR): New macro.
12233         (SYMBOL_REF_STUBVAR_P): Likewise.
12234         * config/i386/winnt.c (stub_list): New structure.
12235         (stub_head): New local variable.
12236         (i386_pe_record_stub): New function.
12237         (i386_pe_file_end): Emit refptr-stubs.
12239 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
12241         PR rtl-optimization/56745
12242         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12243         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12245         PR c++/34949
12246         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12247         and both of them are MEM_REFs, just compare first argument for
12248         equality and attempt to deal even with differing offsets.
12250         PR c++/34949
12251         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12252         of gimple_clobber_p to be MEM_REF.
12253         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12254         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
12255         after gimplification.
12256         * asan.c (get_mem_ref_of_assignment): Don't instrument
12257         gimple_clobber_p stmts.
12258         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12259         gimple_clobber_p stmt if they have MEM_REF lhs and
12260         are dead because of another gimple_clobber_p stmt.
12261         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12262         gimple_clobber_p stmts like debug stmts.
12263         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12264         that refer to unused VAR_DECLs or uninitialized values.
12265         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12266         gimple_clobber_p stmts if they refer to removed parameters.
12267         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12268         formatting.
12270 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12272         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12273         using SWI48 mode attribute.
12275 2013-04-02  Wei Mi  <wmi@google.com>
12277         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12278         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12279         *<rotate_insn><mode>3_mask in i386.md.
12281 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12283         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12285 2013-04-02  Richard Biener  <rguenther@suse.de>
12287         PR tree-optimization/56778
12288         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12289         Runtime alias tests are not supported for gather loads.
12290         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12291         stmts referenced from SSA operands before updating SSA form.
12293 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12294             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12296         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12297         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12298         * config/arm/cortex-a53.md: New file.
12299         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12300         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12301         * config/arm/arm.c (arm_issue_rate): Likewise.
12302         * config/arm/arm-tune.md: Regenerate
12303         * config/arm/arm-tables.opt: Regenerate.
12304         * config/arm/arm-cores.def: Add cortex-a53.
12306 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12308         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12309         non-static link.
12311 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12313         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12314         scalar load/store operations using B/H registers.
12315         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12317 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12319         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12320         scalar move.
12321         * config/aarch64/aarch64.c
12322         (aarch64_simd_scalar_immediate_valid_for_move): New.
12323         * config/aarch64/aarch64-protos.h
12324         (aarch64_simd_scalar_immediate_valid_for_move): New.
12325         * config/aarch64/constraints.md (Dh, Dq): New.
12326         * config/aarch64/iterators.md (hq): New.
12328 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12330         * reorg.c (get_branch_condition): Deal with conditional returns.
12331         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12333 2013-04-01  Wei Mi  <wmi@google.com>
12335         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12336         Truncate operand 2 using %b asm operand modifier.
12337         (*<shift_insn><mode>3_mask): Ditto.
12338         (*<rotate_insn><mode>3_mask): Ditto.
12340 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12342         PR middle-end/56798
12343         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12345 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12347         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12348         of next_real_insn.
12349         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12351 2013-03-30  Lawrence Crowl  <crowl@google.com>
12353         * dse.c (clear_alias_sets): Remove never set.
12354         (disqualified_clear_alias_sets): Remove never set.
12355         (clear_alias_mode_pool): Remove never set.
12356         (dse_step0): Remove condition that is never true.
12357         (canon_address): Remove condition that is never true.
12358         (dse_step7): Remove condition that is never true.
12359         (rest_of_handle_dse): Remove condition that is never true.
12360         (rest_of_handle_dse::did_global): Remove never read from above.
12361         (dse_step2_spill): Remove never called from above.
12362         (dse_step5_spill): Remove never called from above.
12364 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12366         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12367         JUMP_TABLE_DATA changes.
12368         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12369         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12370         (Insns) <jump_table_data>: New entry.
12371         * doc/tm.texi: Regenerate.
12373         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12375         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12376         for table jump at the end of a basic block using tablejump_p.
12377         * targhooks.c (default_invalid_within_doloop): Likewise.
12378         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12379         target hook implementation that is identical to the default hook.
12380         (rs6000_invalid_within_doloop): Remove.
12382         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12383         unused variable from tablejump_p call.
12385         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12386         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12387         (INSN_DELETED_P): Likewise.
12388         (emit_jump_table_data): New prototype.
12389         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12390         after 4th as unused.
12391         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12392         * sched-vis.c (print_insn): Likewise.
12393         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12394         insn for compatibility with back ends that use next_active_insn to
12395         identify jump table data.
12396         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12397         (remove_insn): Likewise.
12398         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12399         to be emitted.
12400         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12401         (emit_jump_table_data): New function.
12403         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12404         basic block, a JUMP_TABLE_DATA never is.
12405         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12406         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12407         off from code handling real insns.
12408         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12409         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12410         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12411         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12412         is not a NONDEBUG_INSN_P.
12413         * ira-costs.c (scan_one_insn): Likewise.
12414         * jump.c (mark_all_labels): Likewise.
12415         (mark_jump_label_1): Likewise.
12416         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12417         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12418         a basic block.
12419         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12420         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12421         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12422         JUMP_TABLE_DATA_P insns.
12423         (calculate_elim_costs_all_insns): Likewise.
12424         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12425         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12426         (delete_output_reload): Code style fixups.
12427         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12428         insn flags on this non-insn.
12429         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12430         as scheduling barriers, for pre-change compatibility.
12431         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12432         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12434         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12435         redundant JUMP_TABLE_DATA_P test.
12436         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12437         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12438         (frv_for_each_packet): Likewise.
12439         * config/i386/i386.c (min_insn_size): Likewise.
12440         (ix86_avoid_jump_mispredicts): Likewise.
12441         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12442         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12443         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12444         (mips16_insn_length): Robustify.
12445         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12446         (mips16_split_long_branches): Likewise.
12447         * config/pa/pa.c (pa_combine_instructions): Likewise.
12448         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12449         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12450         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12451         as contributing to pool range lengths.
12452         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12453         Remove redundant JUMP_TABLE_DATA_P test.
12454         (sh_loop_align): Likewise.
12455         (split_branches): Likewise.
12456         (sh_insn_length_adjustment): Likewise.
12457         * config/spu/spu.c (get_branch_target): Likewise.
12459 2013-03-29  Jan Hubicka  <jh@suse.cz>
12461         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12462         gcov streaming; stream hot bb threshold to ltrans.
12463         * predict.c (get_hot_bb_threshold): Break out from ....
12464         (maybe_hot_count_p): ... here.
12465         (set_hot_bb_threshold): New function.
12466         * lto-section-in.c (lto_section_name): Add profile.
12467         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12468         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12469         and data-streamer.h
12470         (histogram_entry): New structure.
12471         (histogram, histogram_pool): New global vars.
12472         (histogram_hash): New structure.
12473         (histogram_hash::hash): New method.
12474         (histogram_hash::equal): Likewise.
12475         (account_time_size): New function.
12476         (cmp_counts): New function.
12477         (dump_histogram): New function.
12478         (ipa_profile_generate_summary): New function.
12479         (ipa_profile_write_summary): New function.
12480         (ipa_profile_read_summary): New function.
12481         (ipa_profile): Decide on threshold.
12482         (pass_ipa_profile): Add ipa_profile_write_summary and
12483         ipa_profile_read_summary.
12484         * Makefile.in (ipa.o): Update dependencies.
12485         * lto-streamer.h (LTO_section_ipa_profile): New section.
12487 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12489         * tree.h (VAR_P): New.
12491 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12493         PR lto/56777
12494         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12496 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12498         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12499         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12500         (control_flow_insn_p): Likewise.
12501         * cfgrtl.c (duplicate_insn_chain): Likewise.
12502         * final.c (get_attr_length_1): Likewise.
12503         (shorten_branches): Likewise.
12504         (final_scan_insn): Likewise.
12505         * function.c (instantiate_virtual_regs): Likewise.
12506         * gcse.c (insert_insn_end_basic_block): Likewise.
12507         * ira-costs.c (scan_one_insn): Likewise.
12508         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12509         * lra.c (check_rtl): Likewise.
12510         * reload1.c (elimination_costs_in_insn): Likewise.
12511         * reorg.c (follow_jumps): Likewise.
12513         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12514         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12515         (thumb_far_jump_used_p): Likewise.
12516         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12517         (workaround_speculation): Likewise.
12518         (add_sched_insns_for_speculation): Likewise.
12519         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12520         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12521         (frv_for_each_packet): Likewise.
12522         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12523         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12524         (final_emit_insn_group_barriers): Likewise.
12525         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12526         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12527         (mips16_insn_length): Likewise.
12528         * config/pa/pa.c (pa_reorg): Likewise.
12529         (pa_combine_instructions): Likewise.
12530         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12531         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12532         (sh_reorg): Likewise.
12533         (split_branches): Likewise.
12534         * config/spu/spu.c (get_branch_target): Likewise.
12536         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12537         JUMP_TABLE_DATA_P.
12539 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
12541         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12542         Fix declaration name.
12544 2013-03-28  Lawrence Crowl  <crowl@google.com>
12546         * graphds.h (struct graph.indicies): Remove unused.
12547         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12548         (SCOP_ORIGINAL_PDDRS): Remove unused.
12549         * sese.h (extern insert_loop_close_phis): Removed unused.
12550         (extern insert_guard_phis): Removed unused.
12551         (extern ivtype_map_elt_info): Removed unused.
12552         (new_ivtype_map_elt): Removed unused.
12553         * sese.c (ivtype_map_elt_info): Removed unused.
12555 2013-03-28  Lawrence Crowl  <crowl@google.com>
12557         * Makefile.in: Add several missing include dependences.
12558         (DUMPFILE_H): New.
12559         (test-dump.o): New.  This object is not added to any executable,
12560         but is present for ad-hoc testing.
12561         * bitmap.c
12562         (debug (const bitmap_head_def &)): New.
12563         (debug (const bitmap_head_def *)): New.
12564         * bitmap.h
12565         (extern debug (const bitmap_head_def &)): New.
12566         (extern debug (const bitmap_head_def *)): New.
12567         * cfg.c
12568         (debug (edge_def &)): New.
12569         (debug (edge_def *)): New.
12570         * cfghooks.c
12571         (debug (basic_block_def &)): New.
12572         (debug (basic_block_def *)): New.
12573         * dumpfile.h
12574         (dump_node (const_tree, int, FILE *)): Correct source file.
12575         * dwarf2out.c
12576         (debug (die_struct &)): New.
12577         (debug (die_struct *)): New.
12578         * dwarf2out.h
12579         (extern debug (die_struct &)): New.
12580         (extern debug (die_struct *)): New.
12581         * gimple-pretty-print.c
12582         (debug (gimple_statement_d &)): New.
12583         (debug (gimple_statement_d *)): New.
12584         * gimple-pretty-print.h
12585         (extern debug (gimple_statement_d &)): New.
12586         (extern debug (gimple_statement_d *)): New.
12587         * ira-build.c
12588         (debug (ira_allocno_copy &)): New.
12589         (debug (ira_allocno_copy *)): New.
12590         (debug (ira_allocno &)): New.
12591         (debug (ira_allocno *)): New.
12592         * ira-int.h
12593         (extern debug (ira_allocno_copy &)): New.
12594         (extern debug (ira_allocno_copy *)): New.
12595         (extern debug (ira_allocno &)): New.
12596         (extern debug (ira_allocno *)): New.
12597         * ira-lives.c
12598         (debug (live_range &)): New.
12599         (debug (live_range *)): New.
12600         * lra-int.h
12601         (debug (lra_live_range &)): New.
12602         (debug (lra_live_range *)): New.
12603         * lra-lives.c
12604         (debug (lra_live_range &)): New.
12605         (debug (lra_live_range *)): New.
12606         * omega.c
12607         (debug (omega_pb_d &)): New.
12608         (debug (omega_pb_d *)): New.
12609         * omega.h
12610         (extern debug (omega_pb_d &)): New.
12611         (extern debug (omega_pb_d *)): New.
12612         * print-rtl.c
12613         (debug (const rtx_def &)): New.
12614         (debug (const rtx_def *)): New.
12615         * print-tree.c
12616         (debug_tree (tree): Move within file.
12617         (debug_raw (const tree_node &)): New.
12618         (debug_raw (const tree_node *)): New.
12619         (dump_tree_via_hooks (const tree_node *, int)): New.
12620         (debug (const tree_node &)): New.
12621         (debug (const tree_node *)): New.
12622         (debug_verbose (const tree_node &)): New.
12623         (debug_verbose (const tree_node *)): New.
12624         (debug_head (const tree_node &)): New.
12625         (debug_head (const tree_node *)): New.
12626         (debug_body (const tree_node &)): New.
12627         (debug_body (const tree_node *)): New.
12628         (debug_vec_tree (tree): Move and reimplement in terms of dump.
12629         (debug (vec<tree, va_gc> &)): New.
12630         (debug (vec<tree, va_gc> *)): New.
12631         * rtl.h
12632         (extern debug (const rtx_def &)): New.
12633         (extern debug (const rtx_def *)): New.
12634         * sbitmap.c
12635         (debug_raw (simple_bitmap_def &)): New.
12636         (debug_raw (simple_bitmap_def *)): New.
12637         (debug (simple_bitmap_def &)): New.
12638         (debug (simple_bitmap_def *)): New.
12639         * sbitmap.h
12640         (extern debug (simple_bitmap_def &)): New.
12641         (extern debug (simple_bitmap_def *)): New.
12642         (extern debug_raw (simple_bitmap_def &)): New.
12643         (extern debug_raw (simple_bitmap_def *)): New.
12644         * sel-sched-dump.c
12645         (debug (vinsn_def &)): New.
12646         (debug (vinsn_def *)): New.
12647         (debug_verbose (vinsn_def &)): New.
12648         (debug_verbose (vinsn_def *)): New.
12649         (debug (expr_def &)): New.
12650         (debug (expr_def *)): New.
12651         (debug_verbose (expr_def &)): New.
12652         (debug_verbose (expr_def *)): New.
12653         (debug (vec<rtx> &)): New.
12654         (debug (vec<rtx> *)): New.
12655         * sel-sched-dump.h
12656         (extern debug (vinsn_def &)): New.
12657         (extern debug (vinsn_def *)): New.
12658         (extern debug_verbose (vinsn_def &)): New.
12659         (extern debug_verbose (vinsn_def *)): New.
12660         (extern debug (expr_def &)): New.
12661         (extern debug (expr_def *)): New.
12662         (extern debug_verbose (expr_def &)): New.
12663         (extern debug_verbose (expr_def *)): New.
12664         (extern debug (vec<rtx> &)): New.
12665         (extern debug (vec<rtx> *)): New.
12666         * sel-sched-ir.h
12667         (_list_iter_cond_expr): Make inline instead of static.
12668         * sreal.c
12669         (debug (sreal &)): New.
12670         (debug (sreal *)): New.
12671         * sreal.h
12672         (extern debug (sreal &)): New.
12673         (extern debug (sreal *)): New.
12674         * tree.h
12675         (extern debug_raw (const tree_node &)): New.
12676         (extern debug_raw (const tree_node *)): New.
12677         (extern debug (const tree_node &)): New.
12678         (extern debug (const tree_node *)): New.
12679         (extern debug_verbose (const tree_node &)): New.
12680         (extern debug_verbose (const tree_node *)): New.
12681         (extern debug_head (const tree_node &)): New.
12682         (extern debug_head (const tree_node *)): New.
12683         (extern debug_body (const tree_node &)): New.
12684         (extern debug_body (const tree_node *)): New.
12685         (extern debug (vec<tree, va_gc> &)): New.
12686         (extern debug (vec<tree, va_gc> *)): New.
12687         * tree-cfg.c
12688         (debug (struct loop &)): New.
12689         (debug (struct loop *)): New.
12690         (debug_verbose (struct loop &)): New.
12691         (debug_verbose (struct loop *)): New.
12692         * tree-dump.c: Add header dependence.
12693         * tree-flow.h
12694         (extern debug (struct loop &)): New.
12695         (extern debug (struct loop *)): New.
12696         (extern debug_verbose (struct loop &)): New.
12697         (extern debug_verbose (struct loop *)): New.
12698         * tree-data-ref.c
12699         (debug (data_reference &)): New.
12700         (debug (data_reference *)): New.
12701         (debug (vec<data_reference_p> &)): New.
12702         (debug (vec<data_reference_p> *)): New.
12703         (debug (vec<ddr_p> &)): New.
12704         (debug (vec<ddr_p> *)): New.
12705         * tree-data-ref.h
12706         (extern debug (data_reference &)): New.
12707         (extern debug (data_reference *)): New.
12708         (extern debug (vec<data_reference_p> &)): New.
12709         (extern debug (vec<data_reference_p> *)): New.
12710         (extern debug (vec<ddr_p> &)): New.
12711         (extern debug (vec<ddr_p> *)): New.
12712         * tree-ssa-alias.c
12713         (debug (pt_solution &)): New.
12714         (debug (pt_solution *)): New.
12715         * tree-ssa-alias.h
12716         (extern debug (pt_solution &)): New.
12717         (extern debug (pt_solution *)): New.
12718         * tree-ssa-alias.c
12719         (debug (_var_map &)): New.
12720         (debug (_var_map *)): New.
12721         (debug (tree_live_info_d &)): New.
12722         (debug (tree_live_info_d *)): New.
12723         * tree-ssa-alias.h
12724         (extern debug (_var_map &)): New.
12725         (extern debug (_var_map *)): New.
12726         (extern debug (tree_live_info_d &)): New.
12727         (extern debug (tree_live_info_d *)): New.
12729 2013-03-28  Jan Hubicka  <jh@suse.cz>
12731         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12733 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
12735         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12736         record only when desired or required.
12738 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
12740         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12741         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
12743 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12745         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12746         (*andsi3_compare0_uxtw): New pattern.
12747         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12748         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12750 2013-03-28  Jan Hubicka  <jh@suse.cz>
12752         * data-streamer-in.c (streamer_read_gcov_count): New function.
12753         * gimple-streamer-out.c: Include value-prof.h.
12754         (output_gimple_stmt): Output histogram.
12755         (output_bb): Use streamer_write_gcov_count.
12756         * value-prof.c: Include data-streamer.h
12757         (dump_histogram_value): Add HIST_TYPE_MAX.
12758         (stream_out_histogram_value): New function.
12759         (stream_in_histogram_value): New function.
12760         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12761         (stream_out_histogram_value, stream_in_histogram_value): Declare.
12762         * data-streamer-out.c (streamer_write_gcov_count): New function.
12763         (streamer_write_gcov_count_stream): New function.
12764         * lto-cgraph.c (lto_output_edge): Update counter streaming.
12765         (lto_output_node): Likewise.
12766         (input_node, input_edge): Likewise.
12767         * lto-streamer-out.c (output_cfg): Update streaming.
12768         * lto-streamer-in.c (input_cfg): Likewise.
12769         * data-streamer.h (streamer_write_gcov_count,
12770         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12771         * gimple-streamer-in.c: Include value-prof.h
12772         (input_gimple_stmt): Input histograms.
12773         (input_bb): Update profile streaming.
12775 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12777         * genmodes.c (emit_max_int): New function.
12778         (emit_insn_modes_h): Added call to emit_max_function.
12779         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12780         Added doc.
12781         * machmode.def: Fixed comment.
12783 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12785         * combine.c (try_combine): Removed useless assert.
12786         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12788 2013-03-28  Marek Polacek  <polacek@redhat.com>
12789             Richard Biener  <rguenther@suse.de>
12791         PR tree-optimization/56695
12792         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12793         build signed result of a vector comparison.
12794         * tree-cfg.c (verify_gimple_comparison): Check that a result
12795         of a vector comparison has signed type.
12797 2013-03-28  Richard Biener  <rguenther@suse.de>
12799         PR tree-optimization/37021
12800         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12801         do not restrict gaps between groups.
12802         * tree-vect-stmts.c (vectorizable_load): Properly account for
12803         a gap between groups.
12805 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
12807         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12808         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12809         is not enabled.
12811 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
12813         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12814         * doc/extend.texi (Named Address Spaces): Ditto.
12815         (Variable Attributes): Ditto.
12817 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12819         * config.build: Add support for cygwin x64 target.
12820         * config.gcc: Likewise.
12821         * config.host: Likewise.
12822         * configure.ac: Likewise
12823         * configure: Regenerated.
12825 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12827         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12828         * config/i386/t-cygwin-w64: New file.
12829         * config/i386/cygwin-w64.h: New file.
12830         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12831         and add support for x64-cygwin target.
12832         (CPP_SPEC): Likewise.
12833         (CXX_WRAP_SPEC_LIST): Undefine before define.
12834         (LIBGCJ_SONAME): Use 15 as version.
12836 2013-03-27  Richard Biener  <rguenther@suse.de>
12838         PR tree-optimization/56716
12839         * tree-ssa-structalias.c (perform_var_substitution): Adjust
12840         dumping for ref nodes.
12842 2013-03-27  Martin Jambor  <mjambor@suse.cz>
12844         PR tree-optimization/55334
12845         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12846         restricted pointers to arrays.
12848 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12850         * Makefile.in (.SUFFIXES): Add .cc.
12851         (.c.o): Apply same recipe for implicit rule .cc.o.
12853 2013-03-27  Richard Biener  <rguenther@suse.de>
12855         PR tree-optimization/37021
12856         * tree-vect-data-refs.c (vect_check_strided_load): Allow
12857         REALPART/IMAGPART_EXPRs around the supported refs.
12858         * tree-ssa-structalias.c (find_func_aliases): Assume that
12859         floating-point values are not used to transfer pointers.
12861 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12863         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
12864         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12865         * gcc/doc/tm.texi: Regenerate.
12866         * gcc/targhooks.h (default_has_ifunc_p): New.
12867         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
12868         * gcc/config/linux-protos.h: New file.
12869         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12870         this hook for linux which disables support of indirect functions in
12871         android.
12872         * gcc/config/linux-android.c: New file.
12873         * gcc/config/t-linux-android.c: Ditto.
12874         * gcc/config.gcc: Added new object file linux-android.o.
12875         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
12876         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12877         * gcc/varasm.c (do_assemble_alias): Likewise.
12878         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12879         doesn't support indirect functions.
12880         * configure: Regenerate.
12882 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
12884         PR target/56102
12885         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12886         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12887         mult-word mode.
12889 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12891         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12893 2013-03-27  Terry Guo  <terry.guo@arm.com>
12895         * config/arm/arm-cores.def: Added core cortex-r7.
12896         * config/arm/arm-tune.md: Regenerated.
12897         * config/arm/arm-tables.opt: Regenerated.
12898         * doc/invoke.texi: Added entry for core cortex-r7.
12900 2013-03-27  Walter Lee  <walt@tilera.com>
12902         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12903         double-decrement of next_scratch_regno.
12905 2013-03-27  Walter Lee  <walt@tilera.com>
12907         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12908         input operands.
12909         (insn_v1mulus): Ditto.
12910         (insn_v2muls): Ditto.
12912 2013-03-27  Walter Lee  <walt@tilera.com>
12914         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12915         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12917 2013-03-27  Walter Lee  <walt@tilera.com>
12919         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12920         (*sibcall_value): Ditto.
12922 2013-03-27  Walter Lee  <walt@tilera.com>
12924         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12925         (insn_mnz_v8qi): ... this ...
12926         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
12927         vector equivalent.
12928         (insn_v<n>mnz): Replaced by ...
