* target.def (narrow_volatile_bitfield): Note that the default
[official-gcc.git] / gcc / ChangeLog
blobbce291dcfb23f67873f3b938559eaecb9c20ab38
1 2013-08-20  Nick Clifton  <nickc@redhat.com>
3         * target.def (narrow_volatile_bitfield): Note that the default
4         value is false, not !TARGET_STRICT_ALIGN.
5         * doc/tm.texi: Regenerate.
7 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
9         Fix LIB_SPEC for systems without libpthread.
11         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
12         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
13         for Android.
14         * config/i386/linux-common.h: Likewise.
15         * config/mips/linux-common.h: Likewise.
17 2013-08-20  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
19         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
20         checks.
22 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
24         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
25         instances can own GC refs.
27         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
28         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
29         (gcc::context::gt_pch_nx): Likewise.
30         (gcc::context::gt_pch_nx):  Likewise.
31         * ggc.h (gt_ggc_mx <T>): New.
32         (gt_pch_nx_with_op <T>): New.
33         (gt_pch_nx <T>): New.
34         * passes.c (opt_pass::gt_ggc_mx): New.
35         (opt_pass::gt_pch_nx): New.
36         (opt_pass::gt_pch_nx_with_op): New.
37         (pass_manager::gt_ggc_mx): New.
38         (pass_manager::gt_pch_nx): New.
39         (pass_manager::gt_pch_nx_with_op): New.
40         (pass_manager::operator new): Use
41         ggc_internal_cleared_alloc_stat rather than xcalloc.
42         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
43         (pass_manager::gt_ggc_mx): New.
44         (pass_manager::gt_pch_nx): New.
45         (pass_manager::gt_pch_nx_with_op): New.
46         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
47         (opt_pass::operator new): New.
48         (opt_pass::gt_ggc_mx): New.
49         (opt_pass::gt_pch_nx): New.
50         (opt_pass::gt_pch_nx_with_op): New.
52 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
54         * Makefile.in (GTFILES): Add context.h.
55         * context.c (gcc::context::operator new): New.
56         (gcc::context::gt_ggc_mx): New.
57         (gcc::context::gt_pch_nx): New.
58         (gcc::context::gt_pch_nx): New.
59         * context.h (gcc::context): Add GTY((user)) marking.
60         (gcc::context::operator new): New.
61         (gcc::context::gt_ggc_mx): New.
62         (gcc::context::gt_pch_nx): New.
63         (gcc::context::gt_pch_nx): New.
64         (g): Add GTY marking.
65         (gt_ggc_mx (gcc::context *)): New.
66         (gt_pch_nx (gcc::context *)): New.
67         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
68         void *cookie)): New.
69         * gengtype.c (open_base_files) <ifiles>: Add context.h.
71 2013-08-20  Alan Modra  <amodra@gmail.com>
73         PR target/57865
74         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
75         (rs6000_emit_epilogue): Likewise.
77 2013-08-19  Dehao Chen (dehao@google.com)
79         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
81 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
82             Jakub Jelinek  <jakub@redhat.com>
84         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
85         (BUILT_IN_FABSD64): Likewise.
86         (BUILT_IN_FABSD128): Likewise.
87         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
88         (fold_builtin_1): Likewise.
89         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping destination
90         and source operands.
91         (*abstd2_fpr): Likewise.
92         (*nabstd2_fpr): Likewise.
94 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
96         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
97         JUMP_P and INSN_P.
99 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
101         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
102         incomplete.
104 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
106         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
107         * builtins.c (default_libc_has_function): New.
108         (gnu_libc_has_function): Ditto.
109         (no_c99_libc_has_function): Ditto.
110         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
111         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
112         (fold_builtin_sincos): Likewise.
113         (fold_builtin_cexp): Likewise.
114         * builtins.def (DEF_C94_BUILTIN): Likewise.
115         (DEF_C99_BUILTIN): Likewise.
116         (DEF_C99_C90RES_BUILTIN): Likewise.
117         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
118         definitions to using this define.
119         * config/darwin-protos.h (darwin_libc_has_function): New.
120         * config/darwin.c: (darwin_libc_has_function: Ditto.
121         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
122         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
123         * config/darwin.h: Ditto.
124         * config/elfos.h: Ditto.
125         * config/freebsd.h: Ditto.
126         * config/i386/cygming.h: Ditto.
127         * config/i386/djgpp.h: Ditto.
128         * config/i386/i386-interix.h: Ditto.
129         * config/microblaze/microblaze.h: Ditto.
130         * config/mmix/mmix.h: Ditto.
131         * config/gnu-user.h: Ditto.
132         * config/ia64/hpux.h: Ditto.
133         * config/pa/pa-hpux.h: Ditto.
134         * config/pdp11/pdp11.h: Ditto.
135         * config/picochip/picochip.h: Ditto.
136         * config/linux.h: Ditto.
137         * config/netbsd.h: Ditto.
138         * config/openbsd.h: Ditto.
139         * config/rs6000/aix43.h: Ditto.
140         * config/rs6000/aix51.h: Ditto.
141         * config/rs6000/aix52.h: Ditto.
142         * config/rs6000/aix53.h: Ditto.
143         * config/rs6000/aix61.h: Ditto.
144         * config/rs6000/darwin.h: Ditto.
145         * config/rs6000/linux.h: Ditto.
146         * config/rs6000/linux64.h: Ditto.
147         * config/s390/tpf.h: Ditto.
148         * config/sol2-10.h: Ditto.
149         * config/sol2.h: Ditto.
150         * config/vms/vms.h: Ditto.
151         * config/vxworks.h: Ditto.
152         * config/linux-android.c (linux_android_libc_has_function):
153         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
154         * config/linux-protos.h (linux_android_libc_has_function):
155         New declaration.
156         * config/i386/i386.c (ix86_libc_has_function): New.
157         * config/i386/i386-protos.h
158         (ix86_libc_has_function): New declaration.
159         * config/i386/i386.md
160         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
161         ("isinf<mode>2): Likewise.
162         * convert.c (convert_to_integer): Using new target hook
163         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
164         TARGET_C99_FUNCTIONS.
165         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
166         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
167         * coretypes.h (function_class): New enum for different
168         classes of functions.
169         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
170         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
171         (TARGET_HAS_SINCOS): Likewise.
172         (TARGET_LIBC_HAS_FUNCTION): New.
173         * doc/tm.texi: Regenerated.
174         * targhooks.h (default_libc_has_function): New declaration.
175         (no_c99_libc_has_function): Ditto.
176         (gnu_libc_has_function): Ditto.
177         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
178         and TARGET_HAS_SINCOS.
180 2013-08-18  Jan Hubicka  <jh@suse.cz>
182         * Makeifle-in (ipa-devirt.o): New.
183         (GTFILES): Add ipa-utils.h and ipa-devirt.c
184         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
185         (analyze_functions): Look into possible targets of polymorphic call.
186         * dumpfile.c (dump_files): Add type-inheritance dump.
187         * dumpfile.h (TDI_inheritance): New.
188         * ipa-devirt.c: New file.
189         * ipa-utils.h (odr_type_d): Forward declare.
190         (odr_type): New type.
191         (build_type_inheritance_graph): Declare.
192         (possible_polymorphic_call_targets): Declare and introduce inline
193         variant when only edge is pased.
194         (dump_possible_polymorphic_call_targets): Likewise.
195         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
196         * tree.c (type_in_anonymous_namespace_p): Break out from ...
197         (types_same_for_odr): ... here.
198         * tree.h (type_in_anonymous_namespace_p): Declare.
200 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
202         PR tree-optimization/58006
203         * tree-parloops.c (take_address_of): Don't ICE if get_name
204         returns NULL.
205         (eliminate_local_variables_stmt): Remove clobber stmts.
207 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
209         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
210         error message is issued for an alias to undefined symbol.
212 2013-08-18  Jan Hubicka  <jh@suse.cz>
214         * cgraph.c (cgraph_create_indirect_edge): Discover
215         polymorphic calls and record basic info into indirect_info.
216         * gimple-fold.c (gimple_fold_call): When doing BINFO based
217         devirtualization, ignore objc function calls.
218         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
219         call with no parm index info.
220         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
221         * tree.c (virtual_method_call_p): New function.
222         * tree.h (virtual_method_call_p): Declare.
224 2013-08-16  Jan Hubicka  <jh@suse.cz>
226         PR middle-end/58179
227         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
229 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
231         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
232         attribute.
234 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
236         * gengtype.c (type_for_name): Add special-case support for
237         locating types within the "gcc::" namespace.
238         (open_base_files): Emit a "using namespace gcc" directive.
240 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
242         PR target/58160
243         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
244         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
246         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
247         array instead of each individual operand as a separate argument.
248         (emit_fusion_gpr_load): Likewise.
249         (expand_fusion_gpr_load): Add new function declaration.
251         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
252         signature to have the operands passed as an array, instead of as
253         separate arguments.  Allow ZERO_EXTEND to be in the memory
254         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
255         depend on the register live/dead flags when peepholes are run.
256         (expand_fusion_gpr_load): New function to be called from the
257         peephole2 pass, to change the register that addis sets to be the
258         target register.
259         (emit_fusion_gpr_load): Change the calling signature to have the
260         operands passed as an array, instead of as separate arguments.
261         Allow ZERO_EXTEND to be in the memory address, and also
262         SIGN_EXTEND if -mpower8-fusion-sign.
264         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
265         unspec enumeration.
266         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
267         adjust the register addis loads up in the peephole2 pass.  Do not
268         depend on the register live/dead state when the peephole pass is
269         done.
271 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
273         * gengtype.c (create_user_defined_type): Ensure that the kind
274         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
275         declaration is seen before the GTY((user)) marking.
277 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
279         PR target/58105
280         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
282 2013-08-16  Jan Hubicka  <jh@suse.cz>
284         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
285         arugment expected_type.
286         (gimple_fold_call): Use it.
287         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
288         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
289         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
290         (try_make_edge_direct_virtual_call): Likewise.
291         * tree.c (obj_type_ref_class): New.
292         * tree.h (obj_type_ref_class): Use it.
294 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
296         * sched-vis.c (rtl_slim_pp_initialized): Remove.
297         (rtl_slim_pp): Likewise.
298         (init_rtl_slim_pretty_print): Likewise.
299         (dump_value_slim):  Don't call it.  Use local pretty printer.
300         (dump_insn_slim): Likewise.
301         (dump_rtl_slim): Likewise.
302         (str_pattern_slim): Likewise.
303         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
304         Simplify.
306 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
308         PR tree-optimization/58164
309         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
310         walk gimple_goto_dest of GIMPLE_GOTO.
312         PR tree-optimization/58165
313         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
314         bi_call must be the last stmt in a bb, don't split_block, instead
315         use fallthru edge from it and give up if there is none.
316         Release conds vector when returning early.
318 2013-08-14  Xinliang David Li  <davidxl@google.com>
320         * config/i386/i386.c (ix86_option_override_internal):
321         Remove unused variable and field.
323 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
325         PR target/57949
326         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
327         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
328         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
329         and Linux, correct BLKmode alignment when 128-bit alignment is
330         required and compatibility flag is not set.
331         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
332         for zero-size arguments when compatibility flag is not set.
334 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
336         PR tree-optimization/58145
337         * tree-sra.c (build_ref_for_offset): If prev_base has
338         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
340 2013-08-14  Xinliang David Li  <davidxl@google.com>
342         * config/i386/i386.c (ix86_option_override_internal):
343         Fix uninitialized variable error.
345 2013-08-14  Xinliang David Li  <davidxl@google.com>
347         * config/i386/i386.opt: Define two new options.
348         * config/i386/x86-tune.def: Add arch selector field in macros.
349         * config/i386/i386.h: Adjust macro definition.
350         * config/i386/i386.c (ix86_option_override_internal):
351         Refactor the code.
352         (parse_mtune_ctrl_str): New function.
353         (set_ix86_tune_features): New function.
354         (ix86_function_specific_restore): Call the new helper function.
356 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
358         PR rtl-optimization/57662
359         * sel-sched.c (code_motion_process_successors): When the current insn
360         is removed after the recursive traversal, break from the loop.
361         Add comments and debug printouts.
363 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
364             Alexandre Oliva  <aoliva@redhat.com>
366         PR target/58067
367         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
368         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
369         there also UNSPEC_PLTOFF.
371 2013-08-14  Marek Polacek  <polacek@redhat.com>
373         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
374         NUM_CONDITIONS bit positions.
376 2013-08-13  Cary Coutant  <ccoutant@google.com>
378         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
379         (attr_checksum): Hash vector contents instead of pointer.
380         (attr_checksum_ordered): Likewise.
382 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
384         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
385         when Pmode != word_mode.  Add length_address attribute.
386         (sse3_monitor_<mode>): Merge from sse3_monitor and
387         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
388         Pmode != word_mode.  Update insn length attribute.
389         * config/i386/i386.c (ix86_option_override_internal): Update
390         ix86_gen_monitor selection for merged sse3_monitor insn.
392 2013-08-13  Julian Brown  <julian@codesourcery.com>
394         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
395         perform invalid legitimization on greater-than-word-size modes for
396         TARGET_E500_DOUBLE.
398 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
400         * ira.c (setup_class_translate_array): Use aclass instead of cl
401         for classes not fully covered by allocno classes.
403 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
405         PR tree-optimization/57661
406         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
407         * tree-inline.c (tree_function_versioning): Initialize it.
408         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
409         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
410         that is not being copied.
412         PR sanitizer/56417
413         * asan.c (instrument_strlen_call): Fix typo in comment.
414         Use char * type even for the lhs of POINTER_PLUS_EXPR.
416 2013-08-13  Steve Ellcey  <sellcey@mips.com>
418         * config/mips/mips.md (prefetch): Use lw instead of ld on
419         loongson in 32bit mode.
421 2013-08-13  Nick Clifton  <nickc@redhat.com>
423         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
425 2013-08-13  Jan Hubicka  <jh@suse.cz>
427         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
428         introduced edge; fix typo in sanity check.
429         (cgraph_resolve_speculation): Export; improve diagnostic.
430         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
431         speculation at type mismatch.
432         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
433         (cgraph_resolve_speculation): Declare.
434         (symtab_can_be_discarded): New function.
435         * value-prof.c (gimple_ic_transform): Remove actual transform code.
436         * ipa-inline-transform.c (speculation_removed): New global var.
437         (clone_inlined_nodes): See if speculation can be removed.
438         (inline_call): If speculations was removed, we growths may not match.
439         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
440         (speculation_useful_p): New function.
441         (resolve_noninline_speculation): New function.
442         (inline_small_functions): Resolve useless speculations.
443         * ipa-inline.h (speculation_useful_p): Declare
444         * ipa.c (can_replace_by_local_alias): Simplify.
445         (ipa_profile): Produce speculative calls in non-lto, too;
446         add simple cost model; produce local aliases.
448 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
450         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
451         PASS_MANAGER_H.
453 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
455         * config/i386/i386.c (ix86_function_versions): Use error + inform.
457 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
459         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
460         iterator instead of X87MODEF.
462 2013-08-12  Perez Read  <netfirewall@gmail.com>
464         PR target/58132
465         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
466         operand 0 for intel asm alternative.
467         (*movabs<mode>_2): Ditto for operand 1.
469 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
471         * config/aarch64/arm_none.h
472         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
474 2013-08-12  Nick Clifton  <nickc@redhat.com>
476         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
478 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
480         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
481         expand for QI/HImode operand to produce more effictive code for
482         unsigned char(short) --> float(double) conversion.
484 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
486         * doc/invoke.texi: Mention that -ftls-model does not force the final
487         model.
489 2013-08-12  Marek Polacek  <polacek@redhat.com>
490             Marc Glisse  <marc.glisse@inria.fr>
492         PR tree-optimization/57980
493         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
494         when creating -1 constant.
496 2013-08-10  Jan Hubicka  <jh@suse.cz>
498         Workaround binutils PR14342.
499         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
500         (gimple_init_edge_profiler): Likewise.
501         (gimple_gen_ic_func_profiler): Likewise.
503 2013-08-09  Jan Hubicka  <jh@suse.cz>
505         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
507 2013-08-09  Xinliang David Li  <davidxl@google.com>
509         * config/i386/stringop.def: New file.
510         * config/i386/stringop.opt: New file.
511         * config/i386/i386-opts.h: Include stringopt.def.
512         * config/i386/i386.opt: Include stringopt.opt.
513         * config/i386/i386.c (ix86_option_override_internal):
514         Override default size based stringop inline strategies with options.
515         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
516         New function.
518 2013-08-09  Jan Hubicka  <jh@suse.cz>
520         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
522 2013-08-09  Jan Hubicka  <jh@suse.cz>
524         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
525         CGRAPH_FREQ_MAX.
526         (dump_cgraph_node): Dump profile-id.
527         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
528         and common_target_probability.
529         * lto-cgraph.c (lto_output_edge): Stream common targets.
530         (lto_output_node): Stream profile ids.
531         (input_node): Stream profile ids.
532         (input_edge): Stream common targets.
533         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
534         * ipa.c: Include value-prof.h
535         (ipa_profile_generate_summary): Turn indirect call statement histograms
536         into common targets.
537         (ipa_profile): Turn common targets into speculative edges.
539 2013-08-09  Jan Hubicka  <jh@suse.cz>
541         * cgraph.h (cgraph_node): Add profile_id.
542         * value-prof.c (cgraph_node_map): Turn into pointer_map.
543         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
544         (del_node_map): Update.
545         (find_func_by_funcdef_no): Replace by ...
546         (find_func_by_profile_id): ... this one.
547         (gimple_ic_transform): Do not remove useful histograms when
548         speculation is not done; dump info when indirect call removal
549         can happen at LTO.
550         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
551         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
552         (__gcov_indirect_call_profiler_v2): .. this one.
553         * profile.h (init_node_map): Update.
554         * coverage.c (coverage_compute_profile_id): New function.
555         * coverage.h (coverage_compute_profile_id): Declare.
556         * tree-profile.c (init_ic_make_global_vars): Make
557         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
558         (gimple_init_edge_profiler): Update prototype of
559         __gcov_indirect_call_profiler.
560         (gimple_gen_ic_func_profiler): Simplify.
561         (tree_profiling): Use init_node_map
563 2013-08-09  Jan Hubicka  <jh@suse.cz>
565         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
566         non-speculative refs.
567         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
568         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
569         (cgraph_set_call_stmt): Likewise.
570         (cgraph_create_edge_1): Fix release checking compilatoin;
571         clear lto_stmt_uid.
572         (cgraph_free_edge): Free indirect info.
573         (cgraph_turn_edge_to_speculative): New function.
574         (cgraph_speculative_call_info): New function.
575         (cgraph_make_edge_direct): Return direct edge; handle speculation.
576         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
577         (dump_cgraph_node): Dump speculation.
578         (verify_edge_count_and_frequency): Accept speculative edges.
579         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
580         (verify_cgraph_node): Handle speculation.
581         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
582         (cgraph_set_call_stmt): Update prototype.
583         (cgraph_make_edge_direct): Update prototype.
584         (cgraph_speculative_call_info): Declare.
585         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
586         to change; update call of ipa_find_references.
587         * ipa-ref.c (ipa_record_reference): Fix return value; clear
588         lto_stmt_uid and speculative flags.
589         (ipa_dump_references): Dump speculation.
590         (ipa_clone_references): Clone speculative flag.
591         (ipa_clone_referring): Likewise.
592         (ipa_clone_ref): New function.
593         (ipa_find_reference): Look into lto_stmt_uids
594         (ipa_clear_stmts_in_references): Do not clear speculative calls.
595         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
596         (ipa_find_reference): Update declaration.
597         (ipa_clone_ref): Declare.
598         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
599         stream speculative flag.
600         (lto_output_ref): Stream statements uids and speculation.
601         (input_ref): Likewise.
602         (input_edge): Stream speuclation.
603         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
604         (cgraph_set_call_stmt_including_clones): Handle speculation.
605         * ipa-inline.c (heap_edge_removal_hook): New function.
606         (inline_small_functions): Register it.
607         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
608         also initialize refs.
609         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
610         edge to change.
611         (try_make_edge_direct_simple_call): Likewise.
612         (try_make_edge_direct_simple_call): Likewise.
613         (update_indirect_edges_after_inlining): Likewise.
614         (remove_described_reference): Look proper lto_stmt_uid.
615         (propagate_controlled_uses): Likewise.
616         (propagate_controlled_uses): Liekwise.
617         * tree-inline.c (copy_bb): Copy speculative edges.
618         (redirect_all_calls): New function.
619         (copy_cfg_body): Do redirection after loop info is updated.
620         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
622 2013-08-09  Jan Hubicka  <jh@suse.cz>
624         * lto-streamer-out.c (output_function): Renumber PHIs.
625         * lto-streamer-in.c (input_function): Likewise.
627 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
629         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
630         (get_lane_unsigned): Likewise.
631         (dup_lane_scalar): Likewise.
632         (get_lane): enable for VALL.
633         * config/aarch64/aarch64-simd.md
634         (aarch64_dup_lane_scalar<mode>): Remove.
635         (aarch64_get_lane_signed<mode>): Likewise.
636         (aarch64_get_lane_unsigned<mode>): Likewise.
637         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
638         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
639         (aarch64_get_lane<mode>): Enable for all vector modes.
640         (aarch64_get_lanedi): Remove misleading constraints.
641         * config/aarch64/arm_neon.h
642         (__aarch64_vget_lane_any): Define.
643         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
644         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
645         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
646         * config/aarch64/iterators.md (VDQQH): New.
647         (VDQQHS): Likewise.
648         (vwcore): Likewise.
650 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
652         * configure.ac: Add GAS check for LEON instructions on SPARC.
653         * configure: Regenerate.
654         * config.in: Likewise.
655         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
656         sparc*-*-* block.
657         * config/sparc/sparc.opt (LEON, LEON3): New masks.
658         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
659         for LEON or LEON3.
660         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
661         (AS_LEON_FLAG): New macro.
662         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
663         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
664         Deal with LEON and LEON3 for the memory model.
665         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
666         (atomic_compare_and_swap<mode>_1): Likewise.
667         (*atomic_compare_and_swap<mode>_1): Likewise.
669 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
671         * config/arm/neon.md (vcond): Fix floating-point vector
672         comparisons against 0.
674 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
676         * lra-constraints.c (emit_spill_move): Remove assert.
677         (process_alt_operands): Add more debugging
678         output.  Increase reject for spilling into memory.  Decrease
679         reject for reloading scratch.
680         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
682 2013-08-08  Steve Ellcey  <sellcey@mips.com>
684         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
685         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
686         micromips incompatible.  Add nan2008.
687         (MULTILIB_DIRNAMES): Add nan2008.
688         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
689         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
690         and micromips incompatible.  Add nan2008.
691         (MULTILIB_DIRNAMES): Add nan2008.
692         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
694 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
696         PR rtl-optimization/58079
697         * combine.c (combine_simplify_rtx): Avoid using SUBST if
698         simplify_comparison has widened a comparison with an integer.
700 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
702         * config/arm/neon.md (movmisalign<mode>): Disable when we
703         don't allow unaligned accesses.
704         (*movmisalign<mode>_neon_store): Likewise.
705         (*movmisalign<mode>_neon_load): Likewise.
706         (*movmisalign<mode>_neon_store): Likewise.
707         (*movmisalign<mode>_neon_load): Likewise.
709 2013-08-08  Jan Hubicka  <jh@suse.cz>
711         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
712         (make_pass_rebuild_cgraph_edges): Also clear references.
713         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
714         * ipa-inline-transform.c (inline_transform): Remove all references
715         after inlining.
716         * cgraphunit.c (expand_function): Remove all references after
717         expansion.
718         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
719         (ipa_find_reference): Rewrite to iterator.
720         (remove_stmt_references): Likewise.
721         (ipa_clear_stmts_in_references): New function.
722         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
723         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
724         clear references.
725         * ipa-split.c (split_function): Remove references in split function.
727 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
729         PR target/57431
730         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
731         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
733 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
735         PR target/56979
736         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
737         suggested mode for the assignment isn't compatible with the
738         registers required.
740 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
742         PR target/58065
743         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
745 2013-08-07  Xinliang David Li  <davidxl@google.com>
747         * config/i386/i386.opt: New option -mtune-ctrl=.
748         * config/i386/x86-tune.def: New file.
749         * config/i386/i386.h: include x86-tune.def.
750         * config/i386/i386.c (ix86_option_override_internal):
751         Parsing -mtune-ctrl= option and set tune features.
753 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
755         PR other/12081
756         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
757         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
758         to gen_2arg_fn_t.
760 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
762         * rtl.h (update_alignments): Declare.
763         * final.c (grow_label_align): New function extracted from...
764         (shorten_branches): ...here.  Call it.
765         (update_alignments): New function.
766         * reorg.c (sibling_labels): New variable.
767         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
768         the new label along with it onto the sibling_labels vector.
769         (fill_simple_delay_slots): Adjust call to get_label_before.
770         (fill_slots_from_thread): Likewise.
771         (relax_delay_slots): Likewise.
772         (make_return_insns): Likewise.
773         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
775 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
777         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
778         document its semantics.
779         (diagnostic_report_diagnostic): Adjust accordingly.
781 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
783         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
784         (sparc_option_override): ...and port to new C++ pass API.
785         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
787 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
789         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
791 2013-08-06  Caroline Tice  <cmtice@google.com>
793         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
794         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
795         * tree-pass.h: Add pass_vtable_verify.
796         * varasm.c (assemble_variable): Add code to properly set the comdat
797         section and name for the .vtable_map_vars section.
798         (assemble_vtyv_preinit_initializer): New function.
799         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
800         LINK_ONCE flag.
801         * output.h: Add function decl for assemble_vtv_preinit_initializer.
802         * vtable-verify.c: New file.
803         * vtable-verify.h: New file.
804         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
805         initialiation levels.
806         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
807         * passes.def: Insert pass_vtable_verify.
808         * aclocal.m4: Reorder includes.
809         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
810         -fvtv-counts options.
811         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
812         as appropriate, if -fvtable-verify=... is used.
813         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
814         -fvtable-verify=... is used.
815         * Makefile.in (OBJS):  Add vtable-verify.o to list.
816         (vtable-verify.o): Add new build rule.
817         (GTFILES): Add vtable-verify.c to list.
818         * common.opt (fvtable-verify=): New flag.
819         (vtv_priority): Values for fvtable-verify= flag.
820         (fvtv-counts): New flag.
821         (fvtv-debug): New flag.
822         * tree.h (save_vtable_map_decl): New extern function decl.
824 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
826         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
827         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
828         (pass_data_rl78_devirt): ...new pass_data instance and...
829         (make_pass_rl78_devirt): ...new function.
830         (rl78_asm_file_start): Port pass registration to new C++ API.
832 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
834         * coretypes.h (rtl_opt_pass): Add.
835         (gcc::context): Add.
836         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
837         (epiphany_init): Port to new C++ pass API.
838         (epiphany_optimize_mode_switching): Likewise.
839         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
840         (pass_manager::get_pass_mode_switching): New.
841         (pass_manager::get_pass_peephole2): New.
842         * mode-switching.c (pass_mode_switching): Add clone method.
843         * recog.c (pass_peephole2): Add clone method.
844         (pass_split_all_insns): Add clone method.
846 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
848         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
849         (mips_option_override): ...here, porting to new C++ API for passes.
851 2013-08-06  Jan Hubicka  <jh@suse.cz>
853         * cgraph.c (cgraph_get_body): New function based on lto.c
854         implementation.
855         * cgraph.h (cgraph_get_body): Declare.
856         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
857         LTO paths.
858         * cgraphunit.c (expand_function): Get body prior expanding.
859         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
860         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
861         really need.
862         * passes.c (do_per_function_toporder): Get body.
863         * tree-inline.c (expand_call_inline): Get body prior inlining it.
864         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
866 2013-08-06  Martin Jambor  <mjambor@suse.cz>
868         PR fortran/57987
869         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
870         is not re-finalized.  Rename second parameter to no_collect.
872 2013-08-06  Martin Jambor  <mjambor@suse.cz>
874         PR middle-end/58041
875         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
876         MEM_REF has proper alignment information.
878 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
880         PR other/12081
881         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
882         class insn_gen_fn.
883         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
884         rtx (*) (rtx, ...) with insn_gen_fn.
885         * genoutput.c (output_insn_data): Cast gen_? function pointers to
886         insn_gen_fn::stored_funcptr.  Add initializer braces.
888 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
890         Rewrite how instances of passes are cloned to remove assumptions
891         about their sizes (thus allowing pass subclasses to have
892         additional data fields, albeit non-GC-managed ones at this point).
894         * passes.c (make_pass_instance): Now that passes have clone
895         methods, rewrite this function to eliminate XNEW and memcpy
896         calls that used hardcoded sizes.  Since this function no longer
897         creates pass instances, rename it to...
898         (add_pass_instance): ...this.  Document the old way that passes were
899         numbered and flagged, and rework this function to continue using it.
900         (next_pass_1): Add an initial_pass argument for use by
901         add_pass_instance.
902         (position_pass): When adding multiple instances of a pass, use
903         the pass's clone method, rather than relying on the XNEW/memcpy
904         within the former make_pass_instance (now add_pass_instance).
905         (pass_manager::pass_manager): When invoking next_pass_1, also supply
906         the initial instance of the current pass within the pass manager.
908 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
910         This is the automated part of the conversion of passes from C
911         structs to C++ classes.
913         Patch autogenerated by refactor_passes.py from
914         https://github.com/davidmalcolm/gcc-refactoring-scripts
915         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
917         * asan.c (pass_asan): Convert from a global struct to a subclass of
918         gimple_opt_pass along with...
919         (pass_data_asan): ...new pass_data instance and...
920         (make_pass_asan): ...new function.
921         (pass_asan_O0): Convert from a global struct to a subclass of
922         gimple_opt_pass along with...
923         (pass_data_asan_O0): ...new pass_data instance and...
924         (make_pass_asan_O0): ...new function.
925         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
926         subclass of rtl_opt_pass along with...
927         (pass_data_inc_dec): ...new pass_data instance and...
928         (make_pass_inc_dec): ...new function.
929         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
930         a subclass of rtl_opt_pass along with...
931         (pass_data_reorder_blocks): ...new pass_data instance and...
932         (make_pass_reorder_blocks): ...new function.
933         (pass_duplicate_computed_gotos): Convert from a global struct to a
934         subclass of rtl_opt_pass along with...
935         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
936         (make_pass_duplicate_computed_gotos): ...new function.
937         (pass_partition_blocks): Convert from a global struct to a subclass of
938         rtl_opt_pass along with...
939         (pass_data_partition_blocks): ...new pass_data instance and...
940         (make_pass_partition_blocks): ...new function.
941         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
942         struct to a subclass of rtl_opt_pass along with...
943         (pass_data_branch_target_load_optimize1): ...new pass_data instance
944         and...
945         (make_pass_branch_target_load_optimize1): ...new function.
946         (pass_branch_target_load_optimize2): Convert from a global struct to a
947         subclass of rtl_opt_pass along with...
948         (pass_data_branch_target_load_optimize2): ...new pass_data instance
949         and...
950         (make_pass_branch_target_load_optimize2): ...new function.
951         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
952         of rtl_opt_pass along with...
953         (pass_data_jump): ...new pass_data instance and...
954         (make_pass_jump): ...new function.
955         (pass_jump2): Convert from a global struct to a subclass of
956         rtl_opt_pass along with...
957         (pass_data_jump2): ...new pass_data instance and...
958         (make_pass_jump2): ...new function.
959         * cfgexpand.c (pass_expand): Convert from a global struct to a
960         subclass of rtl_opt_pass along with...
961         (pass_data_expand): ...new pass_data instance and...
962         (make_pass_expand): ...new function.
963         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
964         of rtl_opt_pass along with...
965         (pass_data_free_cfg): ...new pass_data instance and...
966         (make_pass_free_cfg): ...new function.