12929         (insn_v1mnz): ... this ...
12930         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
12931         equivalent.
12932         (insn_mz_<mode>): Replaced by ...
12933         (insn_mz_v8qi): ... this ...
12934         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
12935         vector equivalent.
12936         (insn_v<n>mz): Replaced by ...
12937         (insn_v1mz): ... this ...
12938         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
12939         equivalent.
12941 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
12943         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12945 2013-03-26  Roland McGrath  <mcgrathr@google.com>
12947         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12948         than fprintf with a non-constant, non-format string.
12950 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12952         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12953         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
12954         operand 0 predicate.
12955         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12956         attribute.  Use general_x64nomem_operand as operand 1 predicate.
12957         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12958         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12959         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12960         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12961         (mov<mode>_insv_1): Remove expander.  Merge insn with
12962         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12963         Use general_x64nomem_operand as operand 1 predicate.
12964         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12965         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12966         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12967         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12968         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
12969         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12970         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12971         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12972         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
12973         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12974         (general_x64nomem_operand): Ditto.
12976 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12978         * config/rtems.opt: Add -pthread option.
12980 2013-03-26  Richard Biener  <rguenther@suse.de>
12982         * alias.c (find_base_term): Avoid redundant and not used recursion.
12983         (base_alias_check): Get the initial base term from the caller.
12984         (true_dependence_1): Compute and pass base terms to base_alias_check.
12985         (write_dependence_p): Likewise.
12986         (may_alias_p): Likewise.
12988 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
12990         * config/aarch64/aarch64.c (aarch64_classify_address): Support
12991         PC-relative load in SI modes and above only.
12993 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
12995         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12996         * config/arm/iwmmxt.md (WCGR0): Update.
12997         (WCGR1, WCGR2, WCGR3): Likewise.
12999 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13001         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13002         Use x64 and nox64 isa attributes.
13004 2013-03-26  Richard Biener  <rguenther@suse.de>
13006         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13007         alignment computations and rely on get_object_alignment_1
13008         for the !TYPE_P case.
13009         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13011 2013-03-26  Walter Lee  <walt@tilera.com>
13013         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13014         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13016 2013-03-25  Jeff Law  <law@redhat.com>
13018         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13019         check for INTEGRAL_TYPE_P that was missing due to checking in
13020         wrong version of prior patch.
13022 2013-03-25  Walter Lee  <walt@tilera.com>
13024         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13025         TILEGX_INSN_SHUFFLEBYTES1.
13026         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13027         shufflebytes1.
13028         (tilegx_builtins): Ditto.
13029         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13031 2013-03-25  Walter Lee  <walt@tilera.com>
13033         * config/tilegx/tilegx.md (floatsisf2): New pattern.
13034         (floatunssisf2): New pattern.
13035         (floatsidf2): New pattern.
13036         (floatunssidf2): New pattern.
13038 2013-03-25  Walter Lee  <walt@tilera.com>
13040         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13041         tests for constraint J, K, N, P.
13043 2013-03-25  Walter Lee  <walt@tilera.com>
13045         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13046         Use indirect/pcrel encoding.
13047         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13048         Ditto.
13050 2013-03-25  Steve Ellcey  <sellcey@mips.com>
13052         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13053         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13054         * config/mips/mips.c (mips_option_override): Set IMADD default.
13055         * config/mips/mips.h (PTF_AVOID_IMADD): New.
13056         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13057         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13058         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13059         * doc/invoke.texi (-mimadd/-mno-imadd): New.
13061 2013-03-25  Jeff Law  <law@redhat.com>
13063         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13064         slightly to avoid creating and folding useless trees.  Simplify
13065         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13067 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
13069         * config/i386/i386.md (*zero_extendsidi2): Merge with
13070         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
13071         * config/i386/predicates.md (x86_64_zext_operand): Rename from
13072         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
13073         targets.  Clarify comment.
13075 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13077         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13078         pass-through jump functions differently.
13079         (ipa_read_jump_function): Likewise.  Also use setter functions to set
13080         up jump functions.
13082 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13084         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13085         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13086         process it.
13087         (ipa_get_indirect_edge_target): New function.
13088         (devirtualization_time_bonus): New parameter known_aggs, pass it to
13089         ipa_get_indirect_edge_target.  Update all callers.
13090         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
13091         ipa_get_indirect_edge_target_1 instead of calling
13092         ipa_get_indirect_edge_target.
13093         (create_specialized_node): Pass aggvlas to
13094         ipcp_discover_new_direct_edges.
13096 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13098         * config/arm/arm.md (f_sels, f_seld): New types.
13099         (*cmov<mode>): New pattern.
13100         * config/arm/predicates.md (arm_vsel_comparison_operator): New
13101         predicate.
13103 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13105         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13106         POSIX-printf for mingw-hosted  builds.
13108 2013-03-25  Richard Biener  <rguenther@suse.de>
13110         PR middle-end/56694
13111         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13112         must-not-throw stmt location.
13114 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13116         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13117         Emit load-acquire versions when acq is true.
13118         (arm_emit_store_exclusive): Add rel parameter.
13119         Emit store-release versions when rel is true.
13120         (arm_split_compare_and_swap): Use acquire-release instructions
13121         instead.
13122         of barriers when appropriate.
13123         (arm_split_atomic_op): Likewise.
13124         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13125         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13126         (VUNSPEC_SLX): Likewise.
13127         (VUNSPEC_LDA): Likewise.
13128         (VUNSPEC_STL): Likewise.
13129         * config/arm/sync.md (atomic_load<mode>): New pattern.
13130         (atomic_store<mode>): Likewise.
13131         (arm_load_acquire_exclusive<mode>): Likewise.
13132         (arm_load_acquire_exclusivesi): Likewise.
13133         (arm_load_acquire_exclusivedi): Likewise.
13134         (arm_store_release_exclusive<mode>): Likewise.
13136 2013-03-25  Catherine Moore  <clm@codesourcery.com>
13138         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13139         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13140         * config/mip/predicates.md (lwsp_swsp_operand,
13141         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13142         sb16_operand, db4_operand, db7_operand, ib3_operand,
13143         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13144         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13145         andi16_operand): New predicates.
13146         * config/mips/mips.md (compression): New attribute.
13147         (enabled): New attribute.
13148         (length): Consider compression in computing length.
13149         (shift_compression): New code attribute.
13150         (*add<mode>3): New operands. Record compression.
13151         (sub<mode>3): Likewise.
13152         (one_cmpl<mode>2): Likewise.
13153         (*and<mode>3): Likewise.
13154         (*ior<mode>3): Likewise.
13155         (unnamed pattern for xor): Likewise.
13156         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13157         (*<optab><mode>3): Likewise.
13158         (*mov<mode>_internal: Likewise.
13159         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13160         (mips_unsigned_immediate_p): New.
13161         (umips_lwsp_swsp_address_p): New.
13162         (m16_based_address_p): New.
13163         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13164         (mips_unsigned_immediate_p): New prototype.
13165         (lwsp_swsp_address_p): New prototype.
13166         (m16_based_address_p): New prototype.
13167         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13168         (mips_signed_immediate_p): New function.
13169         (m16_based_address_p): New function.
13170         (lwsp_swsp_address_p): New function.
13171         (mips_print_operand_punctuation): Recognize short delay slot insns
13172         for microMIPS.add<mode>3"
13174 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13176         PR target/56720
13177         * config/arm/iterators.md (v_cmp_result): New mode attribute.
13178         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13180 2013-03-25  Richard Biener  <rguenther@suse.de>
13182         PR tree-optimization/56689
13183         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13184         any edge.
13186 2013-03-25  Richard Biener  <rguenther@suse.de>
13188         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13189         of bitmap.
13190         (memory_references): Likewise.
13191         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13192         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13193         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13194         (gather_mem_refs_in_loops): Fold into ...
13195         (analyze_memory_references): ... this.  Move initialization
13196         to tree_ssa_lim_initialize.
13197         (fill_always_executed_in): Rename to ...
13198         (fill_always_executed_in_1): ... this.
13199         (fill_always_executed_in): Move contains_call computation to
13200         this new function from ...
13201         (tree_ssa_lim_initialize): ... here.
13202         (tree_ssa_lim): Call fill_always_executed_in.
13204 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13206         * postreload.c (reload_combine): Fix code detecting returns.
13208 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13210         * function.c (emit_use_return_register_into_block): On cc0 targets,
13211         do not emit the sequence between cc0 setter and user.
13213 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13215         * config/i386/predicates.md (local_symbolic_operand): Interpret
13216         dll-imported symbols as none-local.
13218 2013-03-25  Richard Biener  <rguenther@suse.de>
13220         * tree-ssa-loop-im.c (struct depend): Remove.
13221         (struct lim_aux_data): Make depends a vec of gimples.
13222         (free_lim_aux_data): Adjust.
13223         (add_dependency): Likewise.
13224         (set_level): Likewise.
13226 2013-03-25  Richard Biener  <rguenther@suse.de>
13228         PR middle-end/56434
13229         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13230         the pointer returned by calls with ECF_MALLOC set.
13232 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13234         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13236 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13238         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13239         using MMXMODE mode iterator.
13240         (*move<mode>_internal): Merge with *movv2sf_internal and
13241         *movv2sf_internal_rex64 using MMXMODE mode iterator.
13243 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13245         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13246         (record_last_mem_set_info): Likewise.
13248         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13249         of XNEWVEC followed by memset.
13250         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13252 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13254         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13255         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13256         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13257         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13258         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13259         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13260         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13261         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13262         BARRIER_P instead of GET_CODE.
13264 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13266         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13267         inaccuracy in the probing code.
13269         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13270         (ctrapdi4): Likewise.
13272 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13274         * calls.c (expand_call): Add missing guard to code handling return
13275         of non-BLKmode structures in MSB.
13276         * function.c (expand_function_end): Likewise.
13278 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13280         * combine.c (try_combine): Adjust comment.  Do not add the set of
13281         insn #0 if the destination indirectly is set or dies in insn #2.
13282         Tidy up code to distribute a new note.
13284 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13286         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13287         also for alternatives 16 and 17.
13289 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13291         * config/i386/sse.md (*mov<mode>_internal): Merge with
13292         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13293         Emit insn template depending on type attribute.  Use
13294         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13295         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13296         attribute calculation.  Remove unit attribute calculation.
13297         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13298         Set prefix_data16 attribute for DImode ssemov types.
13299         Use Ym instead of y for SSE-MMX conversion alternatives.
13300         Reorder operand constraints.
13302 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13304         * df.h (df_insn_delete): Adjust prototype.
13305         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13306         and let it decide whether mark the basic block dirty.
13307         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13308         * df-scan.c (df_insn_info_delete): New helper function, split
13309         off from df_insn_delete.
13310         (df_scan_free_bb_info): Use it.
13311         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13312         Likewise.
13313         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13314         that the insn is actually an insn and it has a non-NULL basic block.
13315         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13317 2013-03-22  Richard Biener  <rguenther@suse.de>
13319         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13320         dep_ref members.
13321         (mem_ref_alloc): Do not allocate them.
13322         (refs_independent_p): Do not query or maintain a cache.
13324 2013-03-22  Richard Biener  <rguenther@suse.de>
13326         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13327         (gather_mem_refs_in_loops): Do not compute it.
13328         (analyze_memory_references): Do not allocate it.
13329         (tree_ssa_lim_finalize): Do not free it.
13330         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13332 2013-03-22  Richard Biener  <rguenther@suse.de>
13334         * is-a.h (as_a): Use gcc_checking_assert.
13336 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13338         * config/aarch64/aarch64.c (aarch64_print_operand): New
13339         format specifier for printing a constant in hex.
13340         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13341         format specifier for printing second operand.
13343 2013-03-22  Richard Biener  <rguenther@suse.de>
13345         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13346         bitmaps.
13347         (gather_mem_refs_in_loops): Perform store accumulation here.
13348         (create_vop_ref_mapping_loop): Remove.
13349         (create_vop_ref_mapping): Likewise.
13350         (analyze_memory_references): Initialize refs_stored_in_loop.
13351         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13352         (record_indep_loop): Remove.
13353         (record_dep_loop): New function.
13354         (ref_indep_loop_p_1): Adjust to only walk over references
13355         in the loop, not its subloops.
13356         (ref_indep_loop_p): Rename to ...
13357         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13358         maintaining a more fine-grained cache.
13359         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13360         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13362 2013-03-22  Richard Biener  <rguenther@suse.de>
13364         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13365         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13366         aggregate mem_ref_loc.
13367         (free_mem_ref_locs): Inline into ...
13368         (memref_free): ... this and adjust.
13369         (mem_ref_alloc): Adjust.
13370         (mem_ref_locs_alloc): Remove.
13371         (record_mem_ref_loc): Adjust.
13372         (get_all_locs_in_loop): Rewrite into ...
13373         (for_all_locs_in_loop): ... this iterator.
13374         (rewrite_mem_ref_loc): New functor.
13375         (rewrite_mem_refs): Use for_all_locs_in_loop.
13376         (sm_set_flag_if_changed): New functor.
13377         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13378         (ref_always_accessed): New functor.
13379         (ref_always_accessed_p): Use for_all_locs_in_loop.
13381 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13383         * tree-pass.h (PROP_gimple_lvec): New.
13384         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13385         (init_optimization_passes): Move pass_lower_vector.
13386         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13387         PROP_gimple_lvec.
13388         (pass_lower_vector): Provide PROP_gimple_lvec.
13389         (pass_lower_vector_ssa): Likewise.
13390         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13392 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13394         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13396 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13398         * config/i386/i386.md (*movdi_internal): Disparage slightly
13399         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13400         conversion alternatives.
13402 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13404         PR middle-end/48087
13405         * diagnostic.def (DK_WERROR): New kind.
13406         * diagnostic.h (werrorcount): Define.
13407         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13408         promoted to DK_ERROR, increment DK_WERROR counter instead of
13409         DK_ERROR counter.
13410         * toplev.c (toplev_main): Call print_ignored_options even if
13411         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13412         even if just werrorcount is non-zero.
13414         PR debug/55608
13415         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13416         on failure.
13417         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13418         (string_cst_pool_decl): New function.
13419         (optimize_one_addr_into_implicit_ptr): New function.
13420         (resolve_addr_in_expr): Optimize DWARF location expression
13421         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13422         which doesn't live in memory, but has DW_AT_location or
13423         DW_AT_const_value, or refers to a string literal, into
13424         DW_OP_GNU_implicit_pointer.
13425         (optimize_location_into_implicit_ptr): New function.
13426         (resolve_addr): If removing DW_AT_location of a variable because
13427         it was DW_OP_addr of address of the variable, but the variable doesn't
13428         live in memory, try to emit const value attribute for the initializer.
13430 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13432         * tree.h (VECTOR_TYPE_P): New macro.
13433         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13434         TYPE_MODE): Use it.
13435         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13436         VEC_COND_EXPR cannot be lvalues.
13437         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13439 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13441         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13442         Restrict the transformation to equal modes.
13444 2013-03-21  Richard Biener  <rguenther@suse.de>
13446         PR tree-optimization/39326
13447         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13448         (MEM_ANALYZABLE): Adjust.
13449         (record_mem_ref_loc): Move bitmap ops ...
13450         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13451         unanalyzable refs, do not record locations for it.
13452         (analyze_memory_references): Allocate ref zero as shared
13453         unanalyzable ref.
13454         (refs_independent_p): Do not test for unanalyzed mems here.
13455         (ref_indep_loop_p_1): Special-case disambiguation against
13456         the unanalyzed ref.
13457         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13459 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13461         * config/arm/arm-protos.h (tune_params): Add
13462         prefer_neon_for_64bits field.
13463         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13464         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13465         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13466         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13467         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13468         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13469         (arm_option_override): Handle -mneon-for-64bits new option.
13470         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13471         (prefer_neon_for_64bits): Declare new variable.
13472         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13473         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13474         (arch_enabled): Handle new arch types. Remove support for onlya8
13475         and nota8.
13476         (one_cmpldi2): Use new arch names.
13477         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13478         * config/arm/arm.opt (mneon-for-64bits): Add option.
13479         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13480         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13481         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13482         of onlya8.
13483         * doc/invoke.texi (-mneon-for-64bits): Document.
13485 2013-03-21  Richard Biener  <rguenther@suse.de>
13487         PR tree-optimization/39326
13488         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13489         (sort_bbs_in_loop_postorder_cmp): New function.
13490         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13492 2013-03-21  Richard Biener  <rguenther@suse.de>
13494         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13495         (vect_insert_into_interleaving_chain): Likewise.
13496         (vect_drs_dependent_in_basic_block): Inline ...
13497         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13498         split out from ...
13499         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13500         (vect_check_interleaving): Simplify.
13501         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13502         (vect_slp_analyze_data_ref_dependences): ... this new function.
13503         (dr_group_sort_cmp): New function.
13504         (vect_analyze_data_ref_accesses): Compute data-reference groups
13505         here instead of in vect_analyze_data_ref_dependence.  Use
13506         a more efficient algorithm.
13507         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13508         vect_slp_analyze_data_ref_dependences.  Call
13509         vect_analyze_data_ref_accesses earlier.
13510         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13511         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13512         (vect_slp_analyze_data_ref_dependences): New prototype.
13514 2013-03-21  Richard Biener  <rguenther@suse.de>
13516         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13517         ref is stored in the loop.
13518         (find_refs_for_sm): Walk only over all stores.
13519         (store_motion_loop): Allocate from lim_bitmap_obstack.
13520         (store_motion): Likewise.
13522 2013-03-21  Richard Biener  <rguenther@suse.de>
13524         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13525         Update virtual SSA form.
13527 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13529         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13530         * configure: Regenerate.
13531         * config.in: Regenerate.
13532         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13533         if !HAVE_LD_EH_FRAME_CIEV3.
13535 2013-03-21  Richard Biener  <rguenther@suse.de>
13537         * tree-cfg.c (verify_expr_no_block): New function.
13538         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13539         nor DECL_VALUE_EXPR have locations with associated blocks.
13540         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13541         (clear_unused_block_pointer): Remove code dealing with
13542         blocks in DECL_DEBUG_EXPR locations.
13544 2013-03-21  Richard Biener  <rguenther@suse.de>
13546         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13547         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
13548         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13549         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13550         instead of DECL_DEBUG_EXPR_IS_FROM.
13551         * gimplify.c (gimplify_modify_expr): Likewise.
13552         * tree-cfg.c (verify_expr_location_1): Likewise.
13553         * tree-complex.c (create_one_component_var): Likewise.
13554         * tree-sra.c (create_access_replacement): Likewise.
13555         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13556         (clear_unused_block_pointer): Likewise.
13557         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13558         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13559         * var-tracking.c (var_debug_decl): Likewise.
13560         (track_expr_p): Likewise.
13561         * tree-inline.c (add_local_variables): Likewise.  Set
13562         DECL_HAS_DEBUG_EXPR_P after copying it.
13563         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13564         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
13566 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13568         PR bootstrap/56656
13569         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13570         * configure: Regenerate.
13571         * config.in: Regenerate.
13572         * config/i386/i386.md (*movdf_internal): Use
13573         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13574         movd instead of movq mnemonic for interunit moves.
13575         (*movdi_internal): Ditto.
13577 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13579         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13580         (abd<mode>_3): New pattern.
13581         (aba<mode>_3): New pattern.
13582         (fabd<mode>_3): New pattern.
13584 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13586         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13587         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13588         occurrence of REGISTER_PREFIX as its empty string.
13590 2013-03-20  Jeff Law  <law@redhat.com>
13592         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13593         addititional equivalences for equality comparisons between an SSA_NAME
13594         and a constant where the SSA_NAME was set from a widening conversion.
13596 2013-03-20  Walter Lee  <walt@tilera.com>
13598         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13600 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13602         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13603         depending on type attribute.
13604         (*movti_internal): Ditto.
13605         (*movtf_internal): Ditto.
13606         (*movxf_internal): Ditto.
13607         (*movdf_internal): Ditto.
13608         (*movsf_internal): Ditto.
13610 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13612         * config/i386/i386.md (*movti_internal): Set prefix attribute to
13613         maybe_vex for sselog1 and ssemov types.
13614         (*movdi_internal): Reorder operand constraints.
13615         (*movsi_internal): Ditto.  Set prefix attribute to
13616         maybe_vex for sselog1 and ssemov types.
13617         (*movtf_internal): Set prefix attribute to maybe_vex
13618         for sselog1 and ssemov types.
13619         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
13620         DImode ssemov types.  Reorder operand constraints.
13621         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
13622         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
13623         attribute for SImode ssemov types.  Reorder operand constraints.
13625 2013-03-20  Martin Jambor  <mjambor@suse.cz>
13627         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13628         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13630 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
13632         * config/rs6000/predicates.md (indexed_address, update_address_mem
13633         update_indexed_address_mem): New predicates.
13634         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13635         attribute for load/store instructions.
13636         * config/rs6000/dfp.md (movsd_store): Likewise.
13637         (movsd_load): Likewise.
13638         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13639         (unnamed HI->DI extend define_insn): Likewise.
13640         (unnamed SI->DI extend define_insn): Likewise.
13641         (unnamed QI->SI extend define_insn): Likewise.
13642         (unnamed QI->HI extend define_insn): Likewise.
13643         (unnamed HI->SI extend define_insn): Likewise.
13644         (unnamed HI->SI extend define_insn): Likewise.
13645         (extendsfdf2_fpr): Likewise.
13646         (movsi_internal1): Likewise.
13647         (movsi_internal1_single): Likewise.
13648         (movhi_internal): Likewise.
13649         (movqi_internal): Likewise.
13650         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13651         attribute for load/store instructions.
13652         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13653         instructions.
13654         (mov<mode>_softfloat): Likewise.
13655         (mov<mode>_hardfloat32): Likewise.
13656         (mov<mode>_hardfloat64): Likewise.
13657         (mov<mode>_softfloat64): Likewise.
13658         (movdi_internal32): Likewise.
13659         (movdi_internal64): Likewise.
13660         (probe_stack_<mode>): Likewise.
13662 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
13664         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13665         floating point, and decimal floating point to reload iterator.
13667         * config/rs6000/constraints.md (wl constraint): New constraints to
13668         return FLOAT_REGS if certain options are used to reduce the number
13669         of separate patterns that exist in the file.
13670         (wx constraint): Likewise.
13671         (wz constraint): Likewise.
13673         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13674         -mdebug=reg, print wg, wl, wx, and wz constraints.
13675         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13676         Initialize the reload functions for 64-bit binary/decimal floating
13677         point types.
13678         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13679         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13680         create the buffer on the stack to overcome not having a 32-bit
13681         load and store.
13682         (rs6000_emit_move): Likewise.
13683         (rs6000_secondary_memory_needed_rtx): Likewise.
13684         (rs6000_alloc_sdmode_stack_slot): Likewise.
13685         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13686         via xxlxor, just like DFmode 0.0.
13688         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13689         define as 1 if we are running on a power7 or newer.
13690         (enum r6000_reg_class_enum): Add new constraints.
13692         * config/rs6000/dfp.md (movsd): Delete, combine with binary
13693         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13694         with other moves by using conditional constraits (wg).  Use LFIWZX
13695         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
13696         (movsd splitter): Likewise.
13697         (movsd_hardfloat): Likewise.
13698         (movsd_softfloat): Likewise.
13700         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13701         binary and decimal floating point moves.
13702         (fmove_ok): New attributes to combine binary and decimal floating
13703         point moves, and to combine power6x (mfpgpr) moves along normal
13704         floating moves.
13705         (real_value_to_target): Likewise.
13706         (f32_lr): Likewise.
13707         (f32_lm): Likewise.
13708         (f32_li): Likewise.
13709         (f32_sr): Likewise.
13710         (f32_sm): Likewise.
13711         (f32_si): Likewise.
13712         (movsf): Combine binary and decimal floating point moves.  Combine
13713         power6x (mfpgpr) moves with other moves by using conditional
13714         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
13715         (mov<mode> for SFmode/SDmode); Likewise.
13716         (SFmode/SDmode splitters): Likewise.
13717         (movsf_hardfloat): Likewise.
13718         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13719         (movsf_softfloat): Likewise.
13720         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13722         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13723         wx and wz constraints.
13725         * config/rs6000/constraints.md (wg constraint): New constraint to
13726         return FLOAT_REGS if -mmfpgpr (power6x) was used.
13728         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13729         constraint.
13731         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13732         -mdebug=reg, print wg, wl, wx, and wz constraints.
13733         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13734         Initialize the reload functions for 64-bit binary/decimal floating
13735         point types.
13736         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13737         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13738         create the buffer on the stack to overcome not having a 32-bit
13739         load and store.
13740         (rs6000_emit_move): Likewise.
13741         (rs6000_secondary_memory_needed_rtx): Likewise.
13742         (rs6000_alloc_sdmode_stack_slot): Likewise.
13743         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13744         via xxlxor, just like DFmode 0.0.
13746         * config/rs6000/dfp.md (movdd): Delete, combine with binary
13747         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13748         with other moves by using conditional constraits (wg).  Use LFIWZX
13749         and STFIWX for loading SDmode on power7.
13750         (movdd splitters): Likewise.
13751         (movdd_hardfloat32): Likewise.
13752         (movdd_softfloat32): Likewise.
13753         (movdd_hardfloat64_mfpgpr): Likewise.
13754         (movdd_hardfloat64): Likewise.
13755         (movdd_softfloat64): Likewise.
13757         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13758         64-bit binary and decimal floating point moves.
13759         (FMOVE64X): Likewise.
13760         (movdf): Combine 64-bit binary and decimal floating point moves.
13761         Combine power6x (mfpgpr) moves with other moves by using
13762         conditional constraits (wg).
13763         (mov<mode> for DFmode/DDmode): Likewise.
13764         (DFmode/DDmode splitters): Likewise.
13765         (movdf_hardfloat32): Likewise.
13766         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13767         (movdf_softfloat32): Likewise.
13768         (movdf_hardfloat64_mfpgpr): Likewise.
13769         (movdf_hardfloat64): Likewise.
13770         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13771         (movdf_softfloat64): Likewise.
13772         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13773         (reload_<mode>_load): Move to later in the file so they aren't in
13774         the middle of the floating point move insns.
13775         (reload_<mode>_store): Likewise.
13777         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13778         constraint.
13780         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13781         constraint if -mdebug=reg.
13782         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13783         Enable using dd reload support if needed.
13785         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13786         binary and decimal floating point moves in rs6000.md.
13787         (movtd_internal): Likewise.
13789         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13790         decimal floating point moves.
13791         (movtf): Likewise.
13792         (movtf_internal): Likewise.
13793         (mov<mode>_internal, TDmode/TFmode): Likewise.
13794         (movtf_softfloat): Likewise.
13795         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13797         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13798         movdi_internal64, using wg constraint for move direct operations.
13799         (movdi_internal64): Likewise.
13801         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13802         MODES_TIEABLE_P for selected modes.  Print the numerical value of
13803         the various virtual registers. Use GPR/FPR first/last values,
13804         instead of hard coding the register numbers.  Print which modes
13805         have reload functions registered.
13806         (rs6000_option_override_internal): If -mdebug=reg, trace the options
13807         settings before/after setting cpu, target and subtarget settings.
13808         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13809         and for secondary reload failures in rs6000_secondary_reload_inner.
13810         (rs6000_secondary_reload_fail): Likewise.
13811         (rs6000_secondary_reload_inner): Likewise.
13813         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13814         macros for first/last GPR and FPR registers.
13815         (LAST_GPR_REGNO): Likewise.
13816         (FIRST_FPR_REGNO): Likewise.
13817         (LAST_FPR_REGNO): Likewise.
13819         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13820         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13821         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13822         (vcond<mode><mode>): Likewise.
13823         (vcondu<mode><mode>): Likewise.
13824         (vector_gtu<mode>): Likewise.
13825         (vector_gte<mode>): Likewise.
13826         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13827         to prevent the compiler from converting DImode operations to TImode.
13828         (ior<mode>3): Likewise.
13829         (and<mode>3): Likewise.
13830         (one_cmpl<mode>2): Likewise.
13831         (nor<mode>3): Likewise.
13832         (andc<mode>3): Likewise.
13834         * config/rs6000/constraints.md (wt constraint): New constraint
13835         that returns VSX_REGS if TImode is allowed in VSX registers.
13837         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13838         constant under VSX.
13840         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13841         similar to TImode, but it is restricted to being in the GPRs.
13843         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13844         TImode to occupy a single VSX register.
13846         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13847         -mvsx-timode for power7/power8.
13848         (power7 cpu): Likewise.
13849         (power8 cpu): Likewise.
13851         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13852         sure that TFmode/TDmode take up two registers if they are ever
13853         allowed in the upper VSX registers.
13854         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13855         registers.
13856         (rs6000_init_hard_regno_mode_ok): Likewise.
13857         (rs6000_debug_reg_global): Add debugging for PTImode and wt
13858         constraint.  Print if LRA is turned on.
13859         (rs6000_option_override_internal): Give an error if -mvsx-timode
13860         and VSX is not enabled.
13861         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
13862         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13863         to reg+offset addressing.  Use PTImode when checking offset
13864         addresses for validity.
13865         (reg_offset_addressing_ok_p): Likewise.
13866         (rs6000_legitimate_offset_address_p): Likewise.
13867         (rs6000_legitimize_address): Likewise.
13868         (rs6000_legitimize_reload_address): Likewise.
13869         (rs6000_legitimate_address_p): Likewise.
13870         (rs6000_eliminate_indexed_memrefs): Likewise.
13871         (rs6000_emit_move): Likewise.
13872         (rs6000_secondary_reload): Likewise.
13873         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
13874         reloads to fpr registers to continue to use reg+offset addressing,
13875         but 64-bit reloads to altivec registers need reg+reg addressing.
13876         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13877         it.  Treat LO_SUM like a PLUS operation.
13878         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13879         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13880         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13881         registers to share a register with a smaller sized type, since VSX
13882         puts scalars in the upper 64-bits.
13883         (print_operand): Add support for PTImode.
13884         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13885         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13886         registers, but don't have arithmetic support.
13887         (rs6000_memory_move_cost): Add test for VSX.
13888         (rs6000_opt_masks): Add -mvsx-timode.
13890         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13891         for TImode.
13892         (VSs): Likewise.
13893         (VSr): Use wt constraint for TImode.
13894         (VSv): Drop TImode support.
13895         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13896         (vsx_movti_64bit): Likewise.
13897         (vsx_movti_32bit): Likewise.
13898         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13899         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13900         one '?' on the appropriate output constraint.  Do not allow TImode
13901         logical operations on 32-bit systems.
13902         (vsx_ior<mode>3): Likewise.
13903         (vsx_xor<mode>3): Likewise.
13904         (vsx_one_cmpl<mode>2): Likewise.
13905         (vsx_nor<mode>3): Likewise.
13906         (vsx_andc<mode>3): Likewise.
13907         (vsx_concat_<mode>): Likewise.
13908         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13910         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13911         OPTION_MASK_VSX_TIMODE.
13912         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13913         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13915         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13916         (TI2 iterator): New iterator for TImode, PTImode.
13917         (wd mode attribute): Add values for vector types.
13918         (movti_string): Replace TI move operations with operations for TImode
13919         and PTImode.  Add support for TImode being allowed in VSX registers.
13920         (mov<mode>_string, TImode/PTImode): Likewise.
13921         (movti_ppc64): Likewise.
13922         (mov<mode>_ppc64, TImode/PTImode): Likewise.
13923         (TI mode splitters): Likewise.
13925         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13926         constraint.
13928 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
13930         PR tree-optimization/56355
13931         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13932         Also handle integers with undefined overflow.
13934 2013-03-20  Catherine Moore  <clm@codesourcery.com>
13935             Maciej W. Rozycki  <macro@codesourcery.com>
13936             Tom de Vries  <tom@codesourcery.com>
13937             Nathan Sidwell <nathan@codesourcery.com>
13938             Iain Sandoe  <iain@codesourcery.com>
13939             Nathan Froyd  <froydnj@codesourcery.com>
13940             Chao-ying Fu <fu@mips.com>
13942         * doc/extend.texi: (micromips, nomicromips, nocompression):
13943         Document new function attributes.
13944         * doc/invoke.texi (minterlink-compressed, mmicromips,
13945         m14k, m14ke, m14kec): Document new options.
13946         (minterlink-mips16): Update documentation.
13947         * doc/md.texi (ZC, ZD): Document new constraints.
13948         * configure.ac (gcc_cv_as_micromips): Check if linker
13949         supports the .set micromips directive.
13950         * configure: Regenerate.
13951         * config.in: Regenerate.
13952         * config/mips/mips-tables.opt: Regenerate.
13953         * config/mips/micromips.md: New file.
13954         * constraints.md (ZC, ZD): New constraints.
13955         * config/mips/predicates.md (movep_src_register): New predicate.
13956         (movep_src_operand): New predicate.
13957         (non_volatile_mem_operand): New predicate.
13958         * config/mips/mips.md (multimem): New type.
13959         (length): Differentiate between 17-bit and 18-bit branch offsets.
13960         (MOVEP1, MOVEP2): New mode iterator.
13961         (mov_<load>l): Use ZC constraint.
13962         (mov_<load>r): Likewise.
13963         (mov_<store>l): Likewise.
13964         (mov_<store>r): Likewise.
13965         (*branch_equality<mode>_inverted): Add microMIPS support.
13966         (*branch_equality<mode>): Likewise.
13967         (*jump_absolute): Likewise.
13968         (indirect_jump_<mode>): Likewise.
13969         (tablejump_<mode>): Likewise.
13970         (<optab>_internal): Likewise.
13971         (sibcall_internal): Likewise.
13972         (sibcall_value_internal): Likewise.
13973         (prefetch): Use constraint ZD.
13974         * config/mips/mips.opt (minterlink-compressed): New option.
13975         (minterlink-mips16): Now an alias for minterlink-compressed.
13976         (mmicromips): New option.
13977         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13978         (compare_and_swap_12): Likewise.
13979         (sync_add<mode>): Likewise.
13980         (sync_<optab>_12): Likewise.
13981         (sync_old_<optab>_12): Likewise.
13982         (sync_new_<optab>_12): Likewise.
13983         (sync_nand_12): Likewise.
13984         (sync_old_nand_12): Likewise.
13985         (sync_new_nand_12): Likewise.
13986         (sync_sub<mode>): Likewise.
13987         (sync_old_add<mode>): Likewise.
13988         (sync_old_sub<mode>): Likewise.
13989         (sync_new_add<mode>): Likewise.
13990         (sync_new_sub<mode>): Likewise.
13991         (sync_<optab><mode>): Likewise.
13992         (sync_old_<optab><mode>): Likewise.
13993         (sync_new_<optab><mode>): Likewise.
13994         (sync_nand<mode>): Likewise.
13995         (sync_old_nand<mode>): Likewise.
13996         (sync_new_nand<mode>): Likewise.
13997         (sync_lock_test_and_set<mode>): Likewise.
13998         (test_and_set_12): Likewise.
13999         (atomic_compare_and_swap<mode>): Likewise.
14000         (atomic_exchange<mode>_llsc): Likewise.
14001         (atomic_fetch_add<mode>_llsc): Likewise.
14002         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14003         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14004         (umips_save_restore_pattern_p): Likewise.
14005         (umips_load_store_pair_p): Likewise.
14006         (umips_output_load_store_pair): Likewise.
14007         (umips_movep_target_p): Likewise.
14008         (umips_12bit_offset_address_p): Likewise.
14009         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14010         (mips_base_mips16): Rename this...
14011         (mips_base_compression_flags): ...to this. Update all uses.
14012         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14013         (mips_mips16_decl_p): Delete.
14014         (mips_nomips16_decl_p): Delete.
14015         (mips_get_compress_on_flags): New function.
14016         (mips_get_compress_off_flags): New function.
14017         (mips_get_compress_mode): New function.
14018         (mips_get_compress_on_name): New function.
14019         (mips_get_compress_off_name): New function.
14020         (mips_insert_attributes): Support multiple compression types.
14021         (mips_merge_decl_attributes): Likewise.
14022         (umips_12bit_offset_address_p): New function.
14023         (mips_start_function_definition): Emit .set micromips directive.
14024         (mips_call_may_need_jalx_p): New function.
14025         (mips_function_ok_for_sibcall): Add microMIPS support.
14026         (mips_print_operand_punctuation): Support short delay slots and
14027         compact jumps.
14028         (umips_swm_mask, umips_swm_encoding): New.
14029         (umips_build_save_restore): New function.
14030         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14031         (was_mips16_p): Remove.
14032         (old_compression_mode): New.
14033         (mips_set_compression_mode): New function.
14034         (mips_set_current_function): Add microMIPS support.
14035         (mips_option_override): Likewise.
14036         (umips_save_restore_pattern_p): New function.
14037         (umips_output_save_restore): New function.
14038         (umips_load_store_pair_p_1): New function.
14039         (umips_load_store_pair_p): New function.
14040         (umips_output_load_store_pair_1): New function.
14041         (umips_output_load_store_pair): New function.
14042         (umips_movep_target_p) New function.
14043         (mips_prepare_pch_save): Add microMIPS support.
14044         * config/mips/mips.h (TARGET_COMPRESSION): New.
14045         (TARGET_CPU_CPP_BUILTINS): Update macro
14046         to use new compression flags and to support microMIPS.
14047         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14048         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14049         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14050         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14051         (ASM_SPEC): Support mmicromips and mno-micromips.
14052         (M16STORE_REG_P): New macro.
14053         (MIPS_CALL): Support TARGET_MICROMIPS.
14054         (MICROMIPS_J): New macro.
14055         (mips_base_mips16): Rename this...
14056         (mips_base_compression_flags): ...to this.
14057         (UMIPS_12BIT_OFFSET_P): New macro.
14058         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14059         (MULTILIB_DIRNAMES): Likewise.
14060 2013-03-20  Richard Biener  <rguenther@suse.de>
14062         PR tree-optimization/56661
14063         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14064         the result does not have to be distinct.
14066 2013-03-20  Richard Biener  <rguenther@suse.de>
14068         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14069         remap_gimple_op_r.
14071 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14072             Steven Bosscher <steven@gcc.gnu.org>
14074         PR rtl-optimization/56605
14075         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14077 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14079         PR bootstrap/56656
14080         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14081         that require movd instead of movq.
14083 2013-03-20  Richard Biener  <rguenther@suse.de>
14085         * tree-ssa-structalias.c (struct variable_info): Add pointer
14086         to the first field of an aggregate with sub-vars.  Make
14087         this and the pointer to the next subfield its ID.
14088         (vi_next): New function.
14089         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14090         storedanything_id, integer_id): Increment by one.
14091         (new_var_info, get_call_vi, lookup_call_clobber_vi,
14092         get_call_clobber_vi): Adjust.
14093         (solution_set_expand): Simplify and speedup.
14094         (solution_set_add): Inline into ...
14095         (set_union_with_increment): ... this.  Adjust accordingly.
14096         (do_sd_constraint): Likewise.
14097         (do_ds_constraint): Likewise.
14098         (do_complex_constraint): Simplify.
14099         (build_pred_graph): Adjust.
14100         (solve_graph): Likewise.  Simplify and speedup.
14101         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14102         get_constraint_for_component_ref, get_constraint_for_1,
14103         first_vi_for_offset, first_or_preceding_vi_for_offset,
14104         create_function_info_for, create_variable_info_for_1,
14105         create_variable_info_for, intra_create_variable_infos): Adjust.
14106         (init_base_vars): Push NULL for ID zero.
14107         (compute_points_to_sets): Adjust.
14109 2013-03-20  Richard Biener  <rguenther@suse.de>
14111         * cfgloop.c (verify_loop_structure): Streamline and avoid
14112         ICEing on corrupt loop tree.
14113         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14114         loop tree.
14116 2013-03-20  Richard Biener  <rguenther@suse.de>
14118         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14119         check whether an SSA update is needed.
14121 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
14123         * config/mips/constraints.md (T): Rename to...
14124         (Yf): ...this.
14125         (U): Rename to...
14126         (Yd): ...this.
14127         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14128         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14130 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14132         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14133         (*subsi3_carryin_uxtw): Likewise.
14135 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14137         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14138         (*rorsi3_insn_uxtw): Likewise.
14140 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14142         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14143         (*extrsi5_insn_uxtw): Likewise.
14145 2013-03-19  Richard Biener  <rguenther@suse.de>
14147         PR tree-optimization/56273
14148         * passes.c (init_optimization_passes): Move second VRP after DOM.
14150 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
14152         * config/i386/i386.md (*movti_internal): Merge from
14153         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
14154         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
14155         nox64 isa attributes.
14157 2013-03-18  Richard Biener  <rguenther@suse.de>
14159         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14160         (unite): Likewise.
14161         (merge_node_constraints): Likewise.
14162         (build_succ_graph): Likewise.
14163         (valid_graph_edge): Inline into single caller.
14164         (unify_nodes): Likewise.  Use bitmap_set_bit return value
14165         and cache varinfo.
14166         (scc_visit): Fix formatting and variable use.
14167         (do_sd_constraint): Use gcc_checking_assert.
14168         (do_ds_constraint): Likewise.
14169         (do_complex_constraint): Likewise.
14170         (condense_visit): Likewise.  Cleanup.
14171         (dump_pred_graph): New function.
14172         (perform_var_substitution): Dump the pred-graph before
14173         variable substitution.
14174         (find_equivalent_node): Use gcc_checking_assert.
14175         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14177 2013-03-18  Richard Biener  <rguenther@suse.de>
14179         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14180         Remove cond_expr_stmt_list argument and do not gimplify the
14181         built expression.
14182         (vect_loop_versioning): Adjust.
14183         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14184         Cleanup to use less temporaries.
14185         (vect_create_data_ref_ptr): Cleanup.
14187 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14189         PR tree-optimization/56635
14190         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14191         require types_compatible_p types.
14193 2013-03-18  Nick Clifton  <nickc@redhat.com>
14195         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14196         spurious backslash.
14198         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14199         Add missing line to comment describing function.
14201 2013-03-18  Richard Biener  <rguenther@suse.de>
14203         PR tree-optimization/56210
14204         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14205         Handle string / character search functions.
14206         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14208 2013-03-18  Richard Biener  <rguenther@suse.de>
14210         PR middle-end/56483
14211         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14212         and implement properly.
14213         * gimple.h (gimple_cond_single_var_p): Remove.
14215 2013-03-18  Richard Biener  <rguenther@suse.de>
14217         * tree-data-ref.h (find_data_references_in_loop): Declare.
14218         * tree-data-ref.c (get_references_in_stmt): Use a stack
14219         vector pre-allocated in the callers.
14220         (find_data_references_in_stmt): Adjust.
14221         (graphite_find_data_references_in_stmt): Likewise.
14222         (create_rdg_vertices): Likewise.
14223         (find_data_references_in_loop): Export.
14224         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14225         Compute dependences here...
14226         (vect_analyze_data_refs): ...not here.  When we encounter
14227         a non-vectorizable data reference in basic-block vectorization
14228         truncate the data reference vector.  Do not bother to
14229         fixup data-dependence information for gather loads.
14230         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14231         of data references, as reported.
14233 2013-03-18  Richard Biener  <rguenther@suse.de>
14235         PR tree-optimization/3713
14236         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
14237         has_constants and expr.
14238         (stmt_has_constants): Properly valueize SSA names when deciding
14239         whether the stmt has constants.
14241 2013-03-18  Richard Biener  <rguenther@suse.de>
14243         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14244         whole function when there is nothing to do.
14245         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14246         * tree-vectorizer.c (vectorize_loops): Update virtual and
14247         loop-closed SSA once.
14248         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14250 2013-03-18  Richard Biener  <rguenther@suse.de>
14252         PR middle-end/56113
14253         * domwalk.c (bb_postorder): New global static.
14254         (cmp_bb_postorder): New function.
14255         (walk_dominator_tree): Replace scheme imposing an order for
14256         visiting dominator sons by one sorting them at the time they
14257         are pushed on the stack.
14259 2013-03-18  Richard Biener  <rguenther@suse.de>
14261         PR tree-optimization/39326
14262         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14263         (struct mem_ref): Replace mem member with ao_ref typed member.
14264         (MEM_ANALYZABLE): Adjust.
14265         (memref_eq): Likewise.
14266         (mem_ref_alloc): Likewise.
14267         (gather_mem_refs_stmt): Likewise.
14268         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14269         (execute_sm_if_changed_flag_set): Adjust.
14270         (execute_sm): Likewise.
14271         (ref_always_accessed_p): Likewise.
14272         (refs_independent_p): Likewise.
14273         (can_sm_ref_p): Likewise.
14275 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14277         PR c/56566
14278         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14279         return 1 even for !unsignedp.
14281 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14283         * config/i386/i386.md (isa): Add x64 and nox64.
14284         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14285         (*pushtf): Enable *roF alternative for x64 isa only.
14286         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14287         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14288         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14289         (*movtf_internal): Merge from *movtf_internal_rex64 and
14290         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14291         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14292         nox64 isa attributes.
14293         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14294         nox64 isa attributes.
14295         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14297 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14299         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14301 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14303         PR target/56640
14304         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14305         class names.  Remove trailing comma after "ALL_REGS".
14307 2013-03-16  Jan Hubicka  <jh@suse.cz>
14309         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14310         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14311         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14312         of cgraph_get_create_node.
14313         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14315 2013-03-16  Jason Merrill  <jason@redhat.com>
14317         PR debug/49090
14318         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14319         with DW_AT_default_value.
14321 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14323         * BASE-VER: Set to 4.9.0.
14325 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14327         PR target/56619
14328         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14329         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14330         Document _x* TSX intrinsics.
14332 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14333             David Holsgrove  <david.holsgrove@xilinx.com>
14335         * configure.ac: Add MicroBlaze TLS support detection.
14336         * configure: Regenerate.
14337         * config/microblaze/microblaze-protos.h
14338         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14339         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14340         * config/microblaze/microblaze.c (microblaze_address_type): Add
14341         ADDRESS_TLS and tls_reloc address types.
14342         (microblaze_address_info): Add tls_reloc.
14343         (TARGET_HAVE_TLS): Define.
14344         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14345          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14346          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14347         load_tls_operand,  microblaze_call_tls_get_addr,
14348         microblaze_legitimize_tls_address): New functions.
14349         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14350         (get_base_reg): Use microblaze_tls_symbol_p.
14351         (microblaze_classify_address): Handle TLS.
14352         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14353         label_mentioned_p and microblaze_tls_referenced_p.
14354         (microblaze_legitimize_address): Handle TLS.
14355         (microblaze_address_insns): Handle ADDRESS_TLS.
14356         (pic_address_needs_scratch): Handle TLS.
14357         (print_operand_address): Handle TLS.