967         (pass_into_cfg_layout_mode): Convert from a global struct to a
968         subclass of rtl_opt_pass along with...
969         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
970         (make_pass_into_cfg_layout_mode): ...new function.
971         (pass_outof_cfg_layout_mode): Convert from a global struct to a
972         subclass of rtl_opt_pass along with...
973         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
974         (make_pass_outof_cfg_layout_mode): ...new function.
975         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
976         struct to a subclass of gimple_opt_pass along with...
977         (pass_data_build_cgraph_edges): ...new pass_data instance and...
978         (make_pass_build_cgraph_edges): ...new function.
979         (pass_rebuild_cgraph_edges): Convert from a global struct to a
980         subclass of gimple_opt_pass along with...
981         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
982         (make_pass_rebuild_cgraph_edges): ...new function.
983         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
984         subclass of gimple_opt_pass along with...
985         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
986         and...
987         (make_pass_remove_cgraph_callee_edges): ...new function.
988         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
989         struct to a subclass of rtl_opt_pass along with...
990         (pass_data_stack_adjustments): ...new pass_data instance and...
991         (make_pass_stack_adjustments): ...new function.
992         * combine.c (pass_combine): Convert from a global struct to a subclass
993         of rtl_opt_pass along with...
994         (pass_data_combine): ...new pass_data instance and...
995         (make_pass_combine): ...new function.
996         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
997         global struct to a subclass of rtl_opt_pass along with...
998         (pass_data_compare_elim_after_reload): ...new pass_data instance
999         and...
1000         (make_pass_compare_elim_after_reload): ...new function.
1001         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1002         of rtl_opt_pass along with...
1003         (pass_data_rtl_cprop): ...new pass_data instance and...
1004         (make_pass_rtl_cprop): ...new function.
1005         * cse.c (pass_cse): Convert from a global struct to a subclass of
1006         rtl_opt_pass along with...
1007         (pass_data_cse): ...new pass_data instance and...
1008         (make_pass_cse): ...new function.
1009         (pass_cse2): Convert from a global struct to a subclass of
1010         rtl_opt_pass along with...
1011         (pass_data_cse2): ...new pass_data instance and...
1012         (make_pass_cse2): ...new function.
1013         (pass_cse_after_global_opts): Convert from a global struct to a
1014         subclass of rtl_opt_pass along with...
1015         (pass_data_cse_after_global_opts): ...new pass_data instance and...
1016         (make_pass_cse_after_global_opts): ...new function.
1017         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1018         of rtl_opt_pass along with...
1019         (pass_data_ud_rtl_dce): ...new pass_data instance and...
1020         (make_pass_ud_rtl_dce): ...new function.
1021         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1022         rtl_opt_pass along with...
1023         (pass_data_fast_rtl_dce): ...new pass_data instance and...
1024         (make_pass_fast_rtl_dce): ...new function.
1025         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1026         a subclass of rtl_opt_pass along with...
1027         (pass_data_df_initialize_opt): ...new pass_data instance and...
1028         (make_pass_df_initialize_opt): ...new function.
1029         (pass_df_initialize_no_opt): Convert from a global struct to a
1030         subclass of rtl_opt_pass along with...
1031         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1032         (make_pass_df_initialize_no_opt): ...new function.
1033         (pass_df_finish): Convert from a global struct to a subclass of
1034         rtl_opt_pass along with...
1035         (pass_data_df_finish): ...new pass_data instance and...
1036         (make_pass_df_finish): ...new function.
1037         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1038         rtl_opt_pass along with...
1039         (pass_data_rtl_dse1): ...new pass_data instance and...
1040         (make_pass_rtl_dse1): ...new function.
1041         (pass_rtl_dse2): Convert from a global struct to a subclass of
1042         rtl_opt_pass along with...
1043         (pass_data_rtl_dse2): ...new pass_data instance and...
1044         (make_pass_rtl_dse2): ...new function.
1045         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1046         subclass of rtl_opt_pass along with...
1047         (pass_data_dwarf2_frame): ...new pass_data instance and...
1048         (make_pass_dwarf2_frame): ...new function.
1049         * except.c (pass_set_nothrow_function_flags): Convert from a global
1050         struct to a subclass of rtl_opt_pass along with...
1051         (pass_data_set_nothrow_function_flags): ...new pass_data instance
1052         and...
1053         (make_pass_set_nothrow_function_flags): ...new function.
1054         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1055         subclass of rtl_opt_pass along with...
1056         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1057         and...
1058         (make_pass_convert_to_eh_region_ranges): ...new function.
1059         * final.c (pass_compute_alignments): Convert from a global struct to a
1060         subclass of rtl_opt_pass along with...
1061         (pass_data_compute_alignments): ...new pass_data instance and...
1062         (make_pass_compute_alignments): ...new function.
1063         (pass_final): Convert from a global struct to a subclass of
1064         rtl_opt_pass along with...
1065         (pass_data_final): ...new pass_data instance and...
1066         (make_pass_final): ...new function.
1067         (pass_shorten_branches): Convert from a global struct to a subclass of
1068         rtl_opt_pass along with...
1069         (pass_data_shorten_branches): ...new pass_data instance and...
1070         (make_pass_shorten_branches): ...new function.
1071         (pass_clean_state): Convert from a global struct to a subclass of
1072         rtl_opt_pass along with...
1073         (pass_data_clean_state): ...new pass_data instance and...
1074         (make_pass_clean_state): ...new function.
1075         * function.c (pass_instantiate_virtual_regs): Convert from a global
1076         struct to a subclass of rtl_opt_pass along with...
1077         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1078         (make_pass_instantiate_virtual_regs): ...new function.
1079         (pass_leaf_regs): Convert from a global struct to a subclass of
1080         rtl_opt_pass along with...
1081         (pass_data_leaf_regs): ...new pass_data instance and...
1082         (make_pass_leaf_regs): ...new function.
1083         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1084         subclass of rtl_opt_pass along with...
1085         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1086         and...
1087         (make_pass_thread_prologue_and_epilogue): ...new function.
1088         (pass_match_asm_constraints): Convert from a global struct to a
1089         subclass of rtl_opt_pass along with...
1090         (pass_data_match_asm_constraints): ...new pass_data instance and...
1091         (make_pass_match_asm_constraints): ...new function.
1092         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1093         subclass of rtl_opt_pass along with...
1094         (pass_data_rtl_fwprop): ...new pass_data instance and...
1095         (make_pass_rtl_fwprop): ...new function.
1096         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1097         rtl_opt_pass along with...
1098         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1099         (make_pass_rtl_fwprop_addr): ...new function.
1100         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1101         rtl_opt_pass along with...
1102         (pass_data_rtl_pre): ...new pass_data instance and...
1103         (make_pass_rtl_pre): ...new function.
1104         (pass_rtl_hoist): Convert from a global struct to a subclass of
1105         rtl_opt_pass along with...
1106         (pass_data_rtl_hoist): ...new pass_data instance and...
1107         (make_pass_rtl_hoist): ...new function.
1108         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1109         subclass of gimple_opt_pass along with...
1110         (pass_data_lower_cf): ...new pass_data instance and...
1111         (make_pass_lower_cf): ...new function.
1112         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1113         from a global struct to a subclass of gimple_opt_pass along with...
1114         (pass_data_strength_reduction): ...new pass_data instance and...
1115         (make_pass_strength_reduction): ...new function.
1116         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1117         of rtl_opt_pass along with...
1118         (pass_data_rtl_ifcvt): ...new pass_data instance and...
1119         (make_pass_rtl_ifcvt): ...new function.
1120         (pass_if_after_combine): Convert from a global struct to a subclass of
1121         rtl_opt_pass along with...
1122         (pass_data_if_after_combine): ...new pass_data instance and...
1123         (make_pass_if_after_combine): ...new function.
1124         (pass_if_after_reload): Convert from a global struct to a subclass of
1125         rtl_opt_pass along with...
1126         (pass_data_if_after_reload): ...new pass_data instance and...
1127         (make_pass_if_after_reload): ...new function.
1128         * init-regs.c (pass_initialize_regs): Convert from a global struct to
1129         a subclass of rtl_opt_pass along with...
1130         (pass_data_initialize_regs): ...new pass_data instance and...
1131         (make_pass_initialize_regs): ...new function.
1132         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1133         of ipa_opt_pass_d along with...
1134         (pass_data_ipa_cp): ...new pass_data instance and...
1135         (make_pass_ipa_cp): ...new function.
1136         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1137         global struct to a subclass of gimple_opt_pass along with...
1138         (pass_data_inline_parameters): ...new pass_data instance and...
1139         (make_pass_inline_parameters): ...new function.
1140         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1141         subclass of gimple_opt_pass along with...
1142         (pass_data_early_inline): ...new pass_data instance and...
1143         (make_pass_early_inline): ...new function.
1144         (pass_ipa_inline): Convert from a global struct to a subclass of
1145         ipa_opt_pass_d along with...
1146         (pass_data_ipa_inline): ...new pass_data instance and...
1147         (make_pass_ipa_inline): ...new function.
1148         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1149         struct to a subclass of gimple_opt_pass along with...
1150         (pass_data_local_pure_const): ...new pass_data instance and...
1151         (make_pass_local_pure_const): ...new function.
1152         (pass_ipa_pure_const): Convert from a global struct to a subclass of
1153         ipa_opt_pass_d along with...
1154         (pass_data_ipa_pure_const): ...new pass_data instance and...
1155         (make_pass_ipa_pure_const): ...new function.
1156         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1157         to a subclass of ipa_opt_pass_d along with...
1158         (pass_data_ipa_reference): ...new pass_data instance and...
1159         (make_pass_ipa_reference): ...new function.
1160         * ipa-split.c (pass_split_functions): Convert from a global struct to
1161         a subclass of gimple_opt_pass along with...
1162         (pass_data_split_functions): ...new pass_data instance and...
1163         (make_pass_split_functions): ...new function.
1164         (pass_feedback_split_functions): Convert from a global struct to a
1165         subclass of gimple_opt_pass along with...
1166         (pass_data_feedback_split_functions): ...new pass_data instance and...
1167         (make_pass_feedback_split_functions): ...new function.
1168         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1169         global struct to a subclass of simple_ipa_opt_pass along with...
1170         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1171         instance and...
1172         (make_pass_ipa_function_and_variable_visibility): ...new function.
1173         (pass_ipa_free_inline_summary): Convert from a global struct to a
1174         subclass of simple_ipa_opt_pass along with...
1175         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1176         (make_pass_ipa_free_inline_summary): ...new function.
1177         (pass_ipa_whole_program_visibility): Convert from a global struct to a
1178         subclass of ipa_opt_pass_d along with...
1179         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1180         and...
1181         (make_pass_ipa_whole_program_visibility): ...new function.
1182         (pass_ipa_profile): Convert from a global struct to a subclass of
1183         ipa_opt_pass_d along with...
1184         (pass_data_ipa_profile): ...new pass_data instance and...
1185         (make_pass_ipa_profile): ...new function.
1186         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1187         ipa_opt_pass_d along with...
1188         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1189         (make_pass_ipa_cdtor_merge): ...new function.
1190         * ira.c (pass_ira): Convert from a global struct to a subclass of
1191         rtl_opt_pass along with...
1192         (pass_data_ira): ...new pass_data instance and...
1193         (make_pass_ira): ...new function.
1194         (pass_reload): Convert from a global struct to a subclass of
1195         rtl_opt_pass along with...
1196         (pass_data_reload): ...new pass_data instance and...
1197         (make_pass_reload): ...new function.
1198         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1199         subclass of rtl_opt_pass along with...
1200         (pass_data_cleanup_barriers): ...new pass_data instance and...
1201         (make_pass_cleanup_barriers): ...new function.
1202         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1203         of rtl_opt_pass along with...
1204         (pass_data_loop2): ...new pass_data instance and...
1205         (make_pass_loop2): ...new function.
1206         (pass_rtl_loop_init): Convert from a global struct to a subclass of
1207         rtl_opt_pass along with...
1208         (pass_data_rtl_loop_init): ...new pass_data instance and...
1209         (make_pass_rtl_loop_init): ...new function.
1210         (pass_rtl_loop_done): Convert from a global struct to a subclass of
1211         rtl_opt_pass along with...
1212         (pass_data_rtl_loop_done): ...new pass_data instance and...
1213         (make_pass_rtl_loop_done): ...new function.
1214         (pass_rtl_move_loop_invariants): Convert from a global struct to a
1215         subclass of rtl_opt_pass along with...
1216         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1217         (make_pass_rtl_move_loop_invariants): ...new function.
1218         (pass_rtl_unswitch): Convert from a global struct to a subclass of
1219         rtl_opt_pass along with...
1220         (pass_data_rtl_unswitch): ...new pass_data instance and...
1221         (make_pass_rtl_unswitch): ...new function.
1222         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1223         subclass of rtl_opt_pass along with...
1224         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1225         and...
1226         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1227         (pass_rtl_doloop): Convert from a global struct to a subclass of
1228         rtl_opt_pass along with...
1229         (pass_data_rtl_doloop): ...new pass_data instance and...
1230         (make_pass_rtl_doloop): ...new function.
1231         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1232         a subclass of rtl_opt_pass along with...
1233         (pass_data_lower_subreg): ...new pass_data instance and...
1234         (make_pass_lower_subreg): ...new function.
1235         (pass_lower_subreg2): Convert from a global struct to a subclass of
1236         rtl_opt_pass along with...
1237         (pass_data_lower_subreg2): ...new pass_data instance and...
1238         (make_pass_lower_subreg2): ...new function.
1239         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1240         struct to a subclass of ipa_opt_pass_d along with...
1241         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1242         (make_pass_ipa_lto_gimple_out): ...new function.
1243         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1244         of ipa_opt_pass_d along with...
1245         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1246         (make_pass_ipa_lto_finish_out): ...new function.
1247         * mode-switching.c (pass_mode_switching): Convert from a global struct
1248         to a subclass of rtl_opt_pass along with...
1249         (pass_data_mode_switching): ...new pass_data instance and...
1250         (make_pass_mode_switching): ...new function.
1251         * modulo-sched.c (pass_sms): Convert from a global struct to a
1252         subclass of rtl_opt_pass along with...
1253         (pass_data_sms): ...new pass_data instance and...
1254         (make_pass_sms): ...new function.
1255         * omp-low.c (pass_expand_omp): Convert from a global struct to a
1256         subclass of gimple_opt_pass along with...
1257         (pass_data_expand_omp): ...new pass_data instance and...
1258         (make_pass_expand_omp): ...new function.
1259         (pass_lower_omp): Convert from a global struct to a subclass of
1260         gimple_opt_pass along with...
1261         (pass_data_lower_omp): ...new pass_data instance and...
1262         (make_pass_lower_omp): ...new function.
1263         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1264         of gimple_opt_pass along with...
1265         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1266         (make_pass_diagnose_omp_blocks): ...new function.
1267         * passes.c (pass_early_local_passes): Convert from a global struct to
1268         a subclass of simple_ipa_opt_pass along with...
1269         (pass_data_early_local_passes): ...new pass_data instance and...
1270         (make_pass_early_local_passes): ...new function.
1271         (pass_all_early_optimizations): Convert from a global struct to a
1272         subclass of gimple_opt_pass along with...
1273         (pass_data_all_early_optimizations): ...new pass_data instance and...
1274         (make_pass_all_early_optimizations): ...new function.
1275         (pass_all_optimizations): Convert from a global struct to a subclass
1276         of gimple_opt_pass along with...
1277         (pass_data_all_optimizations): ...new pass_data instance and...
1278         (make_pass_all_optimizations): ...new function.
1279         (pass_all_optimizations_g): Convert from a global struct to a subclass
1280         of gimple_opt_pass along with...
1281         (pass_data_all_optimizations_g): ...new pass_data instance and...
1282         (make_pass_all_optimizations_g): ...new function.
1283         (pass_rest_of_compilation): Convert from a global struct to a subclass
1284         of rtl_opt_pass along with...
1285         (pass_data_rest_of_compilation): ...new pass_data instance and...
1286         (make_pass_rest_of_compilation): ...new function.
1287         (pass_postreload): Convert from a global struct to a subclass of
1288         rtl_opt_pass along with...
1289         (pass_data_postreload): ...new pass_data instance and...
1290         (make_pass_postreload): ...new function.
1291         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1292         subclass of rtl_opt_pass along with...
1293         (pass_data_gcse2): ...new pass_data instance and...
1294         (make_pass_gcse2): ...new function.
1295         * postreload.c (pass_postreload_cse): Convert from a global struct to
1296         a subclass of rtl_opt_pass along with...
1297         (pass_data_postreload_cse): ...new pass_data instance and...
1298         (make_pass_postreload_cse): ...new function.
1299         * predict.c (pass_profile): Convert from a global struct to a subclass
1300         of gimple_opt_pass along with...
1301         (pass_data_profile): ...new pass_data instance and...
1302         (make_pass_profile): ...new function.
1303         (pass_strip_predict_hints): Convert from a global struct to a subclass
1304         of gimple_opt_pass along with...
1305         (pass_data_strip_predict_hints): ...new pass_data instance and...
1306         (make_pass_strip_predict_hints): ...new function.
1307         * recog.c (pass_peephole2): Convert from a global struct to a subclass
1308         of rtl_opt_pass along with...
1309         (pass_data_peephole2): ...new pass_data instance and...
1310         (make_pass_peephole2): ...new function.
1311         (pass_split_all_insns): Convert from a global struct to a subclass of
1312         rtl_opt_pass along with...
1313         (pass_data_split_all_insns): ...new pass_data instance and...
1314         (make_pass_split_all_insns): ...new function.
1315         (pass_split_after_reload): Convert from a global struct to a subclass
1316         of rtl_opt_pass along with...
1317         (pass_data_split_after_reload): ...new pass_data instance and...
1318         (make_pass_split_after_reload): ...new function.
1319         (pass_split_before_regstack): Convert from a global struct to a
1320         subclass of rtl_opt_pass along with...
1321         (pass_data_split_before_regstack): ...new pass_data instance and...
1322         (make_pass_split_before_regstack): ...new function.
1323         (pass_split_before_sched2): Convert from a global struct to a subclass
1324         of rtl_opt_pass along with...
1325         (pass_data_split_before_sched2): ...new pass_data instance and...
1326         (make_pass_split_before_sched2): ...new function.
1327         (pass_split_for_shorten_branches): Convert from a global struct to a
1328         subclass of rtl_opt_pass along with...
1329         (pass_data_split_for_shorten_branches): ...new pass_data instance
1330         and...
1331         (make_pass_split_for_shorten_branches): ...new function.
1332         * ree.c (pass_ree): Convert from a global struct to a subclass of
1333         rtl_opt_pass along with...
1334         (pass_data_ree): ...new pass_data instance and...
1335         (make_pass_ree): ...new function.
1336         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1337         subclass of rtl_opt_pass along with...
1338         (pass_data_stack_regs): ...new pass_data instance and...
1339         (make_pass_stack_regs): ...new function.
1340         (pass_stack_regs_run): Convert from a global struct to a subclass of
1341         rtl_opt_pass along with...
1342         (pass_data_stack_regs_run): ...new pass_data instance and...
1343         (make_pass_stack_regs_run): ...new function.
1344         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1345         subclass of rtl_opt_pass along with...
1346         (pass_data_cprop_hardreg): ...new pass_data instance and...
1347         (make_pass_cprop_hardreg): ...new function.
1348         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1349         subclass of rtl_opt_pass along with...
1350         (pass_data_reginfo_init): ...new pass_data instance and...
1351         (make_pass_reginfo_init): ...new function.
1352         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1353         of rtl_opt_pass along with...
1354         (pass_data_regmove): ...new pass_data instance and...
1355         (make_pass_regmove): ...new function.
1356         * regrename.c (pass_regrename): Convert from a global struct to a
1357         subclass of rtl_opt_pass along with...
1358         (pass_data_regrename): ...new pass_data instance and...
1359         (make_pass_regrename): ...new function.
1360         * reorg.c (pass_delay_slots): Convert from a global struct to a
1361         subclass of rtl_opt_pass along with...
1362         (pass_data_delay_slots): ...new pass_data instance and...
1363         (make_pass_delay_slots): ...new function.
1364         (pass_machine_reorg): Convert from a global struct to a subclass of
1365         rtl_opt_pass along with...
1366         (pass_data_machine_reorg): ...new pass_data instance and...
1367         (make_pass_machine_reorg): ...new function.
1368         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1369         of rtl_opt_pass along with...
1370         (pass_data_sched): ...new pass_data instance and...
1371         (make_pass_sched): ...new function.
1372         (pass_sched2): Convert from a global struct to a subclass of
1373         rtl_opt_pass along with...
1374         (pass_data_sched2): ...new pass_data instance and...
1375         (make_pass_sched2): ...new function.
1376         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1377         to a subclass of rtl_opt_pass along with...
1378         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1379         (make_pass_stack_ptr_mod): ...new function.
1380         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1381         to a subclass of rtl_opt_pass along with...
1382         (pass_data_rtl_store_motion): ...new pass_data instance and...
1383         (make_pass_rtl_store_motion): ...new function.
1384         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1385         of gimple_opt_pass along with...
1386         (pass_data_tracer): ...new pass_data instance and...
1387         (make_pass_tracer): ...new function.
1388         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1389         to a subclass of gimple_opt_pass along with...
1390         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1391         (make_pass_diagnose_tm_blocks): ...new function.
1392         (pass_lower_tm): Convert from a global struct to a subclass of
1393         gimple_opt_pass along with...
1394         (pass_data_lower_tm): ...new pass_data instance and...
1395         (make_pass_lower_tm): ...new function.
1396         (pass_tm_init): Convert from a global struct to a subclass of
1397         gimple_opt_pass along with...
1398         (pass_data_tm_init): ...new pass_data instance and...
1399         (make_pass_tm_init): ...new function.
1400         (pass_tm_mark): Convert from a global struct to a subclass of
1401         gimple_opt_pass along with...
1402         (pass_data_tm_mark): ...new pass_data instance and...
1403         (make_pass_tm_mark): ...new function.
1404         (pass_tm_edges): Convert from a global struct to a subclass of
1405         gimple_opt_pass along with...
1406         (pass_data_tm_edges): ...new pass_data instance and...
1407         (make_pass_tm_edges): ...new function.
1408         (pass_tm_memopt): Convert from a global struct to a subclass of
1409         gimple_opt_pass along with...
1410         (pass_data_tm_memopt): ...new pass_data instance and...
1411         (make_pass_tm_memopt): ...new function.
1412         (pass_ipa_tm): Convert from a global struct to a subclass of
1413         simple_ipa_opt_pass along with...
1414         (pass_data_ipa_tm): ...new pass_data instance and...
1415         (make_pass_ipa_tm): ...new function.
1416         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1417         subclass of gimple_opt_pass along with...
1418         (pass_data_call_cdce): ...new pass_data instance and...
1419         (make_pass_call_cdce): ...new function.
1420         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1421         subclass of gimple_opt_pass along with...
1422         (pass_data_build_cfg): ...new pass_data instance and...
1423         (make_pass_build_cfg): ...new function.
1424         (pass_split_crit_edges): Convert from a global struct to a subclass of
1425         gimple_opt_pass along with...
1426         (pass_data_split_crit_edges): ...new pass_data instance and...
1427         (make_pass_split_crit_edges): ...new function.
1428         (pass_warn_function_return): Convert from a global struct to a
1429         subclass of gimple_opt_pass along with...
1430         (pass_data_warn_function_return): ...new pass_data instance and...
1431         (make_pass_warn_function_return): ...new function.
1432         (pass_warn_function_noreturn): Convert from a global struct to a
1433         subclass of gimple_opt_pass along with...
1434         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1435         (make_pass_warn_function_noreturn): ...new function.
1436         (pass_warn_unused_result): Convert from a global struct to a subclass
1437         of gimple_opt_pass along with...
1438         (pass_data_warn_unused_result): ...new pass_data instance and...
1439         (make_pass_warn_unused_result): ...new function.
1440         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1441         a subclass of gimple_opt_pass along with...
1442         (pass_data_merge_phi): ...new pass_data instance and...
1443         (make_pass_merge_phi): ...new function.
1444         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1445         a subclass of gimple_opt_pass along with...
1446         (pass_data_lower_complex): ...new pass_data instance and...
1447         (make_pass_lower_complex): ...new function.
1448         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1449         gimple_opt_pass along with...
1450         (pass_data_lower_complex_O0): ...new pass_data instance and...
1451         (make_pass_lower_complex_O0): ...new function.
1452         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1453         subclass of gimple_opt_pass along with...
1454         (pass_data_lower_eh): ...new pass_data instance and...
1455         (make_pass_lower_eh): ...new function.
1456         (pass_refactor_eh): Convert from a global struct to a subclass of
1457         gimple_opt_pass along with...
1458         (pass_data_refactor_eh): ...new pass_data instance and...
1459         (make_pass_refactor_eh): ...new function.
1460         (pass_lower_resx): Convert from a global struct to a subclass of
1461         gimple_opt_pass along with...
1462         (pass_data_lower_resx): ...new pass_data instance and...
1463         (make_pass_lower_resx): ...new function.
1464         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1465         of gimple_opt_pass along with...
1466         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1467         (make_pass_lower_eh_dispatch): ...new function.
1468         (pass_cleanup_eh): Convert from a global struct to a subclass of
1469         gimple_opt_pass along with...
1470         (pass_data_cleanup_eh): ...new pass_data instance and...
1471         (make_pass_cleanup_eh): ...new function.
1472         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1473         to a subclass of simple_ipa_opt_pass along with...
1474         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1475         (make_pass_ipa_lower_emutls): ...new function.
1476         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1477         a subclass of gimple_opt_pass along with...
1478         (pass_data_if_conversion): ...new pass_data instance and...
1479         (make_pass_if_conversion): ...new function.
1480         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1481         subclass of gimple_opt_pass along with...
1482         (pass_data_build_ssa): ...new pass_data instance and...
1483         (make_pass_build_ssa): ...new function.
1484         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1485         global struct to a subclass of gimple_opt_pass along with...
1486         (pass_data_loop_distribution): ...new pass_data instance and...
1487         (make_pass_loop_distribution): ...new function.
1488         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1489         subclass of gimple_opt_pass along with...
1490         (pass_data_mudflap_1): ...new pass_data instance and...
1491         (make_pass_mudflap_1): ...new function.
1492         (pass_mudflap_2): Convert from a global struct to a subclass of
1493         gimple_opt_pass along with...
1494         (pass_data_mudflap_2): ...new pass_data instance and...
1495         (make_pass_mudflap_2): ...new function.
1496         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1497         subclass of gimple_opt_pass along with...
1498         (pass_data_mudflap_1): ...new pass_data instance and...
1499         (make_pass_mudflap_1): ...new function.
1500         (pass_mudflap_2): Convert from a global struct to a subclass of
1501         gimple_opt_pass along with...
1502         (pass_data_mudflap_2): ...new pass_data instance and...
1503         (make_pass_mudflap_2): ...new function.
1504         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1505         gimple_opt_pass along with...
1506         (pass_data_nrv): ...new pass_data instance and...
1507         (make_pass_nrv): ...new function.
1508         (pass_return_slot): Convert from a global struct to a subclass of
1509         gimple_opt_pass along with...
1510         (pass_data_return_slot): ...new pass_data instance and...
1511         (make_pass_return_slot): ...new function.
1512         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1513         to a subclass of gimple_opt_pass along with...
1514         (pass_data_object_sizes): ...new pass_data instance and...
1515         (make_pass_object_sizes): ...new function.
1516         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1517         global struct to a subclass of gimple_opt_pass along with...
1518         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1519         and...
1520         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1521         (pass_fixup_cfg): Convert from a global struct to a subclass of
1522         gimple_opt_pass along with...
1523         (pass_data_fixup_cfg): ...new pass_data instance and...
1524         (make_pass_fixup_cfg): ...new function.
1525         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1526         (make_pass_mudflap_1): ...new function.
1527         (pass_mudflap_2): Replace declaration with that of...
1528         (make_pass_mudflap_2): ...new function.
1529         (pass_asan): Replace declaration with that of...
1530         (make_pass_asan): ...new function.
1531         (pass_asan_O0): Replace declaration with that of...
1532         (make_pass_asan_O0): ...new function.
1533         (pass_tsan): Replace declaration with that of...
1534         (make_pass_tsan): ...new function.
1535         (pass_tsan_O0): Replace declaration with that of...
1536         (make_pass_tsan_O0): ...new function.
1537         (pass_lower_cf): Replace declaration with that of...
1538         (make_pass_lower_cf): ...new function.
1539         (pass_refactor_eh): Replace declaration with that of...
1540         (make_pass_refactor_eh): ...new function.
1541         (pass_lower_eh): Replace declaration with that of...
1542         (make_pass_lower_eh): ...new function.
1543         (pass_lower_eh_dispatch): Replace declaration with that of...
1544         (make_pass_lower_eh_dispatch): ...new function.
1545         (pass_lower_resx): Replace declaration with that of...
1546         (make_pass_lower_resx): ...new function.
1547         (pass_build_cfg): Replace declaration with that of...
1548         (make_pass_build_cfg): ...new function.
1549         (pass_early_tree_profile): Replace declaration with that of...
1550         (make_pass_early_tree_profile): ...new function.
1551         (pass_cleanup_eh): Replace declaration with that of...
1552         (make_pass_cleanup_eh): ...new function.
1553         (pass_sra): Replace declaration with that of...
1554         (make_pass_sra): ...new function.
1555         (pass_sra_early): Replace declaration with that of...
1556         (make_pass_sra_early): ...new function.
1557         (pass_early_ipa_sra): Replace declaration with that of...
1558         (make_pass_early_ipa_sra): ...new function.
1559         (pass_tail_recursion): Replace declaration with that of...
1560         (make_pass_tail_recursion): ...new function.
1561         (pass_tail_calls): Replace declaration with that of...
1562         (make_pass_tail_calls): ...new function.
1563         (pass_tree_loop): Replace declaration with that of...
1564         (make_pass_tree_loop): ...new function.
1565         (pass_tree_loop_init): Replace declaration with that of...
1566         (make_pass_tree_loop_init): ...new function.
1567         (pass_lim): Replace declaration with that of...
1568         (make_pass_lim): ...new function.
1569         (pass_tree_unswitch): Replace declaration with that of...
1570         (make_pass_tree_unswitch): ...new function.
1571         (pass_predcom): Replace declaration with that of...
1572         (make_pass_predcom): ...new function.
1573         (pass_iv_canon): Replace declaration with that of...
1574         (make_pass_iv_canon): ...new function.
1575         (pass_scev_cprop): Replace declaration with that of...
1576         (make_pass_scev_cprop): ...new function.
1577         (pass_empty_loop): Replace declaration with that of...
1578         (make_pass_empty_loop): ...new function.
1579         (pass_record_bounds): Replace declaration with that of...
1580         (make_pass_record_bounds): ...new function.
1581         (pass_graphite): Replace declaration with that of...
1582         (make_pass_graphite): ...new function.
1583         (pass_graphite_transforms): Replace declaration with that of...
1584         (make_pass_graphite_transforms): ...new function.
1585         (pass_if_conversion): Replace declaration with that of...
1586         (make_pass_if_conversion): ...new function.
1587         (pass_loop_distribution): Replace declaration with that of...
1588         (make_pass_loop_distribution): ...new function.
1589         (pass_vectorize): Replace declaration with that of...
1590         (make_pass_vectorize): ...new function.
1591         (pass_slp_vectorize): Replace declaration with that of...
1592         (make_pass_slp_vectorize): ...new function.
1593         (pass_complete_unroll): Replace declaration with that of...
1594         (make_pass_complete_unroll): ...new function.
1595         (pass_complete_unrolli): Replace declaration with that of...
1596         (make_pass_complete_unrolli): ...new function.
1597         (pass_parallelize_loops): Replace declaration with that of...
1598         (make_pass_parallelize_loops): ...new function.
1599         (pass_loop_prefetch): Replace declaration with that of...