14358         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14359         (microblaze_expand_move): Handle TLS.
14360         (microblaze_legitimate_constant_p): Check
14361         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14362         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14363         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14364         (PIC_OFFSET_TABLE_REGNUM): Set.
14365         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14366         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14367         (addsi3, movsi_internal2, movdf_internal): Update constraints
14368         * config/microblaze/predicates.md (arith_plus_operand): Define
14369         (move_operand): Redefine as move_src_operand,
14370         check microblaze_tls_referenced_p.
14372 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14374         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14375         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14377 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14379         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14380         CC mode for AND.
14382 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14384         PR tree-optimization/53265
14385         * common.opt (Waggressive-loop-optimizations): New option.
14386         * tree-ssa-loop-niter.c: Include tree-pass.h.
14387         (do_warn_aggressive_loop_optimizations): New function.
14388         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14389         if number_of_latch_executions returned constant.
14390         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14391         early.  If number_of_latch_executions returned constant, set
14392         nb_iterations_upper_bound back to it.
14393         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14394         field.
14395         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14396         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14398         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14399         (MULTILIB_OSDIRNAMES): Set.
14400         * genmultilib: If defaultosdirname doesn't start with :: , set
14401         defaultosdirname2 instead, clear it and emit two . multilib_raw
14402         entries instead of just one.
14404 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14406         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14407         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14408         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14409         (SUBTARGET_OVERRIDE_OPTIONS): New.
14411 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14413         PR target/49880
14414         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14415         (musermode): Convert to Var(TARGET_USERMODE).
14416         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14417         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14418         * config/sh/sh.c (sh_option_override): Use
14419         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14420         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14421         condition.
14422         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14423         TARGET_SH4.
14424         (udivsi3_i4_single, divsi3_i4_single): Use
14425         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14427 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14429         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14430         default setting.
14432 2013-03-13  Richard Biener  <rguenther@suse.de>
14434         PR tree-optimization/56608
14435         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14436         calls when vectorizing basic-blocks.
14438 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14440         PR plugins/45078
14441         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14442         tm_file.
14444 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14446         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14448 2013-03-11  Jan Hubicka  <jh@suse.cz>
14450         PR lto/56557
14451         * lto-streamer-out.c (output_symbol_p): Skip references from
14452         constructors of external variables.
14454 2013-03-11  Jan Hubicka  <jh@suse.cz>
14456         PR middle-end/56571
14457         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14458         from pseudos.
14459         * emit-rtl.c (verify_rtx_sharing): Likewise.
14460         (copy_insn_1): Likewise.
14461         * rtl.c (copy_rtx): Likewise.
14463 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14465         PR target/56591
14466         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14467         output_operand_lossage message.
14469 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14471         PR target/56470
14472         * arm.c (shift_op): Validate RTL pattern on the fly.
14473         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14474         the RTL.
14476 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14478         PR target/56347
14479         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14480         new call patterns that clobber %fr12.
14481         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14482         split and postreload patterns.
14483         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14484         registers %fr12 and %fr12R as call used.
14486 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14488         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14489         (canon_address, record_store, replace_read, check_mem_read_rtx,
14490         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14491         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14492         rest_of_handle_dse): Likewise.
14494 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14496         PR middle-end/56524
14497         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14498         Add base_optabs.
14499         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14500         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14501         (save_optabs_if_changed): Replace with...
14502         (init_tree_optimization_optabs): ...this.
14503         * optabs.c (save_optabs_if_changed): Rename to...
14504         (init_tree_optimization_optabs): ...this.  Take the optimization node
14505         as argument.  Do nothing if the base optabs are already correct.
14506         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14507         to recompute optabs.
14508         * function.h (function): Remove optabs field.
14509         * function.c (invoke_set_current_function_hook): Call
14510         init_tree_optimization_optabs.  Use the result to initialize
14511         this_fn_optabs.
14513 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14515         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14516         if GTMA_HAS_NO_INSTRUMENTATION.
14517         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14518         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14519         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14520         * gimple-pretty-print.c (dump_gimple_transaction): Handle
14521         GTMA_HAS_NO_INSTRUMENTATION.
14523 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
14525         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14526         libasan_preinit.o.
14528 2013-03-08  Marek Polacek  <polacek@redhat.com>
14529             Jakub Jelinek  <jakub@redhat.com>
14531         PR tree-optimization/56478
14532         * predict.c (is_comparison_with_loop_invariant_p): Change the
14533         type of loop_step to tree.
14534         (predict_loops): Adjust.
14535         (predict_iv_comparison): Perform the computations on double_ints.
14537 2013-03-08  Richard Biener  <rguenther@suse.de>
14539         PR tree-optimization/56570
14540         * tree-cfg.c (verify_expr_location_1): Verify locations for
14541         DECL_DEBUG_EXPR.
14542         * tree-sra.c (create_access_replacement): Strip locations
14543         from DECL_DEBUG_EXPRs.
14545 2013-03-08  Richard Biener  <rguenther@suse.de>
14547         * tree-inline.c (expand_call_inline): Do not associate
14548         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14549         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14551 2013-03-08  Richard Biener  <rguenther@suse.de>
14553         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14554         or block changes with -Og.  Fix for location / block encoding
14555         changes and PHI arguments with locations.
14557 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
14559         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14560         for all counters.
14561         (struct output_info): Likewise.
14562         (register_overhead): Remove bad gcc_assert.
14563         (bitmap_find_bit): If there is only a single bitmap element, do not
14564         count a miss as a search.
14565         (print_statistics): Update for counter type changes.
14566         (dump_bitmap_statistics): Likewise.  Print headers such that they
14567         are properly lined up with the printed counters.
14569 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
14571         PR tree-optimization/56559
14572         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14573         check that it has only a single use.
14575 2013-03-07  Richard Biener  <rguenther@suse.de>
14577         * doc/invoke.texi (fwhole-program): Discourage use in combination
14578         with -flto.
14580 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
14582         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14584         PR tree-optimization/56539
14585         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14586         instead of GSI_CONTINUE_LINKING as last argument to
14587         force_gimple_operand_gsi.  Adjust function comment.
14589         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14590         aarch64-cores.def.
14592         PR middle-end/56548
14593         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14594         promoted mode, convert the result back to the original mode.
14596 2013-03-06  Richard Biener  <rguenther@suse.de>
14598         PR middle-end/56294
14599         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14600         (insert_updated_phi_nodes_compare_uids): New function.
14601         (update_ssa): Sort symbols_to_rename after UID before
14602         traversing it to insert PHI nodes.
14604 2013-03-06  Richard Biener  <rguenther@suse.de>
14606         PR middle-end/50494
14607         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14608         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14610         Revert
14611         2013-02-13  Richard Biener  <rguenther@suse.de>
14613         PR lto/50494
14614         * varasm.c (output_constant_def_1): Get the decl representing
14615         the constant as argument.
14616         (output_constant_def): Wrap output_constant_def_1.
14617         (make_decl_rtl): Use output_constant_def_1 with the decl
14618         representing the constant.
14619         (build_constant_desc): Optionally re-use a decl already
14620         representing the constant.
14621         (tree_output_constant_def): Adjust.
14623 2013-03-06  Joey Ye  <joey.ye@arm.com>
14625         PR lto/50293
14626         * gcc.c (convert_white_space): New function.
14627         (main): Handles white space in function name.
14629 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
14631         PR target/56529
14632         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14633         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
14634         to SH_DIV_CALL_TABLE for TARGET_SH2.
14635         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14636         list.
14637         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14638         call-table options.
14640 2013-03-05  Sterling Augustine  <saugustine@google.com>
14641             Cary Coutant  <ccoutant@google.com>
14643         PR debug/55364
14644         * dwarf2out.c (resolve_addr): Don't call
14645         remove_loc_list_addr_table_entries a second time for the same
14646         expression.
14648 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14650         PR debug/56510
14651         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14652         (avoid_complex_debug_insns): New function.
14653         (expand_debug_locations): Call it.
14655         PR rtl-optimization/56484
14656         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14657         lifetimes of hard registers on small register class machines.
14659 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14661         * config/microblaze/microblaze-protos.h: Rename
14662         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14663         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14664         fast_interrupt.
14665         (microblaze_fast_interrupt_function_p): New function.
14666         (microblaze_is_interrupt_handler): Rename to
14667         microblaze_is_interrupt_variant and add fast_interrupt check.
14668         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14669         (save_restore_insns): Likewise.
14670         (compute_frame_size): Likewise.
14671         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14672         (microblaze_globalize_label): Likewise.
14673         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14674         * config/microblaze/microblaze.md: Use wrapper
14675         microblaze_is_interrupt_variant.
14677 2013-03-05  Kai Tietz  <ktietz@redhat.com>
14679         * sdbout.c (sdbout_one_type): Switch to current function's section
14680         supporting cold/hot.
14682 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14684         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14685         -mxl-reorder.
14687 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14689         PR middle-end/56461
14690         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14691         if VALGRIND_GET_VBITS is defined, temporarily make object
14692         memory all defined, and restore previous valgrind addressability
14693         and definability afterwards.  Free this_object at the end.
14695         PR middle-end/56461
14696         * lra.c (lra): Call lra_clear_live_ranges if live_p,
14697         right before calling lra_create_live_ranges, also call it
14698         when clearing live_p.  Only call lra_clear_live_ranges
14699         at the end if live_p.
14701         PR middle-end/56461
14702         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14704 2013-03-05  Richard Biener  <rguenther@suse.de>
14706         PR tree-optimization/56521
14707         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14708         value-id.
14710 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
14712         PR c++/55135
14713         * except.h (remove_unreachable_eh_regions): New prototype.
14714         * except.c (remove_eh_handler_splicer): New function, split out
14715         of remove_eh_handler.
14716         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
14717         warning about running it on many EH regions one at a time.
14718         (remove_unreachable_eh_regions_worker): New function, walk the
14719         EH tree in depth-first order and remove non-marked regions.
14720         (remove_unreachable_eh_regions): New function.
14721         * tree-eh.c (mark_reachable_handlers): New function, split out
14722         from remove_unreachable_handlers.
14723         (remove_unreachable_handlers): Use mark_reachable_handlers and
14724         remove_unreachable_eh_regions.
14725         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14726         and remove_unreachable_eh_regions.
14728 2013-03-05  Richard Biener  <rguenther@suse.de>
14730         PR middle-end/56525
14731         * loop-init.c (fix_loop_structure): Remove loops in two stages,
14732         not freeing them until the end.
14734 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14736         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14738 2013-03-05  Richard Biener  <rguenther@suse.de>
14740         PR tree-optimization/56270
14741         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14742         of loads after scheduling an SLP instance.
14744 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14746         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14747         tic6x.exp.
14748         (check_gcc_parallelize): Run guality.exp as a separate job from
14749         vect.exp with unsorted.exp and $(dg_target_exps) separately from
14750         struct-layout-1.exp with stackalign.exp.
14752         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14754         PR middle-end/56461
14755         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14756         load_index sbitmap even if some bit in it isn't set.
14758         PR middle-end/56461
14759         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14760         (discover_iteration_bound_by_body_walk): Change queues to
14761         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
14762         spelling in comment.  Call safe_push on queues[bound_index] directly.
14763         Release queues[queue_index] in every iteration unconditionally.
14764         Release bounds vector.
14766         PR middle-end/56461
14767         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14768         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14769         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14770         inner_phis vector.
14772 2013-03-05  Richard Biener  <rguenther@suse.de>
14774         PR lto/56515
14775         * tree-inline.c (remap_blocks_to_null): New function.
14776         (expand_call_inline): When expanding a call stmt without
14777         an associated block inline remap all callee blocks to NULL.
14779 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14781         PR rtl-optimization/56494
14782         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14783         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14784         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14786         PR middle-end/56461
14787         * sel-sched-ir.c (free_sched_pools): Release
14788         succs_info_pool.stack[succs_info_pool.max_top] vectors too
14789         if succs_info_pool.max_top isn't -1.
14791         PR bootstrap/56509
14792         * opts.c (opts_obstack, opts_concat): Moved to...
14793         * opts-common.c (opts_obstack, opts_concat): ... here.
14795 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14797         PR middle-end/56461
14798         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14800 2013-03-04  Martin Jambor  <mjambor@suse.cz>
14802         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14803         all appropriate places.
14805 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
14807         PR tree-optimization/56424
14808         * ipa-split.c (split_function): Do not set the RSO flag if result is
14809         not by reference and its type is a register type.
14811 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
14813         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14814         (microblaze_legitimate_pic_operand): Likewise
14815         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14816         new function microblaze_legitimate_pic_operand
14817         * config/microblaze/microblaze-protos.h
14818         (microblaze_legitimate_pic_operand): Declare.
14820 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14822         * config/microblaze/predicates.md (call_insn_simple_operand):
14823         New predicate for supported rtx code types.
14824         * config/microblaze/microblaze.md (call_internal1): Use
14825         call_insn_simple_operand predicate.
14827 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14829         PR middle-end/56461
14830         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14831         partitions.ordered_remove.
14833         PR middle-end/56461
14834         * tree-vect-stmts.c (vectorizable_conversion): Don't call
14835         vec_oprnds0.create (1) for modifier == NONE.
14837         PR middle-end/56461
14838         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14839         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14840         vec_oprnds1 right before pushing anything to it for
14841         scalar_shift_arg.
14843         PR middle-end/56461
14844         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14845         set nbbs to 0 instead of having separate code path.
14846         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14847         instead of false as last argument if returning NULL.
14849 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
14851         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14852         the attribute is now called "target" instead of "option".
14853         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14854         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
14855         attribute/pragma name for TARGET_OPTION_VALID_P and
14856         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
14857         * doc/tm.texi: Regenerated.
14859 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
14861         * config/microblaze/microblaze.c:
14862         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14863         * config/microblaze/microblaze.h: Add -mxl-reorder to
14864         DRIVER_SELF_SPECS.
14865         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14866         instructions emitted if TARGET_REORDER.
14867         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14868         or 0 for -m/-mno case, but initialises as 2 to detect default use case
14869         separately.
14871 2013-03-01  Xinliang David Li  <davidxl@google.com>
14873         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14874         walk length.
14876 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
14878         PR middle-end/56461
14879         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14880         vector even when returning true.  Fix up function comment formatting.
14882         PR middle-end/56461
14883         * ira-build.c (ira_loop_nodes_count): New variable.
14884         (create_loop_tree_nodes): Initialize it.
14885         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14887         PR middle-end/56461
14888         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14889         method on dr_chain and result_chain.
14890         * tree-vect-stmts.c (vectorizable_store): Only call
14891         result_chain.create if j == 0.
14893         PR middle-end/56461
14894         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14895         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14896         before overwriting it.
14898 2013-03-01  Tobias Burnus  <burnus@net-b.de>
14900         * doc/extended.texi (C Extensions): Change order in @menu
14901         to match @node.
14902         (Other MIPS Built-in Functions): Move last MIPS entry before
14903         "picoChip Built-in Functions".
14904         (SH Built-in Functions): Move after RX Built-in Functions.
14905         * doc/gcc.texi (Introduction): Change order in @menu
14906         to match @node.
14907         * doc/md.texi (Constraints): Ditto.
14908         * gty.texi (Type Information): Ditto.
14909         (User-provided marking routines for template types): Make
14910         subsection.
14911         * doc/invoke.texi (AArch64 Options): Move before
14912         "Adapteva Epiphany Options".
14914 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
14915             Jakub Jelinek  <jakub@redhat.com>
14917         PR sanitizer/56454
14918         * asan.c (gate_asan): Lookup no_sanitize_address instead of
14919         no_address_safety_analysis attribute.
14920         * doc/extend.texi (no_address_safety_attribute): Rename to
14921         no_sanitize_address attribute, mention no_address_safety_analysis
14922         attribute as deprecated alias.
14924 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
14926         PR middle-end/56461
14927         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14928         type to vec<vec<tree> > *.
14929         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
14930         to be vec<tree> instead of vec<tree> *, set vec_defs
14931         to vNULL and call vec_defs.create (number_of_vects), adjust other
14932         uses of vec_defs.
14933         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14934         vectorizable_condition): Adjust vect_get_slp_defs callers.
14936 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14938         * config/aarch64/aarch64.c
14939         (aarch64_float_const_representable): Remove unused variable.
14941 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14943         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14945 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14947         * config/aarch64/aarch64-builtins.c
14948         (aarch64_init_simd_builtins): Make static.
14950 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14952         * config/aarch64/aarch64.c
14953         (aarch64_simd_make_constant): Make static.
14955 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14957         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14958         with no initialization to the RHS of debug statements.
14960 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14962         PR tree-optimization/56294
14963         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14964         Adjust dumping.
14965         (get_access_replacement): Do not call create_access_replacement.
14966         Assert a replacement exists.
14967         (get_repl_default_def_ssa_name): Create the replacement declaration
14968         itself.
14970 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14972         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14973         final_end_function.
14975 2013-02-28  Marek Polacek  <polacek@redhat.com>
14977         PR rtl-optimization/56466
14978         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14979         if we're changing a loop.
14980         (peel_loops_completely): Likewise.
14982 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
14984         PR c++/55813
14985         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14987 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14989         PR target/56445
14990         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14991         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14992         INTX_FTYPE_FX, FX_FTYPE_INTX.
14993         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14995 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14997         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14998         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14999         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15000         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15001         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15002         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15003         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15004         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15005         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15006         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15007         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15008         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15009         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15010         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15011         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15012         (avrxmega6): Increase max flash segments from 5 to 6.
15013         * config/avr/t-multilib: Regenerate.
15014         * config/avr/avr-tables.opt: Regenerate.
15015         * doc/avr-mmcu.texi: Regenerate.
15017 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15019         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15020         (avr_device_to_arch): Rename to avr_device_to_ld.
15021         (avr_device_to_as): New prototype.
15022         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15023         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15024         * config/avr/driver-avr.c (avr_device_to_as): New.
15025         (avr_device_to_arch): Rename to avr_device_to_ld.
15027 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15029         PR middle-end/56461
15030         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15031         method on dr_chain and result_chain.
15033         PR middle-end/56461
15034         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15035         pointer_set_destroy on not_executed_last_iteration.
15037         PR middle-end/56461
15038         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15040         PR middle-end/56461
15041         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15042         FOR_EACH_DEFINED_FUNCTION when freeing state.
15044         PR middle-end/56461
15045         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15046         pool_free.
15047         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15048         overwriting it.
15050         PR middle-end/56461
15051         * ipa-cp.c (decide_whether_version_node): Call vec_free on
15052         known_aggs[i].items and release known_aggs vector.
15054         PR middle-end/56461
15055         * ipa-reference.c (propagate): Free node_info even for alias nodes.
15057 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15059         * config/microblaze/microblaze.c (microblaze_emit_compare):
15060         Use xor for EQ/NE comparisions.
15061         * config/microblaze/microblaze.md (cstoresf4): Add constraints
15062         (cbranchsf4): Adjust operator to comparison_operator.
15064 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15066         PR middle-end/56461
15067         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15068         vector.
15069         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15070         vec_safe_push, always update *slot.
15071         (redirect_edge_var_map_clear): Use vec_free.
15072         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15073         (free_var_map_entry): Use vec_free.
15074         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15075         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15077 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
15079         PR middle-end/45472
15080         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15081         when the may_trap_p bit of the exprs being merged differs.
15082         Reorder tests for speculativeness in the logical and operator.
15084 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15086         * incpath.c (add_standard_paths): Use reconcat instead of concat
15087         where appropriate and avoid leaking memory.
15089         * opts.h: Include obstack.h.
15090         (opts_concat): New prototype.
15091         (opts_obstack): New declaration.
15092         * opts.c (opts_concat): New function.
15093         (opts_obstack): New variable.
15094         (init_options_struct): Call gcc_init_obstack on opts_obstack.
15095         (finish_options): Use opts_concat instead of concat
15096         and XOBNEWVEC instead of XNEWVEC.
15097         * opts-common.c (generate_canonical_option, decode_cmdline_option,
15098         generate_option): Likewise.
15099         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15100         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15102         PR target/56455
15103         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15104         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15106 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15108         PR middle-end/56461
15109         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15111 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15113         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15114         (arm_block_move_unaligned_straight): Likewise.
15115         (arm_adjust_block_mem): Likewise.
15117 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15119         PR target/48901
15120         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15121         temp, cond and label.
15122         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15124         PR target/52500
15125         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15126         * config/c6x/c6x.h (dbx_register_map): Update declaration.
15128         PR target/52501
15129         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15130         of prologue/epilogue functions.
15132         PR target/52550
15133         * config/tilegx/tilegx.c (tilegx_expand_prologue):
15134         Remove unused variable cfa_offset.
15135         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15137         PR target/54639
15138         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15139         type promotion to unsigned.
15141         PR target/54640
15142         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15143         for HOST_WIDE_INT of 32 bit / same size as int.
15144         (arm_block_move_unaligned_straight): Likewise.
15145         (arm_adjust_block_mem): Likewise.
15147         PR target/54662
15148         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15149         ALL_CFLAGS.
15151 2013-02-26  Marek Polacek  <polacek@redhat.com>
15153         PR tree-optimization/56426
15154         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15156 2013-02-26  Richard Biener  <rguenther@suse.de>
15158         PR target/56444
15159         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15160         unused variable loops.
15162 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15164         PR tree-optimization/56448
15165         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15166         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15167         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15168         later operands of the references, or even first operand for
15169         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15171         PR tree-optimization/56443
15172         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15173         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15174         to type_for_mode langhook.
15176 2013-02-25  Matt Turner  <mattst88@gmail.com>
15178         * doc/invoke.texi: Document r4700.
15180 2013-02-25  Richard Biener  <rguenther@suse.de>
15182         PR tree-optimization/56175
15183         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15184         split out from ...
15185         (simplify_bitwise_binary): ... here.  Also guard the conversion
15186         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15188 2013-02-25  Catherine Moore  <clm@codesourcery.com>
15190         Revert:
15191         2013-02-24  Catherine Moore  <clm@codesourcery.com>
15192             Maciej W. Rozycki  <macro@codesourcery.com>
15193             Tom de Vries  <tom@codesourcery.com>
15194             Nathan Sidwell  <nathan@codesourcery.com>
15195             Iain Sandoe  <iain@codesourcery.com>
15196             Nathan Froyd  <froydnj@codesourcery.com>
15197             Chao-ying Fu  <fu@mips.com>
15199         * doc/extend.texi: (micromips, nomicromips, nocompression):
15200         Document new function attributes.
15201         * doc/invoke.texi (minterlink-compressed, mmicromips,
15202         m14k, m14ke, m14kec): Document new options.
15203         (minterlink-mips16): Update documentation.
15204         * doc/md.texi (ZC, ZD): Document new constraints.
15205         * configure.ac (gcc_cv_as_micromips): Check if linker
15206         supports the .set micromips directive.
15207         * configure: Regenerate.
15208         * config.in: Regenerate.
15209         * config/mips/mips-tables.opt: Regenerate.
15210         * config/mips/micromips.md: New file.
15211         * constraints.md (ZC, AD): New constraints.
15212         * config/mips/predicates.md (movep_src_register): New predicate.
15213         (movep_src_operand): New predicate.
15214         (non_volatile_mem_operand): New predicate.
15215         * config/mips/mips.md (multimem): New type.
15216         (length): Differentiate between 17-bit and 18-bit branch offsets.
15217         (MOVEP1, MOVEP2): New mode iterator.
15218         (mov_<load>l): Use ZC constraint.
15219         (mov_<load>r): Likewise.
15220         (mov_<store>l): Likewise.
15221         (mov_<store>r): Likewise.
15222         (*branch_equality<mode>_inverted): Add microMIPS support.
15223         (*branch_equality<mode>): Likewise.
15224         (*jump_absolute): Likewise.
15225         (indirect_jump_<mode>): Likewise.
15226         (tablejump_<mode>): Likewise.
15227         (<optab>_internal): Likewise.
15228         (sibcall_internal): Likewise.
15229         (sibcall_value_internal): Likewise.
15230         (prefetch): Use constraint ZD.
15231         * config/mips/mips.opt (minterlink-compressed): New option.
15232         (minterlink-mips16): Now an alias for minterlink-compressed.
15233         (mmicromips): New option.
15234         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15235         (compare_and_swap_12): Likewise.
15236         (sync_add<mode>): Likewise.
15237         (sync_<optab>_12): Likewise.
15238         (sync_old_<optab>_12): Likewise.
15239         (sync_new_<optab>_12): Likewise.
15240         (sync_nand_12): Likewise.
15241         (sync_old_nand_12): Likewise.
15242         (sync_new_nand_12): Likewise.
15243         (sync_sub<mode>): Likewise.
15244         (sync_old_add<mode>): Likewise.
15245         (sync_old_sub<mode>): Likewise.
15246         (sync_new_add<mode>): Likewise.
15247         (sync_new_sub<mode>): Likewise.
15248         (sync_<optab><mode>): Likewise.
15249         (sync_old_<optab><mode>): Likewise.
15250         (sync_new_<optab><mode>): Likewise.
15251         (sync_nand<mode>): Likewise.
15252         (sync_old_nand<mode>): Likewise.
15253         (sync_new_nand<mode>): Likewise.
15254         (sync_lock_test_and_set<mode>): Likewise.
15255         (test_and_set_12): Likewise.
15256         (atomic_compare_and_swap<mode>): Likewise.
15257         (atomic_exchange<mode>_llsc): Likewise.
15258         (atomic_fetch_add<mode>_llsc): Likewise.
15259         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15260         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15261         (umips_save_restore_pattern_p): Likewise.
15262         (umips_load_store_pair_p): Likewise.
15263         (umips_output_load_store_pair): Likewise.
15264         (umips_movep_target_p): Likewise.
15265         (umips_12bit_offset_address_p): Likewise.
15266         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15267         (mips_base_mips16): Rename this...
15268         (mips_base_compression_flags): ...to this. Update all uses.
15269         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15270         (mips_mips16_decl_p): Delete.
15271         (mips_nomips16_decl_p): Delete.
15272         (mips_get_compress_on_flags): New function.
15273         (mips_get_compress_off_flags): New function.
15274         (mips_get_compress_mode): New function.
15275         (mips_get_compress_on_name): New function.
15276         (mips_get_compress_off_name): New function.
15277         (mips_insert_attributes): Support multiple compression types.
15278         (mips_merge_decl_attributes): Likewise.
15279         (umips_12bit_offset_address_p): New function.
15280         (mips_start_function_definition): Emit .set micromips directive.
15281         (mips_call_may_need_jalx_p): New function.
15282         (mips_function_ok_for_sibcall): Add microMIPS support.
15283         (mips_print_operand_punctuation): Support short delay slots and
15284         compact jumps.
15285         (umips_swm_mask, umips_swm_encoding): New.
15286         (umips_build_save_restore): New function.
15287         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15288         (was_mips16_p): Remove.
15289         (old_compression_mode): New.
15290         (mips_set_compression_mode): New function.
15291         (mips_set_current_function): Add microMIPS support.
15292         (mips_option_override): Likewise.
15293         (umips_save_restore_pattern_p): New function.
15294         (umips_output_save_restore): New function.
15295         (umips_load_store_pair_p_1): New function.
15296         (umips_load_store_pair_p): New function.
15297         (umips_output_load_store_pair_1): New function.
15298         (umips_output_load_store_pair): New function.
15299         (umips_movep_target_p) New function.
15300         (mips_prepare_pch_save): Add microMIPS support.
15301         * config/mips/mips.h (TARGET_COMPRESSION): New.
15302         (TARGET_CPU_CPP_BUILTINS): Update macro
15303         to use new compression flags and to support microMIPS.
15304         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15305         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15306         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15307         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15308         (ASM_SPEC): Support mmicromips and mno-micromips.
15309         (M16STORE_REG_P): New macro.
15310         (MIPS_CALL): Support TARGET_MICROMIPS.
15311         (MICROMIPS_J): New macro.
15312         (mips_base_mips16): Rename this...
15313         (mips_base_compression_flags): ...to this.
15314         (UMIPS_12BIT_OFFSET_P): New macro.
15315         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15316         (MULTILIB_DIRNAMES): Likewise.
15318 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15320         PR rtl-optimization/56131
15321         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15322         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15323         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15325 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15327         * doc/invoke.texi (-fsanitize=): Move from optimization
15328         to debugging options.
15330 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15332         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15334 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15335             Alexander Monakov  <amonakov@ispras.ru>
15337         PR middle-end/56077
15338         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15339         flush pending lists also on non-jumps.  Adjust comment.
15341 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15342             Maciej W. Rozycki  <macro@codesourcery.com>
15343             Tom de Vries  <tom@codesourcery.com>
15344             Nathan Sidwell  <nathan@codesourcery.com>
15345             Iain Sandoe  <iain@codesourcery.com>
15346             Nathan Froyd  <froydnj@codesourcery.com>
15347             Chao-ying Fu  <fu@mips.com>
15349         * doc/extend.texi: (micromips, nomicromips, nocompression):
15350         Document new function attributes.
15351         * doc/invoke.texi (minterlink-compressed, mmicromips,
15352         m14k, m14ke, m14kec): Document new options.
15353         (minterlink-mips16): Update documentation.
15354         * doc/md.texi (ZC, ZD): Document new constraints.
15355         * configure.ac (gcc_cv_as_micromips): Check if linker
15356         supports the .set micromips directive.
15357         * configure: Regenerate.
15358         * config.in: Regenerate.
15359         * config/mips/mips-tables.opt: Regenerate.
15360         * config/mips/micromips.md: New file.
15361         * constraints.md (ZC, AD): New constraints.
15362         * config/mips/predicates.md (movep_src_register): New predicate.
15363         (movep_src_operand): New predicate.
15364         (non_volatile_mem_operand): New predicate.
15365         * config/mips/mips.md (multimem): New type.
15366         (length): Differentiate between 17-bit and 18-bit branch offsets.
15367         (MOVEP1, MOVEP2): New mode iterator.
15368         (mov_<load>l): Use ZC constraint.
15369         (mov_<load>r): Likewise.
15370         (mov_<store>l): Likewise.
15371         (mov_<store>r): Likewise.
15372         (*branch_equality<mode>_inverted): Add microMIPS support.
15373         (*branch_equality<mode>): Likewise.
15374         (*jump_absolute): Likewise.
15375         (indirect_jump_<mode>): Likewise.
15376         (tablejump_<mode>): Likewise.
15377         (<optab>_internal): Likewise.
15378         (sibcall_internal): Likewise.
15379         (sibcall_value_internal): Likewise.
15380         (prefetch): Use constraint ZD.
15381         * config/mips/mips.opt (minterlink-compressed): New option.
15382         (minterlink-mips16): Now an alias for minterlink-compressed.
15383         (mmicromips): New option.
15384         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15385         (compare_and_swap_12): Likewise.
15386         (sync_add<mode>): Likewise.
15387         (sync_<optab>_12): Likewise.
15388         (sync_old_<optab>_12): Likewise.
15389         (sync_new_<optab>_12): Likewise.
15390         (sync_nand_12): Likewise.
15391         (sync_old_nand_12): Likewise.
15392         (sync_new_nand_12): Likewise.
15393         (sync_sub<mode>): Likewise.
15394         (sync_old_add<mode>): Likewise.
15395         (sync_old_sub<mode>): Likewise.
15396         (sync_new_add<mode>): Likewise.
15397         (sync_new_sub<mode>): Likewise.
15398         (sync_<optab><mode>): Likewise.
15399         (sync_old_<optab><mode>): Likewise.
15400         (sync_new_<optab><mode>): Likewise.
15401         (sync_nand<mode>): Likewise.
15402         (sync_old_nand<mode>): Likewise.
15403         (sync_new_nand<mode>): Likewise.
15404         (sync_lock_test_and_set<mode>): Likewise.
15405         (test_and_set_12): Likewise.
15406         (atomic_compare_and_swap<mode>): Likewise.
15407         (atomic_exchange<mode>_llsc): Likewise.
15408         (atomic_fetch_add<mode>_llsc): Likewise.
15409         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15410         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15411         (umips_save_restore_pattern_p): Likewise.
15412         (umips_load_store_pair_p): Likewise.
15413         (umips_output_load_store_pair): Likewise.
15414         (umips_movep_target_p): Likewise.
15415         (umips_12bit_offset_address_p): Likewise.
15416         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15417         (mips_base_mips16): Rename this...
15418         (mips_base_compression_flags): ...to this. Update all uses.
15419         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15420         (mips_mips16_decl_p): Delete.
15421         (mips_nomips16_decl_p): Delete.
15422         (mips_get_compress_on_flags): New function.
15423         (mips_get_compress_off_flags): New function.
15424         (mips_get_compress_mode): New function.
15425         (mips_get_compress_on_name): New function.
15426         (mips_get_compress_off_name): New function.
15427         (mips_insert_attributes): Support multiple compression types.
15428         (mips_merge_decl_attributes): Likewise.
15429         (umips_12bit_offset_address_p): New function.
15430         (mips_start_function_definition): Emit .set micromips directive.
15431         (mips_call_may_need_jalx_p): New function.
15432         (mips_function_ok_for_sibcall): Add microMIPS support.
15433         (mips_print_operand_punctuation): Support short delay slots and
15434         compact jumps.
15435         (umips_swm_mask, umips_swm_encoding): New.
15436         (umips_build_save_restore): New function.
15437         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15438         (was_mips16_p): Remove.
15439         (old_compression_mode): New.
15440         (mips_set_compression_mode): New function.
15441         (mips_set_current_function): Add microMIPS support.
15442         (mips_option_override): Likewise.
15443         (umips_save_restore_pattern_p): New function.
15444         (umips_output_save_restore): New function.
15445         (umips_load_store_pair_p_1): New function.
15446         (umips_load_store_pair_p): New function.
15447         (umips_output_load_store_pair_1): New function.
15448         (umips_output_load_store_pair): New function.
15449         (umips_movep_target_p) New function.
15450         (mips_prepare_pch_save): Add microMIPS support.
15451         * config/mips/mips.h (TARGET_COMPRESSION): New.
15452         (TARGET_CPU_CPP_BUILTINS): Update macro
15453         to use new compression flags and to support microMIPS.
15454         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15455         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15456         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15457         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15458         (ASM_SPEC): Support mmicromips and mno-micromips.
15459         (M16STORE_REG_P): New macro.
15460         (MIPS_CALL): Support TARGET_MICROMIPS.
15461         (MICROMIPS_J): New macro.
15462         (mips_base_mips16): Rename this...
15463         (mips_base_compression_flags): ...to this.
15464         (UMIPS_12BIT_OFFSET_P): New macro.
15465         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15466         (MULTILIB_DIRNAMES): Likewise.
15468 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15470         PR target/52555
15471         * target-globals.c (save_target_globals): For init_reg_sets and
15472         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15474 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15476         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15477         * config/aarch64/t-aarch64
15478         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15480 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15482         PR inline-asm/56148
15483         * lra-constraints.c (process_alt_operands): Reload operand
15484         conflicting with earlier clobber only if no more other conflicting
15485         operands.
15487 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15489         PR sanitizer/56393
15490         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15491         if not linking a shared library.
15493 2013-02-22  Seth LaForge  <sethml@google.com>
15495         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15497 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15499         * config/arm/arm.md (split for extendsidi): Update condition.
15500         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15501         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15502         (qhs_zextenddi_cstr): Likewise.
15504 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15506         PR middle-end/56420
15507         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15508         avoid signed wrapping.
15509         (expand_mult): Handle properly multiplication by
15510         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15511         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15512         in the compiler if coeff is HOST_WIDE_INT_MIN.
15513         (expand_divmod): Don't make ext_op1 static, change it's type to
15514         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15516         PR rtl-optimization/50339
15517         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15518         field.
15519         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15520         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15521         into splitting_ashiftrt field.
15522         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15523         ASHIFTRT.
15524         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15525         choices.
15527 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
15529         PR middle-end/56108
15530         * trans-mem.c (execute_tm_mark): Do not expand transactions that
15531         are sure to go irrevocable.
15533 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
15535         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15536         scalars are valid operands.
15538 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15540         PR tree-optimization/56310
15541         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15542         only matching indices and non-negative final offsets.
15543         (intersect_aggregates_with_edge): Pass src_idx to
15544         agg_replacements_to_vector.  Pass src_idx insstead of index to
15545         intersect_with_agg_replacements.
15547 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15549         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15550         instead of hard-wired defaults.
15552 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
15554         * doc/invoke.texi (MIPS Options): Update documentation of the
15555         floating-point multiply-accumulate instruction restrictions.
15557 2013-02-21  Kostya Serebryany  <kcc@google.com>
15559         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15560         asan_shadow_offset on x86_64 linux.
15562 2013-02-21  Richard Biener  <rguenther@suse.de>
15564         PR tree-optimization/56415
15565         Revert
15566         2013-02-11  Richard Biener  <rguenther@suse.de>
15568         PR tree-optimization/56273
15569         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15570         first VRP run.
15572 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15574         PR bootstrap/56258
15575         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15576         instead of @itemx.
15578         PR inline-asm/56405
15579         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15580         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15582 2013-02-20  Jan Hubicka  <jh@suse.cz>
15584         PR tree-optimization/56265
15585         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15586         when target is referenced for first time.
15588 2013-02-20  Richard Biener  <rguenther@suse.de>
15590         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15591         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15592         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15593         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15594         not return anything.
15595         (rename_ssa_copies): Do not remove unused locals.
15596         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15597         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15598         * passes.c (execute_function_todo): Do not schedule unused locals
15599         removal if cleanup_tree_cfg did something.
15600         * tree-ssa-live.c (remove_unused_locals): Dump statistics
15601         about the number of removed locals.
15603 2013-02-20  Richard Biener  <rguenther@suse.de>
15605         PR tree-optimization/56398
15606         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15608 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15610         PR tree-optimization/55334
15611         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15612         restricted pointers to arrays.
15614 2013-02-20  Richard Biener  <rguenther@suse.de>
15615             Jakub Jelinek  <jakub@redhat.com>
15617         PR tree-optimization/56396
15618         * tree-ssa-ccp.c (n_const_val): New static variable.
15619         (get_value): Return NULL for SSA names we don't have a lattice
15620         entry for.
15621         (ccp_initialize): Initialize n_const_val.
15622         * tree-ssa-copy.c (n_copy_of): New static variable.
15623         (init_copy_prop): Initialize n_copy_of.
15624         (get_value): Return NULL_TREE for SSA names we don't have a
15625         lattice entry for.
15627 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15629         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15631 2013-02-20  Richard Biener  <rguenther@suse.de>
15633         * genpreds.c (write_lookup_constraint): Do not compare first
15634         letter of the constraint again.
15636 2013-02-20  Richard Biener  <rguenther@suse.de>
15638         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15639         and ceil_log2.
15640         (get_use_iv_cost): Terminate hashtable walk when coming across
15641         an empty entry.
15643 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
15645         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15646         reassociation for avx2 targets.
15648 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15650         * config/microblaze/microblaze.c: microblaze_has_clz = 0
15651         Add version check for v8.10.a to enable microblaze_has_clz
15652         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15653         version and TARGET_PATTERN_COMPARE check
15654         * config/microblaze/microblaze.md: New clzsi2 instruction
15656 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15658         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15659         function before branching.
15661 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15663         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15664         DUMP_INSN_RTX_UID.
15665         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15667 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15669         PR middle-end/55889
15670         * sel-sched.c: Include ira.h.
15671         (implicit_clobber_conflict_p): New function.
15672         (moveup_expr): Use it.
15673         * Makefile.in (sel-sched.o): Depend on ira.h.
15675 2013-02-19  Richard Biener  <rguenther@suse.de>
15677         PR tree-optimization/56384
15678         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15679         (vn_hash_type): Split out from ...
15680         (vn_hash_constant_with_type): ... here.
15681         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15682         (vn_phi_eq): Compare types from vn_phi_s structure.
15683         (vn_phi_lookup): Populate vn_phi_s type.
15684         (vn_phi_insert): Likewise.
15686 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
15688         PR tree-optimization/56350
15689         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15690         if haven't found reduction or nested cycle operand, rather than
15691         asserting we must find it.
15693         PR tree-optimization/56381
15694         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15695         to fold_build3.
15697 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
15698             Jakub Jelinek  <jakub@redhat.com>
15700         PR target/52555
15701         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15702         (swap_optab_enable): Same.
15703         (init_all_optabs): Use argument instead of global.
15704         * tree.h (struct tree_optimization_option): New field target_optabs.
15705         * expr.h (init_all_optabs): Add argument to prototype.
15706         (TREE_OPTIMIZATION_OPTABS): New.
15707         (save_optabs_if_changed): Protoize.
15708         * optabs.h: Declare this_fn_optabs.
15709         * optabs.c (save_optabs_if_changed): New.
15710         Declare this_fn_optabs.
15711         (init_optabs): Add argument to init_all_optabs() call.
15712         * function.c (invoke_set_current_function_hook): Handle per
15713         function optabs.
15714         * function.h (struct function): New field optabs.
15715         * config/mips/mips.c (mips_set_mips16_mode): Handle when
15716         optimization_current_node has changed.
15717         * target-globals.h (save_target_globals_default_opts): Protoize.
15718         * target-globals.c (save_target_globals_default_opts): New.
15720 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15722         PR target/56347
15723         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15724         registers %fr12 and %fr12R as call used.
15726         PR target/56214
15727         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15728         and HImode, require all displacements to be an integer multiple of
15729         their mode size.
15730         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15731         only allow QImode and HImode when reload is in progress and strict is
15732         true.  Likewise for symbolic addresses.  Use base14_operand to check
15733         displacements in REG+BASE addresses.
15735 2013-02-18  Richard Biener  <rguenther@suse.de>
15737         PR tree-optimization/56366
15738         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15739         handle sign-conversion of outer-loop initial induction value.
15741 2013-02-18  Richard Biener  <rguenther@suse.de>
15743         PR middle-end/56349
15744         * cfghooks.c (merge_blocks): If we merge a latch into another
15745         block adjust references to it.
15746         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15747         (verify_loop_structure): Verify that a recorded latch is in fact
15748         a latch.
15750 2013-02-18  Richard Biener  <rguenther@suse.de>
15752         PR tree-optimization/56321
15753         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15754         order SSA name release and virtual operand unlinking.
15756 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15758         * config/microblaze/microblaze.md (save_stack_block): Define.
15759         (restore_stack_block): Likewise.
15761 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15763         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15764         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15765         * config/microblaze/microblaze.c (microblaze_option_override):
15766         Bail out early for PIC modes when target does not support PIC.
15768 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15770         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15771         Replace with a microblaze version.
15772         (microblaze_trampoline_init): Adapt for microblaze.
15773         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15774         microblaze.
15776 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
15777             Dodji Seketeli  <dodji@redhat.com>
15779         PR asan/56330
15780         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15781         (instrument_mem_region_access): Do not forget to always put
15782         instrumentation of the of 'base' and 'base + len' in a "if (len !=
15783         0) statement, even for cases where either 'base' or 'base + len'
15784         are not instrumented -- because they have been previously
15785         instrumented.  Simplify the logic by putting all the statements
15786         instrument 'base + len' inside a sequence, and then insert that
15787         sequence right before the current insertion point.  Then, to
15788         instrument 'base + len', just get an iterator on that statement.
15789         And do not forget to update the pointer to iterator the function
15790         received as argument.
15792 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
15794         PR rtl-optimization/56348
15795         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15797 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
15799         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15800         (clean_graph_dump_file): Pass base to start_graph_dump.
15802 2013-02-14  Richard Henderson  <rth@redhat.com>
15804         PR target/55941
15805         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15807 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
15809         * collect2-aix.h: Define F_LOADONLY.
15811 2013-02-14  Richard Biener  <rguenther@suse.de>
15813         PR lto/50494
15814         * varasm.c (output_constant_def_1): Get the decl representing
15815         the constant as argument.
15816         (output_constant_def): Wrap output_constant_def_1.
15817         (make_decl_rtl): Use output_constant_def_1 with the decl
15818         representing the constant.
15819         (build_constant_desc): Optionally re-use a decl already
15820         representing the constant.
15821         (tree_output_constant_def): Adjust.
15823 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
15825         Fix an asan crash
15826         * asan.c (instrument_builtin_call):  Really put the length of the
15827         second source argument into src1_len.
15829 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15831         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15832         argument.  If it is false, don't create edge from then_bb to
15833         fallthru_bb.
15834         (insert_if_then_before_iter): Pass true to it.
15835         (build_check_stmt): Pass false to it.
15836         (transform_statements): Flush hash table only on extended basic
15837         block boundaries, rather than at the beginning of every bb.
15838         Don't flush hash table on nonfreeing_call_p calls.
15839         * tree-flow.h (nonfreeing_call_p): New prototype.
15840         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15842 2013-02-13  David S. Miller  <davem@davemloft.net>
15844         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15846 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
15848         PR target/56184
15849         * ira.c (max_regno_before_ira): Move from ...
15850         (ira): ... here.
15851         (fix_reg_equiv_init): Use max_regno_before_ira instead of
15852         vec_safe_length.
15854 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15856         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15858 2013-02-13  Richard Biener  <rguenther@suse.de>
15860         PR lto/56295
15861         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15862         globals in MEM_REFs.
15864 2013-02-13  Richard Biener  <rguenther@suse.de>
15866         * loop-init.c (loop_optimizer_init): Clear loop state when
15867         re-initializing preserved loops.
15868         * loop-unswitch.c (unswitch_single_loop): Return whether
15869         we unswitched the loop.  Do not verify loop state here.
15870         (unswitch_loops): When we unswitched a loop discover new loops.
15872 2013-02-13  Kostya Serebryany  <kcc@google.com>
15874         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15875         on x86_64 linux.
15876         * sanitizer.def: Rename __asan_init to __asan_init_v1.
15878 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
15880         Avoid instrumenting duplicated memory access in the same basic block
15881         * Makefile.in (asan.o): Add new dependency on hash-table.h
15882         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15883         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15884         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15885         (free_mem_ref_resources, has_mem_ref_been_instrumented)
15886         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15887         (get_mem_ref_of_assignment): New functions.
15888         (get_mem_refs_of_builtin_call): Extract from
15889         instrument_builtin_call and tweak a little bit to make it fit with
15890         the new signature.
15891         (instrument_builtin_call): Use the new
15892         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
15893         of is_gimple_builtin_call.
15894         (instrument_derefs, instrument_mem_region_access): Insert the
15895         instrumented memory reference into the hash table.
15896         (maybe_instrument_assignment): Renamed instrument_assignment into
15897         this, and change it to advance the iterator when instrumentation
15898         actually happened and return true in that case.  This makes it
15899         homogeneous with maybe_instrument_assignment, and thus give a
15900         chance to callers to be more 'regular'.