1600         (make_pass_loop_prefetch): ...new function.
1601         (pass_iv_optimize): Replace declaration with that of...
1602         (make_pass_iv_optimize): ...new function.
1603         (pass_tree_loop_done): Replace declaration with that of...
1604         (make_pass_tree_loop_done): ...new function.
1605         (pass_ch): Replace declaration with that of...
1606         (make_pass_ch): ...new function.
1607         (pass_ccp): Replace declaration with that of...
1608         (make_pass_ccp): ...new function.
1609         (pass_phi_only_cprop): Replace declaration with that of...
1610         (make_pass_phi_only_cprop): ...new function.
1611         (pass_build_ssa): Replace declaration with that of...
1612         (make_pass_build_ssa): ...new function.
1613         (pass_build_alias): Replace declaration with that of...
1614         (make_pass_build_alias): ...new function.
1615         (pass_build_ealias): Replace declaration with that of...
1616         (make_pass_build_ealias): ...new function.
1617         (pass_dominator): Replace declaration with that of...
1618         (make_pass_dominator): ...new function.
1619         (pass_dce): Replace declaration with that of...
1620         (make_pass_dce): ...new function.
1621         (pass_dce_loop): Replace declaration with that of...
1622         (make_pass_dce_loop): ...new function.
1623         (pass_cd_dce): Replace declaration with that of...
1624         (make_pass_cd_dce): ...new function.
1625         (pass_call_cdce): Replace declaration with that of...
1626         (make_pass_call_cdce): ...new function.
1627         (pass_merge_phi): Replace declaration with that of...
1628         (make_pass_merge_phi): ...new function.
1629         (pass_split_crit_edges): Replace declaration with that of...
1630         (make_pass_split_crit_edges): ...new function.
1631         (pass_pre): Replace declaration with that of...
1632         (make_pass_pre): ...new function.
1633         (pass_profile): Replace declaration with that of...
1634         (make_pass_profile): ...new function.
1635         (pass_strip_predict_hints): Replace declaration with that of...
1636         (make_pass_strip_predict_hints): ...new function.
1637         (pass_lower_complex_O0): Replace declaration with that of...
1638         (make_pass_lower_complex_O0): ...new function.
1639         (pass_lower_complex): Replace declaration with that of...
1640         (make_pass_lower_complex): ...new function.
1641         (pass_lower_vector): Replace declaration with that of...
1642         (make_pass_lower_vector): ...new function.
1643         (pass_lower_vector_ssa): Replace declaration with that of...
1644         (make_pass_lower_vector_ssa): ...new function.
1645         (pass_lower_omp): Replace declaration with that of...
1646         (make_pass_lower_omp): ...new function.
1647         (pass_diagnose_omp_blocks): Replace declaration with that of...
1648         (make_pass_diagnose_omp_blocks): ...new function.
1649         (pass_expand_omp): Replace declaration with that of...
1650         (make_pass_expand_omp): ...new function.
1651         (pass_expand_omp_ssa): Replace declaration with that of...
1652         (make_pass_expand_omp_ssa): ...new function.
1653         (pass_object_sizes): Replace declaration with that of...
1654         (make_pass_object_sizes): ...new function.
1655         (pass_strlen): Replace declaration with that of...
1656         (make_pass_strlen): ...new function.
1657         (pass_fold_builtins): Replace declaration with that of...
1658         (make_pass_fold_builtins): ...new function.
1659         (pass_stdarg): Replace declaration with that of...
1660         (make_pass_stdarg): ...new function.
1661         (pass_early_warn_uninitialized): Replace declaration with that of...
1662         (make_pass_early_warn_uninitialized): ...new function.
1663         (pass_late_warn_uninitialized): Replace declaration with that of...
1664         (make_pass_late_warn_uninitialized): ...new function.
1665         (pass_cse_reciprocals): Replace declaration with that of...
1666         (make_pass_cse_reciprocals): ...new function.
1667         (pass_cse_sincos): Replace declaration with that of...
1668         (make_pass_cse_sincos): ...new function.
1669         (pass_optimize_bswap): Replace declaration with that of...
1670         (make_pass_optimize_bswap): ...new function.
1671         (pass_optimize_widening_mul): Replace declaration with that of...
1672         (make_pass_optimize_widening_mul): ...new function.
1673         (pass_warn_function_return): Replace declaration with that of...
1674         (make_pass_warn_function_return): ...new function.
1675         (pass_warn_function_noreturn): Replace declaration with that of...
1676         (make_pass_warn_function_noreturn): ...new function.
1677         (pass_cselim): Replace declaration with that of...
1678         (make_pass_cselim): ...new function.
1679         (pass_phiopt): Replace declaration with that of...
1680         (make_pass_phiopt): ...new function.
1681         (pass_forwprop): Replace declaration with that of...
1682         (make_pass_forwprop): ...new function.
1683         (pass_phiprop): Replace declaration with that of...
1684         (make_pass_phiprop): ...new function.
1685         (pass_tree_ifcombine): Replace declaration with that of...
1686         (make_pass_tree_ifcombine): ...new function.
1687         (pass_dse): Replace declaration with that of...
1688         (make_pass_dse): ...new function.
1689         (pass_nrv): Replace declaration with that of...
1690         (make_pass_nrv): ...new function.
1691         (pass_rename_ssa_copies): Replace declaration with that of...
1692         (make_pass_rename_ssa_copies): ...new function.
1693         (pass_sink_code): Replace declaration with that of...
1694         (make_pass_sink_code): ...new function.
1695         (pass_fre): Replace declaration with that of...
1696         (make_pass_fre): ...new function.
1697         (pass_check_data_deps): Replace declaration with that of...
1698         (make_pass_check_data_deps): ...new function.
1699         (pass_copy_prop): Replace declaration with that of...
1700         (make_pass_copy_prop): ...new function.
1701         (pass_vrp): Replace declaration with that of...
1702         (make_pass_vrp): ...new function.
1703         (pass_uncprop): Replace declaration with that of...
1704         (make_pass_uncprop): ...new function.
1705         (pass_return_slot): Replace declaration with that of...
1706         (make_pass_return_slot): ...new function.
1707         (pass_reassoc): Replace declaration with that of...
1708         (make_pass_reassoc): ...new function.
1709         (pass_rebuild_cgraph_edges): Replace declaration with that of...
1710         (make_pass_rebuild_cgraph_edges): ...new function.
1711         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1712         (make_pass_remove_cgraph_callee_edges): ...new function.
1713         (pass_build_cgraph_edges): Replace declaration with that of...
1714         (make_pass_build_cgraph_edges): ...new function.
1715         (pass_local_pure_const): Replace declaration with that of...
1716         (make_pass_local_pure_const): ...new function.
1717         (pass_tracer): Replace declaration with that of...
1718         (make_pass_tracer): ...new function.
1719         (pass_warn_unused_result): Replace declaration with that of...
1720         (make_pass_warn_unused_result): ...new function.
1721         (pass_diagnose_tm_blocks): Replace declaration with that of...
1722         (make_pass_diagnose_tm_blocks): ...new function.
1723         (pass_lower_tm): Replace declaration with that of...
1724         (make_pass_lower_tm): ...new function.
1725         (pass_tm_init): Replace declaration with that of...
1726         (make_pass_tm_init): ...new function.
1727         (pass_tm_mark): Replace declaration with that of...
1728         (make_pass_tm_mark): ...new function.
1729         (pass_tm_memopt): Replace declaration with that of...
1730         (make_pass_tm_memopt): ...new function.
1731         (pass_tm_edges): Replace declaration with that of...
1732         (make_pass_tm_edges): ...new function.
1733         (pass_split_functions): Replace declaration with that of...
1734         (make_pass_split_functions): ...new function.
1735         (pass_feedback_split_functions): Replace declaration with that of...
1736         (make_pass_feedback_split_functions): ...new function.
1737         (pass_strength_reduction): Replace declaration with that of...
1738         (make_pass_strength_reduction): ...new function.
1739         (pass_ipa_lower_emutls): Replace declaration with that of...
1740         (make_pass_ipa_lower_emutls): ...new function.
1741         (pass_ipa_function_and_variable_visibility): Replace declaration with
1742         that of...
1743         (make_pass_ipa_function_and_variable_visibility): ...new function.
1744         (pass_ipa_tree_profile): Replace declaration with that of...
1745         (make_pass_ipa_tree_profile): ...new function.
1746         (pass_early_local_passes): Replace declaration with that of...
1747         (make_pass_early_local_passes): ...new function.
1748         (pass_ipa_whole_program_visibility): Replace declaration with that
1749         of...
1750         (make_pass_ipa_whole_program_visibility): ...new function.
1751         (pass_ipa_lto_gimple_out): Replace declaration with that of...
1752         (make_pass_ipa_lto_gimple_out): ...new function.
1753         (pass_ipa_increase_alignment): Replace declaration with that of...
1754         (make_pass_ipa_increase_alignment): ...new function.
1755         (pass_ipa_inline): Replace declaration with that of...
1756         (make_pass_ipa_inline): ...new function.
1757         (pass_ipa_free_lang_data): Replace declaration with that of...
1758         (make_pass_ipa_free_lang_data): ...new function.
1759         (pass_ipa_free_inline_summary): Replace declaration with that of...
1760         (make_pass_ipa_free_inline_summary): ...new function.
1761         (pass_ipa_cp): Replace declaration with that of...
1762         (make_pass_ipa_cp): ...new function.
1763         (pass_ipa_reference): Replace declaration with that of...
1764         (make_pass_ipa_reference): ...new function.
1765         (pass_ipa_pure_const): Replace declaration with that of...
1766         (make_pass_ipa_pure_const): ...new function.
1767         (pass_ipa_pta): Replace declaration with that of...
1768         (make_pass_ipa_pta): ...new function.
1769         (pass_ipa_lto_finish_out): Replace declaration with that of...
1770         (make_pass_ipa_lto_finish_out): ...new function.
1771         (pass_ipa_tm): Replace declaration with that of...
1772         (make_pass_ipa_tm): ...new function.
1773         (pass_ipa_profile): Replace declaration with that of...
1774         (make_pass_ipa_profile): ...new function.
1775         (pass_ipa_cdtor_merge): Replace declaration with that of...
1776         (make_pass_ipa_cdtor_merge): ...new function.
1777         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1778         of...
1779         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1780         (pass_init_datastructures): Replace declaration with that of...
1781         (make_pass_init_datastructures): ...new function.
1782         (pass_fixup_cfg): Replace declaration with that of...
1783         (make_pass_fixup_cfg): ...new function.
1784         (pass_expand): Replace declaration with that of...
1785         (make_pass_expand): ...new function.
1786         (pass_instantiate_virtual_regs): Replace declaration with that of...
1787         (make_pass_instantiate_virtual_regs): ...new function.
1788         (pass_rtl_fwprop): Replace declaration with that of...
1789         (make_pass_rtl_fwprop): ...new function.
1790         (pass_rtl_fwprop_addr): Replace declaration with that of...
1791         (make_pass_rtl_fwprop_addr): ...new function.
1792         (pass_jump): Replace declaration with that of...
1793         (make_pass_jump): ...new function.
1794         (pass_jump2): Replace declaration with that of...
1795         (make_pass_jump2): ...new function.
1796         (pass_lower_subreg): Replace declaration with that of...
1797         (make_pass_lower_subreg): ...new function.
1798         (pass_cse): Replace declaration with that of...
1799         (make_pass_cse): ...new function.
1800         (pass_fast_rtl_dce): Replace declaration with that of...
1801         (make_pass_fast_rtl_dce): ...new function.
1802         (pass_ud_rtl_dce): Replace declaration with that of...
1803         (make_pass_ud_rtl_dce): ...new function.
1804         (pass_rtl_dce): Replace declaration with that of...
1805         (make_pass_rtl_dce): ...new function.
1806         (pass_rtl_dse1): Replace declaration with that of...
1807         (make_pass_rtl_dse1): ...new function.
1808         (pass_rtl_dse2): Replace declaration with that of...
1809         (make_pass_rtl_dse2): ...new function.
1810         (pass_rtl_dse3): Replace declaration with that of...
1811         (make_pass_rtl_dse3): ...new function.
1812         (pass_rtl_cprop): Replace declaration with that of...
1813         (make_pass_rtl_cprop): ...new function.
1814         (pass_rtl_pre): Replace declaration with that of...
1815         (make_pass_rtl_pre): ...new function.
1816         (pass_rtl_hoist): Replace declaration with that of...
1817         (make_pass_rtl_hoist): ...new function.
1818         (pass_rtl_store_motion): Replace declaration with that of...
1819         (make_pass_rtl_store_motion): ...new function.
1820         (pass_cse_after_global_opts): Replace declaration with that of...
1821         (make_pass_cse_after_global_opts): ...new function.
1822         (pass_rtl_ifcvt): Replace declaration with that of...
1823         (make_pass_rtl_ifcvt): ...new function.
1824         (pass_into_cfg_layout_mode): Replace declaration with that of...
1825         (make_pass_into_cfg_layout_mode): ...new function.
1826         (pass_outof_cfg_layout_mode): Replace declaration with that of...
1827         (make_pass_outof_cfg_layout_mode): ...new function.
1828         (pass_loop2): Replace declaration with that of...
1829         (make_pass_loop2): ...new function.
1830         (pass_rtl_loop_init): Replace declaration with that of...
1831         (make_pass_rtl_loop_init): ...new function.
1832         (pass_rtl_move_loop_invariants): Replace declaration with that of...
1833         (make_pass_rtl_move_loop_invariants): ...new function.
1834         (pass_rtl_unswitch): Replace declaration with that of...
1835         (make_pass_rtl_unswitch): ...new function.
1836         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1837         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1838         (pass_rtl_doloop): Replace declaration with that of...
1839         (make_pass_rtl_doloop): ...new function.
1840         (pass_rtl_loop_done): Replace declaration with that of...
1841         (make_pass_rtl_loop_done): ...new function.
1842         (pass_web): Replace declaration with that of...
1843         (make_pass_web): ...new function.
1844         (pass_cse2): Replace declaration with that of...
1845         (make_pass_cse2): ...new function.
1846         (pass_df_initialize_opt): Replace declaration with that of...
1847         (make_pass_df_initialize_opt): ...new function.
1848         (pass_df_initialize_no_opt): Replace declaration with that of...
1849         (make_pass_df_initialize_no_opt): ...new function.
1850         (pass_reginfo_init): Replace declaration with that of...
1851         (make_pass_reginfo_init): ...new function.
1852         (pass_inc_dec): Replace declaration with that of...
1853         (make_pass_inc_dec): ...new function.
1854         (pass_stack_ptr_mod): Replace declaration with that of...
1855         (make_pass_stack_ptr_mod): ...new function.
1856         (pass_initialize_regs): Replace declaration with that of...
1857         (make_pass_initialize_regs): ...new function.
1858         (pass_combine): Replace declaration with that of...
1859         (make_pass_combine): ...new function.
1860         (pass_if_after_combine): Replace declaration with that of...
1861         (make_pass_if_after_combine): ...new function.
1862         (pass_ree): Replace declaration with that of...
1863         (make_pass_ree): ...new function.
1864         (pass_partition_blocks): Replace declaration with that of...
1865         (make_pass_partition_blocks): ...new function.
1866         (pass_match_asm_constraints): Replace declaration with that of...
1867         (make_pass_match_asm_constraints): ...new function.
1868         (pass_regmove): Replace declaration with that of...
1869         (make_pass_regmove): ...new function.
1870         (pass_split_all_insns): Replace declaration with that of...
1871         (make_pass_split_all_insns): ...new function.
1872         (pass_fast_rtl_byte_dce): Replace declaration with that of...
1873         (make_pass_fast_rtl_byte_dce): ...new function.
1874         (pass_lower_subreg2): Replace declaration with that of...
1875         (make_pass_lower_subreg2): ...new function.
1876         (pass_mode_switching): Replace declaration with that of...
1877         (make_pass_mode_switching): ...new function.
1878         (pass_sms): Replace declaration with that of...
1879         (make_pass_sms): ...new function.
1880         (pass_sched): Replace declaration with that of...
1881         (make_pass_sched): ...new function.
1882         (pass_ira): Replace declaration with that of...
1883         (make_pass_ira): ...new function.
1884         (pass_reload): Replace declaration with that of...
1885         (make_pass_reload): ...new function.
1886         (pass_clean_state): Replace declaration with that of...
1887         (make_pass_clean_state): ...new function.
1888         (pass_branch_prob): Replace declaration with that of...
1889         (make_pass_branch_prob): ...new function.
1890         (pass_value_profile_transformations): Replace declaration with that
1891         of...
1892         (make_pass_value_profile_transformations): ...new function.
1893         (pass_postreload_cse): Replace declaration with that of...
1894         (make_pass_postreload_cse): ...new function.
1895         (pass_gcse2): Replace declaration with that of...
1896         (make_pass_gcse2): ...new function.
1897         (pass_split_after_reload): Replace declaration with that of...
1898         (make_pass_split_after_reload): ...new function.
1899         (pass_branch_target_load_optimize1): Replace declaration with that
1900         of...
1901         (make_pass_branch_target_load_optimize1): ...new function.
1902         (pass_thread_prologue_and_epilogue): Replace declaration with that
1903         of...
1904         (make_pass_thread_prologue_and_epilogue): ...new function.
1905         (pass_stack_adjustments): Replace declaration with that of...
1906         (make_pass_stack_adjustments): ...new function.
1907         (pass_peephole2): Replace declaration with that of...
1908         (make_pass_peephole2): ...new function.
1909         (pass_if_after_reload): Replace declaration with that of...
1910         (make_pass_if_after_reload): ...new function.
1911         (pass_regrename): Replace declaration with that of...
1912         (make_pass_regrename): ...new function.
1913         (pass_cprop_hardreg): Replace declaration with that of...
1914         (make_pass_cprop_hardreg): ...new function.
1915         (pass_reorder_blocks): Replace declaration with that of...
1916         (make_pass_reorder_blocks): ...new function.
1917         (pass_branch_target_load_optimize2): Replace declaration with that
1918         of...
1919         (make_pass_branch_target_load_optimize2): ...new function.
1920         (pass_leaf_regs): Replace declaration with that of...
1921         (make_pass_leaf_regs): ...new function.
1922         (pass_split_before_sched2): Replace declaration with that of...
1923         (make_pass_split_before_sched2): ...new function.
1924         (pass_compare_elim_after_reload): Replace declaration with that of...
1925         (make_pass_compare_elim_after_reload): ...new function.
1926         (pass_sched2): Replace declaration with that of...
1927         (make_pass_sched2): ...new function.
1928         (pass_stack_regs): Replace declaration with that of...
1929         (make_pass_stack_regs): ...new function.
1930         (pass_stack_regs_run): Replace declaration with that of...
1931         (make_pass_stack_regs_run): ...new function.
1932         (pass_df_finish): Replace declaration with that of...
1933         (make_pass_df_finish): ...new function.
1934         (pass_compute_alignments): Replace declaration with that of...
1935         (make_pass_compute_alignments): ...new function.
1936         (pass_duplicate_computed_gotos): Replace declaration with that of...
1937         (make_pass_duplicate_computed_gotos): ...new function.
1938         (pass_variable_tracking): Replace declaration with that of...
1939         (make_pass_variable_tracking): ...new function.
1940         (pass_free_cfg): Replace declaration with that of...
1941         (make_pass_free_cfg): ...new function.
1942         (pass_machine_reorg): Replace declaration with that of...
1943         (make_pass_machine_reorg): ...new function.
1944         (pass_cleanup_barriers): Replace declaration with that of...
1945         (make_pass_cleanup_barriers): ...new function.
1946         (pass_delay_slots): Replace declaration with that of...
1947         (make_pass_delay_slots): ...new function.
1948         (pass_split_for_shorten_branches): Replace declaration with that of...
1949         (make_pass_split_for_shorten_branches): ...new function.
1950         (pass_split_before_regstack): Replace declaration with that of...
1951         (make_pass_split_before_regstack): ...new function.
1952         (pass_convert_to_eh_region_ranges): Replace declaration with that
1953         of...
1954         (make_pass_convert_to_eh_region_ranges): ...new function.
1955         (pass_shorten_branches): Replace declaration with that of...
1956         (make_pass_shorten_branches): ...new function.
1957         (pass_set_nothrow_function_flags): Replace declaration with that of...
1958         (make_pass_set_nothrow_function_flags): ...new function.
1959         (pass_dwarf2_frame): Replace declaration with that of...
1960         (make_pass_dwarf2_frame): ...new function.
1961         (pass_final): Replace declaration with that of...
1962         (make_pass_final): ...new function.
1963         (pass_rtl_seqabstr): Replace declaration with that of...
1964         (make_pass_rtl_seqabstr): ...new function.
1965         (pass_release_ssa_names): Replace declaration with that of...
1966         (make_pass_release_ssa_names): ...new function.
1967         (pass_early_inline): Replace declaration with that of...
1968         (make_pass_early_inline): ...new function.
1969         (pass_inline_parameters): Replace declaration with that of...
1970         (make_pass_inline_parameters): ...new function.
1971         (pass_update_address_taken): Replace declaration with that of...
1972         (make_pass_update_address_taken): ...new function.
1973         (pass_convert_switch): Replace declaration with that of...
1974         (make_pass_convert_switch): ...new function.
1975         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1976         to a subclass of simple_ipa_opt_pass along with...
1977         (pass_data_ipa_tree_profile): ...new pass_data instance and...
1978         (make_pass_ipa_tree_profile): ...new function.
1979         * tree-sra.c (pass_sra_early): Convert from a global struct to a
1980         subclass of gimple_opt_pass along with...
1981         (pass_data_sra_early): ...new pass_data instance and...
1982         (make_pass_sra_early): ...new function.
1983         (pass_sra): Convert from a global struct to a subclass of
1984         gimple_opt_pass along with...
1985         (pass_data_sra): ...new pass_data instance and...
1986         (make_pass_sra): ...new function.
1987         (pass_early_ipa_sra): Convert from a global struct to a subclass of
1988         gimple_opt_pass along with...
1989         (pass_data_early_ipa_sra): ...new pass_data instance and...
1990         (make_pass_early_ipa_sra): ...new function.
1991         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1992         subclass of gimple_opt_pass along with...
1993         (pass_data_ccp): ...new pass_data instance and...
1994         (make_pass_ccp): ...new function.
1995         (pass_fold_builtins): Convert from a global struct to a subclass of
1996         gimple_opt_pass along with...
1997         (pass_data_fold_builtins): ...new pass_data instance and...
1998         (make_pass_fold_builtins): ...new function.
1999         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2000         subclass of gimple_opt_pass along with...
2001         (pass_data_copy_prop): ...new pass_data instance and...
2002         (make_pass_copy_prop): ...new function.
2003         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2004         global struct to a subclass of gimple_opt_pass along with...
2005         (pass_data_rename_ssa_copies): ...new pass_data instance and...
2006         (make_pass_rename_ssa_copies): ...new function.
2007         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2008         subclass of gimple_opt_pass along with...
2009         (pass_data_dce): ...new pass_data instance and...
2010         (make_pass_dce): ...new function.
2011         (pass_dce_loop): Convert from a global struct to a subclass of
2012         gimple_opt_pass along with...
2013         (pass_data_dce_loop): ...new pass_data instance and...
2014         (make_pass_dce_loop): ...new function.
2015         (pass_cd_dce): Convert from a global struct to a subclass of
2016         gimple_opt_pass along with...
2017         (pass_data_cd_dce): ...new pass_data instance and...
2018         (make_pass_cd_dce): ...new function.
2019         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2020         subclass of gimple_opt_pass along with...
2021         (pass_data_dominator): ...new pass_data instance and...
2022         (make_pass_dominator): ...new function.
2023         (pass_phi_only_cprop): Convert from a global struct to a subclass of
2024         gimple_opt_pass along with...
2025         (pass_data_phi_only_cprop): ...new pass_data instance and...
2026         (make_pass_phi_only_cprop): ...new function.
2027         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2028         subclass of gimple_opt_pass along with...
2029         (pass_data_dse): ...new pass_data instance and...
2030         (make_pass_dse): ...new function.
2031         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2032         a subclass of gimple_opt_pass along with...
2033         (pass_data_forwprop): ...new pass_data instance and...
2034         (make_pass_forwprop): ...new function.
2035         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2036         struct to a subclass of gimple_opt_pass along with...
2037         (pass_data_tree_ifcombine): ...new pass_data instance and...
2038         (make_pass_tree_ifcombine): ...new function.
2039         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2040         subclass of gimple_opt_pass along with...
2041         (pass_data_ch): ...new pass_data instance and...
2042         (make_pass_ch): ...new function.
2043         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2044         subclass of gimple_opt_pass along with...
2045         (pass_data_tree_loop): ...new pass_data instance and...
2046         (make_pass_tree_loop): ...new function.
2047         (pass_tree_loop_init): Convert from a global struct to a subclass of
2048         gimple_opt_pass along with...
2049         (pass_data_tree_loop_init): ...new pass_data instance and...
2050         (make_pass_tree_loop_init): ...new function.
2051         (pass_lim): Convert from a global struct to a subclass of
2052         gimple_opt_pass along with...
2053         (pass_data_lim): ...new pass_data instance and...
2054         (make_pass_lim): ...new function.
2055         (pass_tree_unswitch): Convert from a global struct to a subclass of
2056         gimple_opt_pass along with...
2057         (pass_data_tree_unswitch): ...new pass_data instance and...
2058         (make_pass_tree_unswitch): ...new function.
2059         (pass_predcom): Convert from a global struct to a subclass of
2060         gimple_opt_pass along with...
2061         (pass_data_predcom): ...new pass_data instance and...
2062         (make_pass_predcom): ...new function.
2063         (pass_vectorize): Convert from a global struct to a subclass of
2064         gimple_opt_pass along with...
2065         (pass_data_vectorize): ...new pass_data instance and...
2066         (make_pass_vectorize): ...new function.
2067         (pass_graphite): Convert from a global struct to a subclass of
2068         gimple_opt_pass along with...
2069         (pass_data_graphite): ...new pass_data instance and...
2070         (make_pass_graphite): ...new function.
2071         (pass_graphite_transforms): Convert from a global struct to a subclass
2072         of gimple_opt_pass along with...
2073         (pass_data_graphite_transforms): ...new pass_data instance and...
2074         (make_pass_graphite_transforms): ...new function.
2075         (pass_check_data_deps): Convert from a global struct to a subclass of
2076         gimple_opt_pass along with...
2077         (pass_data_check_data_deps): ...new pass_data instance and...
2078         (make_pass_check_data_deps): ...new function.
2079         (pass_iv_canon): Convert from a global struct to a subclass of
2080         gimple_opt_pass along with...
2081         (pass_data_iv_canon): ...new pass_data instance and...
2082         (make_pass_iv_canon): ...new function.
2083         (pass_scev_cprop): Convert from a global struct to a subclass of
2084         gimple_opt_pass along with...
2085         (pass_data_scev_cprop): ...new pass_data instance and...
2086         (make_pass_scev_cprop): ...new function.
2087         (pass_record_bounds): Convert from a global struct to a subclass of
2088         gimple_opt_pass along with...
2089         (pass_data_record_bounds): ...new pass_data instance and...
2090         (make_pass_record_bounds): ...new function.
2091         (pass_complete_unroll): Convert from a global struct to a subclass of
2092         gimple_opt_pass along with...
2093         (pass_data_complete_unroll): ...new pass_data instance and...
2094         (make_pass_complete_unroll): ...new function.
2095         (pass_complete_unrolli): Convert from a global struct to a subclass of
2096         gimple_opt_pass along with...
2097         (pass_data_complete_unrolli): ...new pass_data instance and...
2098         (make_pass_complete_unrolli): ...new function.
2099         (pass_parallelize_loops): Convert from a global struct to a subclass
2100         of gimple_opt_pass along with...
2101         (pass_data_parallelize_loops): ...new pass_data instance and...
2102         (make_pass_parallelize_loops): ...new function.
2103         (pass_loop_prefetch): Convert from a global struct to a subclass of
2104         gimple_opt_pass along with...
2105         (pass_data_loop_prefetch): ...new pass_data instance and...
2106         (make_pass_loop_prefetch): ...new function.
2107         (pass_iv_optimize): Convert from a global struct to a subclass of
2108         gimple_opt_pass along with...
2109         (pass_data_iv_optimize): ...new pass_data instance and...
2110         (make_pass_iv_optimize): ...new function.
2111         (pass_tree_loop_done): Convert from a global struct to a subclass of
2112         gimple_opt_pass along with...
2113         (pass_data_tree_loop_done): ...new pass_data instance and...
2114         (make_pass_tree_loop_done): ...new function.
2115         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2116         struct to a subclass of gimple_opt_pass along with...
2117         (pass_data_cse_reciprocals): ...new pass_data instance and...
2118         (make_pass_cse_reciprocals): ...new function.
2119         (pass_cse_sincos): Convert from a global struct to a subclass of
2120         gimple_opt_pass along with...
2121         (pass_data_cse_sincos): ...new pass_data instance and...
2122         (make_pass_cse_sincos): ...new function.
2123         (pass_optimize_bswap): Convert from a global struct to a subclass of
2124         gimple_opt_pass along with...
2125         (pass_data_optimize_bswap): ...new pass_data instance and...
2126         (make_pass_optimize_bswap): ...new function.
2127         (pass_optimize_widening_mul): Convert from a global struct to a
2128         subclass of gimple_opt_pass along with...
2129         (pass_data_optimize_widening_mul): ...new pass_data instance and...
2130         (make_pass_optimize_widening_mul): ...new function.
2131         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2132         subclass of gimple_opt_pass along with...
2133         (pass_data_phiopt): ...new pass_data instance and...
2134         (make_pass_phiopt): ...new function.
2135         (pass_cselim): Convert from a global struct to a subclass of
2136         gimple_opt_pass along with...
2137         (pass_data_cselim): ...new pass_data instance and...
2138         (make_pass_cselim): ...new function.
2139         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2140         subclass of gimple_opt_pass along with...
2141         (pass_data_phiprop): ...new pass_data instance and...
2142         (make_pass_phiprop): ...new function.
2143         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2144         subclass of gimple_opt_pass along with...
2145         (pass_data_pre): ...new pass_data instance and...
2146         (make_pass_pre): ...new function.
2147         (pass_fre): Convert from a global struct to a subclass of
2148         gimple_opt_pass along with...
2149         (pass_data_fre): ...new pass_data instance and...
2150         (make_pass_fre): ...new function.
2151         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2152         subclass of gimple_opt_pass along with...
2153         (pass_data_reassoc): ...new pass_data instance and...
2154         (make_pass_reassoc): ...new function.
2155         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2156         subclass of gimple_opt_pass along with...
2157         (pass_data_sink_code): ...new pass_data instance and...
2158         (make_pass_sink_code): ...new function.
2159         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2160         subclass of gimple_opt_pass along with...
2161         (pass_data_strlen): ...new pass_data instance and...
2162         (make_pass_strlen): ...new function.
2163         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2164         struct to a subclass of gimple_opt_pass along with...
2165         (pass_data_build_alias): ...new pass_data instance and...
2166         (make_pass_build_alias): ...new function.
2167         (pass_build_ealias): Convert from a global struct to a subclass of
2168         gimple_opt_pass along with...
2169         (pass_data_build_ealias): ...new pass_data instance and...
2170         (make_pass_build_ealias): ...new function.
2171         (pass_ipa_pta): Convert from a global struct to a subclass of
2172         simple_ipa_opt_pass along with...
2173         (pass_data_ipa_pta): ...new pass_data instance and...
2174         (make_pass_ipa_pta): ...new function.
2175         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2176         subclass of gimple_opt_pass along with...
2177         (pass_data_uncprop): ...new pass_data instance and...
2178         (make_pass_uncprop): ...new function.
2179         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2180         global struct to a subclass of gimple_opt_pass along with...
2181         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2182         (make_pass_late_warn_uninitialized): ...new function.
2183         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2184         to a subclass of gimple_opt_pass along with...