15901         (transform_statements): Clear the memory reference hash table
15902         whenever we enter a new BB, when we cross a function call, or when
15903         we are done transforming statements.  Use
15904         maybe_instrument_assignment instead of instrumentation.  No more
15905         need to special case maybe_instrument_assignment and advance the
15906         iterator after calling it; it's now handled just like
15907         maybe_instrument_call.  Update comment.
15909 2013-02-13  Richard Biener  <rguenther@suse.de>
15911         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15912         Fix loop discovery code.
15914 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
15916         PR inline-asm/56148
15917         * lra-constraints.c (process_alt_operands): Match early clobber
15918         operand with itself.  Check conflicts with earlyclobber only if
15919         the operand is not reloaded.  Prefer to reload conflicting operand
15920         if earlyclobber and matching operands are the same.
15922 2013-02-12  Richard Biener  <rguenther@suse.de>
15924         PR lto/56297
15925         * lto-streamer-out.c (write_symbol): Do not output symbols
15926         for hard register variables.
15928 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
15930         PR target/54222
15931         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15932         (umulsidi3_insn, mulsidi3_insn): New insns.
15934 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
15936         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15937         (struct tune_params): Add vec_costs field.
15938         * config/arm/arm.c (arm_builtin_vectorization_cost)
15939         (arm_add_stmt_cost): New functions.
15940         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15941         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15942         (arm_default_vec_cost): New struct of type cpu_vec_costs.
15943         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15944         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15945         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15946         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15948 2013-02-12  Richard Biener  <rguenther@suse.de>
15950         PR lto/56295
15951         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15952         decls again if possible.
15954 2013-02-12  Richard Biener  <rguenther@suse.de>
15956         PR middle-end/56288
15957         * tree-ssa.c (verify_ssa_name): Fix check, move
15958         SSA_NAME_IN_FREE_LIST check up.
15960 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
15961             Steven Bosscher   <steven@gcc.gnu.org>
15963         PR rtl-optimization/56151
15964         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15965         equal to op0 or op1, and last_insn pattern is CODE operation
15966         with MEM dest and one of the operands matches that MEM.
15968 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
15970         * doc/extend.texi: Document Function Multiversioning and "default"
15971         parameter string to target attribute.
15972         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15973         target attribute parameter is "default".
15974         (ix86_compare_version_priority): Remove checks for target attribute.
15975         (ix86_mangle_function_version_assembler_name): Change error to sorry.
15976         Remove check for target attribute equal to NULL. Add assert.
15977         (ix86_generate_version_dispatcher_body): Change error to sorry.
15979 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
15980             Jack Howarth  <howarth@bromo.med.uc.edu>
15981             Patrick Marlier  <patrick.marlier@gmail.com>
15983         PR libitm/55693
15984         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15985         define ENDFILE_SPEC as TM_DESTRUCTOR.
15986         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15988 2013-02-11  Alexander Potapenko  <glider@google.com>
15989             Jack Howarth  <howarth@bromo.med.uc.edu>
15990             Jakub Jelinek  <jakub@redhat.com>
15992         PR sanitizer/55617
15993         * config/darwin.c (cdtor_record): Rename ctor_record.
15994         (sort_cdtor_records): Rename sort_ctor_records.
15995         (finalize_dtors): New routine to sort destructors by
15996         priority before use in assemble_integer.
15997         (machopic_asm_out_destructor): Use finalize_dtors if needed.
15999 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16001         PR rtl-optimization/56275
16002         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16003         offset is non-negative and less than cmode size before
16004         calling simplify_subreg.
16006 2013-02-11  Richard Biener  <rguenther@suse.de>
16008         PR tree-optimization/56264
16009         * cfgloop.h (fix_loop_structure): Adjust prototype.
16010         * loop-init.c (fix_loop_structure): Return the number of
16011         newly discovered loops.
16012         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16013         are discovered, do a full loop-closed SSA rewrite.
16015 2013-02-11  Richard Biener  <rguenther@suse.de>
16017         PR tree-optimization/56273
16018         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16019         first VRP run.
16020         (check_array_ref): Fix missing newline in dumps.
16021         (search_for_addr_array): Likewise.
16023 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
16025         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16027 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
16029         PR target/56256
16030         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16032 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
16034         PR rtl-optimization/56246
16035         * lra-constraints.c (simplify_operand_subreg): Try to reuse
16036         reload pseudo.
16037         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16038         constraints are satisfied.
16040 2013-02-08  Jeff Law  <law@redhat.com>
16042         PR debug/53948
16043         * emit-rtl.c (reg_is_parm_p): New function.
16044         * regs.h (reg_is_parm_p): New prototype.
16045         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16046         callee-clobbered registers.
16048 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
16050         PR target/56043
16051         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16052         If there is no implicit builtin declaration, just return NULL.
16054 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16056         * config/i386/sse.md (FMAMODEM): New mode iterator.
16057         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16058         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16060 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16062         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16063         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16064         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16066 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16068         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16069         (microblaze*-*-elf): Likewise.
16070         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16071         LINK_SPEC.
16072         * config/microblaze/microblaze-c.c: Add builtin defines for
16073         _LITTLE_ENDIAN and _BIG_ENDIAN.
16074         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16075         add to TARGET_DEFAULT flags.
16076         Expand ASM_SPEC and LINK_SPEC.
16077         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16078         * config/microblaze/microblaze.md: Update extendsidi2 and
16079         movdi_internal instructions to use low-order / high-order reg
16080         print_operands.
16081         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16082         options and inversemask / mask of LITTLE_ENDIAN.
16083         * config/microblaze/t-microblaze: Expand multilib options to
16084         include mlittle-endian (le) and update exceptions patterns.
16086 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
16088         PR rtl-optimization/56195
16089         * lra-constraints.c (get_reload_reg): Don't reuse regs
16090         if they have smaller mode than requested, if they have
16091         wider mode than requested, try to return a SUBREG.
16093         PR tree-optimization/56250
16094         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16095         if type is unsigned and code isn't MULT_EXPR.
16097 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16099         PR tree-optimization/56064
16100         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16101         bits according to mode.
16102         * fixed-value.h (fixed_from_double_int)
16103         (const_fixed_from_double_int): Adjust comments.
16105 2013-02-08  Richard Biener  <rguenther@suse.de>
16107         PR lto/56231
16108         * lto-streamer.h (struct data_in): Remove current_file, current_line
16109         and current_col members.
16110         * lto-streamer-out.c (lto_output_location): Stream changed bits
16111         en-block for efficiency.
16112         * lto-streamer-in.c (clear_line_info): Remove.
16113         (lto_input_location): Cache current file, line and column
16114         globally via local statics.  Read changed bits en-block.
16115         (input_function): Do not call clear_line_info.
16116         (lto_read_body): Likewise.
16117         (lto_input_toplevel_asms): Likewise.
16119 2013-02-08  Michael Matz  <matz@suse.de>
16121         PR tree-optimization/52448
16122         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16123         (nt_call_phase): New static.
16124         (add_or_mark_expr): Only mark accesses with newer phase than any
16125         call seen.
16126         (nonfreeing_call_p): New.
16127         (nt_init_block): Update nt_call_phase, mark blocks as visited.
16128         (nt_fini_block): Keep blocks marked as visited.
16129         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16131 2013-02-08  Richard Biener  <rguenther@suse.de>
16133         * ira.c (ira): Free broken dominator information.
16135 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16137         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16139 2013-02-08  Marek Polacek  <polacek@redhat.com>
16141         * cfgloop.c (verify_loop_structure): Add more checking of headers.
16143 2013-02-08  Richard Biener  <rguenther@suse.de>
16145         PR middle-end/56181
16146         * cfgloop.h (flow_loops_find): Adjust.
16147         (bb_loop_header_p): Declare.
16148         * cfgloop.c (bb_loop_header_p): New function split out from ...
16149         (flow_loops_find): ... here.  Adjust function signature,
16150         support incremental loop structure update.
16151         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
16152         * cfgloopmanip.c (fix_loop_structure): Move ...
16153         * loop-init.c (fix_loop_structure): ... here.
16154         (apply_loop_flags): Split out from ...
16155         (loop_optimizer_init): ... here.
16156         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
16157         in incremental mode, only remove dead loops here.
16159 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16161         PR target/54222
16162         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16163         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16164         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16165         (*round<mode>3.libgcc): New insns for fixed-modes.
16166         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16167         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16168         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16169         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16170         implementations.  Define to __builtin_avr_absFX,
16171         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16172         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16173         __builtin_avr_countlsFX, respectively.
16174         * config/avr/avr-c.c (target.h): Include it.
16175         (enum avr_builtin_id): New enum.
16176         (avr_resolve_overloaded_builtin): New static function.
16177         (avr_register_target_pragmas): Use it to set
16178         targetm.resolve_overloaded_builtin.
16179         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16180         tree nodes used by DEF_BUILTIN.
16181         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16182         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16183         <AVR_BUILTIN_xxBITS>: Same.
16185 2013-02-08  Richard Biener  <rguenther@suse.de>
16187         * cfgloop.c (verify_loop_structure): Properly handle
16188         a loop exiting to another loop header.
16189         * ira-int.h (ira_loops): Remove.
16190         * ira.c (ira_loops): Remove.
16191         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16192         (do_reload): Use loop_optimizer_finalize.
16193         * ira-build.c (create_loop_tree_nodes): Use get_loops and
16194         number_of_loops to access the loop tree.
16195         (more_one_region_p): Likewise.
16196         (finish_loop_tree_nodes): Likewise.
16197         (rebuild_regno_allocno_maps): Likewise.
16198         (mark_loops_for_removal): Likewise.
16199         (mark_all_loops_for_removal): Likewise.
16200         (remove_unnecessary_regions): Likewise.
16201         (ira_build): Likewise.
16202         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16204 2013-02-08  Richard Biener  <rguenther@suse.de>
16206         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16207         * ipa-pure-const.c (analyze_function): Avoid calling
16208         mark_irreducible_loops twice.
16209         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16211 2013-02-07  David S. Miller  <davem@davemloft.net>
16213         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16214         on 'reg'.
16215         * var-tracking.c (vt_add_function_parameter): Test the presence of
16216         HAVE_window_save properly and do not remap argument registers when
16217         we have a leaf function.
16219 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
16221         PR bootstrap/56227
16222         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16223         instead of "ll".
16224         * config/i386/i386.c (ix86_print_operand): Ditto.
16226 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16228         * lra-constraints.c (process_alt_operands): Fix recently added comment.
16230 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16232         PR rtl-optimization/56225
16233         * lra-constraints.c (process_alt_operands): Check that reload hard
16234         reg can hold value for strict_low_part.
16236 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16238         PR debug/56154
16239         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16240         dwarf2out_end_function.
16241         (in_first_function_p, maybe_at_text_label_p,
16242         first_loclabel_num_not_at_text_label): New variables.
16243         (dwarf2out_var_location): In the first function find out
16244         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16245         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16246         functions.
16248 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
16250         PR rtl-optimization/56178
16251         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16252         SUBREG of a register.  Tidy up related block of code.
16253         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16254         note if the source is a register or a SUBREG of a register.
16256 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16258         PR target/56228
16259         * config/rs6000/rs6000.md (ptrm): New mode attr.
16260         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16261         call_value_indirect_aix<pttrsize>,
16262         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16263         m in constraints.
16265 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16267         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16268         if -bnortl. Convert to strcmp and strncmp.
16270 2013-02-07  Alan Modra  <amodra@gmail.com>
16272         PR target/54009
16273         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16274         addresses won't wrap when offsetting.
16275         (rs6000_secondary_reload): Provide secondary reloads needed for
16276         wrapping LO_SUM addresses.
16278 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16280         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16281         MACH, just __MACH__.
16283 2013-02-06  Richard Biener  <rguenther@suse.de>
16285         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16286         instead of calling fix_loop_structure.
16288 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16290         PR middle-end/56217
16291         * omp-low.c (use_pointer_for_field): Return false if
16292         lower_send_shared_vars doesn't generate any copy-out code.
16294 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16296         PR rtl-optimization/56131
16297         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16298         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16299         of the label is NULL.  Add comment.
16301 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16303         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16305         PR sanitizer/55374
16306         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16307         (STATIC_LIBTSAN_LIBS): Likewise.
16308         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16309         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16310         is defined, don't add anything else beyond that.
16311         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16312         (LINK_COMMAND_SPEC): Use them.
16314         PR tree-optimization/56205
16315         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16316         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16317         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16319 2013-02-05  Richard Biener  <rguenther@suse.de>
16321         PR tree-optimization/53342
16322         PR tree-optimization/53185
16323         * tree-vectorizer.h (vect_check_strided_load): Remove.
16324         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16325         not disallow peeling for vectorized strided loads.
16326         (vect_check_strided_load): Make static and simplify.
16327         (vect_analyze_data_refs): Adjust.
16328         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16329         correctly when vectorizing strided loads.
16331 2013-02-05  Richard Biener  <rguenther@suse.de>
16333         * doc/install.texi: Refer to ISL, not PPL.
16335 2013-02-05  Jan Hubicka  <jh@suse.cz>
16337         PR tree-optimization/55789
16338         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16340 2013-02-05  Jan Hubicka  <jh@suse.cz>
16342         PR tree-optimization/55789
16343         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16344         the dead call anyway.
16346 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16348         PR sanitizer/55374
16349         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16351 2013-02-04  Alexander Potapenko  <glider@google.com>
16352             Jack Howarth  <howarth@bromo.med.uc.edu>
16353             Jakub Jelinek  <jakub@redhat.com>
16355         PR sanitizer/55617
16356         * config/darwin.c (sort_ctor_records): Stabilized qsort
16357         on constructor priority by using original position.
16358         (finalize_ctors): New routine to sort constructors by
16359         priority before use in assemble_integer.
16360         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16362 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16364         PR libstdc++/54314
16365         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16366         about visibility on artificial decls.
16367         * config/sol2.c (solaris_assemble_visibility): Likewise.
16369 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16371         PR target/56186
16372         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16373         argument and improve checking of return-argument types for 16-byte
16374         modes.
16375         (ix86_function_value_1): Add additional valtype argument on call
16376         of function_value_64.
16377         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16378         handling infunction_value_64 function.
16380 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16382         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16384 2013-02-04  Richard Biener  <rguenther@suse.de>
16386         PR tree-optimization/56188
16387         * tree-ssa-structalias.c (label_visit): Consider case with
16388         initially non-empty points-to set.
16389         (perform_var_substitution): Dump node mapping and clean up.
16391 2013-02-04  Richard Guenther  <rguenther@suse.de>
16393         PR lto/56168
16394         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16395         node prevail as last resort.
16396         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16397         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16399 2013-02-04  Richard Biener  <rguenther@suse.de>
16401         PR tree-optimization/56113
16402         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16403         Merge into ...
16404         (equiv_class_lookup_or_add): ... this.
16405         (label_visit): Adjust and fix error in previous patch.
16406         (perform_var_substitution): Adjust.
16408 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16410         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16411         * config/sh/sh4-300.md: Likewise.
16412         * config/sh/sh4a.md: Likewise.
16413         * config/sh/constraints.md: Likewise.
16414         * config/sh/sh.md: Likewise.
16415         * config/sh/netbsd-elf.h: Likewise.
16416         * config/sh/predicates.md: Likewise.
16417         * config/sh/sh-protos.h: Likewise.
16418         * config/sh/ushmedia.h: Likewise.
16419         * config/sh/linux.h: Likewise.
16420         * config/sh/sh.c: Likewise.
16421         * config/sh/superh.h: Likewise.
16422         * config/sh/elf.h: Likewise.
16423         * config/sh/sh4.md: Likewise.
16424         * config/sh/sh.h: Likewise.
16426 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16428         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16429         constraint to match_test floating_point_store_memory_operand().
16430         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16431         (base14_operand): New.
16432         (floating_point_store_memory_operand): New.
16433         (integer_store_memory_operand): Revise to use base14_operand and
16434         reg_plus_base_memory_operand.
16435         (move_dest_operand): Allow symbolic_memory_operands.
16436         (symbolic_memory_operand): Check for LO_SOM.
16437         (symbolic_operand): Change default case to break.
16438         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16439         CONST_DOUBLE values to be reloaded by putting them into memory when
16440         the destination is a floating point register.
16441         (movdf): Remove code to handle CONST_DOUBLE.
16442         (movsf): Likewise.
16443         (reload_indf_r1): New.
16444         (reload_insf_r1): New.
16445         Consistently use "Q" and "T" constraints with integer and floating
16446         point move instructions, respectively.
16447         (movdi): Remove FAIL.
16448         Change predicate for source operand unamed DImode move from
16449         general_operand to move_src_operand.
16450         (umulsidi3): Change predicate for destination operand to
16451         register_operand.
16452         Likewise for similar unamed patterns.
16453         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16454         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16455         (hppa_legitimize_address): Simplify mask calculation.
16456         (pa_emit_move_sequence): Revised handling of secondary reloads from
16457         REG+D addresses for floating point loads and stores.  Directly handle
16458         loading CONST0_RTX (mode) to a floating point register.
16459         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16460         to floating point registers.  Don't restrict secondary reloads to
16461         floating point registers to integer modes.  Revise some comments and
16462         cleanup some code.
16463         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16464         (pa_legitimate_address_p): New.
16465         (pa_legitimize_reload_address): New.
16466         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16467         (STRICT_REG_OK_FOR_BASE_P): New.
16468         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16469         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16471 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16472             Andrew Dixie  <andrewd@gentrack.com>
16474         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16475         flag set.
16477 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16479         * expmed.c (extract_bit_field_1): Pass the full width of the
16480         structure to get_best_reg_extraction_insn.
16482 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16484         PR target/54601
16485         * configure.ac (use_cxa_atexit): Add AIX.
16486         * configure: Regenerate.
16488         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16490 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16492         PR debug/54793
16493         * final.c (need_profile_function): New variable.
16494         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16495         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16496         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16497         notes, targetm.asm_out.function_prologue doesn't emit anything,
16498         HAVE_prologue and profiler should be emitted before prologue,
16499         set need_profile_function instead of emitting it.
16500         (final_scan_insn): If need_profile_function, emit
16501         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16502         NOTE_INSN_FUNCTION_BEG note.
16504 2013-02-01  Richard Henderson  <rth@redhat.com>
16506         * config/rs6000/rs6000.md (smulditi3): New.
16507         (umulditi3): New.
16509         * config/alpha/alpha.md (umulditi3): New.
16511 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16513         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16514         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16516 2013-02-01  Richard Biener  <rguenther@suse.de>
16518         PR tree-optimization/56113
16519         * tree-ssa-structalias.c (label_visit): Reduce work for
16520         single-predecessor nodes.
16522 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
16524         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16525         range isn't testing for zero.
16527 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
16529         PR middle-end/56113
16530         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16532 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
16533             Nick Clifton  <nickc@redhat.com>
16535         * config/v850/constraints.md (Q): Define as a memory constraint.
16536         * config/v850/predicates.md (label_ref_operand): New predicate.
16537         (e3v5_shift_operand): New predicate.
16538         (ior_operator): New predicate.
16539         * config/v850/t-v850: Add e3v5 multilib.
16540         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16541         (v850_gen_movdi): Prototype.
16542         * config/v850/v850.c: Add support for e3v5 architecture.
16543         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16544         TARGET_V850E_UP.
16545         (construct_save_jarl): Add e3v5 long JARL support.
16546         (v850_adjust_insn_length): New function.  Adjust length of call
16547         insns when using e3v5 instructions.
16548         (v850_gen_movdi): New function: Generate instructions to move a
16549         DImode value.
16550         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16551         (CPP_SPEC): Define __v850e3v5__ as appropriate.
16552         (TARGET_USE_FPU): Enable for e3v5.
16553         (CONST_OK_FOR_W): New macro.
16554         (ADJUST_INSN_LENGTH): Define.
16555         * config/v850/v850.md (UNSPEC_LOOP): Define.
16556         (attr cpu): Add v850e3v5.
16557         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16558         (movdi): New pattern.
16559         (movdi_internal): New pattern.
16560         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16561         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16562         (cstoresf4): Likewise.
16563         (cstoredf4): Likewise.
16564         (insv): New pattern.
16565         (rotlso3_a): New pattern.
16566         (rotlsi3_b): New pattern
16567         (rotlsi3_v850e3v5): New pattern.
16568         (doloop_begin): New pattern.
16569         (fix_loop_counter): New pattern.
16570         (doloop_end): New pattern.
16571         (branch_normal): Add e3v5 long branch support.
16572         (branch_invert): Likewise.
16573         (branch_z_normal): Likewise.
16574         (branch_z_invert): Likewise.
16575         (branch_nz_normal): Likewise.
16576         (branch_nz_invert): Likewise.
16577         (call_internal_short): Add e3v5 register-indirect JARL support.
16578         (call_internal_long): Likewise.
16579         (call_value_internal_short): Likewise.
16580         (call_value_internal_long): Likewise.
16581         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16582         (mloop): New option.
16583         * config.gcc: Add support for configuring v840e3v5 target.
16584         * doc/invoke.texi: Document new v850 specific command line options.
16586 2013-01-31  Paul Koning  <ni1d@arrl.net>
16588         PR debug/55059
16589         PR debug/54508
16590         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16591         children if parent is a class.
16592         (prune_unused_types_prune): Don't add DW_AT_declaration.
16594 2013-01-31  Richard Biener  <rguenther@suse.de>
16596         PR tree-optimization/56157
16597         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16598         match up operand with SLP child.
16600 2013-01-31  Jason Merrill  <jason@redhat.com>
16602         PR debug/54410
16603         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16604         parameters the first time.
16605         (gen_scheduled_generic_parms_dies): Check completeness here.
16607 2013-01-31  Richard Biener  <rguenther@suse.de>
16609         PR middle-end/53073
16610         * common.opt (faggressive-loop-optimizations): New flag,
16611         enabled by default.
16612         * doc/invoke.texi (faggressive-loop-optimizations): Document.
16613         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16614         infer_loop_bounds_from_undefined by it.
16616 2013-01-31  Richard Biener  <rguenther@suse.de>
16618         PR tree-optimization/56150
16619         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16620         visit virtual operands.
16621         (find_uses_to_rename_bb): Likewise.
16623 2013-01-31  Richard Biener  <rguenther@suse.de>
16625         PR tree-optimization/56150
16626         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16627         mixed store non-store stmts.
16629 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
16631         PR sanitizer/55374
16632         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16633         LIBASAN_EARLY_SPEC is defined.
16634         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16635         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16636         before %o.
16637         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16639         PR c++/55742
16640         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16641         invalid args instead of ICEing on it.
16642         (ix86_valid_target_attribute_tree): Return error_mark_node if
16643         ix86_valid_target_attribute_inner_p failed.
16644         (ix86_valid_target_attribute_p): Return false only if
16645         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
16646         target("default") attribute.
16647         (sorted_attr_string): Change argument from const char * to tree,
16648         merge in all target attribute arguments rather than just one.
16649         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
16650         instead of free.  Avoid using strcat.
16651         (ix86_mangle_function_version_assembler_name): Mangle
16652         target("default") as if no target attribute is present.  Adjust
16653         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
16654         instead of xmalloc and XDELETEVEC instead of free.
16655         (ix86_function_versions): Don't return true if one of the decls
16656         doesn't have target attribute.  If they don't and one of the decls
16657         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
16658         sorted_attr_string caller.  Use XDELETEVEC instead of free.
16659         (ix86_supports_function_versions): Remove.
16660         (make_name): Fix up formatting.
16661         (make_dispatcher_decl): Remove resolver_name and its initialization.
16662         Avoid leaking memory.
16663         (is_function_default_version): Return true if there is
16664         target("default") attribute rather than no target attribute at all.
16665         (make_resolver_func): Avoid leaking memory.
16666         (ix86_generate_version_dispatcher_body): Likewise.
16667         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16668         * target.def (supports_function_versions): Remove.
16669         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16670         * doc/tm.texi: Regenerated.
16672 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
16674         PR rtl-optimization/56144
16675         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16676         for values with side effects.
16678 2013-01-30  Richard Biener  <rguenther@suse.de>
16680         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16681         (sparseset_pop): Likewise.
16682         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
16683         to be able to use quick_push in the worker loop.
16685 2013-01-30  Marek Polacek  <polacek@redhat.com>
16687         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16689 2013-01-30  Richard Biener  <rguenther@suse.de>
16691         PR lto/56147
16692         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16694 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
16696         PR tree-optimization/56064
16697         * fixed-value.c (fixed_from_double_int): New function.
16698         * fixed-value.h (fixed_from_double_int): New prototype.
16699         (const_fixed_from_double_int): New static inline function.
16700         * fold-const.c (native_interpret_fixed): New static function.
16701         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16702         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16703         (native_encode_fixed): New static function.
16704         (native_encode_expr) <FIXED_CST>: Use it.
16705         (native_interpret_int): Move double_int worker code to...
16706         * double-int.c (double_int::from_buffer): ...this new static method.
16707         * double-int.h (double_int::from_buffer): Prototype it.
16709 2013-01-30  Richard Biener  <rguenther@suse.de>
16711         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16712         New pointer-map and obstack.
16713         (init_alias_vars): Allocate pointer-map and obstack.
16714         (delete_points_to_sets): Free them.
16715         (find_what_var_points_to): Cache result.
16716         (find_what_p_points_to): Adjust for changed interface of
16717         find_what_var_points_to.
16718         (compute_points_to_sets): Likewise.
16719         (ipa_pta_execute): Likewise.
16721 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16723         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16724         * configure: Regenerate.
16725         * config.in: Regenerate.
16726         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16727         #nobits/#progbits if supported.
16729 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
16731         PR target/56121
16732         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16733         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16734         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16736 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16738         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16739         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16741 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16743         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16744         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16746 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16748         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16749         declaration.
16750         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16751         * config/arm/cortex-a7.md: New bypasses using
16752         arm_mac_accumulator_is_result.
16754 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16756         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
16757         (cortex_a7_neon_mla): Likewise.
16758         (cortex_a7_fpfmad): New reservation.
16759         (cortex_a7_fpmacs): Use ffmas and update required units.
16760         (cortex_a7_fpmuld): Update required units and latency.
16761         (cortex_a7_fpmacd): Likewise.
16762         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16763         (cortex_a7_neon). Likewise.
16764         (bypass) Update participating units.
16766 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16768         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16769         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16770         from fmac to ffma.
16771         * config/arm/vfp11.md (vfp_farith): Use ffmas.
16772         (vfp_fmul): Use ffmad.
16773         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16774         (cortex_r4_fmacd): Use ffmad.
16775         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16776         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
16777         (cortex_a9_fmacd): Use ffmad.
16778         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16779         (cortex_a8_vfp_macd): Use ffmad.
16780         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16781         (cortex_a5_fpmacd): Use ffmad.
16782         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16783         (cortex_a15_vfp_macd): Use ffmad.
16784         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16786 2013-01-29  Jason Merrill  <jason@redhat.com>
16788         PR libstdc++/54314
16789         * varasm.c (default_assemble_visibility): Don't warn about
16790         visibility on artificial decls.
16792 2013-01-29  Richard Biener  <rguenther@suse.de>
16794         PR tree-optimization/56113
16795         * tree-ssa-structalias.c (equiv_class_lookup): Also return
16796         the bitmap leader.
16797         (label_visit): Free duplicate bitmaps and record the leader instead.
16798         (perform_var_substitution): Adjust.
16800 2013-01-29  Richard Biener  <rguenther@suse.de>
16802         PR tree-optimization/55270
16803         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16804         the CFG, schedule loops for fixup.
16806 2013-01-29  Nick Clifton  <nickc@redhat.com>
16808         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16809         SP_REG.
16811 2013-01-28  Leif Ekblad  <leif@rdos.net>
16813         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16814         * config/i386/i386.h (TARGET_RDOS): New macro.
16815         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16816         * config/i386/i386.c (ix86_option_override_internal): For 64bit
16817         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16818         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16819         DEFAULT_LARGE_SECTION_THRESHOLD.
16820         * config/i386/i386.md (R14_REG, R15_REG): New constants.
16821         * config/i386/rdos.h: New file.
16822         * config/i386/rdos64.h: New file.
16824 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
16826         PR other/54814
16827         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16828         TEST_HARD_REG_BIT.
16830 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16832         PR rtl-optimization/56117
16833         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16834         call cselib_lookup_from_insn on the MEM before calling
16835         add_insn_mem_dependence.
16837 2013-01-28  Richard Biener  <rguenther@suse.de>
16839         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16840         to a stmt that didn't have one.
16841         (copy_phis_for_bb): Likewise for PHI arguments.
16842         (copy_debug_stmt): Likewise for debug stmts.
16844 2013-01-28  Richard Biener  <rguenther@suse.de>
16846         PR tree-optimization/56034
16847         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16848         (partition_builtin_p): Adjust.
16849         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
16850         it is the last partition.
16851         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16852         up the vertex for the definition.
16853         (classify_partition): Classify whether a partition is a
16854         PKIND_REDUCTION, thus has uses outside of the loop.
16855         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16856         Merge all PKIND_REDUCTION partitions into the last partition.
16857         (tree_loop_distribution): Seed partitions from reductions as well.
16859 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16861         PR tree-optimization/56125
16862         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16863         pow(x,c) into sqrt(x) * powi(x, n/2) or
16864         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16865         optimizing for size.
16866         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16867         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16868         integer.
16870         PR tree-optimization/56094
16871         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16872         to UNKNOWN_LOCATION while gimplifying expr.
16874 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
16876         PR target/56114
16877         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16878         operand 0 in movabs insn template for -masm=intel asm alternative.
16879         (*movabs<mode>_2): Ditto for operand 1.
16881 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
16883         PR target/54663
16884         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16885         of microblaze-c.o
16887 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16889         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16890         tm_file.
16892 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16894         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16895         Undef to avoid warning.
16897 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16899         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16900         * configure: Regenerate.
16902 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
16904         PR tree-optimization/56098
16905         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16906         for stmts with volatile ops.
16907         (cond_store_replacement): Don't optimize if assign has volatile ops.
16908         (cond_if_else_store_replacement_1): Don't optimize if either
16909         then_assign or else_assign have volatile ops.
16910         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16911         volatile ops.
16913 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16915         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16917 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16919         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16920         missing ':' in asm example.
16922 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
16924         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16925         entries into lane and laneq entries.
16926         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16927         Remove AdvSIMD scalar modes.
16928         (aarch64_sq<r>dmulh_laneq<mode>): New.
16929         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16930         modes.
16931         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16932         builtin implementations to relfect changes in RTL in aarch64-simd.md.
16933         * config/aarch64/iterators.md (VCOND): New.
16934         (VCONQ): New.
16936 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16938         PR target/54222
16939         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16940         Add NULL LIBNAME argument to existing definitions.
16941         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16942         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16943         * config/avr/avr.c (DEF_BUILTIN): Same.
16944         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16945         (avr_expand_builtin): Expand to a vanilla call if a libgcc
16946         implementation is available (DECL_ASSEMBLER_NAME is set).
16947         (avr_fold_absfx): New static function.
16948         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16949         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16950         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16951         AVR_BUILTIN_ABSLLK.
16952         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16953         (abshk, absk, abslk, absllk): Provide as static inline functions.
16955 2013-01-25  Marek Polacek  <polacek@redhat.com>
16957         PR tree-optimization/56035
16958         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16960 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
16962         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16963         (*movtf_internal_rex64): Add (!o,C) alternative
16964         (*movxf_internal_rex64): Ditto.
16965         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16967 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
16969         * doc/invoke.texi: fix typo.
16970         * doc/objc.texi: fix typo.
16972 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
16974         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16975         for the first two alternatives.
16977 2013-01-24  Diego Novillo  <dnovillo@google.com>
16979         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
16980         (ggc-zone.o): Remove.
16981         * configure.ac: Remove option --with-gc.
16982         * configure: Re-generate.
16983         * doc/install.texi: Remove documentation for --with-gc.
16984         * gengtype.c (write_enum_defn): Remove.  Update all users.
16985         (write_Types_process_field): Remove generation of gt_e_* argument.
16986         (output_type_enum): Remove.  Update all users.
16987         (write_enum_defn): Remove.  Update all users.
16988         (enum alloc_zone): Remove.  Update all users.
16989         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16990         * ggc-common.c (ggc_splay_alloc): Remove first argument.
16991         Update all callers.
16992         (struct ptr_data): Remove field TYPE.  Update all users.
16993         (gt_pch_note_object): Remove argument TYPE.  Update all users.
16994         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16995         Update all users.
16996         * ggc-none.c (ggc_alloc_typed_stat): Remove.
16997         (struct alloc_zone): Remove.
16998         (ggc_internal_alloc_zone_stat): Remove.
16999         (ggc_internal_cleared_alloc_zone_stat): Remove.
17000         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17001         (ggc_pch_count_object): Remove last argument.  Update all users.
17002         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17003         (struct alloc_zone): Remove.
17004         * ggc-zone.c: Remove.
17005         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17006         (struct alloc_zone): Remove.
17007         (ggc_alloc_typed_stat): Remove.
17008         (ggc_alloc_typed): Remove.
17009         (ggc_splay_alloc): Remove first argument.
17010         (rtl_zone): Remove.  Update all users.
17011         (tree_zone): Remove.  Update all users.
17012         (tree_id_zone): Remove.  Update all users.
17013         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17014         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17015         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17016         * tree-ssanames.c: Remove references to zone allocator in comments.
17018 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
17020         * config/avr/avr.c (avr_out_fract): Make register numbers that
17021         might be outside of source operand signed.
17023 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
17025         * config/i386/constraints.md (Yf): New constraint.
17026         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17027         of f constraint to conditionaly disable x87 register preferences.
17028         (*movdf_internal): Ditto.
17029         (*movsf_internal): Ditto.
17031 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
17033         PR inline-asm/55934
17034         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17035         that have operands with impossible constraints.
17036         Add a FIXME for a speed-up opportunity.
17037         * lra-constraints.c (process_alt_operands): Verify that a class
17038         selected from constraints on asms is valid for the operand mode.
17039         (curr_insn_transform): Remove incorrect comment.
17041 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
17043         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17044         TOC operand is a valid symbol ref in the constant pool.
17046 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17048         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17050 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
17052         PR target/54222
17053         * config/avr/stdfix.h: New file.
17054         * t-avr (stdfix-gcc.h): New rule to build it.
17055         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17057 2013-01-23  Kostya Serebryany  <kcc@google.com>
17059         * config/darwin.h: remove dependency on
17060         CoreFoundation (asan on Mac OS).
17062 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
17064         PR target/49069
17065         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17066         instead of cmpdi_operand for first comparison operand.
17067         Don't assert that comparison operands aren't both constants.
17069 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17071         * doc/install.texi (Downloading the Source): Update references to
17072         downloading separate components.
17074 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17076         * doc/extend.texi (__int128): Improve grammar.
17078 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
17080         PR target/56028
17081         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17082         alternative to (o,r).
17083         (*movdi_internal_rex64): Remove (!o,n) alternative.
17084         (DImode immediate->memory splitter): Remove.
17085         (DImode immediate->memory peephole2): Remove.
17086         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17087         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17088         alternative to (!o,*r).
17089         (*movtf_internal_sse): New pattern.
17090         (*movxf_internal_rex64): New pattern.
17091         (*movxf_internal): Disable for TARGET_64BIT.
17092         (*movdf_internal_rex64): Remove (!o,F) alternative.
17094 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
17096         PR middle-end/56074
17097         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17098         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17099         * tree-vect-loop-manip.c (find_loop_location): Also ignore
17100         stmt locations where LOCATION_LOCUS of the stmt location is
17101         UNKNOWN_LOCATION or BUILTINS_LOCATION.
17103         PR target/55686
17104         * config/i386/i386.md (UNSPEC_STOS): New.
17105         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17106         *strsetqi_1): Add UNSPEC_STOS.
17108 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
17110         PR c++/56067
17111         * doc/invoke.texi: Remove left over -Wsynth example.
17113 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17115         PR tree-optimization/56051
17116         * fold-const.c (fold_binary_loc): Don't fold
17117         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17118         a narrowing conversion, or widening conversion from signed
17119         to unsigned.
17121 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17123         PR rtl-optimization/56023
17124         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17125         dependent on debug instruction.
17127 2013-01-21  Martin Jambor  <mjambor@suse.cz>
17129         PR middle-end/56022
17130         * function.c (allocate_struct_function): Call
17131         invoke_set_current_function_hook earlier.
17133 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17135         * reload1.c (init_reload): Only initialize reload_obstack
17136         during the first call.
17138 2013-01-21  Marek Polacek  <polacek@redhat.com>
17140         * cfgloop.c (verify_loop_structure): Fix up grammar.
17142 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
17144         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17145         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17147 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17149         PR target/56058
17150         * config/arm/marvell-pj4.md: Update copyright year.
17151         Fix up use of alu to alu_reg and simple_alu_imm.
17153 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17155         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17157 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
17159         PR target/55433
17160         * lra-constraints.c (curr_insn_transform): Don't reuse original
17161         insn for secondary memory move when memory mode should be different.
17163 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17165         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17166         atomic_storedi_1): New patterns.
17168 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17170         btver2 pipeline descriptions.
17171         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17172         descriptions.
17173         * config/i386/i386.md (btver2_decode): New type attributes.
17174         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17175         type attributes.
17176         * config/i386/btver2.md: New file describing btver2 pipelines.
17178 2013-01-19  Andrew Pinski  <apinski@cavium.com>
17180         PR tree-optimization/52631
17181         * tree-ssa-sccvn (visit_use): Before looking up the original
17182         statement, try looking up the simplified expression.
17184 2013-01-19  Anthony Green  <green@moxielogic.com>
17186         * config/moxie/moxie.c (moxie_expand_prologue): Set
17187         current_function_static_stack_size.
17189 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17191         PR tree-optimization/56029
17192         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17193         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17195 2013-01-18  Sharad Singhai  <singhai@google.com>
17197         PR tree-optimization/55995
17198         * dumpfile.c (dump_loc): Print location only if available.
17199         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17201 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
17203         PR target/55433
17204         * lra-constraints.c (curr_insn_transform): Reuse original insn for
17205         secondary memory move.
17206         (inherit_reload_reg): Use rclass instead of cl for
17207         check_secondary_memory_needed_p.
17209 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17211         PR middle-end/56015
17212         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17213         the case where writing real complex part of target modifies op1.
17215 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
17217         * config/aarch64/aarch64-simd.md
17218         (aarch64_vcond_internal<mode>): Handle unordered cases.
17219         * config/aarch64/iterators.md (v_cmp_result): New.
17221 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
17222             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17224         * config/arm/marvell-pj4.md: New file.
17225         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17226         * config/arm/arm.md (generic_sched): Add marvell_pj4.
17227         (generic_vfp): Likewise.
17228         * config/arm/arm-cores.def: Add marvell-pj4.
17229         * config/arm/arm-tune.md: Regenerate.
17230         * config/arm/arm-tables.opt: Regenerate.
17231         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17232         * doc/invoke.texi: Document marvell-pj4.
17234 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
17236         * config/aarch64/arm_neon.h: Map scalar types to standard types.
17238 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17240         PR debug/54114
17241         PR debug/54402
17242         PR debug/49888
17243         * var-tracking.c (negative_power_of_two_p): New.
17244         (global_get_addr_cache, local_get_addr_cache): New.
17245         (get_addr_from_global_cache, get_addr_from_local_cache): New.
17246         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
17247         heading comment.
17248         (vt_stack_offset_p): Remove.
17249         (vt_canon_true_dep): Always canonicalize loc's address.
17250         (clobber_overlapping_mems): Make sure we have a MEM.
17251         (local_get_addr_clear_given_value): New.
17252         (val_reset): Clear local cached entries.
17253         (compute_bb_dataflow): Create and release the local cache.
17254         Disable duplicate MEMs clobbering.
17255         (emit_notes_in_bb): Clobber MEMs likewise.
17256         (vt_emit_notes): Create and release the local cache.
17257         (vt_initialize, vt_finalize): Create and release the global
17258         cache, respectively.
17259         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17261 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17263         PR libmudflap/53359
17264         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17265         not found in the symtab.
17267 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17269         PR debug/56006
17270         PR rtl-optimization/55547
17271         PR rtl-optimization/53827
17272         PR debug/53671
17273         PR debug/49888
17274         * alias.c (offset_overlap_p): New, factored out of...
17275         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17276         the conservative special case for symbolic constants.  Don't
17277         adjust zero sizes on alignment.
17279 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17281         PR rtl-optimization/52573
17282         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17283         REG_UNUSED for the same register.
17285 2013-01-17  Richard Biener  <rguenther@suse.de>
17286             Marek Polacek  <polacek@redhat.com>
17288         PR rtl-optimization/55833
17289         * loop-unswitch.c (unswitch_loops): Move loop verification...
17290         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17291         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17292         Set it to true when we're removing a loop from hierarchy tree in
17293         an irreducible region.
17294         (fix_bb_placements): Adjust caller.
17295         (fix_loop_placements): Likewise.
17297 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17299         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17300         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17301         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17302         Remove ID.  Adjust comments.
17303         * config/avr/avr-c.c (avr_builtin_name): Remove.
17304         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17305         * config/avr/avr.c (avr_tolower): New static function.
17306         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17307         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17308         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17309         default expansion.
17311 2013-01-17  Jan Hubicka  <jh@suse.cz>
17313         PR tree-optimization/55273
17314         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17316 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17318         PR target/55981
17319         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17320         store through atomic_store<mode>_1.
17321         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17323 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17325         PR tree-optimizations/55264
17326         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17327         for virtual methods.
17328         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17329         virtual methods before inlining is over.
17330         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17331         virtual functions.
17332         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17333         non-virtual.
17335 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17337         PR rtl-optimization/56005
17338         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17339         pending reads for prefetch.
17341 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17343         * config/aarch64/aarch64.md
17344         (*cstoresi_neg_uxtw): New pattern.
17345         (*cmovsi_insn_uxtw): New pattern.
17346         (*<optab>si3_uxtw): New pattern.
17347         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17348         (*<optab>si3_insn_uxtw): New pattern.
17349         (*bswapsi2_uxtw): New pattern.
17351 2013-01-16  Richard Biener  <rguenther@suse.de>
17353         * tree-inline.c (tree_function_versioning): Remove set but
17354         never used variable.
17356 2013-01-16  Richard Biener  <rguenther@suse.de>
17358         PR tree-optimization/55964
17359         * tree-flow.h (rename_variables_in_loop): Remove.
17360         (rename_variables_in_bb): Likewise.
17361         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17362         (copy_loop_before): Adjust and delete update-ssa status.
17363         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17364         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17365         (rename_variables_in_loop): Remove.
17366         (slpeel_update_phis_for_duplicate_loop): Likewise.
17367         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17368         use available cfg machinery instead of duplicating it.
17369         Update PHI nodes and perform poor-mans SSA update here.
17370         (slpeel_tree_peel_loop_to_edge): Adjust.
17372 2013-01-16  Richard Biener  <rguenther@suse.de>
17374         PR tree-optimization/54767
17375         PR tree-optimization/53465
17376         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17377         (vrp_visit_phi_node): For PHI arguments coming via backedges
17378         drop all symbolical range information.
17379         (execute_vrp): Compute backedges.
17381 2013-01-16  Richard Biener  <rguenther@suse.de>
17383         * doc/install.texi: Update CLooG and ISL requirements to
17384         0.18.0 and 0.11.1.
17386 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17388         PR target/55301
17389         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17390         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17391         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17393 2013-01-16  DJ Delorie  <dj@redhat.com>
17395         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17396         (UNSPECV_SP_SWITCH_E): New.
17397         (sp_switch_1): Change to an unspec.
17398         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17399         replace $r15.
17401 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17403         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17404         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17405         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17406         (expand_mem_thread_fence): Ditto.
17407         (expand_mem_signal_fence): Ditto.
17408         (expand_atomic_load): Ditto.
17409         (expand_atomic_store): Ditto.
17411 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17413         PR rtl-optimization/55547
17414         PR rtl-optimization/53827
17415         PR debug/53671
17416         PR debug/49888
17417         * alias.c (memrefs_conflict_p): Set sizes to negative after
17418         AND adjustments.
17420 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17422         PR target/55940
17423         * function.c (thread_prologue_and_epilogue_insns): Always
17424         add crtl->drap_reg to set_up_by_prologue.set, even if
17425         stack_realign_drap is false.
17427 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17429         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17430         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17431         *call): Fix indention.
17433 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17435         PR target/55876
17436         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17437         Update comment.
17439 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17441         PR rtl-optimization/55153
17442         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17444 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17446         PR tree-optimization/55920
17447         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17448         accesses as grp_to_be_debug_replaced.
17450 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17452         PR tree-optimization/55920
17453         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17454         there is non-useless type conversion needed from debug rhs to lhs,
17455         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17457 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17458             Mikael Pettersson  <mikpe@it.uu.se>
17460         PR target/43961
17461         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17462         Thumb.
17463         (ASM_OUTPUT_CASE_LABEL): Remove.
17464         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17465         * final.c (shorten_branches): Update alignment of labels before
17466         jump tables if CASE_VECTOR_SHORTEN_MODE.
17468 2013-01-15  Richard Biener  <rguenther@suse.de>
17470         PR bootstrap/55961
17471         * system.h: Do not include gmp.h for building host tools.
17473 2013-01-15  Richard Biener  <rguenther@suse.de>
17475         PR middle-end/55882
17476         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17477         account for bitpos when computing alignment.
17479 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17481         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17482         (ix86_target_macros_internal): Likewise.
17484         * config/i386/i386.c (m_CORE2I7): Removed.
17485         (m_CORE_HASWELL): New macro.
17486         (m_CORE_ALL): Likewise.
17487         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17488         (initial_ix86_arch_features): Likewise.
17489         (processor_target_table): Initializations for Core avx2.
17490         (cpu_names): New names "core-avx2".
17491         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17492         PROCESSOR_CORE_HASWELL.
17493         (ix86_issue_rate): New case.
17494         (ia32_multipass_dfa_lookahead): Likewise.
17495         (ix86_sched_init_global): Likewise.
17497         * config/i386/i386.h (TARGET_HASWELL): New macro.
17498         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17499         (processor_type): New PROCESSOR_HASWELL.
17501 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17503         PR tree-optimization/55955
17504         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17505         *SHIFT_EXPR and *ROTATE_EXPR codes.
17507         PR tree-optimization/48766
17508         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17509         -ftrapv disable -fwrapv.
17511 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17513         PR target/55974
17514         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17515         etc. to 1 and not to __flash.
17516         Use LL suffix for __INT24_MAX__ with -mint8.
17517         Use ULL suffix for __UINT24_MAX__ with -mint8.
17519 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17521         * config/avr/avr-arch.h
17522         (struct base_arch_s): Use typedef avr_arch_t instead.