2185         (pass_data_init_datastructures): ...new pass_data instance and...
2186         (make_pass_init_datastructures): ...new function.
2187         (pass_early_warn_uninitialized): Convert from a global struct to a
2188         subclass of gimple_opt_pass along with...
2189         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2190         (make_pass_early_warn_uninitialized): ...new function.
2191         (pass_update_address_taken): Convert from a global struct to a
2192         subclass of gimple_opt_pass along with...
2193         (pass_data_update_address_taken): ...new pass_data instance and...
2194         (make_pass_update_address_taken): ...new function.
2195         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2196         struct to a subclass of gimple_opt_pass along with...
2197         (pass_data_release_ssa_names): ...new pass_data instance and...
2198         (make_pass_release_ssa_names): ...new function.
2199         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2200         subclass of gimple_opt_pass along with...
2201         (pass_data_stdarg): ...new pass_data instance and...
2202         (make_pass_stdarg): ...new function.
2203         * tree-switch-conversion.c (pass_convert_switch): Convert from a
2204         global struct to a subclass of gimple_opt_pass along with...
2205         (pass_data_convert_switch): ...new pass_data instance and...
2206         (make_pass_convert_switch): ...new function.
2207         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2208         to a subclass of gimple_opt_pass along with...
2209         (pass_data_tail_recursion): ...new pass_data instance and...
2210         (make_pass_tail_recursion): ...new function.
2211         (pass_tail_calls): Convert from a global struct to a subclass of
2212         gimple_opt_pass along with...
2213         (pass_data_tail_calls): ...new pass_data instance and...
2214         (make_pass_tail_calls): ...new function.
2215         * tree-vect-generic.c (pass_lower_vector): Convert from a global
2216         struct to a subclass of gimple_opt_pass along with...
2217         (pass_data_lower_vector): ...new pass_data instance and...
2218         (make_pass_lower_vector): ...new function.
2219         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2220         gimple_opt_pass along with...
2221         (pass_data_lower_vector_ssa): ...new pass_data instance and...
2222         (make_pass_lower_vector_ssa): ...new function.
2223         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2224         to a subclass of gimple_opt_pass along with...
2225         (pass_data_slp_vectorize): ...new pass_data instance and...
2226         (make_pass_slp_vectorize): ...new function.
2227         (pass_ipa_increase_alignment): Convert from a global struct to a
2228         subclass of simple_ipa_opt_pass along with...
2229         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2230         (make_pass_ipa_increase_alignment): ...new function.
2231         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2232         gimple_opt_pass along with...
2233         (pass_data_vrp): ...new pass_data instance and...
2234         (make_pass_vrp): ...new function.
2235         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2236         subclass of simple_ipa_opt_pass along with...
2237         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2238         (make_pass_ipa_free_lang_data): ...new function.
2239         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2240         gimple_opt_pass along with...
2241         (pass_data_tsan): ...new pass_data instance and...
2242         (make_pass_tsan): ...new function.
2243         (pass_tsan_O0): Convert from a global struct to a subclass of
2244         gimple_opt_pass along with...
2245         (pass_data_tsan_O0): ...new pass_data instance and...
2246         (make_pass_tsan_O0): ...new function.
2247         * var-tracking.c (pass_variable_tracking): Convert from a global
2248         struct to a subclass of rtl_opt_pass along with...
2249         (pass_data_variable_tracking): ...new pass_data instance and...
2250         (make_pass_variable_tracking): ...new function.
2251         * web.c (pass_web): Convert from a global struct to a subclass of
2252         rtl_opt_pass along with...
2253         (pass_data_web): ...new pass_data instance and...
2254         (make_pass_web): ...new function.
2255         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2256         declaration with that of...
2257         (make_pass_mode_switch_use): ...new function.
2258         (pass_resolve_sw_modes): Replace declaration with that of...
2259         (make_pass_resolve_sw_modes): ...new function.
2260         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2261         from a global struct to a subclass of rtl_opt_pass along with...
2262         (pass_data_mode_switch_use): ...new pass_data instance and...
2263         (make_pass_mode_switch_use): ...new function.
2264         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2265         from a global struct to a subclass of rtl_opt_pass along with...
2266         (pass_data_resolve_sw_modes): ...new pass_data instance and...
2267         (make_pass_resolve_sw_modes): ...new function.
2268         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2269         struct to a subclass of rtl_opt_pass along with...
2270         (pass_data_insert_vzeroupper): ...new pass_data instance and...
2271         (make_pass_insert_vzeroupper): ...new function.
2272         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2273         global struct to a subclass of rtl_opt_pass along with...
2274         (pass_data_work_around_errata): ...new pass_data instance and...
2275         (make_pass_work_around_errata): ...new function.
2276         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2277         struct to a subclass of rtl_opt_pass along with...
2278         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2279         (make_pass_mips_machine_reorg2): ...new function.
2281 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2283         * passes.c (pass_manager::operator new): New.
2285 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2287         Handwritten part of conversion of passes to C++ classes.
2289         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2290         (toplev.o): Add dep on PASS_MANAGER_H.
2291         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2292         of early local pases to reflect this moving from a global to a
2293         member of gcc::pass_manager.
2294         (cgraph_add_new_function): Likewise.
2295         * lto-cgraph.c (lto_output_node): Update for conversion of
2296         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2297         * passes.c (opt_pass::clone): New.
2298         (opt_pass::gate): New.
2299         (opt_pass::execute): New.
2300         (opt_pass::opt_pass): New.
2301         (pass_manager::execute_early_local_passes): New.
2302         (pass_manager::execute_pass_mode_switching): new.
2303         (finish_optimization_passes): Convert to...
2304         (pass_manager::finish_optimization_passes): ...this.
2305         (finish_optimization_passes): Update for conversion of passes to
2306         C++ classes.
2307         (register_dump_files_1): Use has_gate since we cannot portably
2308         check a vtable entry against NULL.
2309         (dump_one_pass): Likewise.
2310         (ipa_write_summaries_2): Likewise.
2311         (ipa_write_optimization_summaries_1): Likewise.
2312         (ipa_read_summaries_1): Likewise.
2313         (ipa_read_optimization_summaries_1): Likewise.
2314         (execute_ipa_stmt_fixups): Likewise.
2315         (pass_manager::pass_manager): Rewrite pass-creation, invoking
2316         pass-creation functions rather than wiring up globals, and
2317         storing the results in fields of pass_manager generated using
2318         pass-instances.def.
2319         (pass_manager::dump_profile_report): Update for conversion of
2320         passes to C++ classes.
2321         (pass_manager::execute_ipa_summary_passes): Likewise.
2322         (execute_one_ipa_transform_pass): Likewise.
2323         (execute_one_pass): Use has_gate and has_execute since we cannot
2324         portably check a vtable entry against NULL.
2325         * pass_manager.h (pass_manager::finish_optimization_passes): New.
2326         (pass_manager): Use pass-instances.def to add fields for the
2327         various pass instances.
2328         * toplev.c (finalize): Update for move of
2329         finish_optimization_passes to a method of gcc::pass_manager.
2330         * toplev.h (finish_optimization_passes): Move to method of class
2331         pass_manager.
2332         * tree-pass.h (struct pass_data): New.
2333         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2334         (opt_pass::gate): Convert to virtual function.
2335         (opt_pass::~opt_pass): New.
2336         (opt_pass::clone): New.
2337         (opt_pass::execute): Convert to virtual function.
2338         (opt_pass::opt_pass): New.
2339         (opt_pass::ctxt_): new.
2340         (gimple_opt_pass): Convert to subclass of opt_pass.
2341         (gimple_opt_pass::gimple_opt_pass): New.
2342         (rtl_opt_pass): Convert to subclass of opt_pass.
2343         (rtl_opt_pass::rtl_opt_pass): New.
2344         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2345         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2346         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2347         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2348         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2349         invocation of pass_mode_switching to reflect this moving from a
2350         global to a member of gcc::pass_manager.
2351         (ix86_option_override): Rework how pass_insert_vzeroupper is
2352         added to the pass_manager to reflect autogenerated changes.
2353         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2355 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2357         PR rtl-optimization/57708
2358         * recog.c (peep2_find_free_register): Validate all regs in a
2359         multi-reg mode.
2361 2013-08-05  Jan Hubicka  <jh@suse.cz>
2363         PR lto/57602
2364         * cgraph.c (verify_cgraph_node): Accept local flags from other
2365         partitions.
2366         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2367         (function_and_variable_visibility): Likewise.
2368         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2370 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2372         * graph.c (init_graph_slim_pretty_print): Remove.
2373         (print_graph_cfg): Do not call it.  Use local pretty printer.
2374         (start_graph_dump): Likewise.
2376 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2378         * gimple-pretty-print.c (buffer): Remove.
2379         (initialized): Likewise.
2380         (maybe_init_pretty_print): Likewise.
2381         (print_gimple_stmt): Do not call it.  Use non-static local
2382         pretty_printer variable.
2383         (print_gimple_expr): Likewise.
2384         (print_gimple_seq): Likewise.
2385         (gimple_dump_bb): Likewise.
2387 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2389         * asan.c (asan_pp): Remove.
2390         (asan_pp_initialized): Likewise.
2391         (asan_pp_initialize): Likewise.
2392         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2393         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2394         (asan_add_global): Likewise.
2396 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2398         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2399         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2400         * pretty-print.c (pp_formatted_text_data): Likewise.
2401         (pp_indent): Rename from pp_base_indent.
2402         (pp_format): Rename from pp_base_format.
2403         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2404         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2405         (pp_flush): Rename from pp_base_flush.
2406         (pp_set_line_maximum_length): Rename from
2407         pp_base_set_line_maximum_length.
2408         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2409         (pp_set_prefix): Rename from pp_base_set_prefix.
2410         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2411         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2412         (pp_append_text): Rename from pp_base_append_text.
2413         (pp_formatted_text): Rename from pp_base_formatted_text.
2414         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2415         (pp_remaining_character_count_for_line): Rename from
2416         pp_base_remaining_character_count_for_line.
2417         (pp_newline): Rename from pp_base_newline.
2418         (pp_character): Rename from pp_base_character.
2419         (pp_string): Rename from pp_base_string.
2420         (pp_maybe_space): Rename from pp_base_maybe_space.
2421         * asan.c (asan_pp_string): Adjust.
2422         (asan_emit_stack_protection): Likewise.
2423         (asan_add_global): Likewise.
2424         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2425         * tree-mudflap.c (mf_varname_tree): Likewise.
2426         * tree-pretty-print.c (pp_tree_identifier): Rename from
2427         pp_base_tree_identifier.
2428         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2429         Declare as function.
2431 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2433         * pretty-print.h (pp_bar_bar): New.
2434         (pp_ampersand_ampersand): Likewise.
2435         (pp_less_equal): Likewise.
2436         (pp_greater_equal): Likewise.
2437         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2438         printer functions instead of pp_string or operators and punctuators.
2439         (dump_gimple_call): Likewise.
2440         (dump_gimple_omp_for): Likewise.
2441         (dump_gimple_transaction): Likewise.
2442         (dump_gimple_phi): Likewise.
2443         (pp_gimple_stmt_1): Likewise.
2444         * sched-vis.c (print_insn): Likewise.
2445         * tree-mudflap.c (mf_varname_tree): Likewise.
2446         * tree-pretty-print.c (dump_block_node): Likewise.
2447         (dump_generic_node): Likewise.
2449 2013-08-02  Jan Hubicka  <jh@suse.cz>
2451         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2452         boundaries.
2453         * lto-streamer-out.c (tree_is_indexable): Results decls and
2454         parm decls are not indexable.
2455         (DFS_write_tree_body): Do not follow args and results.
2456         (hash_tree): Likewise.
2457         (output_functions): Rearrange so struct function is needed
2458         only when real body is output; be able to also ouptut abstract
2459         functions; output DECL_ARGUMENTS and DECL_RESULT.
2460         (lto_output): When not in WPA, ale store abstract functions.
2461         (write_symbol): Do not care about RESULT_DECL.
2462         (output_symbol_p): Handle correctly sbtract decls.
2463         * lto-streamer-in.c (input_function): Rearrange so struct
2464         function can be NULL at entry; allow streaming of
2465         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2466         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2467         sanity check during LTO.
2468         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2469         RESULT_DECl and DECL_ARGUMENTS.
2470         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2471         Likewise.
2473 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2475         * pretty-print.h (pp_underscore): New.
2476         (pp_comma): Tidy.
2477         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2478         printer functions instead of pp_character.
2479         (dump_binary_rhs): Likewise.
2480         (dump_ternary_rhs): Likewise.
2481         (dump_gimple_call_args): Likewise.
2482         (pp_points_to_solution): Likewise.
2483         (dump_gimple_call): Likewise.
2484         (dump_gimple_switch): Likewise.
2485         (dump_gimple_cond): Likewise.
2486         (dump_gimple_bind): Likewise.
2487         (dump_gimple_try): Likewise.
2488         (dump_gimple_omp_for): Likewise.
2489         (dump_gimple_omp_continue): Likewise.
2490         (dump_gimple_omp_single): Likewise.
2491         (dump_gimple_omp_sections): Likewise.
2492         (dump_gimple_omp_block): Likewise.
2493         (dump_gimple_omp_critical): Likewise.
2494         (dump_gimple_transaction): Likewise.
2495         (dump_gimple_asm): Likewise.
2496         (dump_gimple_phi): Likewise.
2497         (dump_gimple_omp_parallel): Likewise.
2498         (dump_gimple_omp_task): Likewise.
2499         (dump_gimple_omp_atomic_load): Likewise.
2500         (dump_gimple_omp_atomic_store): Likewise.
2501         (dump_gimple_mem_ops): Likewise.
2502         (pp_gimple_stmt_1): Likewise.
2503         (pp_cfg_jump): Likewise.
2504         (dump_implicit_edges): Likewise.
2505         (gimple_dump_bb_for_graph): Likewise.
2506         * graph.c (draw_cfg_node): Likewise.
2507         * langhooks.c (lhd_print_error_function): Likewise.
2508         * sched-vis.c (print_exp): Likewise.
2509         (print_value): Likewise.
2510         (print_pattern): Likewise.
2511         (print_insn): Likewise.
2512         (rtl_dump_bb_for_graph): Likewise.
2513         * tree-pretty-print.c (dump_function_declaration): Likewise.
2514         (dump_array_domain): Likewise.
2515         (dump_omp_clause): Likewise.
2516         (dump_location): Likewise.
2517         (dump_generic_node): Likewise.
2518         (print_struct_decl): Likewise.
2519         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2521 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2523         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2524         candidate table when replacing a candidate statement.
2525         (replace_rhs_if_not_dup): Likewise.
2526         (replace_one_candidate): Likewise.
2528 2013-08-02  Jan Hubicka  <jh@suse.cz>
2529             Martin Liska  <marxin.liska@gmail.com>
2531         * cgraphunit.c (add_new_function): Fix logic when adding from
2532         late IPA pass.
2533         (assemble_thunk): Rename to ...
2534         (expand_thunk); .. this one; export; get it working with
2535         general functions; make produced gimple valid.
2536         * cgraph.h (expand_thunk): Declare.
2538 2013-08-02  Jan Hubicka  <jh@suse.cz>
2540         * ipa-cp.c (gather_context_independent_values): Use
2541         ipa_get_param_move_cost.
2542         (get_replacement_map): Remove PARAM; move parameter folding
2543         into tree-inline.c
2544         (create_specialized_node): Update.
2545         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2546         assert that we have gimple body; update move_cost.
2547         (count_formal_params): Assert that we have gimple body.
2548         (ipa_dump_param): New function.
2549         (ipa_alloc_node_params): Break out from ...
2550         (ipa_initialize_node_params): ... here.
2551         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2552         (ipa_write_node_info): Stream move costs.
2553         (ipa_read_node_info): Read move costs.
2554         (ipa_update_after_lto_read): Do not recompute node params.
2555         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2556         (ipa_get_param): Check we are not in WPA.
2557         (ipa_get_param_move_cost): New.
2558         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2559         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2560         parm numbers to be present.
2562 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2564         PR rtl-optimization/58048
2565         * lra-constraints.c (process_alt_operands): Don't check asm
2566         operand on register.
2568 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2570         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2571         the implied StoreLoad barrier for atomic operations if before.
2573 2013-08-02  Jan Hubicka  <jh@suse.cz>
2574             Martin Liska  <marxin.liska@gmail.com>
2576         * cgraph.c (cgraph_function_body_availability): Do not check
2577         cgraph flags.
2578         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2579         symtab_node_availability): Declare.
2580         * ipa.c (can_replace_by_local_alias): New.
2581         (function_and_variable_visibility): Use it.
2582         * symtab.c (symtab_for_node_and_aliases,
2583         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2585 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2587         PR rtl-optimization/57963
2588         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2589         (lra_constraints): Use them.
2591 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2593         * config/arm/types.md (define_attr "type"): Add "load_acq"
2594         and "store_rel".
2595         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2596         changes.
2597         (cortex_a53_store1): Likewise.
2599 2013-08-01  Jan Hubicka  <jh@suse.cz>
2601         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2602         partitions are not needed.
2604 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
2606         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2607         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2608         MAYBE_NON_Q_CLASS_P where appropriate.
2610 2013-08-01  Jan Hubicka  <jh@suse.cz>
2612         * cgraph.h (release_function_body): Declare.
2613         * tree.c (free_lang_data_in_decl): Free, parameters and return values
2614         of unused delcarations.
2616 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2618         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2619         RTL form when subtracting a constant.
2621 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2623         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2624         Generate canonical plus rtx with negated immediate instead of minus
2625         where appropriate.
2626         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2628 2013-08-01  Jan Hubicka  <jh@suse.cz>
2630         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2631         (cgraph_release_function_body): Likewise.
2632         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2633         * cgraph.h (cgrpah_node): Rename abstract_and_needed
2634         to used_as_abstract_origin.
2635         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2636         symbols used as abstract origins.
2637         * cgraphunit.c (analyze_functions): Update.
2638         * ipa.c (symtab_remove_unreachable_nodes): Recompute
2639         used_as_abstract_origin.
2640         * tree-inline.c (tree_function_versioning): Update
2641         used_as_abstract_origin; be ready for DECL_RESULT and
2642         DECL_ARGUMENTS to be NULL.
2644         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2645         for abstract functions.
2646         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2647         real symbols.
2649 2013-08-01  Jan Hubicka  <jh@suse.cz>
2651         * profile.c (compute_value_histograms): Fix thinko.
2653 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2655         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
2656         aarch-common-protos.h to extra_headers.
2657         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2658         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2659         * config/aarch64/t-aarch64 (aarch-common.o): Define.
2661 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2663         * config/aarch64/aarch64.md (define_attr "type"): Delete.
2664         Include "../arm/types.md".  Define "type" attribute for all patterns.
2665         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2666         attribute changes.
2668 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
2670         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2671         to support power8 load fusion.
2672         (fusion_gpr_mem_load): Likewise.
2674         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2676         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2677         declarations for power8 load fusion.
2678         (emit_fusion_gpr_load): Likewise.
2680         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2681         tuning for power8, turn on fusion mode by default.  Turn on sign
2682         extending fusion mode if normal fusion mode is on, and we are at
2683         -O2 or -O3.
2684         (fusion_gpr_load_p): New function, return true if we can fuse an
2685         addis instruction with a dependent load to a GPR.
2686         (emit_fusion_gpr_load): Emit the instructions for power8 load
2687         fusion to GPRs.
2689         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2690         (VSX load fusion peepholes): New peepholes to fuse together an
2691         addi instruction with a VSX load instruction.
2693         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2694         peepholes to fuse an addis instruction with a load to a GPR base
2695         register.  If we are supporting sign extending fusions, convert
2696         sign extending loads to zero extending loads and add an explicit
2697         sign extension.
2699 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2701         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
2702         aarch-common-protos.h to extra_headers.
2703         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2704         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2705         (arm_early_store_addr_dep): Likewise.
2706         (arm_no_early_alu_shift_dep): Likewise.
2707         (arm_no_early_alu_shift_value_dep): Likewise.
2708         (arm_no_early_mul_dep): Likewise.
2709         (arm_no_early_store_addr_dep): Likewise.
2710         (arm_mac_accumulator_is_mul_result): Likewise.
2711         (arm_mac_accumulator_is_result): Likewise.
2712         * config/arm/aarch-common.c: ... here.  New file.
2713         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2714         here to ...
2715         (arm_early_store_addr_dep): Likewise.
2716         (arm_no_early_alu_shift_dep): Likewise.
2717         (arm_no_early_alu_shift_value_dep): Likewise.
2718         (arm_no_early_mul_dep): Likewise.
2719         (arm_no_early_store_addr_dep): Likewise.
2720         (arm_mac_accumulator_is_mul_result): Likewise.
2721         (arm_mac_accumulator_is_result): Likewise.
2722         * config/arm/aarch-common-protos.h: ... here.  New file.
2723         * config/arm/t-arm (aarch-common.o): Define.
2725 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2727         * config/arm/arm.md: Include new file "types.md".
2728         (define_attr "type"): Move from here to ...
2729         (define_attr "mul32"): Likewise.
2730         (define_attr "mul64"): Likewise.
2731         * config/arm/types.md: ... here.  New file.
2733 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2735         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2736         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2738 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2740         * gen-pass-instances.awk: Fix offset of substr().
2742 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
2744         * Makefile.in (pass-instances.def): New.
2745         (passes.o): Replace dependency on passes.def with one on
2746         pass-instances.def
2748         * gen-pass-instances.awk: New.
2750         * passes.c (pass_manager::pass_manager): Use pass-instances.def
2751         rather than passes.def, updating local definition of NEXT_PASS
2752         macro to add an extra NUM parameter (currently unused).
2754 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
2756         * Makefile.in (PASS_MANAGER_H): New.
2757         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2758         (passes.o): Likewise.
2759         (statistics.o): Likewise.
2760         (cgraphunit.o): Likewise.
2761         (context.o): Depend on PASS_MANAGER_H.
2763         * pass_manager.h: New.
2765         * cgraphunit.c (cgraph_add_new_function): Update for moves
2766         of globals to fields of pass_manager.
2767         (analyze_function): Likewise.
2768         (expand_function): Likewise.
2769         (ipa_passes): Likewise.
2770         (compile): Likewise.
2772         * context.c (context::context): New.
2773         * context.h  (context::context): New.
2774         (context::get_passes): New.
2775         (context::passes_): New.
2777         * lto-cgraph.c (input_node): Update for moves of globals to
2778         fields of pass_manager.
2780         * passes.c (all_passes): Remove, in favor of a field of the
2781         same name within the new class pass_manager.
2782         (all_small_ipa_passes): Likewise.
2783         (all_lowering_passes): Likewise.
2784         (all_regular_ipa_passes): Likewise.
2785         (all_late_ipa_passes): Likewise.
2786         (all_lto_gen_passes): Likewise.
2787         (passes_by_id): Likewise.
2788         (passes_by_id_size): Likewise.
2789         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2790         the new class pass_manager.
2791         (set_pass_for_id): Convert to...
2792         (pass_manager::set_pass_for_id): ...method.
2793         (get_pass_for_id): Convert to...
2794         (pass_manager::get_pass_for_id): ...method.
2795         (register_one_dump_file): Move body of implementation into...
2796         (pass_manager::register_one_dump_file): ...here.
2797         (register_dump_files_1): Convert to...
2798         (pass_manager::register_dump_files_1): ...method.
2799         (register_dump_files): Convert to...
2800         (pass_manager::register_dump_files): ...method.
2801         (create_pass_tab): Update for moves of globals to fields of
2802         pass_manager.
2803         (dump_passes): Move body of implementation into...
2804         (pass_manager::dump_passes): ...here.
2805         (register_pass): Move body of implementation into...
2806         (pass_manager::register_pass): ...here.
2807         (init_optimization_passes): Convert into...
2808         (pass_manager::pass_manager): ...constructor for new
2809         pass_manager class, and initialize the pass_lists array.
2810         (check_profile_consistency): Update for moves of globals to
2811         fields of pass_manager.
2812         (dump_profile_report): Move body of implementation into...
2813         (pass_manager::dump_profile_report): ...here.
2814         (ipa_write_summaries_1): Update for moves of pass lists from
2815         being globals to fields of pass_manager.
2816         (ipa_write_optimization_summaries): Likewise.
2817         (ipa_read_summaries):  Likewise.
2818         (ipa_read_optimization_summaries): Likewise.
2819         (execute_all_ipa_stmt_fixups): Likewise.
2821         * statistics.c (statistics_fini): Update for moves of globals to
2822         fields of pass_manager.
2824         * toplev.c (general_init): Replace call to
2825         init_optimization_passes with construction of the pass_manager
2826         instance.
2828         * tree-pass.h (all_passes): Remove, in favor of a field of the
2829         same name within the new class pass_manager.
2830         (all_small_ipa_passes): Likewise.
2831         (all_lowering_passes): Likewise.
2832         (all_regular_ipa_passes): Likewise.
2833         (all_lto_gen_passes): Likewise.
2834         (all_late_ipa_passes): Likewise.
2835         (passes_by_id): Likewise.
2836         (passes_by_id_size): Likewise.
2837         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2838         the new class pass_manager.
2839         (get_pass_for_id): Remove.
2841 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2843         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2844         configs.
2846 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2848         * arm.md (mulhi3): New expand pattern.
2850 2013-07-30  Jan Hubicka  <jh@suse.cz>
2851             Martin Liska  <marxin.liska@gmail.com>
2853         * profile.c (compute_value_histograms): Do not ICE when
2854         there is mismatch only on some counters.
2856 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2858         PR rtl-optimization/57637
2859         * function.c (move_insn_for_shrink_wrap): Also check the
2860         GEN set of the LIVE problem for the liveness analysis
2861         if it exists, otherwise give up.
2863 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2865         PR tree-optimization/57993
2866         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2867         replaced statement in the candidate table.
2868         (phi_add_costs): Return infinite cost when the hidden basis does
2869         not dominate all phis on which the candidate is dependent.
2870         (replace_one_candidate): Record replaced statement in the
2871         candidate table.
2873 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
2875         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2876         (ashlv2si3): New expander.
2877         (*ashlv2si3_i): New define_insn_and_split.
2878         * predicates.md (float_operation): Allow patterns with three
2879         basic sub-patterns.
2881         PR rtl-optimization/58021
2882         * mode-switching.c (create_pre_exit): Always split off preceding
2883         insns if we are not at the basic block head.
2885 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
2887         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2888         (UCLIBC_DYNAMIC_LINKER): New macro.
2889         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2890         `-mnan=2008'.
2891         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2892         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
2893         `-mnan=2008'.
2894         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2895         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2896         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2897         for SF and DF modes.  Use ieee_quad_format for TF mode.
2898         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2899         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2900         (mips_option_override): Handle `-mnan=legacy'.
2901         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2902         `-mabs=2008' and `-mnan=2008'.
2903         (OPTION_DEFAULT_SPECS): Add "nan" default.
2904         (ASM_SPEC): Handle `-mnan='.
2905         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2906         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2907         comment accordingly.
2908         (neg<mode>2): Likewise.
2909         * config/mips/mips.opt (mabs, mnan): New options.
2910         * doc/install.texi (Configuration): Document `--with-nan=' option.
2911         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2912         `-mnan=' options.
2913         (MIPS Options): Document them.
2914         * config.gcc <mips*-*-*>: Handle `--with-nan='.
2915         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2916         * configure: Regenerate.
2917         * config.in: Regenerate.
2919 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2921         * config/i386/i386.md (float post-reload splitters): Do not check
2922         for subregs of SSE registers.
2924 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2925             H.J. Lu  <hongjiu.lu@intel.com>
2927         PR target/57954
2928         PR target/57988
2929         * config/i386/i386.md (post-reload splitter
2930         to avoid partial SSE reg dependency stalls): New pattern.
2932 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2934         * config/s390/s390.md ("movcc"): Swap load and store instructions.
2936 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
2938         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2939         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2941 2013-07-26  Cary Coutant  <ccoutant@google.com>
2943         * dwarf2out.c (die_checksum_ordered): Don't include template
2944         instantiations in signature.
2945         (is_template_parameter): New function.
2946         (is_template_instantiation): New function.
2947         (generate_skeleton_bottom_up): Don't include template instantiations
2948         in type unit DIE.
2949         (generate_skeleton): Likewise.
2950         (break_out_comdat_types): Move recursive call to break out nested
2951         types earlier.
2952         (prune_unused_types_mark_generic_parms_dies): Call
2953         is_template_parameter.
2955 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
2957         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2958         uses vector registers.
2959         * config/aarch64/iterators.md: Add attributes rtn and vas.
2961 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2962             Richard Earnshaw  <richard.earnshaw@arm.com>
2964         * combine.c (simplify_comparison): Re-canonicalize operands
2965         where appropriate.
2966         * config/arm/arm.md (movcond_addsi): New splitter.
2968 2013-07-25  Sterling Augustine  <saugustine@google.com>
2970         * dwarf2out.c (size_of_pubnames): Move code to...
2971         (include_pubname_in_output): ...here.  New.
2972         (want_pubnames): Rearrange.
2973         (output_pubnames): Call include_pubname_in_output.  Move assertion.
2975 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2977         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2979 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2981         PR target/38836
2982         * doc/extend.texi: Remove obsolete builtins. Fix
2983         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2985 2013-07-25  Jan Hubicka  <jh@suse.cz>
2987         * cgraph.c (release_function_body): Break out from ...
2988         (cgraph_release_function_body): ... this one; also release DECL_RESULT
2989         and DECL_ARGUMENTS.
2990         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2991         old_tree in the map.
2992         (create_specialized_node): Update.
2993         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2994         into index.
2995         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2996         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2997         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2998         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2999         DECL_RESULT.
3001 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3003         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3004         addsi3_carryin_alt2_<optab>): Correct output template.
3006 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3008         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3009         Adjust for arm_restrict_it.
3010         Remove trailing whitespace.
3012 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
3014         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3015         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3017         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3019 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
3021         PR rtl-optimization/57960
3022         * lra-constraints.c (process_alt_operands): Use the right mode
3023         when checking strict_low.
3025 2013-07-25  Jan Hubicka  <jh@suse.cz>
3027         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3028         * cgraph.c (cgraph_remove_node): Do not release function body
3029         when in cgraph streaming.
3030         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3031         in other partitions are not considered reachable; fix handling of
3032         clones.
3034 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3036         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3038 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3040         PR target/19599
3041         PR target/57731
3042         PR target/57837
3043         * config/arm/arm.md ("*sibcall_insn): Replace use of
3044         Ss with US. Adjust output for v5 and v4t.
3045         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3047         * config/arm/constraints.md ("Ss"): Rename to US.
3049 2013-07-25  Terry Guo  <terry.guo@arm.com>
3051         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3052         shift_add/shift_sub0/shift_sub1 RTXs.
3054 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
3055             Anton Blanchard <anton@au1.ibm.com>
3057         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3058         (altivec_vpks<VI_char>ss): Likewise.
3059         (altivec_vpks<VI_char>us): Likewise.
3060         (altivec_vpku<VI_char>us): Likewise.
3061         (altivec_vpku<VI_char>um): Likewise.
3063 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
3065         Introduce context class.
3067         * Makefile.in (CONTEXT_H): New.
3068         (OBJS): Add context.o.
3069         (toplev.o): Add CONTEXT_H to dependencies.
3070         (context.o): New.
3072         * toplev.c (general_init): Create the singleton gcc::context instance.
3074         * context.c: New.
3076         * context.h: New.
3078 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3080         PR rtl-optimization/57968
3081         * mode-switching.c (create_pre_exit): Allow instructions that
3082         don't set a return register to need a non-exit mode.
3084 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3085             Anton Blanchard <anton@au1.ibm.com>
3087         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3088         operands to vperm for little endian.
3089         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3090         of lvsl to create the control mask for a vperm for little endian.
3092 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3093             Anton Blanchard <anton@au1.ibm.com>
3095         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3096         two operands for little-endian.