17523         (struct arch_info_s): Use typedef avr_arch_info_t instead.
17524         (struct mcu_type_s): Use typedef avr_mcu_t instead.
17525         * config/avr/avr.c: Same.
17526         * config/avr/avr-devices.c: Same.
17527         * config/avr/driver-avr.c: Same.
17528         * config/avr/gen-avr-mmcu-texi.c: Same.
17529         * config/avr/avr-mcus.def: Adjust comment.
17531 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
17533         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17534         * config/aarch64/iterators.md (VALLDI): New.
17536 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
17537             Andi Kleen  <ak@linux.intel.com>
17539         PR target/55948
17540         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17541         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17542         memmodel flag.
17544 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17546         * config/avr/avr-stdint.h: Remove trailing blanks.
17547         * config/avr/avr-log.h: Same.
17548         * config/avr/avr-arch.h: Same.
17549         * config/avr/avr-devices.c: Same.
17550         * config/avr/avr-dimode.md: Same.
17551         * config/avr/predicates.md: Same.
17552         * config/avr/avr-c.c: Same.  And fix typo.
17554         * config/avr/avr-protos.h: Same.  And:
17555         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17556         (init_cumulative_args): Rename to avr_init_cumulative_args.
17557         (expand_prologue): Rename to avr_expand_prologue.
17558         (expand_epilogue): Rename to avr_expand_epilogue.
17559         (adjust_insn_length): Rename to avr_adjust_insn_length.
17560         (notice_update_cc): Rename to avr_notice_update_cc.
17561         (final_prescan_insn): Rename to avr_final_prescan_insn.
17562         * config/avr/avr.c: Same.
17563         * config/avr/avr.h: Same.
17564         * config/avr/avr.md: Remove trailing blanks.
17565         (prologue): Use avr_expand_prologue.
17566         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17568 2013-01-14  Richard Biener  <rguenther@suse.de>
17570         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17571         verify_location, collect_subblocks): New functions.
17572         (verify_gimple_in_cfg): Verify that locations only reference
17573         BLOCKs in the functions BLOCK tree.
17575 2013-01-14  Richard Biener  <rguenther@suse.de>
17577         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17578         PHI argument.
17579         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17580         unshare reference.
17581         (insert_out_of_ssa_copy_on_edge): Likewise.
17582         (rewrite_close_phi_out_of_ssa): Likewise.
17583         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17584         debug expressions.
17585         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17586         propagated constants.
17587         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17588         can not be shared.
17590 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17592         * config/avr/avr-modes.def: Add GPL copyright notice.
17594 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
17596         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17597         MEMMODEL_MASK to determine memory model.
17598         (atomic_store<mode>): Ditto from operands[2].
17599         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17601 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
17603         PR fortran/55935
17604         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17605         (fold_gimple_assign): Don't call unshare_expr here.
17606         (fold_ctor_reference): Call unshare_expr.
17608 2013-01-13  Terry Guo  <terry.guo@arm.com>
17610         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17611         * doc/fragments.texi: Document MULTILIB_REUSE.
17612         * gcc.c (multilib_reuse): New internal spec.
17613         (set_multilib_dir): Also search multilib from multilib_reuse.
17614         * genmultilib (tmpmultilib3): Refactor code.
17615         (tmpmultilib4): Ditto.
17616         (multilib_reuse): New multilib argument.
17618 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
17620         * Makefile.in: Update copyright.
17622 2013-01-12  Tom de Vries  <tom@codesourcery.com>
17624         PR middle-end/55890
17625         * calls.c (expand_call): Check if arg_nr is valid.
17627 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17629         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17630         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17631         documentation.  Add missing '__' in front of
17632         __builtin_ia32_packssdw256.
17634 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17636         PR target/55719
17637         * config/s390/s390.c (s390_preferred_reload_class): Do not return
17638         NO_REGS for larl operands.
17639         (s390_reload_larl_operand): Use s390_load_address instead of
17640         emit_move_insn.
17642 2013-01-11  Richard Biener  <rguenther@suse.de>
17644         * tree-cfg.c (verify_node_sharing_1): Split out from ...
17645         (verify_node_sharing): ... here.
17646         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17648 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
17650         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17651         Substitute TREECHECKING.
17652         * configure: Regenerate.
17653         * Makefile.in (TREECHECKING): New.
17655 2013-01-11  Richard Guenther  <rguenther@suse.de>
17657         PR tree-optimization/44061
17658         * tree-vrp.c (extract_range_basic): Compute zero as
17659         value-range for __builtin_constant_p of function parameters.
17661 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
17663         Update copyright years.
17665 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
17667         PR rtl-optimization/55672
17668         * lra-eliminations.c (mark_not_eliminable): Permit addition with
17669         const to be eliminable.
17671 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
17673         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17674         * configure: Regenerate.
17676 2013-01-10  Richard Biener  <rguenther@suse.de>
17678         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17680 2013-01-10  Richard Biener  <rguenther@suse.de>
17682         PR bootstrap/55792
17683         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17684         locations for virtual PHI arguments.
17685         (rewrite_update_phi_arguments): Likewise.
17687 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
17689         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17690         on to assembler.
17692 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
17694         PR tree-optimization/55921
17695         * tree-complex.c (expand_complex_asm): New function.
17696         (expand_complex_operations_1): Call it for GIMPLE_ASM.
17698 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17700         PR target/55718
17701         * config/s390/s390.c (s390_symref_operand_p)
17702         (s390_loadrelative_operand_p): Merge the two functions.
17703         (s390_check_qrst_address, print_operand_address): Add parameters
17704         to s390_loadrelative_operand_p invokation.
17705         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17706         (s390_reload_larl_operand, s390_secondary_reload): Use
17707         s390_loadrelative_operand_p instead of s390_symref_operand_p.
17708         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17710 2013-01-09  Mike Stump  <mikestump@comcast.net>
17712         * dse.c (record_store): Remove unnecessary assert.
17714 2013-01-09  Jan Hubicka  <jh@suse.cz>
17716         PR tree-optimization/55569
17717         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17718         * cfgloop.h (scale_loop_profile): Likewise.
17720 2013-01-09  Jan Hubicka  <jh@suse.cz>
17722         PR lto/45375
17723         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17724         functions.
17725         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17727 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
17729         PR middle-end/55114
17730         * expr.h (maybe_emit_group_store): Declare.
17731         * expr.c (maybe_emit_group_store): New function.
17732         * builtins.c (expand_builtin_int_roundingfn): Call it.
17733         (expand_builtin_int_roundingfn_2): Likewise.
17735 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
17737         PR rtl-optimization/55829
17738         * lra-constraints.c (match_reload): Add code for absent output.
17739         (curr_insn_transform): Add code for reloads of matched inputs
17740         without output.
17742 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
17744         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17745         attribute of movddup insn to DF.
17746         (*vec_interleave_lowv2df): Ditto.
17747         (vec_dupv2df): Ditto.
17749 2013-01-09  Jan Hubicka  <jh@suse.cz>
17751         PR tree-optimiation/55875
17752         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17753         EVERY_ITERATION parameter.
17754         (number_of_iterations_exit): Check if exit is executed every iteration.
17755         (idx_infer_loop_bounds): Similarly here.
17756         (n_of_executions_at_most): Simplify
17757         to only test for cases where statement is dominated by the
17758         particular bound; handle correctly the "postdominance" test.
17759         (scev_probably_wraps_p): Use max loop iterations info
17760         as a global bound first.
17762 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
17763             Nick Clifton  <nickc@redhat.com>
17765         * config/v850/v850.md (cbranchsf4): New pattern.
17766         (cstoresf4): New pattern.
17767         (cbranchdf4): New pattern.
17768         (cstoredf4): New pattern.
17769         (movsicc): Disallow floating point comparisons.
17770         (cmpsf_le_insn): Fix order of operators.
17771         (cmpsf_lt_insn): Likewise.
17772         (cmpsf_eq_insn): Likewise.
17773         (cmpdf_le_insn): Likewise.
17774         (cmpdf_lt_insn): Likewise.
17775         (cmpdf_eq_insn): Likewise.
17776         (cmpsf_ge_insn): Use LE comparison.
17777         (cmpdf_ge_insn): Likewise.
17778         (cmpsf_gt_insn): Use LT comparison.
17779         (cmpdf_gt_insn): Likewise.
17780         (cmpsf_ne_insn): Delete pattern.
17781         (cmpdf_ne_insn): Delete pattern.
17782         * config/v850/v850.c (v850_gen_float_compare): Use
17783         gen_cmpdf_eq_insn for NE comparison.
17784         (v850_float_z_comparison_operator)
17785         (v850_float_nz_comparison_operator): Move from here ...
17786         * config/v850/predicates.md: ... to here.  Move GT and GE
17787         comparisons into v850_float_z_comparison_operator.
17788         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17789         Delete prototype.
17790         (v850_float_nz_comparison_operator): Likewise.
17792 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17794         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17795         with calls to gen_insvsi/gen_insvdi.
17797 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17799         * config/i386/i386.c (initial_ix86_tune_features): Set up
17800         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17802 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
17803             Jakub Jelinek  <jakub@redhat.com>
17805         PR tree-optimization/48189
17806         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17807         If nitercst is 0, don't predict the exit edge.
17809 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17811         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17812         in asm_fprintf with reg_names.
17813         (aarch64_print_operand_address): Likewise.
17814         (aarch64_return_addr): Likewise.
17815         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17817 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17819         * config/pa/pa.h (VAL_U6_BITS_P): Define.
17820         (INT_U6_BITS): Likewise.
17821         * config/pa/predicates.md (uint6_operand): New predicate.
17822         (shift5_operand, shift6_operand): Likewise.
17823         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17824         arith32_operand.
17825         (lshrdi3): Use shift6_operand.
17826         (shrpsi4, shrpdi4): New insn patterns.
17827         (extzv): Delete expander.
17828         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
17829         predicates in unamed zero extract patterns.  Tighten common constraint.
17830         (extv): Delete expander.
17831         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
17832         predicates in unamed sign extract patterns.  Tighten common constraint.
17833         (insv): Delete expander.
17834         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
17835         predicates in unamed insert patterns.  Tighten common constraint.
17836         Change uint32_operand predicate to uint6_operand predicate in unamed
17837         DImode pattern to insert constant values of type 1...1xxxx.
17839 2013-01-04  Jan Hubicka  <jh@suse.cz>
17841         PR tree-optimization/55823
17842         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17843         issue.
17845 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17846             Uros Bizjak  <ubizjak@gmail.com>
17848         PR rtl-optimization/55845
17849         * df-problems.c (can_move_insns_across): Stop scanning at
17850         volatile_insn_p source instruction or give up if
17851         across_from .. across_to range contains any volatile_insn_p
17852         instructions.
17854 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17856         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17857         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17858         Declare.
17859         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17860         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17862 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17864         PR fortran/55341
17865         * asan.c (asan_clear_shadow): New function.
17866         (asan_emit_stack_protection): Use it.
17868 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17870         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17871         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17872         with tab instead of space.
17874 2013-01-08  Nick Clifton  <nickc@redhat.com>
17876         * config/rl78/rl78.c (rl78_expand_prologue): Always select
17877         register bank 0 at the start of an interrupt handler.
17878         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17879         MDBH registers.
17881 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17883         * config/aarch64/aarch64-simd.md
17884         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17885         (aarch64_simd_bsl): Likewise.
17886         (aarch64_vcond_internal<mode>): Likewise.
17887         (vcond<mode><mode>): Likewise.
17888         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17889         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17891 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17893         * config/aarch64/aarch64-builtins.c
17894         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17896 2013-01-08  Martin Jambor  <mjambor@suse.cz>
17898         PR debug/55579
17899         * tree-sra.c (analyze_access_subtree): Return true also after
17900         potentially creating a debug-only replacement.
17902 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17904         PR middle-end/55890
17905         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17907         PR tree-optimization/54120
17908         * tree-vrp.c (range_fits_type_p): Don't allow
17909         src_precision < precision from signed vr to unsigned_p
17910         if vr->min or vr->max is negative.
17911         (simplify_float_conversion_using_ranges): Test can_float_p
17912         against CODE_FOR_nothing.
17914 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17915             Richard Biener  <rguenther@suse.de>
17917         PR middle-end/55851
17918         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17919         types instead of just INTEGER_TYPE types.
17921 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
17923         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17924         TF_SIZE): Define.
17926 2013-01-07  Steve Ellcey  <sellcey@mips.com>
17928         PR target/42661
17929         * config/mips/mips.opt: Change mad to mmad to match documentation.
17931 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17933         PR target/55897
17934         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17935         .progmemx.data now.
17937 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17939         PR target/55897
17940         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17941         (avr_addrspace_t): Add .section_name field.
17942         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17943         array size.
17944         (avr_addrspace): Same.  Initialize .section_name.  Remove last
17945         NULL entry.  Put __memx into .progmemx.data.
17946         (progmem_section_prefix): Remove.
17947         (avr_asm_init_sections): No need to initialize progmem_section.
17948         (avr_asm_named_section): Use avr_addrspace[].section_name to get
17949         section name prefix.
17950         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
17951         retrieve the progmem section.
17952         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17953         boundary to run over avr_addrspace[].
17954         (avr_register_target_pragmas): Ditto.
17956 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
17958         * varasm.c (output_constant_def_contents): For asan_protect_global
17959         protected strings, adjust DECL_ALIGN if needed, before testing for
17960         anchored symbols.
17961         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17962         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
17963         normal decls.
17964         (output_object_block): For asan protected decls, emit asan padding
17965         after their contents.
17966         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17967         (asan_finish_file): Test it here instead.
17969 2013-01-07  Nick Clifton  <nickc@redhat.com>
17970             Matthias Klose  <doko@debian.org>
17971             Doug Kwan  <dougkwan@google.com>
17972             H.J. Lu  <hongjiu.lu@intel.com>
17974         PR driver/55470
17975         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17977         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17979         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17981         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17983         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17985 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17987         PR target/54461
17988         * doc/install.texi (Cross-Compiler-Specific Options): Document
17989         --with-avrlibc.
17991 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
17993         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17994         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17995         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17996         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17997         vqmovun_high_s64): Fix source operand number and update copyright.
17999 2013-01-07  Richard Biener  <rguenther@suse.de>
18001         PR middle-end/55890
18002         * gimple.h (gimple_call_builtin_p): New overload.
18003         * gimple.c (validate_call): New function.
18004         (gimple_call_builtin_p): Likewise.
18005         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18006         Use gimple_call_builtin_p.
18007         (find_func_clobbers): Likewise.
18008         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18009         (strlen_optimize_stmt): Likewise.
18011 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18013         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18014         (vld1q_dup_*): Likewise.
18015         (vld1_*): Likewise.
18016         (vld1q_*): Likewise.
18017         (vld1_lane_*): Likewise.
18018         (vld1q_lane_*): Likewise.
18020 2013-01-07  Richard Biener  <rguenther@suse.de>
18022         * lto-streamer.h (LTO_minor_version): Bump to 2.
18024 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18026         * config/aarch64/aarch64-protos.h
18027         (aarch64_const_double_zero_rtx_p): Rename to...
18028         (aarch64_float_const_zero_rtx_p): ...this.
18029         (aarch64_float_const_representable_p): New.
18030         (aarch64_output_simd_mov_immediate): Likewise.
18031         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18032         move immediate case.
18033         * config/aarch64/aarch64.c
18034         (aarch64_const_double_zero_rtx_p): Rename to...
18035         (aarch64_float_const_zero_rtx_p): ...this.
18036         (aarch64_print_operand): Allow printing of new constants.
18037         (aarch64_valid_floating_const): New.
18038         (aarch64_legitimate_constant_p): Check for valid floating-point
18039         constants.
18040         (aarch64_simd_valid_immediate): Likewise.
18041         (aarch64_vect_float_const_representable_p): New.
18042         (aarch64_float_const_representable_p): Likewise.
18043         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18044         (aarch64_output_simd_mov_immediate): New.
18045         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18046         (*movdf_aarch64): Likewise.
18047         * config/aarch64/constraints.md (Ufc): New.
18048         (Y): call aarch64_float_const_zero_rtx.
18049         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18051 2013-01-07  Richard Biener  <rguenther@suse.de>
18053         PR tree-optimization/55888
18054         PR tree-optimization/55862
18055         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18056         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18057         not if it is contained therein.
18059 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18061         * config/avr/t-avr: Typo.
18063 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18065         PR55243
18066         * config/avr/t-avr: Don't automatically rebuild
18067         $(srcdir)/config/avr/t-multilib
18068         $(srcdir)/config/avr/avr-tables.opt
18069         $(srcdir)/doc/avr-mmcu.texi
18070         (avr-mcus): New phony target to build them on request.
18071         (s-avr-mlib, s-avr-mmcu-texi): Remove.
18072         * avr/avr-mcus.def: Adjust comments.
18074 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
18076         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18078 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18080         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18082 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18084         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18086 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
18088         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18089         to generate profiling.
18090         * config/rs6000/aix64.h (LIB_SPEC): Same.
18092 2013-01-04  Andrew Pinski  <apinski@cavium.com>
18094         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18095         New function.
18096         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18098 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
18100         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18101         unconditionally.
18102         (ix86_expand_move): Ditto.
18103         (ix86_zero_extend_to_Pmode): Ditto.
18104         (ix86_expand_call): Ditto.
18105         (ix86_expand_special_args_builtin): Ditto.
18106         (ix86_expand_builtin): Ditto.
18108 2013-01-04  Richard Biener  <rguenther@suse.de>
18110         PR tree-optimization/55862
18111         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18112         translating them through PHI nodes.
18114 2013-01-04  Martin Jambor  <mjambor@suse.cz>
18116         PR tree-optimization/55755
18117         * tree-sra.c (sra_modify_assign): Do not check that an access has no
18118         children when trying to avoid producing a VIEW_CONVERT_EXPR.
18120 2013-01-04  Marek Polacek  <polacek@redhat.com>
18122         PR middle-end/55859
18123         * opts.c (default_options_optimization): Clarify error message.
18125 2013-01-04  Richard Biener  <rguenther@suse.de>
18127         PR middle-end/55863
18128         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18129         reassociation.
18131 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18133         PR target/53789
18134         * config/pa/pa.md (movsi): Revert previous change.
18135         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18136         references.
18138 2013-01-03  Richard Henderson  <rth@redhat.com>
18140         * config/i386/i386.c (ix86_expand_move): Always assign to op1
18141         after eliminating TLS symbols.
18143 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
18145         PR bootstrap/50167
18146         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18147         * graphite-poly.c (debug_gmp_value): Likewise.
18149 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
18151         PR target/55712
18152         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18153         selected code model, define __code_mode_small__, __code_model_medium__,
18154         __code_model_large__, __code_model_32__ or __code_model_kernel__.
18155         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18156         xchg temporary register with %k.  Declare temporary register as
18157         early clobbered.
18158         [__x86_64__]: For medium and large code models, preserve %rbx register.
18160 2013-01-03  Richard Biener  <rguenther@suse.de>
18162         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18163         (dump_subscript): Adjust.
18164         (finalize_ddr_dependent): Do not dump redundant info.
18165         (analyze_siv_subscript): Adjust.
18166         (subscript_dependence_tester): Likewise.
18167         (compute_affine_dependence): Likewise.
18169 2013-01-03  Richard Biener  <rguenther@suse.de>
18171         Revert
18172         2013-01-03  Richard Biener  <rguenther@suse.de>
18174         PR tree-optimization/55857
18175         * tree-vect-stmts.c (vectorizable_load): Do not setup
18176         re-alignment for invariant loads.
18178         2013-01-02  Richard Biener  <rguenther@suse.de>
18180         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18181         invariant load do not generate a vector load from the scalar location.
18183 2013-01-03  Richard Biener  <rguenther@suse.de>
18185         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18186         for not vectorizing.
18187         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18188         not build INDIRECT_REFs, call get_name once only.
18189         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
18190         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18192 2013-01-03  Richard Biener  <rguenther@suse.de>
18194         PR tree-optimization/55857
18195         * tree-vect-stmts.c (vectorizable_load): Do not setup
18196         re-alignment for invariant loads.
18198 2013-01-03  Richard Biener  <rguenther@suse.de>
18200         PR lto/55848
18201         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18202         prefer a built-in decl.
18204 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18206         * gcc.c (process_command): Update copyright notice dates.
18207         * gcov.c (print_version): Likewise.
18208         * gcov-dump.c (print_version): Likewise.
18210         PR rtl-optimization/55838
18211         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18212         iv0.step, iv1.step and step.
18214 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18215             Marc Glisse  <marc.glisse@inria.fr>
18217         PR tree-optimization/55832
18218         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18219         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18220         integer_{one,zero}_node.
18222 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18224         PR debug/54402
18225         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18226         * var-tracking.c (reverse_op): Don't add reverse ops to
18227         VALUEs that have already
18228         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18230 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
18232         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18234 2013-01-02  Teresa Johnson  <tejohnson@google.com>
18236         * dumpfile.c (dump_loc): Print filename with location.
18237         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18238         new location_t parameter to emit complete unroll message with
18239         new dump framework.
18240         (canonicalize_loop_induction_variables): Compute loops location
18241         and pass to try_unroll_loop_completely.
18242         * loop-unroll.c (report_unroll_peel): New function.
18243         (peel_loops_completely): Use new dump format with location
18244         for main dumpfile message, and invoke report_unroll_peel on success.
18245         (decide_unrolling_and_peeling): Ditto.
18246         (decide_peel_once_rolling): Remove old dumpfile message subsumed
18247         by report_unroll_peel.
18248         (decide_peel_completely): Ditto.
18249         (decide_unroll_constant_iterations): Ditto.
18250         (decide_unroll_runtime_iterations): Ditto.
18251         (decide_peel_simple): Ditto.
18252         (decide_unroll_stupid): Ditto.
18253         * cfgloop.c (get_loop_location): New function.
18254         * cfgloop.h (get_loop_location): Declare.
18256 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18258         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18259         NULL.
18261 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18263         PR middle-end/55198
18264         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18265         BLKmode objects when EXPAND_MEMORY is specified.
18267 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18269         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18270         in loop predicate.
18271         (fold_builtin_cpu): Do not share cpu model decls across statements.
18273 2013-01-02  Jason Merrill  <jason@redhat.com>
18275         PR c++/55804
18276         * tree.c (build_array_type_1): Revert earlier change.
18278 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18280         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18281         "cortex-a57".
18282         * config/aarch64/aarch64-tune.md: Re-generate.
18284 2013-01-02  Richard Biener  <rguenther@suse.de>
18286         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18287         invariant load do not generate a vector load from the scalar location.
18289 2013-01-02  Richard Biener  <rguenther@suse.de>
18291         PR bootstrap/55784
18292         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18293         * configure: Regenerate.
18295 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18297         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18298         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18299         (expand_builtin_int_roundingfn_2): Keep the original target around
18300         for the fallback case.
18302 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18304         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18305         to be clear for sign changes.
18307 2013-01-01  Jan Hubicka  <jh@suse.cz>
18309         * ipa-inline-analysis.c: Fix formatting.
18311 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18313         PR tree-optimization/55831
18314         * tree-vect-loop.c (get_initial_def_for_induction): Use
18315         gsi_after_labels instead of gsi_start_bb.
18317 Copyright (C) 2013 Free Software Foundation, Inc.
18319 Copying and distribution of this file, with or without modification,
18320 are permitted in any medium without royalty provided the copyright
18321 notice and this notice are preserved.