3098 2013-07-23  Steve Ellcey  <sellcey@mips.com>
3100         * config/mips/mips.c (mips_case_values_threshold): New.
3101         (TARGET_CASE_VALUES_THRESHOLD): Define.
3103 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3104             Anton Blanchard <anton@au1.ibm.com>
3106         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3107         selection of field for vector splat in little endian mode.
3109 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
3111         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3112         expanders to rs6000.md.
3113         (ior<mode>3): Likewise.
3114         (and<mode>3): Likewise.
3115         (one_cmpl<mode>2): Likewise.
3116         (nor<mode>3): Likewise.
3117         (andc<mode>3): Likewise.
3118         (eqv<mode>3): Likewise.
3119         (nand<mode>3): Likewise.
3120         (orc<mode>3): Likewise.
3122         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3123         declaration.
3125         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3126         to split multi-word logical operations.
3127         (rs6000_split_logical_di): Likewise.
3128         (rs6000_split_logical): Likewise.
3130         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3131         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3132         and allow TImode operations in 32-bit.
3133         (vsx_and<mode>3_64bit): Likewise.
3134         (vsx_ior<mode>3_32bit): Likewise.
3135         (vsx_ior<mode>3_64bit): Likewise.
3136         (vsx_xor<mode>3_32bit): Likewise.
3137         (vsx_xor<mode>3_64bit): Likewise.
3138         (vsx_one_cmpl<mode>2_32bit): Likewise.
3139         (vsx_one_cmpl<mode>2_64bit): Likewise.
3140         (vsx_nor<mode>3_32bit): Likewise.
3141         (vsx_nor<mode>3_64bit): Likewise.
3142         (vsx_andc<mode>3_32bit): Likewise.
3143         (vsx_andc<mode>3_64bit): Likewise.
3144         (vsx_eqv<mode>3_32bit): Likewise.
3145         (vsx_eqv<mode>3_64bit): Likewise.
3146         (vsx_nand<mode>3_32bit): Likewise.
3147         (vsx_nand<mode>3_64bit): Likewise.
3148         (vsx_orc<mode>3_32bit): Likewise.
3149         (vsx_orc<mode>3_64bit): Likewise.
3151         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3152         logical types in GPRs.
3154         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3155         logical insns to rs6000.md, and allow TImode operations in
3156         32-bit.
3157         (altivec_ior<mode>3): Likewise.
3158         (altivec_xor<mode>3): Likewise.
3159         (altivec_one_cmpl<mode>2): Likewise.
3160         (altivec_nor<mode>3): Likewise.
3161         (altivec_andc<mode>3): Likewise.
3163         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3164         attributes for moving the 128-bit logical operations into
3165         rs6000.md.
3166         (BOOL_REGS_OUTPUT): Likewise.
3167         (BOOL_REGS_OP1): Likewise.
3168         (BOOL_REGS_OP2): Likewise.
3169         (BOOL_REGS_UNARY): Likewise.
3170         (BOOL_REGS_AND_CR0): Likewise.
3171         (one_cmpl<mode>2): Add support for DI logical operations on
3172         32-bit, splitting the operations to 32-bit.
3173         (anddi3): Likewise.
3174         (iordi3): Likewise.
3175         (xordi3): Likewise.
3176         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3177         changes to combine the 32/64-bit code, allow logical operations on
3178         TI mode in 32-bit, and to use similar match_operator patterns like
3179         scalar mode uses.  Combine the Altivec and VSX code for logical
3180         operations, and move it here.
3181         (ior<mode>3, 128-bit types): Likewise.
3182         (xor<mode>3, 128-bit types): Likewise.
3183         (one_cmpl<mode>3, 128-bit types): Likewise.
3184         (nor<mode>3, 128-bit types): Likewise.
3185         (andc<mode>3, 128-bit types): Likewise.
3186         (eqv<mode>3, 128-bit types): Likewise.
3187         (nand<mode>3, 128-bit types): Likewise.
3188         (orc<mode>3, 128-bit types): Likewise.
3189         (and<mode>3_internal): Likewise.
3190         (bool<mode>3_internal): Likewise.
3191         (boolc<mode>3_internal1): Likewise.
3192         (boolc<mode>3_internal2): Likewise.
3193         (boolcc<mode>3_internal1): Likewise.
3194         (boolcc<mode>3_internal2): Likewise.
3195         (eqv<mode>3_internal1): Likewise.
3196         (eqv<mode>3_internal2): Likewise.
3197         (one_cmpl1<mode>3_internal): Likewise.
3199 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3201         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3202         Rename flag_stack_usage to flag_stack_usage_info.
3204 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3206         * config/microblaze/sync.md: New file.
3207         * config/microblaze/microblaze.md: Include sync.md
3208         * config/microblaze/microblaze.c: Add print_operand 'y'.
3209         * config/microblaze/constraints.md: Add memory_contraint
3210         'Q' which is a single register.
3212 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
3214         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3216 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3218         * reload.c (find_reloads): Exit loop once we find this operand
3219         cannot be reloaded somehow for this alternative.
3221         * reload.c (find_reloads): Exit loop once we find a hard register.
3223         * rtlanal.c (computed_jump_p): Exit loop once we find label
3224         reference is used.
3226         * i386.c (ix86_pad_returns): Exit loop after setting replace.
3228         * cfgloopmanip.c (remove_path): Exit loop after setting
3229         irred_invalidated.
3231         * gensupport.c (subst_dup): Avoid loop if code is not
3232         MATCH_DUP nor MATCH_OP_DUP.
3234 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
3236         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3238 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3240         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3241         true for SP_REGNUM if mode == ptr_mode.
3242         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3243         with value R0_REGNUM + 31.
3245 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3247         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3248         pad pointer-typed argument downward.
3250 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3252         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3253         and __ILP32__ when the ILP32 model is in use.
3255 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3257         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3258         (aarch64_load_symref_appropriately): In the case of
3259         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3260         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3261         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3262         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3263         if 'mode' doesn't equal to 'ptr_mode'.
3264         (aarch64_output_mi_thunk): Add an assertion on the alignment of
3265         'vcall_offset'; change to call aarch64_emit_move differently depending
3266         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3267         to calculate the upper bound of 'vcall_offset'.
3268         (aarch64_cannot_force_const_mem): Change to also return true if
3269         mode != ptr_mode.
3270         (aarch64_legitimize_reload_address): In the case of large
3271         displacements, add new local variable 'xmode' and an assertion
3272         based on it; change to use 'xmode' to generate the new rtx and
3273         reload.
3274         (aarch64_asm_trampoline_template): Change to generate the template
3275         differently depending on TARGET_ILP32 or not; change to use
3276         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3277         (aarch64_trampoline_size): Removed.
3278         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3279         and replace immediate literals with it.  Change to use 'ptr_mode'
3280         instead of 'DImode' and call convert_memory_address if the mode
3281         of 'fnaddr' doesn't equal to 'ptr_mode'.
3282         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3283         to output symbol.
3284         (aarch64_elf_asm_destructor): Likewise.
3285         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3286         on TARGET_ILP32 instead of aarch64_trampoline_size.
3287         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3288         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3289         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3290         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
3291         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3292         (storewb_pair<GPI:mode>_<P:mode>): ... this.
3293         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3294         depending on the value of 'mode'.
3295         (add_losym_<mode>): New.
3296         (ldr_got_small_<mode>): New, based on ldr_got_small.
3297         (ldr_got_small): Remove.
3298         (ldr_got_small_sidi): New.
3299         * config/aarch64/iterators.md (P): New.
3300         (PTR): Change to 'ptr_mode' in the condition.
3302 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3304         * config.gcc (aarch64*-*-*): Support --with-abi.
3305         (aarch64*-*-elf): Support --with-multilib-list.
3306         (aarch64*-*-linux*): Likewise.
3307         (supported_defaults): Add abi to aarch64*-*-*.
3308         * configure.ac: Mention AArch64 for --with-multilib-list.
3309         * configure: Re-generated.
3310         * config/aarch64/biarchilp32.h: New file.
3311         * config/aarch64/biarchlp64.h: New file.
3312         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3313         (ABI_SPEC): Ditto.
3314         (MULTILIB_DEFAULTS): Ditto.
3315         (DRIVER_SELF_SPECS): Ditto.
3316         (ASM_SPEC): Update to also substitute -mabi.
3317         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3318         file whose name depends on -mabi= and -mbig-endian.
3319         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3320         TARGET_ILP32.
3321         (POINTER_SIZE): New define.
3322         (POINTERS_EXTEND_UNSIGNED): Ditto.
3323         (enum aarch64_abi_type): New enumeration tag.
3324         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3325         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3326         (TARGET_ILP32): New define.
3327         * config/aarch64/aarch64.opt (mabi): New.
3328         (aarch64_abi): New.
3329         (ilp32, lp64): New values for -mabi.
3330         * config/aarch64/t-aarch64 (comma): New define.
3331         (MULTILIB_OPTIONS): Ditto.
3332         (MULTILIB_DIRNAMES): Ditto.
3333         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3334         * doc/invoke.texi: Document -mabi for AArch64.
3336 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
3338         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3340 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3341             Anton Blanchard <anton@au1.ibm.com>
3343         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3344         endianness when selecting field to splat.
3346 2013-07-22  Eric Christopher  <echristo@gmail.com>
3348         * dwarf2out.c (die_odr_checksum): New function to use
3349         CHECKSUM_ macros and ULEB128 for DIE tag.
3350         (generate_type_signature): Use.
3352 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3354         * config.gcc (sparc*-*-*): Accept leon3 processor.
3355         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3356         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3357         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3358         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3359         (mfix-ut699): Adjust comment.
3360         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3361         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3362         (CPP_CPU_SPEC): Likewise.
3363         (ASM_CPU_SPEC): Likewise.
3364         * config/sparc/sparc.c (leon3_cost): New constant.
3365         (sparc_option_override): Add leon3 support.
3366         (mem_ref): New function.
3367         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3368         (sparc_do_work_around_errata): Look into the instruction in the delay
3369         slot and adjust accordingly.  Add fix for the data cache nullify issues
3370         of the UT699.  Change insertion position for the NOP.
3371         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3372         (leon3_load): New reservation.
3373         (leon_store): Bump latency to 2.
3374         (grfpu): New automaton.
3375         (grfpu_alu): New unit.
3376         (grfpu_ds): Likewise.
3377         (leon_fp_alu): Adjust.
3378         (leon_fp_mult): Delete.
3379         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3380         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3381         * config/sparc/sparc.md (cpu): Add leon3.
3382         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3383         (swapsi): Likewise.
3384         (atomic_test_and_set): Likewise.
3385         (ldstub): Likewise.
3387 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3389         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3390         default for R5900 targets.
3391         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3392         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3393         * config/mips/mips.c (mips_option_override): Report an error for
3394         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3395         for -march=r5900 -mhard-float.
3397 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3399         * df-problems.c (can_move_insns_across): Exit loop once we
3400         find a non-fixed, non-global register.
3402         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3403         setting can_throw.
3405         * omega.c (omega_eliminate_red): Break after setting red_found.
3406         (omega_problem_has_red_equations): Similarly after setting found.
3407         (omega_query_variable): Similarly after setting coupled.
3409 2013-07-22  Marek Polacek  <polacek@redhat.com>
3411         * gimplify.c: Don't include gimple.h twice.
3413 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3415         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3416         instead of TARGET_THUMB1.
3417         (Pz): New constraint.
3418         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3419         encodings.
3420         (compare_negsi_si): Likewise.
3421         (compare_addsi2_op0): Likewise.
3422         (compare_addsi2_op1): Likewise.
3423         (addsi3_carryin_<optab>): Likewise.
3424         (addsi3_carryin_alt2_<optab>): Likewise.
3425         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3426         for arm_restrict_it.
3427         (subsi3_carryin): Likewise.
3428         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3429         (minmax_arithsi): Disable for arm_restrict_it.
3430         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3431         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3432         (satsi_<SAT:code>_shift): Likewise.
3433         (arm_shiftsi3): Add alternative for 16-bit encoding.
3434         (arm32_movhf): Disable for arm_restrict_it.
3435         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3436         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3438 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3440         * config/arm/arm.md (attribute "insn"): Delete.
3441         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3442         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3443         (not_shiftsi): Update for attribute change.
3444         (not_shiftsi_compare0): Likewise.
3445         (not_shiftsi_compare0_scratch): Likewise.
3446         (arm_one_cmplsi2): Likewise.
3447         (thumb1_one_cmplsi2): Likewise.
3448         (notsi_compare0): Likewise.
3449         (notsi_compare0_scratch): Likewise.
3450         (thumb1_movdi_insn): Likewise.
3451         (arm_movsi_insn): Likewise.
3452         (movhi_insn_arch4): Likewise.
3453         (movhi_bytes): Likewise.
3454         (arm_movqi_insn): Likewise.
3455         (thumb1_movqi_insn): Likewise.
3456         (arm32_movhf): Likewise.
3457         (thumb1_movhf): Likewise.
3458         (arm_movsf_soft_insn): Likewise.
3459         (thumb1_movsf_insn): Likewise.
3460         (thumb_movdf_insn): Likewise.
3461         (movsicc_insn): Likewise.
3462         (movsfcc_soft_insn): Likewise.
3463         (and_scc): Likewise.
3464         (cond_move): Likewise.
3465         (if_move_not): Likewise.
3466         (if_not_move): Likewise.
3467         (if_shift_move): Likewise.
3468         (if_move_shift): Likewise.
3469         (if_shift_shift): Likewise.
3470         (if_not_arith): Likewise.
3471         (if_arith_not): Likewise.
3472         (cond_move_not): Likewise.
3473         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3474         (neon_mov<mode>): Likewise.
3475         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3476         (thumb2_movsi_vfp): Likewise.
3477         (movsf_vfp): Likewise.
3478         (thumb2_movsf_vfp): Likewise.
3479         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3480         change.
3481         (cortexa7_older_only): Likewise.
3482         (cortexa7_younger): Likewise.
3483         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3484         (1020alu_shift_op): Likewise.
3485         (1020alu_shift_reg_op): Likewise.
3486         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3487         (alu_shift_op): Likewise.
3488         (alu_shift_reg_op): Likewise.
3489         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3490         (11_alu_shift_op): Likewise.
3491         (11_alu_shift_reg_op): Likewise.
3492         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3493         (9_alu_shift_reg_op): Likewise.
3494         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3495         change.
3496         (cortex_a15_alu_shift): Likewise.
3497         (cortex_a15_alu_shift_reg): Likewise.
3498         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3499         (cortex_a5_alu_shift): Likewise.
3500         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3501         change.
3502         (cortex_a53_alu_shift): Likewise.
3503         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3504         change.
3505         (cortex_a7_alu_reg): Likewise.
3506         (cortex_a7_alu_shift): Likewise.
3507         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3508         (cortex_a8_alu_shift): Likewise.
3509         (cortex_a8_alu_shift_reg): Likewise.
3510         (cortex_a8_mov): Likewise.
3511         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3512         (cortex_a9_dp_shift): Likewise.
3513         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3514         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3515         (cortex_r4_mov): Likewise.
3516         (cortex_r4_alu_shift): Likewise.
3517         (cortex_r4_alu_shift_reg): Likewise.
3518         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3519         (526_alu_shift_op): Likewise.
3520         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3521         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3522         (626te_alu_shift_op): Likewise.
3523         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3524         (726te_alu_op): Likewise.
3525         (726te_alu_shift_op): Likewise.
3526         (726te_alu_shift_reg_op): Likewise.
3527         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3528         (mp626_alu_shift_op): Likewise.
3529         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3530         (pj4_alu_e1_conds): Likewise.
3531         (pj4_alu): Likewise.
3532         (pj4_alu_conds): Likewise.
3533         (pj4_shift): Likewise.
3534         (pj4_shift_conds): Likewise.
3535         (pj4_alu_shift): Likewise.
3536         (pj4_alu_shift_conds): Likewise.
3538 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3540         * config/arm/predicates.md (shiftable_operator_strict_it):
3541         New predicate.
3542         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3543         Disable cond_exec version for arm_restrict_it.
3544         (thumb2_smaxsi3): Convert to generate cond_exec.
3545         (thumb2_sminsi3): Likewise.
3546         (thumb32_umaxsi3): Likewise.
3547         (thumb2_uminsi3): Likewise.
3548         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3549         (thumb2_neg_abssi2): Likewise.
3550         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3551         (thumb2_movsicc_insn): Adjust alternatives.
3552         (thumb2_mov_negscc): Disable for arm_restrict_it.
3553         (thumb2_mov_negscc_strict_it): New pattern.
3554         (thumb2_mov_notscc_strict_it): New pattern.
3555         (thumb2_mov_notscc): Disable for arm_restrict_it.
3556         (thumb2_ior_scc): Likewise.
3557         (thumb2_ior_scc_strict_it): New pattern.
3558         (thumb2_cond_move): Adjust for arm_restrict_it.
3559         (thumb2_cond_arith): Disable for arm_restrict_it.
3560         (thumb2_cond_arith_strict_it): New pattern.
3561         (thumb2_cond_sub): Adjust for arm_restrict_it.
3562         (thumb2_movcond): Likewise.
3563         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3564         (thumb2_zero_extendhisi2_v6): Likewise.
3565         (thumb2_zero_extendqisi2_v6): Likewise.
3566         (orsi_notsi_si): Likewise.
3567         (orsi_not_shiftsi_si): Likewise.
3569 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3571         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3572         instruction sequence is 1 byte shorter.
3574 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3576         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3577         it is not needed after split.
3579 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3581         PR target/51784
3582         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3583         second label for nonlocal goto receivers. Don't output pic base labels
3584         unless we're producing PIC; mark that action unreachable().
3585         (ix86_save_reg): If the function contains a nonlocal label, save the
3586         PIC base reg.
3587         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3588         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3589         (update_pic_label_number_if_needed): New.
3590         (machopic_output_function_base_name): Adjust for nonlocal receiver
3591         case.
3592         (machopic_should_output_picbase_label): New.
3593         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3594         (nonlocal_goto_receiver): New insn and split.
3596 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3598         * config/aarch64/aarch64-builtins.c
3599         (aarch64_fold_builtin): Fold abs in all modes.
3600         * config/aarch64/aarch64-simd-builtins.def
3601         (abs): Enable for all modes.
3602         * config/aarch64/arm_neon.h
3603         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3604         (vabs_f64): Add missing intrinsic.
3606 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
3608         * config/aarch64/arm_neon.h (vabs_s64): New function
3610 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
3612         PR target/57516
3613         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3614         * config/avr/avr.md (adjust_len): Add `round'.
3615         * config/avr/avr-protos.h (avr_out_round): New prototype.
3616         (avr_out_plus): Add `out_label' argument.
3617         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3618         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3619         Handle the case where `insn' is just a pattern.
3620         (avr_out_bitop): Handle the case where `insn' is just a pattern.
3621         (avr_out_round): New function.
3622         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3624 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
3626         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3627         Add check for flag_stack_usage to handle -fstack-usage support
3629 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
3631         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3632         interaction for new Power8 flags and VSX.
3634 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
3636         PR middle-end/57698
3637         * tree-inline.c (expand_call_inline): Emit errors during
3638         early_inlining only if optimization is not turned on.
3640 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3642         * passes.def: New.
3644         * passes.c (init_optimization_passes): Move the construction of
3645         the pass hierarchy into a new passes.def file.
3647         * Makefile.in (passes.o): Add dependency on passes.def.
3649 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3651         * passes.c (init_optimization_passes): Introduce macros for
3652         constructing the tree of passes (INSERT_PASSES_AFTER,
3653         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3654         TERMINATE_PASS_LIST).
3656 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
3657             Wei Mi  <wmi@google.com>
3659         PR rtl-optimization/57878
3660         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3661         top.
3662         (reload_pseudo_compare_func): Check nregs first for reload
3663         pseudos.
3665 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3667         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3669 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
3671         * read-rtl.c (validate_const_int): Once an invalid character is
3672         seen, quit the loop.
3674         * gengtype.c (write_roots): Similarly once we find the "deletable"
3675         or "if_marked" option.
3677 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3679         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3680         "xtab" and "sat".  Move value "clz" from here to ...
3681         (attriubte "type"): ... here.
3682         (satsi_<SAT:code>): Delete "insn" attribute.
3683         (satsi_<SAT:code>_shift): Likewise.
3684         (arm_zero_extendqisi2addsi): Likewise.
3685         (arm_extendqisi2addsi): Likewise.
3686         (clzsi2): Update for attribute changes.
3687         (rbitsi2): Likewise.
3688         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3689         attribute.
3690         (arm_usatsihi): Likewise.
3691         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3693 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3695         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3696         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
3697         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
3698         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
3699         in alphabetical order.
3700         (attribute "core_cycles"): Update for attribute changes.
3701         (arm_addsi3): Likewise.
3702         (addsi3_compare0): Likewise.
3703         (addsi3_compare0_scratch): Likewise.
3704         (addsi3_compare_op1): Likewise.
3705         (addsi3_compare_op2): Likewise.
3706         (compare_addsi2_op0): Likewise.
3707         (compare_addsi2_op1): Likewise.
3708         (addsi3_carryin_shift_<optab>): Likewise.
3709         (subsi3_carryin_shift): Likewise.
3710         (rsbsi3_carryin_shift): Likewise.
3711         (arm_subsi3_insn): Likewise.
3712         (subsi3_compare0): Likewise.
3713         (subsi3_compare): Likewise.
3714         (arm_andsi3_insn): Likewise.
3715         (thumb1_andsi3_insn): Likewise.
3716         (andsi3_compare0): Likewise.
3717         (andsi3_compare0_scratch): Likewise.
3718         (zeroextractsi_compare0_scratch
3719         (andsi_not_shiftsi_si): Likewise.
3720         (iorsi3_insn): Likewise.
3721         (iorsi3_compare0): Likewise.
3722         (iorsi3_compare0_scratch): Likewise.
3723         (arm_xorsi3): Likewise.
3724         (thumb1_xorsi3_insn): Likewise.
3725         (xorsi3_compare0): Likewise.
3726         (xorsi3_compare0_scratch): Likewise.
3727         (satsi_<SAT:code>_shift): Likewise.
3728         (rrx): Likewise.
3729         (arm_shiftsi3): Likewise.
3730         (shiftsi3_compare0): Likewise.
3731         (not_shiftsi): Likewise.
3732         (not_shiftsi_compare0): Likewise.
3733         (not_shiftsi_compare0_scratch): Likewise.
3734         (arm_one_cmplsi2): Likewise.
3735         (thumb_one_complsi2): Likewise.
3736         (notsi_compare0): Likewise.
3737         (notsi_compare0_scratch): Likewise.
3738         (thumb1_zero_extendhisi2): Likewise.
3739         (arm_zero_extendhisi2): Likewise.
3740         (arm_zero_extendhisi2_v6): Likewise.
3741         (arm_zero_extendhisi2addsi): Likewise.
3742         (thumb1_zero_extendqisi2): Likewise.
3743         (thumb1_zero_extendqisi2_v6): Likewise.
3744         (arm_zero_extendqisi2): Likewise.
3745         (arm_zero_extendqisi2_v6): Likewise.
3746         (arm_zero_extendqisi2addsi): Likewise.
3747         (thumb1_extendhisi2): Likewise.
3748         (arm_extendhisi2): Likewise.
3749         (arm_extendhisi2_v6): Likewise.
3750         (arm_extendqisi): Likewise.
3751         (arm_extendqisi_v6): Likewise.
3752         (arm_extendqisi2addsi): Likewise.
3753         (thumb1_extendqisi2): Likewise.
3754         (thumb1_movdi_insn): Likewise.
3755         (arm_movsi_insn): Likewise.
3756         (movsi_compare0): Likewise.
3757         (movhi_insn_arch4): Likewise.
3758         (movhi_bytes): Likewise.
3759         (arm_movqi_insn): Likewise.
3760         (thumb1_movqi_insn): Likewise.
3761         (arm32_movhf): Likewise.
3762         (thumb1_movhf): Likewise.
3763         (arm_movsf_soft_insn): Likewise.
3764         (thumb1_movsf_insn): Likewise.
3765         (movdf_soft_insn): Likewise.
3766         (thumb_movdf_insn): Likewise.
3767         (arm_cmpsi_insn): Likewise.
3768         (cmpsi_shiftsi): Likewise.
3769         (cmpsi_shiftsi_swp): Likewise.
3770         (arm_cmpsi_negshiftsi_si): Likewise.
3771         (movsicc_insn): Likewise.
3772         (movsfcc_soft_insn): Likewise.
3773         (arith_shiftsi): Likewise.
3774         (arith_shiftsi_compare0
3775         (arith_shiftsi_compare0_scratch
3776         (sub_shiftsi): Likewise.
3777         (sub_shiftsi_compare0
3778         (sub_shiftsi_compare0_scratch
3779         (and_scc): Likewise.
3780         (cond_move): Likewise.
3781         (if_plus_move): Likewise.
3782         (if_move_plus): Likewise.
3783         (if_move_not): Likewise.
3784         (if_not_move): Likewise.
3785         (if_shift_move): Likewise.
3786         (if_move_shift): Likewise.
3787         (if_shift_shift): Likewise.
3788         (if_not_arith): Likewise.
3789         (if_arith_not): Likewise.
3790         (cond_move_not): Likewise.
3791         (thumb1_ashlsi3): Set type attribute.
3792         (thumb1_ashrsi3): Likewise.
3793         (thumb1_lshrsi3): Likewise.
3794         (thumb1_rotrsi3): Likewise.
3795         (shiftsi3_compare0_scratch): Likewise.
3796         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3797         (neon_mov<mode>): Likewise.
3798         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3799         attribute changes.
3800         (thumb2_movsi_insn): Likewise.
3801         (thumb2_cmpsi_neg_shiftsi): Likewise.
3802         (thumb2_extendqisi_v6): Likewise.
3803         (thumb2_zero_extendhisi2_v6): Likewise.
3804         (thumb2_zero_extendqisi2_v6): Likewise.
3805         (thumb2_shiftsi3_short): Likewise.
3806         (thumb2_addsi3_compare0_scratch): Likewise.
3807         (orsi_not_shiftsi_si): Likewise.
3808         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3809         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3810         changes.
3811         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3812         (1020alu_shift_op): Likewise.
3813         (1020alu_shift_reg_op): Likewise.
3814         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3815         (alu_shift_op): Likewise.
3816         (alu_shift_reg_op): Likewise.
3817         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3818         (11_alu_shift_op): Likewise.
3819         (11_alu_shift_reg_op): Likewise.
3820         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3821         (9_alu_shift_reg_op): Likewise.
3822         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3823         changes.
3824         (cortex_a15_alu_shift): Likewise.
3825         (cortex_a15_alu_shift_reg): Likewise.
3826         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3827         changes.
3828         (cortex_a5_alu_shift): Likewise.
3829         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3830         changes.
3831         (cortex_a53_alu_shift): Likewise.
3832         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3833         changes.
3834         (cortex_a7_alu_reg): Likewise.
3835         (cortex_a7_alu_shift): Likewise.
3836         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3837         changes.
3838         (cortex_a8_alu_shift): Likewise.
3839         (cortex_a8_alu_shift_reg): Likewise.
3840         (cortex_a8_mov): Likewise.
3841         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3842         (cortex_a9_dp_shift): Likewise.
3843         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3844         changes.
3845         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3846         changes.
3847         (cortex_r4_mov): Likewise.
3848         (cortex_r4_alu_shift): Likewise.
3849         (cortex_r4_alu_shift_reg): Likewise.
3850         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3851         (526_alu_shift_op): Likewise.
3852         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3853         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3854         (626te_alu_shift_op): Likewise.
3855         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3856         (726te_alu_op): Likewise.
3857         (726te_alu_shift_op): Likewise.
3858         (726te_alu_shift_reg_op): Likewise.
3859         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3860         (mp626_alu_shift_op): Likewise.
3861         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3862         (pj4_alu_e1_conds): Likewise.
3863         (pj4_alu): Likewise.
3864         (pj4_alu_conds): Likewise.
3865         (pj4_shift): Likewise.
3866         (pj4_shift_conds): Likewise.
3867         (pj4_alu_shift): Likewise.
3868         (pj4_alu_shift_conds): Likewise.
3869         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3870         changes.
3871         (cortexa7_older_only): Likewise.
3872         (cortexa7_younger): Likewise.
3874 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3876         * ipa-pure-const.c (generate_summary): Rename to...
3877         (pure_const_generate_summary): ... this.
3879 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
3881         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3883 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
3885         PR target/57909
3886         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3887         usage in HI mode.
3889 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3891         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3892         enabled without -march=zEC12.
3893         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3894         flags to be set.
3896 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3898         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3899         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3900         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3901         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3902         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3903         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3904         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3905         ISA_HAS_FP4.
3906         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3907         and ISA_HAS_NMADD3_NMSUB3.
3908         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3909         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3910         (nmsub4<mode>, nmsub3<mode>): Likewise.
3911         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3913 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3915         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3916         TARGET_MIPS5400 checking.
3918 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
3919             Peter Bergner  <bergner@vnet.ibm.com>
3921         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3922         registers in the comment.
3923         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3924         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3925         rather than FIRST_PSEUDO_REGISTERS.
3927 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
3929         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3930         enable extra ISA flags with TARGET_HTM.
3932 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3934         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3935         Fix comment typos.
3937 2013-07-15  Cong Hou  <congh@google.com>
3939         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3940         in compare function for sorting.
3942 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
3944         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3945         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3946         * config/rs6000/rs6000.opt: Add -mhtm option.
3947         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3948         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3949         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3950         __HTM__ if the HTM instructions are available.
3951         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3952         htm_spr_reg_operand): New define_predicates.
3953         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3954         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3955         Include htm.md.
3956         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3957         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3958         HTM builtin functions.
3959         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3960         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3961         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3962         (rs6000_builtin_mask_calculate): Likewise.
3963         (rs6000_option_override_internal): Likewise.
3964         (bdesc_htm): Add new HTM builtin support.
3965         (htm_spr_num): New function.
3966         (htm_spr_regno): Likewise.
3967         (rs6000_htm_spr_icode): Likewise.
3968         (htm_expand_builtin): Likewise.
3969         (htm_init_builtins): Likewise.
3970         (rs6000_expand_builtin): Add support for HTM builtin functions.
3971         (rs6000_init_builtins): Likewise.
3972         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3973         option.
3974         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3975         (TARGET_HTM, MASK_HTM): Define macros.
3976         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3977         (FIXED_REGISTERS): Likewise.
3978         (CALL_USED_REGISTERS): Likewise.
3979         (CALL_REALLY_USED_REGISTERS): Likewise.
3980         (REG_ALLOC_ORDER): Likewise.
3981         (enum reg_class): Likewise.
3982         (REG_CLASS_NAMES): Likewise.
3983         (REG_CLASS_CONTENTS): Likewise.
3984         (REGISTER_NAMES): Likewise.
3985         (ADDITIONAL_REGISTER_NAMES): Likewise.
3986         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3987         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3988         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3989         * config/rs6000/htm.md: New file.
3990         * config/rs6000/htmintrin.h: New file.
3991         * config/rs6000/htmxlintrin.h: New file.
3993 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3995         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3996         Define SYMBOL_TINY_GOT, update comment.
3997         * config/aarch64/aarch64.c
3998         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3999         (aarch64_expand_mov_immediate): Likewise.
4000         (aarch64_print_operand): Likewise.
4001         (aarch64_classify_symbol): Likewise.
4002         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4003         (ldr_got_tiny): Define.
4005 2013-07-13  Tobias Grosser  <tobias@grosser.es>
4007         PR tree-optimization/54094
4008         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4009           scheduling dimension for the parallelism check from the polyhedral
4010           information in the AST.
4011         * graphite-dependences.c (carries_deps): Do not assume the schedule is
4012           in 2D + 1 form.
4014 2013-07-13  Jason Merrill  <jason@redhat.com>
4016         * print-tree.c (debug_vec_tree): Use debug_raw.
4017         (debug_raw (vec<tree, va_gc> &)): New.
4018         (debug_raw (vec<tree, va_gc> *)): New.
4019         * tree.h: Declare them.
4021 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
4023         * ifcvt.c (ifcvt_after_combine): New static variable.
4024         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4025         for size.
4026         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
4027         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4028         rest_of_handle_if_after_reload): Pass new argument for if_convert.
4030 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
4032         * config/mips/mips.c (mips_expand_call): Remove empty statement.
4034 2013-07-12  Michael Matz  <matz@suse.de>
4036         PR middle-end/55771
4037         * convert.c (convert_to_real): Reject non-float inner types.
4039 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
4041         * config/aarch64/aarch64-protos.h
4042         (aarch64_simd_immediate_valid_for_move): Remove.
4043         * config/aarch64/aarch64.c (simd_immediate_info): New member.
4044         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4045         cases.
4046         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4048 2013-07-11  Steve Ellcey  <sellcey@mips.com>
4050         * config/mips/mips.c (mips_conditional_register_usage): Do not
4051         use t[0-7] registers in MIPS16 mode when optimizing for size.
4053 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
4055         * config/i386/i386.c (dispatch_function_versions): Fix array
4056         indexing of function_version_info to match actual_versions.
4058 2013-07-11  Teresa Johnson  <tejohnson@google.com>
4060         * vec.h (struct va_gc): Move release out-of-line.
4061         (va_gc::release): Call ggc_free on released vec.
4063 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4065         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4066         Require GOT register as additional operand in UNSPEC.
4067         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4068         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4069         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4070         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4071         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4072         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4073         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4075 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
4077         PR target/57631
4078         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4079         name seen by assembler/linker rather if available.
4081 2013-07-11  Andreas Schwab  <schwab@suse.de>
4083         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4085 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4087         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4089 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4091         * doc/tm.texi.in: Move hook documentation to ....
4092         * target.def: ... here.
4094         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4095         text on @hook line.
4096         * doc/tm.texi: Regenerate.
4098 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
4100         PR c++/57869
4101         * doc/invoke.texi: Document -Wconditionally-supported.
4103 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4105         PR target/57844
4106         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4107         of my_fp.
4109 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4111         PR target/57506
4112         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4113         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4114         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4115         Remove duplicate devices.
4116         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4117         * config/avr/t-multilib: Regenerate.
4118         * config/avr/avr-tables.opt: Regenerate.
4119         * doc/avr-mmcu.texi: Regenerate.
4121 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4123         PR target/56987
4124         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4126 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
4128         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4129         the cost of MULT when optimizing for size.
4131 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4133         * config/cr16/cr16-protos.h: Don't include target.h.
4135 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
4137         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4138         adjust register size for TDmode and TFmode for VSX registers.
4140 2013-07-08  Kai Tietz  <ktietz@redhat.com>
4142         PR target/56892
4143         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4144         hook_bool_const_tree_true.
4146 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4148         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4149         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4150         * config/s390/s390.md: Define FPR*_REGNUM constants.
4151         Fix FPR2_REGNUM constant (18 -> 17).
4152         ("*trunc<BFP:mode><DFP_ALL:mode>2")
4153         ("*trunc<DFP_ALL:mode><BFP:mode>2")
4154         ("trunc<BFP:mode><DFP_ALL:mode>2")
4155         ("trunc<DFP_ALL:mode><BFP:mode>2")
4156         ("*extend<BFP:mode><DFP_ALL:mode>2")
4157         ("*extend<DFP_ALL:mode><BFP:mode>2")
4158         ("extend<BFP:mode><DFP_ALL:mode>2")
4159         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4160         FPR4_REGNUM.
4162 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
4164         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4166 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4168         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4169         and cfun_fpr_bit_p to cfun_fpr_save_p.
4170         (s390_frame_area, s390_register_info, s390_frame_info)
4171         (s390_emit_prologue, s390_emit_epilogue)
4172         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4173         register numbers.
4174         * config/s390/s390.h: Define *_REGNUM macros for floating point
4175         register numbers.
4177 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
4179         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4181 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
4183         PR rtl-optimization/57786
4184         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4185         and break out of the loop when it is set to false.
4187 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
4189         PR target/57819
4190         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4191         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4192         (const_int 63)) 0)).
4193         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4194         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4195         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4197         PR rtl-optimization/57829
4198         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4199         mask bits outside of mode are just sign-extension from mode to HWI.
4201 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
4203         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4204         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4205         adjust_address instead of change_address to keep info about alignment.
4206         (emit_strmov): Remove.
4207         (emit_memmov): New function.
4208         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4209         (expand_movmem_epilogue): Likewise and return updated rtx for
4210         destination.
4211         (expand_constant_movmem_prologue): Likewise and return updated rtx for
4212         destination and source.
4213         (decide_alignment): Refactor, handle vector_loop.
4214         (ix86_expand_movmem): Likewise.
4215         (ix86_expand_setmem): Likewise.
4216         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4218 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
4220         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4221         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4223 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4225         * config/i386/sse.md (sse_movlhps): Change alternative 3
4226         of operand 2 to "m".
4228 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4230         PR target/57807
4231         * config/i386/sse.md (iptr): New mode attribute.
4232         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4233         (<sse>_vm<plusminus_insn><mode>3): Ditto.
4234         (<sse>_vmmul<mode>3): Ditto.
4235         (<sse>_vmdiv<mode>3): Ditto.
4236         (sse_vmrcpv4sf2): Ditto.
4237         (<sse>_vmsqrt<mode>2): Ditto.
4238         (sse_vmrsqrtv4sf2): Ditto.
4239         (<sse>_vm<code><mode>3): Ditto.
4240         (avx_vmcmp<mode>3): Ditto.
4241         (<sse>_vmmaskcmp<mode>3): Ditto.
4242         (<sse>_comi): Ditto.
4243         (<sse>_ucomi): Ditto.
4244         (*xop_vmfrcz_<mode>): Ditto.
4245         (*fmai_fmadd_<mode>): Ditto.
4246         (*fmai_fmsub_<mode>): Ditto.
4247         (*fmai_fnmadd_<mode>): Ditto.
4248         (*fmai_fnmsub_<mode>): Ditto.
4249         (*fma4i_vmfmadd_<mode>): Ditto.
4250         (*fma4i_vmfmsub_<mode>): Ditto.
4251         (*fma4i_vmfnmadd_<mode>): Ditto.
4252         (*fma4i_vmfnmsub_<mode>): Ditto.
4253         (*xop_vmfrcz_<mode>): Ditto.
4254         (sse_cvtps2pi): Ditto.
4255         (sse_cvttps2pi): Ditto.
4256         (sse_cvtss2si): Ditto.
4257         (sse_cvtss2si_2): Ditto.
4258         (sse_cvtss2siq_2): Ditto.
4259         (sse_cvttss2si): Ditto.
4260         (sse_cvttss2siq): Ditto.
4261         (sse_cvtsd2si): Ditto.
4262         (sse_cvtsd2si_2): Ditto.
4263         (sse_cvtsd2siq_2): Ditto.
4264         (sse_cvttsd2si): Ditto.
4265         (sse_cvttsd2siq): Ditto.
4266         (sse_cvtsd2ss): Ditto.
4267         (sse_cvtss2sd): Ditto.
4268         (avx2_pbroadcast<mode>): Ditto.
4269         (avx2_pbroadcast<mode>_1): Ditto.
4270         (*avx_vperm_broadcast_v4sf): Ditto.
4272         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4273         (sse_movlhps): Ditto.
4274         (sse_storehps): Ditto.
4275         (sse_loadhps): Ditto.
4276         (sse_storelps): Ditto.
4277         (sse_loadlps): Ditto.
4278         (*vec_concatv4sf): Ditto.
4279         (*vec_interleave_highv2df): Ditto.
4280         (*vec_interleave_lowv2df): Ditto.
4281         (*vec_extractv2df_1_sse): Ditto.
4282         (*vec_extractv2df_0_sse): Ditto.
4283         (sse2_storelpd): Ditto.
4284         (sse2_loadlpd): Ditto.
4285         (sse2_movsd): Ditto.
4286         (*vec_concatv4si): Ditto.
4287         (vec_concatv2di): Ditto.
4289         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4290         for Intel asm dialect.
4291         (mmx_punpcklwd): Ditto.
4292         (mmx_punpckldq): Ditto.
4294         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4295         for intel assembler dialect.
4297 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
4299         PR target/29776
4300         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4301         for BUILT_IN_C{LZ,LRSB}*.
4302         * tree.h (CASE_INT_FN): Add FN##IMAX case.
4303         * tree-vrp.c (extract_range_basic): Handle
4304         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
4305         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4306         fall thru to code calling set_value*.
4307         * builtins.c (expand_builtin): Remove *IMAX cases.
4308         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4309         if width is bigger than 2*HWI.
4311 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
4313         PR rtl-optimization/55342
4314         * lra-int.h (lra_subreg_reload_pseudos): New.
4315         * lra.c: Add undoing optional reloads to the block diagram.
4316         (lra_subreg_reload_pseudos): New.
4317         (lra_optional_reload_pseudos): Change comments.
4318         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
4319         lra_optional_reload_pseudos after undo transformations.
4320         * lra-assigns.c (pseudo_prefix_title): New.
4321         (lra_setup_reg_renumber): Use it.
4322         (spill_for): Ditto.  Check subreg reload pseudos too.
4323         (assign_by_spills): Consider subreg reload pseudos too.
4324         * lra-constraints.c (simplify_operand_subreg): Use
4325         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4326         (curr_insn_transform): Recognize and do optional reloads.
4327         (undo_optional_reloads): New.
4328         (lra_undo_inheritance): Call undo_optional_reloads.
4330 2013-07-05  Thomas Quinot  <quinot@adacore.com>
4332         * tree-complex.c (expand_complex_operations_1): Fix typo.
4334 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
4336         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4337         (tune_params): New member 'const vec_costs'.
4338         * config/aarch64/aarch64.c (generic_vector_cost): New.
4339         (generic_tunings): New member 'generic_vector_cost'.
4340         (aarch64_builtin_vectorization_cost): New.
4341         (aarch64_add_stmt_cost): New.
4342         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4343         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4345 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4347         PR target/57777
4348         * config/i386/predicates.md (vsib_address_operand): Disallow
4349         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4351 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4353         PR middle-end/55030
4354         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4355         expand_builtin_setjmp_receiver.
4356         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4357         with NULL for the label parameter.
4358         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4359         the frame-pointer.  Adjust comments.
4360         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4361         only if LABEL is non-NULL.
4363 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4365         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4366         (ARM_ABI_AAPCS64): Ditto.
4367         (arm_abi): Ditto.
4368         (ARM_DEFAULT_ABI): Ditto.
4370 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4372         * config/aarch64/aarch64-builtins.c
4373         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4374         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4375         (st1): Likewise.
4376         * config/aarch64/aarch64-simd.md
4377         (aarch64_ld1<VALL:mode>): New.
4378         (aarch64_st1<VALL:mode>): Likewise.
4379         * config/aarch64/arm_neon.h
4380         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4382 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4384         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4385         target ISA is AVX.
4386         (ix86_option_override_internal):Turn on all -mavx target flags by
4387         default as they are dependent on AVX anyway.
4389 2013-07-02  Cary Coutant  <ccoutant@google.com>
4391         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4392         deterministic hash.
4393         (loc_checksum_ordered): Likewise.
4394         (hash_loc_operands): Remove inline keyword.
4396 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4398         PR tree-optimization/57741
4399         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4400         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4401         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4402         Allow REAL_CST step_exprs if flag_associative_math.
4403         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4405 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4407         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4409 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4411         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4413 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4415         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4416         encoding.
4417         (iorsi3_insn): Likewise.
4418         (arm_xorsi3): Likewise.
4420 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4422         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4423         to ...
4424         (attribute "type"): ... here, and prefix with "wmmx_".
4425         (attribute "core_cycles"): Update for attribute changes.
4426         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4427         (tbcstv4hi): Likewise.
4428         (tbcstv2si): Likewise.
4429         (iwmmxt_iordi3): Likewise.
4430         (iwmmxt_xordi3): Likewise.
4431         (iwmmxt_anddi3): Likewise.
4432         (iwmmxt_nanddi3): Likewise.
4433         (iwmmxt_arm_movdi): Likewise.
4434         (iwmmxt_movsi_insn): Likewise.
4435         (mov<mode>_internal): Likewise.
4436         (and<mode>3_iwmmxt): Likewise.
4437         (ior<mode>3_iwmmxt): Likewise.
4438         (xor<mode>3_iwmmxt): Likewise.
4439         (add<mode>3_iwmmxt): Likewise.
4440         (ssaddv8qi3): Likewise.
4441         (ssaddv4hi3): Likewise.
4442         (ssaddv2si3): Likewise.
4443         (usaddv8qi3): Likewise.
4444         (usaddv4hi3): Likewise.
4445         (usaddv2si3): Likewise.
4446         (sub<mode>3_iwmmxt): Likewise.
4447         (sssubv8qi3): Likewise.
4448         (sssubv4hi3): Likewise.
4449         (sssubv2si3): Likewise.
4450         (ussubv8qi3): Likewise.
4451         (ussubv4hi3): Likewise.
4452         (ussubv2si3): Likewise.
4453         (mulv4hi3_iwmmxt): Likewise.
4454         (smulv4hi3_highpart): Likewise.
4455         (umulv4hi3_highpart): Likewise.
4456         (iwmmxt_wmacs): Likewise.
4457         (iwmmxt_wmacsz): Likewise.
4458         (iwmmxt_wmacu): Likewise.
4459         (iwmmxt_wmacuz): Likewise.
4460         (iwmmxt_clrdi): Likewise.
4461         (iwmmxt_clrv8qi): Likewise.
4462         (iwmmxt_clr4hi): Likewise.
4463         (iwmmxt_clr2si): Likewise.
4464         (iwmmxt_uavgrndv8qi3): Likewise.
4465         (iwmmxt_uavgrndv4hi3): Likewise.
4466         (iwmmxt_uavgv8qi3): Likewise.
4467         (iwmmxt_uavgv4hi3): Likewise.
4468         (iwmmxt_tinsrb): Likewise.
4469         (iwmmxt_tinsrh): Likewise.
4470         (iwmmxt_tinsrw): Likewise.
4471         (iwmmxt_textrmub): Likewise.
4472         (iwmmxt_textrmsb): Likewise.
4473         (iwmmxt_textrmuh): Likewise.
4474         (iwmmxt_textrmsh): Likewise.
4475         (iwmmxt_textrmw): Likewise.
4476         (iwmxxt_wshufh): Likewise.
4477         (eqv8qi3): Likewise.
4478         (eqv4hi3): Likewise.
4479         (eqv2si3): Likewise.
4480         (gtuv8qi3): Likewise.
4481         (gtuv4hi3): Likewise.
4482         (gtuv2si3): Likewise.
4483         (gtv8qi3): Likewise.
4484         (gtv4hi3): Likewise.
4485         (gtv2si3): Likewise.
4486         (smax<mode>3_iwmmxt): Likewise.
4487         (umax<mode>3_iwmmxt): Likewise.
4488         (smin<mode>3_iwmmxt): Likewise.
4489         (umin<mode>3_iwmmxt): Likewise.
4490         (iwmmxt_wpackhss): Likewise.
4491         (iwmmxt_wpackwss): Likewise.
4492         (iwmmxt_wpackdss): Likewise.
4493         (iwmmxt_wpackhus): Likewise.
4494         (iwmmxt_wpackwus): Likewise.
4495         (iwmmxt_wpackdus): Likewise.
4496         (iwmmxt_wunpckihb): Likewise.
4497         (iwmmxt_wunpckihh): Likewise.
4498         (iwmmxt_wunpckihw): Likewise.
4499         (iwmmxt_wunpckilb): Likewise.
4500         (iwmmxt_wunpckilh): Likewise.
4501         (iwmmxt_wunpckilw): Likewise.
4502         (iwmmxt_wunpckehub): Likewise.
4503         (iwmmxt_wunpckehuh): Likewise.
4504         (iwmmxt_wunpckehuw): Likewise.
4505         (iwmmxt_wunpckehsb): Likewise.
4506         (iwmmxt_wunpckehsh): Likewise.
4507         (iwmmxt_wunpckehsw): Likewise.
4508         (iwmmxt_wunpckelub): Likewise.
4509         (iwmmxt_wunpckeluh): Likewise.
4510         (iwmmxt_wunpckeluw): Likewise.
4511         (iwmmxt_wunpckelsb): Likewise.
4512         (iwmmxt_wunpckelsh): Likewise.
4513         (iwmmxt_wunpckelsw): Likewise.
4514         (ror<mode>3): Likewise.
4515         (ashr<mode>3_iwmmxt): Likewise.
4516         (lshr<mode>3_iwmmxt): Likewise.
4517         (ashl<mode>3_iwmmxt): Likewise.
4518         (ror<mode>3_di): Likewise.
4519         (ashr<mode>3_di): Likewise.
4520         (lshr<mode>3_di): Likewise.
4521         (ashl<mode>3_di): Likewise.
4522         (iwmmxt_wmadds): Likewise.
4523         (iwmmxt_wmaddu): Likewise.
4524         (iwmmxt_tmia): Likewise.
4525         (iwmmxt_tmiaph): Likewise.
4526         (iwmmxt_tmiabb): Likewise.
4527         (iwmmxt_tmiatb): Likewise.
4528         (iwmmxt_tmiabt): Likewise.
4529         (iwmmxt_tmiatt): Likewise.
4530         (iwmmxt_tmovmskb): Likewise.
4531         (iwmmxt_tmovmskh): Likewise.
4532         (iwmmxt_tmovmskw): Likewise.
4533         (iwmmxt_waccb): Likewise.
4534         (iwmmxt_wacch): Likewise.
4535         (iwmmxt_waccw): Likewise.
4536         (iwmmxt_waligni): Likewise.
4537         (iwmmxt_walignr): Likewise.
4538         (iwmmxt_walignr0): Likewise.
4539         (iwmmxt_walignr1): Likewise.
4540         (iwmmxt_walignr2): Likewise.
4541         (iwmmxt_walignr3): Likewise.
4542         (iwmmxt_wsadb): Likewise.
4543         (iwmmxt_wsadh): Likewise.
4544         (iwmmxt_wsadbz): Likewise.
4545         (iwmmxt_wsadhz): Likewise.
4546         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4547         (iwmmxt_wabsdiffb): Likewise.
4548         (iwmmxt_wabsdiffh): Likewise.
4549         (iwmmxt_wabsdiffw): Likewise.
4550         (iwmmxt_waddsubhx): Likewise
4551         (iwmmxt_wsubaddhx): Likewise.
4552         (addc<mode>3): Likewise.
4553         (iwmmxt_avg4): Likewise.
4554         (iwmmxt_avg4r): Likewise.
4555         (iwmmxt_wmaddsx): Likewise.
4556         (iwmmxt_wmaddux): Likewise.
4557         (iwmmxt_wmaddsn): Likewise.
4558         (iwmmxt_wmaddun): Likewise.
4559         (iwmmxt_wmulwsm): Likewise.
4560         (iwmmxt_wmulwum): Likewise.
4561         (iwmmxt_wmulsmr): Likewise.
4562         (iwmmxt_wmulumr): Likewise.
4563         (iwmmxt_wmulwsmr): Likewise.
4564         (iwmmxt_wmulwumr): Likewise.
4565         (iwmmxt_wmulwl): Likewise.
4566         (iwmmxt_wqmulm): Likewise.
4567         (iwmmxt_wqmulwm): Likewise.
4568         (iwmmxt_wqmulmr): Likewise.
4569         (iwmmxt_wqmulwmr): Likewise.
4570         (iwmmxt_waddbhusm): Likewise.
4571         (iwmmxt_waddbhusl): Likewise.
4572         (iwmmxt_wqmiabb): Likewise.
4573         (iwmmxt_wqmiabt): Likewise.
4574         (iwmmxt_wqmiatb): Likewise.
4575         (iwmmxt_wqmiatt): Likewise.
4576         (iwmmxt_wqmiabbn): Likewise.
4577         (iwmmxt_wqmiabtn): Likewise.
4578         (iwmmxt_wqmiatbn): Likewise.
4579         (iwmmxt_wqmiattn): Likewise.
4580         (iwmmxt_wmiabb): Likewise.
4581         (iwmmxt_wmiabt): Likewise.
4582         (iwmmxt_wmiatb): Likewise.
4583         (iwmmxt_wmiatt): Likewise.
4584         (iwmmxt_wmiabbn): Likewise.
4585         (iwmmxt_wmiabtn): Likewise.
4586         (iwmmxt_wmiatbn): Likewise.
4587         (iwmmxt_wmiattn): Likewise.
4588         (iwmmxt_wmiawbb): Likewise.
4589         (iwmmxt_wmiawbt): Likewise.
4590         (iwmmxt_wmiawtb): Likewise.
4591         (iwmmxt_wmiawtt): Likewise.
4592         (iwmmxt_wmiawbbn): Likewise.
4593         (iwmmxt_wmiawbtn): Likewise.
4594         (iwmmxt_wmiawtbn): Likewise.
4595         (iwmmxt_wmiawttn): Likewise.
4596         (iwmmxt_wmerge): Likewise.
4597         (iwmmxt_tandc<mode>3): Likewise.
4598         (iwmmxt_torc<mode>3): Likewise.
4599         (iwmmxt_torvsc<mode>3): Likewise.
4600         (iwmmxt_textrc<mode>3): Likewise.
4601         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4602         (wmmxt_pack): Likewise.
4603         (wmmxt_mult_c1): Likewise.
4604         (wmmxt_mult_c2): Likewise.
4605         (wmmxt_alu_c1): Likewise.
4606         (wmmxt_alu_c2): Likewise.
4607         (wmmxt_alu_c3): Likewise.
4608         (wmmxt_transfer_c1): Likewise.
4609         (wmmxt_transfer_c2): Likewise.
4610         (wmmxt_transfer_c3): Likewise.
4611         (marvell_f_iwmmxt_wstr): Likewise.
4612         (marvell_f_iwmmxt_wldr): Likewise.
4614 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
4616         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4618 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4620         Revert:
4621         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4622         * lra-constraints.c (need_for_split_p): Check call used hard regs
4623         living through calls.
4625         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4626         call used regs for call insn.
4628 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
4630         PR target/57736
4631         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4632         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4634 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
4636         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4638 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4640         * lra-constraints.c (need_for_split_p): Check call used hard regs
4641         living through calls.
4643 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
4645         PR target/57744
4646         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4647         to tie with any other modes.  Eliminate Altivec vector mode tests,
4648         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
4649         code, to return 0 if testing MODE2 for a condition, if we've
4650         already tested MODE1 for the same condition.
4652 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4654         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4655         layout.
4657 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4659         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4660         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4662 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4664         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4665         Define.
4666         (aarch64_symbolic_constant_p): Remove.
4667         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4668         static.  Fix line length and white space.
4669         (aarch64_symbolic_constant_p): Remove.
4670         * config/aarch64/predicates.md (aarch64_valid_symref):
4671         Use aarch64_classify_symbol_expression.
4673 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4675         * config/arm/constraints.md (Ts): New constraint.
4676         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4677         16-bit encodings.
4678         (compare_scc): Use "Ts" constraint for operand 0.
4679         (ior_scc_scc): Likewise.
4680         (and_scc_scc): Likewise.
4681         (and_scc_scc_nodom): Likewise.
4682         (ior_scc_scc_cmp): Likewise for operand 7.
4683         (and_scc_scc_cmp): Likewise.
4684         * config/arm/thumb2.md (thumb2_movsi_insn):
4685         Add alternatives for 16-bit encodings.
4686         (thumb2_movhi_insn): Likewise.
4687         (thumb2_movsicc_insn): Likewise.
4688         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
4689         (thumb2_negscc): Use "Ts" constraint.
4690         Move mvn instruction outside cond_exec block.
4691         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4692         for 16-bit encodings.
4694 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4696         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4697         encoding.
4698         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4699         (mulsi3subsi): Likewise.
4700         (mulsidi3adddi): Likewise.
4701         (mulsidi3_v6): Likewise.
4702         (umulsidi3_v6): Likewise.
4703         (umulsidi3adddi_v6): Likewise.
4704         (smulsi3_highpart_v6): Likewise.
4705         (umulsi3_highpart_v6): Likewise.
4706         (mulhisi3tb): Likewise.
4707         (mulhisi3bt): Likewise.
4708         (mulhisi3tt): Likewise.
4709         (maddhisi4): Likewise.
4710         (maddhisi4tb): Likewise.
4711         (maddhisi4tt): Likewise.
4712         (maddhidi4): Likewise.
4713         (maddhidi4tb): Likewise.
4714         (maddhidi4tt): Likewise.
4715         (zeroextractsi_compare0_scratch): Likewise.
4716         (insv_zero): Likewise.
4717         (insv_t2): Likewise.
4718         (anddi_notzesidi_di): Likewise.
4719         (anddi_notsesidi_di): Likewise.
4720         (andsi_notsi_si): Likewise.
4721         (iordi_zesidi_di): Likewise.
4722         (xordi_zesidi_di): Likewise.
4723         (andsi_iorsi3_notsi): Likewise.
4724         (smax_0): Likewise.
4725         (smax_m1): Likewise.
4726         (smin_0): Likewise.
4727         (not_shiftsi): Likewise.
4728         (unaligned_loadsi): Likewise.
4729         (unaligned_loadhis): Likewise.
4730         (unaligned_loadhiu): Likewise.
4731         (unaligned_storesi): Likewise.
4732         (unaligned_storehi): Likewise.
4733         (extv_reg): Likewise.
4734         (extzv_t2): Likewise.
4735         (divsi3): Likewise.
4736         (udivsi3): Likewise.
4737         (arm_zero_extendhisi2addsi): Likewise.
4738         (arm_zero_extendqisi2addsi): Likewise.
4739         (compareqi_eq0): Likewise.
4740         (arm_extendhisi2_v6): Likewise.
4741         (arm_extendqisi2addsi): Likewise.
4742         (arm_movt): Likewise.
4743         (thumb2_ldrd): Likewise.
4744         (thumb2_ldrd_base): Likewise.
4745         (thumb2_ldrd_base_neg): Likewise.
4746         (thumb2_strd): Likewise.
4747         (thumb2_strd_base): Likewise.
4748         (thumb2_strd_base_neg): Likewise.
4749         (arm_negsi2): Add alternative for 16-bit encoding.
4750         (arm_one_cmplsi2): Likewise.
4752 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4754         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4755         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4756         (movdfcc): Likewise.
4757         * config/arm/vfp.md (*thumb2_movsf_vfp):
4758         Disable predication for arm_restrict_it.
4759         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4760         (*thumb2_movdfcc_vfp): Likewise.
4761         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4762         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4763         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4764         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4765         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4766         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4767         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4768         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4769         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4770         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4771         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4772         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4773         Disable predication for arm_restrict_it.
4775 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
4777         * config/i386/bmiintrin.h (_bextr_u32): New.
4778         (_bextr_u64): Ditto.
4780 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
4782         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4783         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4784         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4785         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4786         * config/mips/n32-elf.h: ...this new file.
4788 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
4790         PR target/57224
4791         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4792         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4794 2013-06-27  Catherine Moore <clm@codesourcery.com>
4796         * config/mips/mips-tables.opt: Regenerate.
4797         * config/mips/mips-cpus.def: Add m14ke and m14kec.
4798         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4799         * doc/invoke.texi: Add -m14kc.
4801 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4803         PR target/57623
4804         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4805         constraints of operand 1 and 2.
4807         PR target/57623
4808         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4809         to match RTL canonicalization.  Swap predicates and
4810         constraints of operand 1 and 2.
4812 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
4814         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4815         Process OP_INOUT regs for splitting too.
4817 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4819         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4820         decl before the loop, initialize to NULL.
4821         (vectorizable_load): Initialize ptr_incr to NULL.
4823 2013-06-27  Martin Jambor  <mjambor@suse.cz>
4825         PR lto/57208
4826         * ipa-ref.h (ipa_maybe_record_reference): Declare.
4827         * ipa-ref.c (ipa_maybe_record_reference): New function.
4828         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4829         * ipa-cp.c (create_specialized_node): Record potential references from
4830         aggvals.
4831         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4833 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4835         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4836         parameter 'mode' of type 'enum machine_mode mode'; change to pass
4837         'mode' to force_reg.
4838         (aarch64_add_offset): Update calls to aarch64_force_temporary.
4839         (aarch64_expand_mov_immediate): Likewise.
4841 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4843         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4844         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4846 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4848         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4849         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4850         (struct machine_function): Add tbegin_p.
4851         (s390_canonicalize_comparison): Fold CC mode compares to
4852         conditional jump if possible.
4853         (s390_emit_jump): Return the emitted jump.
4854         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4855         Handle CCRAWmode compares.
4856         (s390_option_override): Default to -mhtm if available.
4857         (s390_reg_clobbered_rtx): Handle floating point regs as well.
4858         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4859         FPRs instead of df_regs_ever_live_p.
4860         (s390_optimize_nonescaping_tx): New function.
4861         (s390_init_frame_layout): Extend clobbered_regs array to cover
4862         FPRs as well.
4863         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4864         (s390_expand_tbegin): New function.
4865         (enum s390_builtin): New enum definition.
4866         (code_for_builtin): New array definition.
4867         (s390_init_builtins): New function.
4868         (s390_expand_builtin): New function.
4869         (TARGET_INIT_BUILTINS): Define.
4870         (TARGET_EXPAND_BUILTIN): Define.
4871         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4872         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4873         (s390_alc_comparison): Likewise.
4874         * config/s390/s390-modes.def: Add CCRAWmode.
4875         * config/s390/s390.h (processor_flags): Add PF_TX.
4876         (TARGET_CPU_HTM): Define macro.
4877         (TARGET_HTM): Define macro.
4878         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4879         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4880         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4881         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4882         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4883         (TBEGIN_MASK, TBEGINC_MASK): New constants.
4884         ("*cc_to_int"): Move up.
4885         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4886         constants other than 0.
4887         ("*ccraw_to_int"): New insn and splitter definition.
4888         ("tbegin", "tbegin_nofloat", "tbegin_retry")
4889         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4890         ("tx_assist"): New expander.
4891         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4892         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4893         * config/s390/s390.opt: Add -mhtm option.
4894         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4895         * config/s390/htmxlintrin.h: New file.
4896         * config/s390/htmintrin.h: New file.
4897         * config/s390/s390intrin.h: New file.
4898         * doc/extend.texi: Document htm builtins.
4899         * config.gcc: Add the new header files to extra_headers.
4901 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
4903         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4904         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4906 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
4907             Pat Haugen <pthaugen@us.ibm.com>
4908             Peter Bergner <bergner@vnet.ibm.com>
4910         * config/rs6000/power8.md: New.
4911         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4912         setting for power8 entry.
4913         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4914         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4915         test for Power4/Power5 only.
4916         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4917         support.
4918         (force_new_group): Adjust comment.
4919         * config/rs6000/rs6000.md: Include power8.md.
4921 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
4923         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4924         * config/arm/arm-protos.h (arm_max_conditional_execute): New
4925         declaration.
4926         (tune_params): Update comment.
4927         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4928         (arm_max_conditional_execute): New function.
4929         (thumb2_final_prescan_insn): Use max_insn_skipped and
4930         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4932 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
4934         PR tree-optimization/57705
4935         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4936         SSA_NAME step, provided that it is not defined inside the loop.
4937         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4938         (get_initial_def_for_induction): Handle SSA_NAME IV step.
4940 2013-06-25  Martin Jambor  <mjambor@suse.cz>
4942         PR middle-end/57670
4943         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4944         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4945         calls in the dump.
4946         (ipa_note_param_call): Initialize member_ptr flag.
4947         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4948         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4949         (ipa_write_indirect_edge_info): Stream member_ptr flag.
4950         (ipa_read_indirect_edge_info): Likewise.
4952 2013-06-25  Richard Biener  <rguenther@suse.de>
4954         PR middle-end/56977
4955         * passes.c (init_optimization_passes): Move pass_fold_builtins
4956         and pass_dce earlier with -Og.
4958 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
4960         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4961         <BIT_FIELD_REF>: Remove trailing TAB.
4962         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4963         remove blank line.
4965 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4967         PR tree-optimization/57358
4968         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4969         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4970         (ipa_analyze_params_uses): Generate pessimistic info when true.
4972 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4974         PR tree-optimization/57539
4975         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4976         global.inlined_to of the new node to it.  All callers changed.
4977         * ipa-inline-transform.c (clone_inlined_nodes): New variable
4978         inlining_into, pass it to cgraph_clone_node.
4979         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4980         ipa_free_edge_args_substructures.
4981         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4982         rdesc linked list.  Do not assert rdesc edges have inlined caller.
4983         Assert we have found an rdesc in the rdesc list.
4985 2013-06-24  Richard Biener  <rguenther@suse.de>
4987         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4988         (pointer_set_lookup): Declare.
4989         (class pointer_map): New template class implementing a
4990         generic pointer to T map.
4991         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4992         pointer_map<T>::contains, pointer_map<T>::insert,
4993         pointer_map<T>::traverse): New functions.
4994         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4995         (pointer_set_lookup): New function.
4996         (pointer_set_contains): Use pointer_set_lookup.
4997         (pointer_set_insert): Likewise.
4998         (insert_aux): Remove.
4999         (struct pointer_map_t): Embed a pointer_set_t.
5000         (pointer_map_create): Adjust.
5001         (pointer_map_destroy): Likewise.
5002         (pointer_map_contains): Likewise.
5003         (pointer_map_insert): Likewise.
5004         (pointer_map_traverse): Likewise.
5005         * tree-streamer.h (struct streamer_tree_cache_d): Use a
5006         pointer_map<unsigned> instead of a pointer_map_t.
5007         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5008         (streamer_tree_cache_lookup): Likewise.
5009         (streamer_tree_cache_create): Likewise.
5010         (streamer_tree_cache_delete): Likewise.
5011         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5012         pointer_map<unsigned> instead of a pointer_map_t.
5013         (lto_init_tree_ref_encoder): Adjust.
5014         (lto_destroy_tree_ref_encoder): Likewise.
5015         * lto-section-out.c (lto_output_decl_index): Likewise.
5016         (lto_record_function_out_decl_state): Likewise.
5017         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5019 2013-06-24  Richard Biener  <rguenther@suse.de>
5021         PR tree-optimization/57488
5022         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5024 2013-06-24  Alan Modra  <amodra@gmail.com>
5026         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5027         (gen_easy_altivec_constant): Likewise.
5028         * config/rs6000/predicates.md (easy_vector_constant_add_self,
5029         easy_vector_constant_msb): Likewise.
5031 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
5033         PR target/57688
5034         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5035         add missing return true.
5037 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
5039         PR target/52483
5040         * config/sh/predicates.md (general_extend_operand): Invoke
5041         general_movsrc_operand for memory operands.
5042         (general_movsrc_operand): Allow reg+reg addressing, do not use
5043         general_operand for memory operands.
5045 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
5047         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
5048         when current target options does not apply.
5049         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5050         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5051         * config/i386/bmiintrin.h: Pass appropriate target
5052         attributes to header.
5053         * config/i386/mmintrin.h: Ditto.
5054         * config/i386/nmmintrin.h: Ditto.
5055         * config/i386/avx2intrin.h: Ditto.
5056         * config/i386/fxsrintrin.h: Ditto.
5057         * config/i386/tbmintrin.h: Ditto.
5058         * config/i386/xsaveintrin.h: Ditto.
5059         * config/i386/f16cintrin.h: Ditto.
5060         * config/i386/xtestintrin.h: Ditto.
5061         * config/i386/xsaveoptintrin.h: Ditto.
5062         * config/i386/bmi2intrin.h: Ditto.
5063         * config/i386/lzcntintrin.h: Ditto.
5064         * config/i386/smmintrin.h: Ditto.
5065         * config/i386/wmmintrin.h: Ditto.
5066         * config/i386/x86intrin.h: Remove all header include guards.
5067         * config/i386/prfchwintrin.h: Ditto.
5068         * config/i386/pmmintrin.h: Ditto.
5069         * config/i386/tmmintrin.h: Ditto.
5070         * config/i386/xmmintrin.h: Ditto.
5071         * config/i386/popcntintrin.h: Ditto.
5072         * config/i386/rdseedintrin.h: Ditto.
5073         * config/i386/ammintrin.h: Ditto.
5074         * config/i386/emmintrin.h: Ditto.
5075         * config/i386/immintrin.h: Remove all header include guards.
5076         * config/i386/fma4intrin.h: Ditto.
5077         * config/i386/lwpintrin.h: Ditto.
5078         * config/i386/xopintrin.h: Ditto.
5079         * config/i386/ia32intrin.h: Ditto.
5080         * config/i386/avxintrin.h: Ditto.
5081         * config/i386/rtmintrin.h: Ditto.
5082         * config/i386/fmaintrin.h: Ditto.
5083         * config/i386/mm3dnow.h: Ditto.
5085 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5087         * common/config/i386/i386-common.c: Handle LZCNT.
5089 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5091         * doc/extend.texi: Use __atomic_store_n instead of
5092         __atomic_store in HLE example.
5094 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
5096         * config/sh/sh.c: Remove <cstdlib> workaround.
5098 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5100         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
5102 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5104         * doc/extend.texi: Document that __atomic_clear and
5105           __atomic_test_and_set should only be used with bool.
5107 2013-06-20  Jan Hubicka  <jh@suse.cz>
5109         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5110         types_same_for_odr.
5111         * tree.c (decls_same_for_odr): New function.
5112         (same_for_edr): New function.
5113         (types_same_for_odr): New function.
5114         (get_binfo_at_offset): Use it.
5115         * tree.h (types_same_for_odr): Declare.
5117 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
5118             Jason Merrill  <jason@redhat.com>
5120         * system.h: Include <cstdlib> as well as <stdlib.h>.
5122 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
5124         PR target/57655
5125         * config/i386/i386.c (construct_container): Report error if
5126         long double is used with disabled x87 float returns.
5128 2013-06-20  Jan Hubicka  <jh@suse.cz>
5130         * lto-cgraph.c (input_symtab): Do not set cgraph state.
5132 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
5134         PR rtl-optimization/57425
5135         PR rtl-optimization/57569
5136         * alias.c (write_dependence_p): Remove parameters mem_mode and
5137         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
5138         Changed all callers.
5139         (canon_anti_dependence): Get comments and semantics in sync.
5140         Add parameter mem_canonicalized.  Changed all callers.
5141         * rtl.h (canon_anti_dependence): Update prototype.
5143 2013-06-20  Richard Biener  <rguenther@suse.de>
5145         * data-streamer-in.c (streamer_read_uhwi): Optimize single
5146         byte case, inline streamer_read_uchar and defer section
5147         overrun check.
5149 2013-06-20  Richard Biener  <rguenther@suse.de>
5151         PR tree-optimization/57584
5152         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5153         SSA names into the expanded expression that take part in
5154         abnormal coalescing.
5156 2013-06-19  Sharad Singhai  <singhai@google.com>
5158         * gcov.c (print_usage): Handle new option.
5159         (process_args): Ditto.
5160         (get_gcov_intermediate_filename): New function.
5161         (output_intermediate_file): New function.
5162         (output_gcov_file): New function
5163         (generate_results): Handle new option.
5164         (release_function): Relase demangled name.
5165         (read_graph_file): Handle demangled name.
5166         (output_lines): Ditto.
5167         * doc/gcov.texi: Document gcov intermediate format.
5169 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
5171         PR bootstrap/57604
5172         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5173         (lra_emit_add): Use the functions.  Add comment about Y as an
5174         address segment.
5176 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
5178         PR driver/57652
5179         * collect2.c (collect_atexit): New.
5180         (collect_exit): Delete.
5181         (main): Register collect_atexit with atexit.
5182         (collect_wait): Change collect_exit to exit.
5183         (do_wait): Same.
5184         * collect2.h (collect_exit): Delete.
5185         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5187 2013-06-19  Wei Mi  <wmi@google.com>
5189         PR rtl-optimization/57518
5190         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5191         if regno is used in paradoxical subreg.
5192         (update_equiv_regs): Check pdx_subregs[regno] before
5193         set a reg to be equivalent with a mem.
5195 2013-06-19  Matthias Klose  <doko@ubuntu.com>
5197         PR driver/57651
5198         * file-find.h (find_a_file): Add a mode parameter.
5199         * file-find.c (find_a_file): Likewise.
5200         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5201         with X_OK for the executables.
5202         * collect2.c (main): Call find_a_file with X_OK.
5204 2013-06-19  Steve Ellcey  <sellcey@mips.com>
5206         PR target/56942
5207         * config/mips/mips.md (casesi_internal_mips16_<mode>):
5208         Use NEXT_INSN instead of next_real_insn.
5210 2013-06-19  Jan Hubicka  <jh@suse.cz>
5212         * cgraph.h (const_value_known_p): Replace by ...
5213         (ctor_for_folding): .. this one.
5214         * cgraphunit.c (process_function_and_variable_attributes): Use it.
5215         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5216         * expr.c (expand_expr_real_1): Likewise.
5217         (string_constant): Likewise.
5218         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5219         * ipa.c (process_references): Likewise.
5220         (symtab_remove_unreachable_nodes): Likewise.
5221         * ipa-inline-analysis.c (param_change_prob): Likewise.
5222         * gimple-fold.c (canonicalize_constructor_val): Likewise.
5223         (get_base_constructor): Likwise.
5224         * varpool.c (varpool_remove_node): Likewise.
5225         (varpool_remove_initializer): LIkewise.
5226         (dump_varpool_node): LIkwise.
5227         (const_value_known_p): Rewrite to ...
5228         (ctor_for_folding): ... this one.
5230 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
5232         PR driver/57651
5233         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5234         PERSONALITY in $PATH derived prefixes.
5236 2013-06-19  Jeff Law  <law@redhat.com>
5238         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5239         in comment.
5241         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5242         (simplify_bitwise_binary): Use it to simpify certain binary ops on
5243         booleans.
5245 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
5247         * config/arm/vfp.md: Move VFP instruction classification documentation
5248         to ...
5249         * config/arm/arm.md: ... here.  Update instruction classification
5250         documentation.
5252 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
5254         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5255         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5256         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
5258 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
5260         PR target/57609
5261         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5262         with NEXT_INSN.  Use tablejump_p to check for jump table data
5263         insns.
5265 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
5267         PR c++/56544
5268         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5269         that now in C++ the value is correct per the C++ standards.
5271 2013-06-19  Richard Biener  <rguenther@suse.de>
5273         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5274         for global context.
5276 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5278         Revert:
5279         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5281         PR target/57609
5282         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5283         with next_active_insn.
5285 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
5287         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5288         functions are inlined during failures to flag an error.
5289         * tree-inline.c (expand_call_inline): Allow the error to be flagged
5290         in early inline pass.
5292 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
5294         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5295         in comments.
5297 2013-06-18  Julian Brown  <julian@codesourcery.com>
5299         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5300         Permit virtual register pre-reload if !strict.
5301         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5302         change.
5303         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5304         prototype.
5305         * config/arm/neon.md (movmisalign<mode>): Use
5306         neon_perm_struct_or_reg_operand instead of
5307         neon_struct_or_register_operand.
5308         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5309         neon_permissive_struct_operand instead of neon_struct_operand.
5310         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5311         neon_vector_mem_operand.
5312         * config/arm/predicates.md (neon_struct_operand): Adjust call to
5313         neon_vector_mem_operand.
5314         (neon_permissive_struct_operand): New.
5315         (neon_struct_or_register_operand): Rename to...
5316         (neon_perm_struct_or_reg_operand): This. Adjust call to
5317         neon_vector_mem_operand.
5319 2013-06-18  Richard Biener  <rguenther@suse.de>
5321         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5322         * lto-streamer.h: Include pointer-set.h.
5323         (struct lto_decl_slot): Remove.
5324         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5325         Remove next_index entry.
5326         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5327         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5328         (lto_init_tree_ref_encoder): Adjust.
5329         (lto_destroy_tree_ref_encoder): Likewise.
5330         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5331         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5332         (lto_output_decl_index): Adjust.
5333         (lto_new_out_decl_state): Likewise.
5334         (lto_record_function_out_decl_state): Likewise.
5335         * lto-streamer-out.c (copy_function): Likewise.
5337 2013-06-18  Richard Biener  <rguenther@suse.de>
5339         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5340         * cgraphunit.c: Include cfgloop.h.
5341         (init_lowered_empty_function): Initialize the loop tree.
5342         (assemble_thunk): Insert new BBs into loops.
5344 2013-06-18  Richard Biener  <rguenther@suse.de>
5346         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5347         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5348         the map from cache entry to cache index optional.
5349         (streamer_tree_cache_replace_tree): Adjust accordingly.
5350         (streamer_tree_cache_append): Likewise.
5351         (streamer_tree_cache_delete): Likewise.
5352         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5353         streamer cache map from cache entry to cache index.
5354         * lto-streamer-out.c (create_output_block): Adjust.
5356 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5358         * config/arm/arm.md (attribute "insn"): Move multiplication and
5359         division attributes to...
5360         (attribute "type"): ... here.  Remove mult.
5361         (attribute "mul32"): New attribute.
5362         (attribute "mul64"): Add umaal.
5363         (*arm_mulsi3): Update attributes.
5364         (*arm_mulsi3_v6): Likewise.
5365         (*thumb_mulsi3): Likewise.
5366         (*thumb_mulsi3_v6): Likewise.
5367         (*mulsi3_compare0): Likewise.
5368         (*mulsi3_compare0_v6): Likewise.
5369         (*mulsi_compare0_scratch): Likewise.
5370         (*mulsi_compare0_scratch_v6): Likewise.
5371         (*mulsi3addsi): Likewise.
5372         (*mulsi3addsi_v6): Likewise.
5373         (*mulsi3addsi_compare0): Likewise.
5374         (*mulsi3addsi_compare0_v6): Likewise.
5375         (*mulsi3addsi_compare0_scratch): Likewise.
5376         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5377         (*mulsi3subsi): Likewise.
5378         (*mulsidi3adddi): Likewise.
5379         (*mulsi3addsi_v6): Likewise.
5380         (*mulsidi3adddi_v6): Likewise.
5381         (*mulsidi3_nov6): Likewise.
5382         (*mulsidi3_v6): Likewise.
5383         (*umulsidi3_nov6): Likewise.
5384         (*umulsidi3_v6): Likewise.
5385         (*umulsidi3adddi): Likewise.
5386         (*umulsidi3adddi_v6): Likewise.
5387         (*smulsi3_highpart_nov6): Likewise.
5388         (*smulsi3_highpart_v6): Likewise.
5389         (*umulsi3_highpart_nov6): Likewise.
5390         (*umulsi3_highpart_v6): Likewise.
5391         (mulhisi3): Likewise.
5392         (*mulhisi3tb): Likewise.
5393         (*mulhisi3bt): Likewise.
5394         (*mulhisi3tt): Likewise.
5395         (maddhisi4): Likewise.
5396         (*maddhisi4tb): Likewise.
5397         (*maddhisi4tt): Likewise.
5398         (maddhidi4): Likewise.
5399         (*maddhidi4tb): Likewise.
5400         (*maddhidi4tt): Likewise.
5401         (divsi3): Likewise.
5402         (udivsi3): Likewise.
5403         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5404         (thumb2_mulsi_short_compare0): Likewise.
5405         (thumb2_mulsi_short_compare0_scratch): Likewise.
5406         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5407         (1020mult2): Likewise.
5408         (1020mult3): Likewise.
5409         (1020mult4): Likewise.
5410         (1020mult5): Likewise.
5411         (1020mult6): Likewise.
5412         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5413         change.
5414         (cortex_a15_mult64): Likewise.
5415         (cortex_a15_sdiv): Likewise.
5416         (cortex_a15_udiv): Likewise.
5417         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5418         (mult2): Likewise.
5419         (mult3): Likewise.
5420         (mult4): Likewise.
5421         (mult5): Likewise.
5422         (mult6): Likewise.
5423         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5424         (pj4_ir_div): Likewise.
5425         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5426         (11_mult2): Likewise.
5427         (11_mult3): Likewise.
5428         (11_mult4): Likewise.
5429         (11_mult5): Likewise.
5430         (11_mult6): Likewise.
5431         (11_mult7): Likewise.
5432         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5433         (cortex_a8_mla): Likewise.
5434         (cortex_a8_mull): Likewise.
5435         (cortex_a8_smulwy): Likewise.
5436         (cortex_a8_smlald): Likewise.
5437         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5438         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5439         (cortex_r4_mul_3): Likewise.
5440         (cortex_r4_mla_4): Likewise.
5441         (cortex_r4_mla_3): Likewise.
5442         (cortex_r4_smlald): Likewise.
5443         (cortex_r4_mull): Likewise.
5444         (cortex_r4_sdiv): Likewise.
5445         (cortex_r4_udiv): Likewise.
5446         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5447         (cortex_a7_idiv): Likewise.
5448         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5449         (9_mult2): Likewise.
5450         (9_mult3): Likewise.
5451         (9_mult4): Likewise.
5452         (9_mult5): Likewise.
5453         (9_mult6): Likewise.
5454         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5455         (cortex_a53_sdiv): Likewise.
5456         (cortex_a53_udiv): Likewise.
5457         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5458         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5459         (mp626_mult2): Likewise.
5460         (mp626_mult3): Likewise.
5461         (mp626_mult4): Likewise.
5462         * config/arm/fa526.md (526_mult1): Update attribute change.
5463         (526_mult2): Likewise.
5464         * config/arm/arm-generic.md (mult): Update attribute change.
5465         (mult_ldsched_strongarm): Likewise.
5466         (mult_ldsched): Likewise.
5467         (multi_cycle): Likewise.
5468         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5469         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5470         (606te_mult2): Likewise.
5471         (606te_mult3): Likewise.
5472         (606te_mult4): Likewise.
5473         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5474         (cortex_a9_mac16): Likewise.
5475         (cortex_a9_multiply): Likewise.
5476         (cortex_a9_mac): Likewise.
5477         (cortex_a9_multiply_long): Likewise.
5478         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5479         (626te_mult2): Likewise.
5480         (626te_mult3): Likewise.
5481         (626te_mult4): Likewise.
5483 2013-06-18  Richard Biener  <rguenther@suse.de>
5485         PR lto/57334
5486         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5488 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5490         PR target/57609
5491         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5492         with next_active_insn.
5494 2013-06-18  Alan Modra  <amodra@gmail.com>
5496         * config/rs6000/rs6000.h (enum data_align): New.
5497         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5498         (DATA_ABI_ALIGNMENT): Define.
5499         (CONSTANT_ALIGNMENT): Correct comment.
5500         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5501         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5503 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5505         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5506         ATTRIBUTE_UNUSED marking.
5508 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5510         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5511         alternative and update.
5512         (aarch64_dup_lanedi): Delete.
5513         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5514         * config/aarch64/aarch64-simd-builtins.def: Update.
5516 2013-06-17  Richard Biener  <rguenther@suse.de>
5518         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5519         (lto_input_scc): Declare.
5520         (lto_input_tree_1): Likewise.
5521         (struct lto_stats_d): Add num_tree_bodies_output and
5522         num_pickle_refs_output.
5523         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5524         (lto_read_tree_1): Split out from ...
5525         (lto_read_tree): ... this.
5526         (lto_input_scc): New function.
5527         (lto_input_tree_1): Split out from ...
5528         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5529         (lto_data_in_create): Create the streamer cache without hashes.
5530         * lto-streamer-out.c (create_output_block): Create the streamer
5531         cache with hashes when not doing WPA.
5532         (lto_write_tree_1): Split out from ...
5533         (lto_write_tree): ... this.
5534         (get_symbol_initial_value): New function.
5535         (lto_output_tree_1): Split out from ...
5536         (lto_output_tree): ... this.  Write trees as series of SCCs
5537         using a DFS walk via DFS_write_tree.
5538         (struct sccs, struct scc_entry): New types.
5539         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5540         (DFS_write_tree_body): New function.
5541         (DFS_write_tree): Likewise.
5542         (hash_tree): Likewise.
5543         (scc_entry_compare): Likewise.
5544         (hash_scc): Likewise.
5545         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5546         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5547         TREE_CHAIN as regular reference.
5548         (streamer_read_integer_cst): Remove.
5549         (streamer_get_pickled_tree): Adjust.
5550         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5551         of DECL_EXTERNALs in BLOCK_VARS for now.
5552         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5553         reference.
5554         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5555         Add hash value argument and record that if hashes are recorded
5556         in the cache.
5557         (streamer_tree_cache_insert_1): Adjust.
5558         (streamer_tree_cache_insert): Likewise.
5559         (streamer_tree_cache_insert_at): Rename to ...
5560         (streamer_tree_cache_replace_tree): ... this and adjust.
5561         (streamer_tree_cache_append): Adjust.
5562         (record_common_node): Likewise.
5563         (streamer_tree_cache_create): Add argument whether to
5564         record hash values together with trees.
5565         (streamer_tree_cache_delete): Adjust.
5566         * tree-streamer.h (struct streamer_tree_cache_d): Add
5567         vector of hashes.
5568         (streamer_read_integer_cst): Remove.
5569         (streamer_tree_cache_insert): Adjust.
5570         (streamer_tree_cache_append): Likewise.
5571         (streamer_tree_cache_insert_at): Rename to ...
5572         (streamer_tree_cache_replace_tree): ... this and adjust.
5573         (streamer_tree_cache_create): Add argument whether to record hashes.
5574         (streamer_tree_cache_get): Rename to ...
5575         (streamer_tree_cache_get_tree): ... this.
5576         (streamer_tree_cache_get_hash): New function.
5577         * tree.c (cache_integer_cst): New function.
5578         * tree.h (cache_integer_cst): Declare.
5579         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5580         * lto-symtab.c (lto_varpool_replace_node): Only release
5581         DECL_INITIAL of non-prevailing decls.
5582         * varpool.c (varpool_remove_initializer): Do not release
5583         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5585 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5587         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5588         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5589         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5590         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5591         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5592         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5593         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5594         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5595         instead of TARGET_64BIT.
5596         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5597         Require ISA_HAS_<D>DIV.
5599 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
5601         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5602         (mips*-*-linux*): Move default with_llsc setting to where other
5603         defaults are set.
5604         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5605         with_arch block.
5606         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5607         Likewise.  Remove default with_tune setting.  Move default float
5608         setting to its own block.  Handle with_llsc in the same block as above.
5610 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
5612         PR rtl-optimization/57425
5613         PR rtl-optimization/57569
5614         * alias.c (write_dependence_p): Add new parameters mem_mode,
5615         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
5616         Changed all callers.
5617         (canon_anti_dependence): New function.
5618         * cse.c (check_dependence): Use canon_anti_dependence.
5619         * cselib.c (cselib_invalidate_mem): Likewise.
5620         * rtl.h (canon_anti_dependence): Declare.
5622 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5624         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5625         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5626         ".set mips3" for 64-bit targets.
5628 2013-06-15  Dehao Chen  <dehao@google.com>
5630         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5631         * gimple-low.c (gimple_check_call_matching_types): Likewise.
5632         (gimple_check_call_args): Likewise.
5633         * value-prof.c (check_ic_target): Likewise.
5634         * ipa-inline.c (early_inliner): Likewise.
5635         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5636         * cgraph.c (cgraph_create_edge_1): Likewise.
5637         (cgraph_make_edge_direct): Likewise.
5639 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
5641         PR target/57615
5642         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5643         rs6000_output_move_128bit to handle emitting quad memory
5644         operations.  Set attribute length to 8 bytes.
5646 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
5648         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5649         New pattern.
5650         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5651         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5652         (aarch64_<su>mlsl<mode>): Likewise.
5654 2013-06-14  Mike Stump  <mikestump@comcast.net>
5656         * Makefile.in (TARGET_H): Add insn-codes.h.
5658 2013-06-14  Alan Modra  <amodra@gmail.com>
5660         PR middle-end/57134
5661         PR middle-end/57586
5662         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5663         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
5664         bitfield expansion when EXPAND_MEMORY.
5665         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5667 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
5669         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5670         test for clearing quad memory on 32-bit later.
5672 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5674         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5675         (fold_negate_expr): Likewise.
5676         (fold_real_zero_addition_p): Handle vectors.
5677         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5679 2013-06-14  Alan Modra  <amodra@gmail.com>
5681         * varasm.c (force_const_mem): Revert 2013-06-07 change.
5683 2013-06-13  Jan Hubicka  <jh@suse.cz>
5685         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5686         Local comdats are not externally visible.
5687         * symtab.c (dump_symtab_base): Dump externally visible.
5688         (verify_symtab_base): Verify back links in the symtab hash.
5690 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5692         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5693         CONVERT_EXPR as equal nodes.
5695 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5697         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5699 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5701         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5702         Generalize to complex and vector.
5703         * tree.c (build_all_ones_cst): New function.
5704         * tree.h (build_all_ones_cst): Declare it.
5706 2013-06-13  Alan Modra  <amodra@gmail.com>
5708         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5709         * config/rs6000/rs6000.md (signbittf2): New insn.
5710         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5711         (abstf2_internal, cmptf_internal2): Likewise.
5712         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5714 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5715             Pat Haugen <pthaugen@us.ibm.com>
5716             Peter Bergner <bergner@vnet.ibm.com>
5718         * config/rs6000/rs6000.c (emit_load_locked): Add support for
5719         power8 byte, half-word, and quad-word atomic instructions.
5720         (emit_store_conditional): Likewise.
5721         (rs6000_expand_atomic_compare_and_swap): Likewise.
5722         (rs6000_expand_atomic_op): Likewise.
5724         * config/rs6000/sync.md (larx): Add new modes for power8.
5725         (stcx): Likewise.
5726         (AINT): New mode iterator to include TImode as well as normal
5727         integer modes on power8.
5728         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5729         that VSX registers are not considered.  Use AINT mode iterator
5730         instead of INT1 to allow inclusion of quad word atomic operations
5731         on power8.
5732         (load_locked<mode>): Likewise.
5733         (store_conditional<mode>): Likewise.
5734         (atomic_compare_and_swap<mode>): Likewise.
5735         (atomic_exchange<mode>): Likewise.
5736         (atomic_nand<mode>): Likewise.
5737         (atomic_fetch_<fetchop_name><mode>): Likewise.
5738         (atomic_nand_fetch<mode>): Likewise.
5739         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5740         each type.
5741         (ATOMIC): On power8, add QImode, HImode modes.
5742         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5743         modes that promote to SImode.
5744         (load_lockedti): Convert TImode arguments to PTImode, so that we
5745         get a guaranteed even/odd register pair.
5746         (load_lockedpti): Likewise.
5747         (store_conditionalti): Likewise.
5748         (store_conditionalpti): Likewise.
5750         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5751         atomic load/store instructions.
5752         (HSI): Likewise.
5754 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
5756         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5757         loads.
5758         (insn_count): New attribute, with most cases extracted from...
5759         (length): ...here.  Redefine most cases in terms of insn_count.
5760         (single_insn): Delete.
5761         (can_delay): Use insn_count to check for single instructions.
5762         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5763         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5764         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5765         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5766         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5767         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5768         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5769         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5770         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5771         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5772         rather than "length".
5773         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
5774         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5775         Use "insn_count" rather than "length".
5776         * config/mips/mips-dsp.md
5777         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5778         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5779         length attributes.
5781 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
5783         PR tree-optimization/57361
5784         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5786 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
5788         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5789         to split.
5790         (aarch64_simd_combine<mode>): New instruction expansion.
5791         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5792         function prototype.
5793         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5794         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5796 2013-06-12  Jan Hubicka  <jh@suse.cz>
5798         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5799         decl has when in streaming stage.
5800         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5801         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5803 2013-06-12  Roland Stigge <stigge@antcom.de>
5805         PR target/57578
5806         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5808 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5810         PR tree-optimization/57537
5811         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5812         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5814 2013-06-12  Richard Biener  <rguenther@suse.de>
5816         * data-streamer.h (streamer_write_char_stream): CSE
5817         obs->current_pointer.
5818         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5819         streamer_write_char_stream manually and optimize the resulting loop.
5820         (streamer_write_hwi_stream): Likewise.
5822 2013-06-12  Jan Hubicka  <jh@suse.cz>
5824         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5825         * cgraph.h (varpool_create_empty_node): Declare.
5826         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5827         duplicated nodes.
5828         * symtab.c (symtab_unregister_node): Be lax about missin entries
5829         in node hash.
5830         (symtab_get_node): Update comment.
5831         * varpool.c (varpool_create_empty_node): Break out from ...
5832         (varpool_node_for_decl): ... here.
5833         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5835 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
5837         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5838         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5839         part.  Use straight-line flow at the end.
5840         <COMPONENT_REF>: Remove superfluous else.
5841         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5843 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5845         PR target/56564
5846         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5847         target hook even for !TREE_PUBLIC decls.  If no resolution info
5848         is available, return false for common and external decls.
5850 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
5852         * config/rl78/constraints.md (U): New constraint.
5853         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5854         valloc attribute.
5856 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5858         PR target/57589
5859         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5860         to allow returning address to AT_PLATFORM name.
5862 2013-06-11  Jan Hubicka  <jh@suse.cz>
5864         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5865         * cgraph.h (symtab_node_base): Add weakref flag.
5866         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5867         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5868         (output_weakrefs): Use weakref flag.
5869         * fold-const.c (simple_operand_p): Handle WEAK.
5870         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5871         * ipa.c (varpool_externally_visible_p): Drop weakref.
5872         (function_and_variable_visibility): Update comment; fix weakref
5873         sanity checks; do not clear DECL_WEAK on them.
5874         * lto-cgraph.c (lto_output_node): update.
5875         (lto_output_varpool_node): Update.
5876         (input_overwrite_node): Update.
5877         (input_node): Update.
5878         (input_varpool_node): Update.
5879         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5880         (lto_symtab_merge_symbols): Add sanity check.
5881         (lto_symtab_prevailing_decl): Do not special case weakrefs.
5882         * passes.c (rest_of_decl_compilation): Set static flag, too.
5883         * symtab.c (dump_symtab_base): Dump weakref.
5884         (verify_symtab_base): Sanity check weakrefs.
5885         (symtab_make_decl_local): Remove duplicated code.
5886         (symtab_alias_ultimate_target): Simplify.
5887         * varpool.c (varpool_create_variable_alias): Set weakref flag.
5889 2013-06-11  DJ Delorie  <dj@redhat.com>
5891         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5892         (rl78_unwind_word_mode): New.
5894 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5896         * final.c (debug_prefix_maps): Make static.
5898 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5900         * function.c (initial_trampoline): Remove stray copy.
5902 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
5904         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5906 2013-06-11  Martin Jambor  <mjambor@suse.cz>
5908         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5909         within bounds at the beginning of the function.
5911 2013-06-11  Alan Modra  <amodra@gmail.com>
5913         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
5914         reporting.
5915         (get_named_section): Don't NULL !DECL_P decl.
5917 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
5919         * doc/invoke.texi (core-avx2): Document.
5920         (slm): Likewise.
5921         (atom): Updated with MOVBE.
5923 2013-06-11  Richard Biener  <rguenther@suse.de>
5925         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5927 2013-06-11  Anton Blanchard  <anton@samba.org>
5929         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5930         correct shift value in little-endian mode.
5932 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
5934         PR target/56564
5935         * varasm.c (get_variable_align): Move #endif to the right place.
5937 2013-06-10  Cary Coutant  <ccoutant@google.com>
5939         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5940         for hash so that hash table traversal order is deterministic.
5942 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
5943             Pat Haugen <pthaugen@us.ibm.com>
5944             Peter Bergner <bergner@vnet.ibm.com>
5946         * config/rs6000/vector.md (GPR move splitter): Do not split moves
5947         of vectors in GPRS if they are direct moves or quad word load or
5948         store moves.
5950         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5951         declaration.
5952         (direct_move_p): Likewise.
5953         (quad_load_store_p): Likewise.
5955         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5956         classes into bins based on the physical register type.
5957         (reg_class_to_reg_type): Likewise.
5958         (IS_STD_REG_TYPE): Likewise.
5959         (IS_FP_VECT_REG_TYPE): Likewise.
5960         (reload_fpr_gpr): Arrays to determine what insn to use if we can
5961         use direct move instructions.
5962         (reload_gpr_vsx): Likewise.
5963         (reload_vsx_gpr): Likewise.
5964         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5965         information that is a simplification of register classes.  Also
5966         precalculate direct move reload helpers.
5967         (direct_move_p): New function to return true if the operation can
5968         be done as a direct move instruciton.
5969         (quad_load_store_p): New function to return true if the operation
5970         is a quad memory operation.
5971         (rs6000_legitimize_address): If quad memory, only allow register
5972         indirect for TImode addresses.
5973         (rs6000_legitimate_address_p): Likewise.
5974         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5975         (rs6000_reload_register_type): Likewise.
5976         (register_to_reg_type): Return register type.
5977         (rs6000_secondary_reload_simple_move): New helper function for
5978         secondary reload and secondary memory needed to identify anything
5979         that is a simple move, and does not need reloading.
5980         (rs6000_secondary_reload_direct_move): New helper function for
5981         secondary reload to identify cases that can be done with several
5982         instructions via the direct move instructions.
5983         (rs6000_secondary_reload_move): New helper function for secondary
5984         reload to identify moves between register types that can be done.
5985         (rs6000_secondary_reload): Add support for quad memory operations
5986         and for direct move.
5987         (rs6000_secondary_memory_needed): Likewise.
5988         (rs6000_debug_secondary_memory_needed): Change argument names.
5989         (rs6000_output_move_128bit): New function to return the move to
5990         use for 128-bit moves, including knowing about the various
5991         limitations of quad memory operations.
5993         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5994         memory operations.  call rs6000_output_move_128bit for the actual
5995         instruciton(s) to generate.
5996         (vsx_movti_64bit): Likewise.
5998         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5999         (UNSPEC_P8V_MTVSRWZ): Likewise.
6000         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6001         (UNSPEC_P8V_MTVSRD): Likewise.
6002         (UNSPEC_P8V_XXPERMDI): Likewise.
6003         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6004         (UNSPEC_FUSION_GPR): Likewise.
6005         (FMOVE128_GPR): New iterator for direct move.
6006         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6007         (f32_sv): Likewise.
6008         (f32_dm): Likewise.
6009         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6010         loads and direct move instructions.
6011         (zero_extendsidi2_lfiwzx): Likewise.
6012         (extendsidi2_lfiwax): Likewise.
6013         (extendsidi2_nocell): Likewise.
6014         (floatsi<mode>2_lfiwax): Likewise.
6015         (lfiwax): Likewise.
6016         (floatunssi<mode>2_lfiwzx): Likewise.
6017         (lfiwzx): Likewise.
6018         (fix_trunc<mode>_stfiwx): Likewise.
6019         (fixuns_trunc<mode>_stfiwx): Likewise.
6020         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6021         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6022         (parity<mode>2_cmpb): Set length/type attr.
6023         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6024         for 'mr.' to fast_compare.
6025         (bpermd_<mode>): Change type attr to popcnt.
6026         (p8_fmrgow_<mode>): New insns for power8 direct move support.
6027         (p8_mtvsrwz_1): Likewise.
6028         (p8_mtvsrwz_2): Likewise.
6029         (reload_fpr_from_gpr<mode>): Likewise.
6030         (p8_mtvsrd_1): Likewise.
6031         (p8_mtvsrd_2): Likewise.
6032         (p8_xxpermdi_<mode>): Likewise.
6033         (reload_vsx_from_gpr<mode>): Likewise.
6034         (reload_vsx_from_gprsf): Likewise.
6035         (p8_mfvsrd_3_<mode>): LIkewise.
6036         (reload_gpr_from_vsx<mode>): Likewise.
6037         (reload_gpr_from_vsxsf): Likewise.
6038         (p8_mfvsrd_4_disf): Likewise.
6039         (multi-word GPR splits): Do not split direct moves or quad memory
6040         operations.
6042 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
6044         * tree-into-ssa.c (interesting_blocks): Make static.
6046 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
6048         PR target/56564
6049         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6050         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6051         Use DATA_ABI_ALIGNMENT for that case instead if defined.
6052         (get_variable_align): New function.
6053         (get_variable_section, emit_bss, emit_common,
6054         assemble_variable_contents, place_block_symbol): Use
6055         get_variable_align instead of DECL_ALIGN.
6056         (assemble_noswitch_variable): Add align argument, use it
6057         instead of DECL_ALIGN.
6058         (assemble_variable): Adjust caller.  Use get_variable_align
6059         instead of DECL_ALIGN.
6060         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6061         caller.
6062         (DATA_ABI_ALIGNMENT): Define.
6063         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6064         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
6065         opt is false, only return the psABI mandated alignment increase.
6066         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6067         (DATA_ABI_ALIGNMENT): ... this.
6068         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6069         (DATA_ABI_ALIGNMENT): ... this.
6070         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6071         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6072         (DATA_ABI_ALIGNMENT): ... this.
6073         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6074         * doc/tm.texi: Regenerated.
6076 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
6078         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6079         cmp_code to construct REG_EQUAL note.
6081 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6083         PR target/57568
6084         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6085         that operands[2] doesn't overlap with operands[0].
6087 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6088             Jan Hubicka  <jh@suse.cz>
6090         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6091         hack to mark symbols as used.
6093 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6095         PR rtl-optimization/57559
6096         * lra-constraints.c (process_alt_operands): Don't discourage
6097         memory with known offset for offsetable memory constraint.
6098         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6100 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
6102         * varasm.c (struct oc_local_state): Reorder fields.
6103         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6104         and adjust accordingly.
6105         (output_constructor): Reorder initialization code and adjust call to
6106         output_constructor_bitfield.
6108 2013-06-07  Jan Hubicka  <jh@suse.cz>
6110         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6112 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
6114         * tree-object-size.c (unknown): Make const.
6116 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6118         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6119         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6120         for last alternative in the cpu_facility attribute.
6122 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6124         PR target/56315
6125         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6126         (xordi3): Change operand 2 constraint to arm_xordi_operand.
6127         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6128         * config/arm/constraints.md (Dg): New constraint.
6129         * config/arm/neon.md (xordi3_neon): Remove.
6130         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6131         * config/arm/predicates.md (arm_xordi_operand): New predicate.
6133 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6135         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6136         Clean up alternatives.
6138 2013-06-07  Alan Modra  <amodra@gmail.com>
6140         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6141         va_list_gpr_size.
6143 2013-06-07  Alan Modra  <amodra@gmail.com>
6145         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6147 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6149         * config/arm/constraints.md (Df): New constraint.
6150         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6151         Correct length attribute for last two alternatives.
6153 2013-06-07  Alan Modra  <amodra@gmail.com>
6155         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6156         override user -mfp-in-toc.
6157         (offsettable_ok_by_alignment): Consider just the current access
6158         rather than the whole object, unless BLKmode.  Handle
6159         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6160         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6161         for -mcmodel=medium.
6162         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6163         override user -mfp-in-toc or -msum-in-toc.  Default to
6164         -mno-fp-in-toc for -mcmodel=medium.
6166 2013-06-06  DJ Delorie  <dj@redhat.com>
6168         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6169         TARGET_VALID_POINTER_MODE.
6171 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
6172             Pat Haugen <pthaugen@us.ibm.com>
6173             Peter Bergner <bergner@vnet.ibm.com>
6175         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6176         Document new power8 builtins.
6178         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6179         condition code register, to allow 128-bit logical operations to be
6180         done in the VSX or GPR registers.
6181         (nor<mode>3): Use the canonical form for nor.
6182         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6183         vclz*, and vpopcnt* vector instructions.
6184         (nand<mode>3): Likewise.
6185         (orc<mode>3): Likewise.
6186         (clz<mode>2): LIkewise.
6187         (popcount<mode>2): Likewise.
6189         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6190         that only the GPRs are recognized.
6192         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6193         support for new power8 builtins.
6195         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6196         builtin functions.
6197         (xscvdpspn): Likewise.
6198         (vclz): Likewise.
6199         (vclzb): Likewise.
6200         (vclzh): Likewise.
6201         (vclzw): Likewise.
6202         (vclzd): Likewise.
6203         (vpopcnt): Likewise.
6204         (vpopcntb): Likewise.
6205         (vpopcnth): Likewise.
6206         (vpopcntw): Likewise.
6207         (vpopcntd): Likewise.
6208         (vgbbd): Likewise.
6209         (vmrgew): Likewise.
6210         (vmrgow): Likewise.
6211         (eqv): Likewise.
6212         (eqv_v16qi3): Likewise.
6213         (eqv_v8hi3): Likewise.
6214         (eqv_v4si3): Likewise.
6215         (eqv_v2di3): Likewise.
6216         (eqv_v4sf3): Likewise.
6217         (eqv_v2df3): Likewise.
6218         (nand): Likewise.
6219         (nand_v16qi3): Likewise.
6220         (nand_v8hi3): Likewise.
6221         (nand_v4si3): Likewise.
6222         (nand_v2di3): Likewise.
6223         (nand_v4sf3): Likewise.
6224         (nand_v2df3): Likewise.
6225         (orc): Likewise.
6226         (orc_v16qi3): Likewise.
6227         (orc_v8hi3): Likewise.
6228         (orc_v4si3): Likewise.
6229         (orc_v2di3): Likewise.
6230         (orc_v4sf3): Likewise.
6231         (orc_v2df3): Likewise.
6233         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6234         allow power8 quad mode in 64-bit.
6235         (rs6000_builtin_vectorized_function): Add support to vectorize
6236         ISA 2.07 count leading zeros, population count builtins.
6237         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6238         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6239         (builtin_function_type): Add vgbbd builtin function which takes an
6240         unsigned argument.
6241         (altivec_expand_vec_perm_const): Add support for new power8 merge
6242         instructions.
6244         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6245         that does not include TImdoe for use with 32-bit.
6246         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6247         instructions.
6248         (UNSPEC_VSX_CVDPSPN): Likewise.
6249         (vsx_xscvdpspn): Likewise.
6250         (vsx_xscvspdpn): Likewise.
6251         (vsx_xscvdpspn_scalar): Likewise.
6252         (vsx_xscvspdpn_directmove): Likewise.
6253         (vsx_and<mode>3): Split logical operations into 32-bit and
6254         64-bit. Add support to do logical operations on TImode as well as
6255         VSX vector types.  Allow logical operations to be done in either
6256         VSX registers or in general purpose registers in 64-bit mode.  Add
6257         splitters if GPRs were used. For AND, add clobber of CCmode to
6258         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
6259         encoding.
6260         (vsx_and<mode>3_32bit): Likewise.
6261         (vsx_and<mode>3_64bit): Likewise.
6262         (vsx_ior<mode>3): Likewise.
6263         (vsx_ior<mode>3_32bit): Likewise.
6264         (vsx_ior<mode>3_64bit): Likewise.
6265         (vsx_xor<mode>3): Likewise.
6266         (vsx_xor<mode>3_32bit): Likewise.
6267         (vsx_xor<mode>3_64bit): Likewise.
6268         (vsx_one_cmpl<mode>2): Likewise.
6269         (vsx_one_cmpl<mode>2_32bit): Likewise.
6270         (vsx_one_cmpl<mode>2_64bit): Likewise.
6271         (vsx_nor<mode>3): Likewise.
6272         (vsx_nor<mode>3_32bit): Likewise.
6273         (vsx_nor<mode>3_64bit): Likewise.
6274         (vsx_andc<mode>3): Likewise.
6275         (vsx_andc<mode>3_32bit): Likewise.
6276         (vsx_andc<mode>3_64bit): Likewise.
6277         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6278         and xxlorc instructions.
6279         (vsx_eqv<mode>3_64bit): Likewise.
6280         (vsx_nand<mode>3_32bit): Likewise.
6281         (vsx_nand<mode>3_64bit): Likewise.
6282         (vsx_orc<mode>3_32bit): Likewise.
6283         (vsx_orc<mode>3_64bit): Likewise.
6285         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6287         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6288         instruction.
6289         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6290         (p8_vmrgow): Likewise.
6291         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6292         GPRs to be split under VSX.
6293         (p8v_clz<mode>2): Add power8 count leading zero support.
6294         (p8v_popcount<mode>2): Add power8 population count support.
6295         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6296         support.
6298         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6299         instruction.
6301         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6302         builtin functions.
6303         (vec_nand): Likewise.
6304         (vec_vclz): Likewise.
6305         (vec_vclzb): Likewise.
6306         (vec_vclzd): Likewise.
6307         (vec_vclzh): Likewise.
6308         (vec_vclzw): Likewise.
6309         (vec_vgbbd): Likewise.
6310         (vec_vmrgew): Likewise.
6311         (vec_vmrgow): Likewise.
6312         (vec_vpopcnt): Likewise.
6313         (vec_vpopcntb): Likewise.
6314         (vec_vpopcntd): Likewise.
6315         (vec_vpopcnth): Likewise.
6316         (vec_vpopcntw): Likewise.
6318 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6320         PR rtl-optimization/57468
6321         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6322         spilled pseudos.
6324 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6326         PR rtl-optimization/57459
6327         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6328         type when setting live regs.
6330 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6332         * config/s390/s390.opt (mlra): New option.
6333         * config/s390/s390.c (s390_decompose_address): Check displacement
6334         for all registers for LRA.
6335         (s390_secondary_reload): Don't used secondary reloads for LRA.
6336         (s390_lra_p): New function.
6337         (TARGET_LRA_P): Define.
6338         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6339         of attribute cpu_facility to zarch for the last alternative.
6340         (*cmpmem_short): Ditto.
6342 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6344         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6345         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6346         (arm_expand_prologue): Likewise.
6348 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6350         PR c++/53743
6351         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6352         as this is now done by redirect_edge_and_branch_force.
6353         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6354         barriers, and fix interaction with splitting.
6355         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6356         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6357         reflect changes made in the routine.
6358         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6359         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6360         since this is called in cfglayout mode, and replace partition fixup
6361         with assert as that is now done by force_nonfallthru_and_redirect.
6362         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6363         already be marked with region crossing note.
6364         (insert_section_boundary_note): Make non-static, gate on flag
6365         has_bb_partition, rewrite to also check for multiple partitions.
6366         (rest_of_handle_reorder_blocks): Remove call to
6367         insert_section_boundary_note, now done later during free_cfg.
6368         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6369         * bb-reorder.h (insert_section_boundary_note): Declare.
6370         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6371         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6372         invoke insert_section_boundary_note.
6373         (try_redirect_by_replacing_jump): Remove unnecessary
6374         check for region crossing note.
6375         (fixup_partition_crossing): New function.
6376         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6377         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6378         in non-cfglayout mode.
6379         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6380         remove old code that tried to do this. Emit barrier correctly
6381         when we are in cfglayout mode.
6382         (last_bb_in_partition): New function.
6383         (rtl_split_edge): Correctly fixup partition boundaries.
6384         (commit_one_edge_insertion): Remove old code that tried to
6385         fixup region crossing edge since this is now handled in
6386         split_block, and set up insertion point correctly since
6387         block may now end in a jump.
6388         (verify_hot_cold_block_grouping): Guard against checking when not in
6389         linearized RTL mode.
6390         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6391         notes.
6392         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6393         rtl_verify_flow_info, so not called in cfglayout mode.
6394         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6395         (fixup_reorder_chain): Remove old code that attempted to fixup region
6396         crossing note as this is now handled in force_nonfallthru_and_redirect.
6397         (duplicate_insn_chain): Don't duplicate switch section notes.
6398         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6399         note.
6400         * basic-block.h (emit_barrier_after_bb): Declare.
6402 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6404         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6405         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6406         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6408 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6410         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6411         where appropriate.
6412         * config/arm/ldmstm.md: Regenerate.
6414 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6416         * config/arm/sync.md (atomic_loaddi_1):
6417         Disable predication for arm_restrict_it.
6418         (arm_load_exclusive<mode>): Likewise.
6419         (arm_load_exclusivesi): Likewise.
6420         (arm_load_exclusivedi): Likewise.
6421         (arm_load_acquire_exclusive<mode>): Likewise.
6422         (arm_load_acquire_exclusivesi): Likewise.
6423         (arm_load_acquire_exclusivedi): Likewise.
6424         (arm_store_exclusive<mode>): Likewise.
6425         (arm_store_exclusive<mode>): Likewise.
6426         (arm_store_release_exclusivedi): Likewise.
6427         (arm_store_release_exclusive<mode>): Likewise.
6429 2013-06-06  Richard Biener  <rguenther@suse.de>
6431         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6432         after LTO_null.
6433         (lto_tag_is_tree_code_p): Adjust.
6434         (lto_tag_is_gimple_code_p): Likewise.
6435         (lto_gimple_code_to_tag): Likewise.
6436         (lto_tag_to_gimple_code): Likewise.
6437         (lto_tree_code_to_tag): Likewise.
6438         (lto_tag_to_tree_code): Likewise.
6439         * data-streamer.h (streamer_write_hwi_in_range): Use
6440         uhwi streaming to stream the normalized range.
6441         (streamer_read_hwi_in_range): Likewise.
6443 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6445         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6446         (predicable_short_it): Likewise.
6447         (predicated): Likewise.
6448         (enabled): Handle above.
6449         (define_cond_exec): Set predicated attribute to yes.
6451 2013-06-05  Mike Stump  <mikestump@comcast.net>
6453         * gdbinit.in (__FUNCTION__): Add.
6455 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6457         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6458         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6460 2013-06-05  Jan Hubicka  <jh@suse.cz>
6462         * varasm.c (mark_decl_referenced): Revert the removal until targets
6463         are fixed.
6465 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6467         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6468         instead of mark_decl_referenced.
6470 2013-06-05  Jan Hubicka  <jh@suse.cz>
6472         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6473         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6474         and symtab_used_from_object_file_p.
6475         (cgraph_make_node_local_1): Clear forced_by_abi.
6476         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6477         * cgraph.h (symtab_node_base): Add forced_by_abi.
6478         (decide_is_variable_needed): Remove.
6479         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6480         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6481         (decide_is_symbol_needed): ... this one; handle symbols in general;
6482         always analyze virtuals; honnor forced_by_abi.
6483         (cgraph_finalize_function): Update.
6484         (varpool_finalize_decl): Update.
6485         (symbol_defined_and_needed): Remove.
6486         (analyze_functions): Update.
6487         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6488         output_refs, input_overwrite_node): Handle forced_by_abi.
6489         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6490         (address_taken_from_non_vtable_p): ... this one.
6491         (comdat_can_be_unshared_p_1): New function.
6492         (cgraph_comdat_can_be_unshared_p): Rename to ...
6493         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6494         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6495         (function_and_variable_visibility): Clear forced_by_abi as needed.
6496         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6497         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6498         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6499         * varpool.c (decide_is_variable_needed): Remove.
6501 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6503         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6504         (arm_option_override): Override arm_restrict_it where appropriate.
6505         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6506         * config/arm/arm.opt (mrestrict-it): New command-line option.
6507         * doc/invoke.texi: Document -mrestrict-it.
6509 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6511         * tsan.c (tsan_atomic_table): Make const.
6513 2013-06-05  Richard Biener  <rguenther@suse.de>
6515         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6516         index associated with the tree we are supposed to replace.
6517         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6518         TREE_ASM_WRITTEN as zero for everything but SSA names.
6520 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6522         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6524 2013-06-04  Jan Hubicka  <jh@suse.cz>
6526         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6527         (input_node, input_varpool_node): Handle correctly external same
6528         body aliases.
6529         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6530         nodes at ltrans stage.
6532 2013-06-04  Jan Hubicka  <jh@suse.cz>
6534         * ipa-inline.c (update_caller_keys): Fix availability test.
6535         (update_callee_keys): Likewise.
6536         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6537         to follow ELF standard.
6539 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6541         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6542         (mips64r5900el-*-elf*): New configurations.
6543         * config/mips/mips-cpus.def (r5900): New processor.
6544         * config/mips/mips-tables.opt: Regenerate.
6545         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6546         (mips_issue_rate): Handle PROCESSOR_R5900.
6547         (mips_reorg_process_insns): Force reorder mode for the R5900.
6548         * config/mips/mips.h (TARGET_MIPS5900): Define.
6549         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6550         TARGET_MIPS5900.
6551         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6552         TARGET_MIPS5900.
6553         * config/mips/mips.md (processor): Add r5900.
6554         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6556 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6558         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6559         into function to generate MOVI instruction.
6560         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6561         (aarch64_preferred_simd_mode): Turn into wrapper.
6562         (aarch64_output_scalar_simd_mov_immediate): New function.
6563         * config/aarch64/aarch64-protos.h: Add prototype for above.
6565 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6567         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6568         element_char member.
6569         (sizetochar): Return signed char.
6570         (aarch64_simd_valid_immediate): Remove elchar and other
6571         unnecessary variables.
6572         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6573         Calculate element_char as required.
6574         * config/aarch64/aarch64-protos.h: Update and move prototype
6575         for aarch64_output_simd_mov_immediate.
6576         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6577         Update arguments.
6579 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6581         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6582         information completed by aarch64_simd_valid_immediate.
6583         (aarch64_legitimate_constant_p): Update arguments.
6584         (aarch64_simd_valid_immediate): Work with struct rather than many
6585         pointers.
6586         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6587         (aarch64_simd_make_constant): Update arguments.
6588         (aarch64_output_simd_mov_immediate): Work with struct rather than
6589         many pointers.  Output immediate directly rather than as operand.
6590         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6591         Update prototype.
6592         * config/aarch64/constraints.md (Dn): Update arguments.
6594 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6596         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6597         longer static.
6598         (aarch64_simd_immediate_valid_for_move): Remove.
6599         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6600         (aarch64_simd_make_constant): Update call.
6601         (aarch64_output_simd_mov_immediate): Update call.
6602         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6603         Add prototype.
6604         * config/aarch64/constraints.md (Dn): Update call.
6606 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6608         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6609         return type to bool for prototype.
6610         (aarch64_legitimate_constant_p): Check for true instead of not -1.
6611         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6612         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6614 2013-06-04  Catherine Moore  <clm@codesourcery.com>
6616         * config/mips/mips.opt (meva): New.
6617         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6618         (ASM_SPEC): Handle -meva.
6619         * doc/invoke.texi (meva):  Document.
6621 2013-06-04  Alan Modra  <amodra@gmail.com>
6623         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6624         constant output.
6626 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6628         * rtl.def: Add extra fourth optional field to define_cond_exec.
6629         * gensupport.c (process_one_cond_exec): Process attributes from
6630         define_cond_exec.
6631         * doc/md.texi: Document fourth field in define_cond_exec.
6633 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
6635         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6636         out the processing order as in store_bit_field_1.
6638 2013-06-04  Jan Hubicka  <jh@suse.cz>
6640         PR middle-end/57500
6641         * cgraphunit.c (cgraph_process_same_body_aliases): Create
6642         non-VAR_DECL node if it does not exist yet.
6644 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
6646         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6647         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6648         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6649         target_cpu_default setting.
6651 2013-06-03  Teresa Johnson  <tejohnson@google.com>
6653         * dumpfile.c (opt_info_switch_p): Change -fopt-info
6654         default to -fopt-info=optimized instead of all.
6655         * doc/invoke.texi: Ditto.
6656         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6657         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6658         (execute_vect_slp): Emit BB vectorization success under
6659         MSG_OPTIMIZED_LOCATIONS.
6660         * tree-vect-slp.c (vect_slp_transform_bb): Change
6661         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6662         * tree-vect-loop.c (vect_transform_loop): Ditto.
6664 2013-06-03  Jason Merrill  <jason@redhat.com>
6666         PR c++/57415
6667         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6668         Use TARGET_EXPR for C++.
6670 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
6672         PR rtl-optimization/57268
6673         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6674         if DEBUG_INSN_P (insn).
6676         Reapply
6677         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6679         PR rtl-optimization/57268
6680         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6681         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6683 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6685         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6686         (ix86_avoid_lea_for_addr): Likewise.
6687         (exact_dependency_1): Likewise.
6688         (ix86_adjust_cost): Likewise.
6689         (swap_top_of_ready_list): Fix formatting and !reload_completed check
6690         removed.
6691         (do_reorder_for_imul): Fix typo, formatting and
6692         !reload_completed check removed.
6693         (ix86_sched_reorder): Fix typo and formatting.
6694         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6695         list.
6697 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
6699         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6701 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
6703         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6704         <STRING_CST>: Likewise.
6705         <VECTOR_CST>: Likewise.
6707 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
6708             Mikael Morin  <mikael@gcc.gnu.org>
6710         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6711         * config.in: Regenerated.
6712         * configure: Regenerated.
6714 2013-06-01  Jan Hubicka  <jh@suse.cz>
6716         PR middle-end/57366
6717         * cgraphunit.c (compile): When weakref is not supported,
6718         set up transparent aliases before final output pass.
6719         * varasm.c (assemble_alias): Do not try to do it here.
6721 2013-06-01  Jan Hubicka  <jh@suse.cz>
6723         PR middle-end/57467
6724         * passes.c (for_per_function): Skip unanalyzed functions.
6726 2013-06-01  Jan Hubicka  <jh@suse.cz>
6728         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6729         (lto_symtab_merge_symbols_1): ... this one.
6730         (lto_symtab_merge_cgraph_nodes): Rename to ...
6731         (lto_symtab_merge_symbols): ... this one; simplify.
6732         * cgraph.c (same_body_aliases_done): Rename to ...
6733         (cpp_implicit_aliases_done): ... this one.
6734         (cgraph_create_function_alias): Update.
6735         (cgraph_same_body_alias): Update.
6736         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6737         (verify_edge_corresponds_to_fndecl): Simplify.
6738         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6739         (cgraph_node): Remove same_body_alias.
6740         (varpool_node): Remove alias_of and extra_name_alias.
6741         (same_body_aliases_done): Rename to ..
6742         (cpp_implicit_aliases_done): ... this one.
6743         (symtab_alias_ultimate_target): Add default parameter.
6744         (symtab_resolve_alias): New function.
6745         (fixup_same_cpp_alias_visibility): Declare.
6746         (cgraph_function_node): Add default parameter.
6747         (cgraph_node_asm_name): Likewise.
6748         (cgraph_function_or_thunk_node): Add default parameter; do
6749         not ICE when it is NULL.
6750         (varpool_variable_node): Likewise.
6751         * tree-emutls.c (create_emultls_var): Update.
6752         (ipa_lower_emutls): Update.
6753         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6754         (cgraph_reset_node): Reset alias info.
6755         (cgraph_finalize_function): Update.
6756         (fixup_same_cpp_alias_visibility): Move to symtab.c.
6757         (analyze_function): Simplify.
6758         (cgraph_process_same_body_aliases): Simplify.
6759         (analyze_functions): Fixup same body aliases.
6760         (handle_alias_pairs): Simplify.
6761         (assemble_thunk): Update.
6762         (assemble_thunks_and_aliases): Update.
6763         (output_weakrefs): Rewrite.
6764         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6765         (lto_output_varpool_node): Likewise.
6766         (compute_ltrans_boundary): Remve assert.
6767         (get_alias_symbol): New functoin.
6768         (input_node): Rewrite alias handling.
6769         (input_varpool_node): Likewise.
6770         * ipa-pure-const.c (propagate_pure_const): Fix formating.
6771         * ipa.c (process_references): Handle weakrefs correctly.
6772         (symtab_remove_unreachable_nodes): Likewise.
6773         * trans-mem.c (get_cg_data): Update.
6774         (ipa_tm_create_version_alias): Update.
6775         (ipa_tm_execute): Update.
6776         * symtab.c (dump_symtab_base): Dump aliases.
6777         (verify_symtab_base): Verify aliases.
6778         (symtab_node_availability): New function.
6779         (symtab_alias_ultimate_target): Simplify.
6780         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6781         handle all the fixup cases.
6782         (symtab_resolve_alias): New function.
6783         * passes.c (ipa_write_summaries): Handle weakrefs.
6784         * varpool.c (varpool_analyze_node): Simplify.
6785         (assemble_aliases): Update.
6786         (varpool_create_variable_alias): Simplify.
6787         (varpool_extra_name_alias): Simplify.
6788         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6789         (lto_symtab_merge_symbols): ... this one.
6791 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
6793         Revert
6794         PR rtl-optimization/57268
6795         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6796         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6798 2013-06-01  Tobias Burnus  <burnus@net-b.de>
6800         Partially reverted:
6801         2013-05-31  Tobias Burnus  <burnus@net-b.de>
6803         PR middle-end/57073
6804         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6805         further up.
6807 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6809         PR rtl-optimization/57268
6810         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6811         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6813 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6815         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6816         unordered comparison operators when -fno-trapping-math is in effect
6817         on the e500.
6818         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6819         and implement unordered comparison operators properly on the e500.
6821 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6823         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6824         for constant scalar integers.
6825         (simplify_relational_operation_1): Likewise.
6827 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
6829         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6830         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
6831         Fix comment.
6833 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6834             Igor Zamyatin  <igor.zamyatin@intel.com>
6836         Silvermont (SLM) architecture performance tuning.
6837         * config/i386/i386.h (enum ix86_tune_indices): Add
6838         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6839         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6841         * config/i386/i386.c (initial_ix86_tune_features)
6842         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6843         (ix86_lea_outperforms): Handle Silvermont tuning.
6844         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6845         call.
6846         (ix86_use_lea_for_mov): Likewise.
6847         (ix86_avoid_lea_for_addr): Likewise.
6848         (ix86_lea_for_add_ok): Likewise.
6849         (exact_dependency_1): New function.
6850         (exact_store_load_dependency): Likewise.
6851         (ix86_adjust_cost): Handle Silvermont tuning.
6852         (do_reoder_for_imul): Likewise.
6853         (swap_top_of_ready_list): New function.
6854         (ix86_sched_reorder): Changed to handle Silvermont tuning.
6856         * config/i386/i386.md (peepholes that split memory operand in fp
6857         converts): New.
6859 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6861         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6862         Remove un-necessary braces.
6864 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6866         * config/aarch64/aarch64.c (aarch64_classify_symbol):
6867         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6869 2013-05-31  Tobias Burnus  <burnus@net-b.de>
6871         PR middle-end/57073
6872         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6874 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6876         PR target/56315
6877         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6878         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6879         * config/arm/neon.md (iordi3_neon): Remove.
6880         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6881         * config/arm/predicates.md (imm_for_neon_logic_operand):
6882         Move to earlier in the file.
6883         (neon_logic_op2): Likewise.
6884         (arm_iordi_operand_neon): New predicate.
6886 2013-05-31  Richard Biener  <rguenther@suse.de>
6888         PR tree-optimization/57478
6889         PR tree-optimization/57453
6890         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6891         are life as well.
6893 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6895         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6896         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6898 2013-05-30  Tobias Burnus  <burnus@net-b.de>
6899             Thomas Koenig  <tkoenig@gcc.gnu.org>
6901         PR middle-end/57073
6902         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6903         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6905 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
6907         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6909 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
6911         * target.def (register_usage_leveling_p): New hook.
6912         * targhooks.c (default_register_usage_leveling_p): New.
6913         * targhooks.h (default_register_usage_leveling_p): New prototype.
6914         * lra-assigns.c (register_usage_leveling_p): Use the hook.
6915         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6916         * doc/tm.texi: Update.
6917         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6919 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
6921         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6922         (*insv_reg<mode>): New define_insn.
6924 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
6926         PR rtl-optimization/57439
6927         * postreload.c (move2add_valid_value_p): Check that we have
6928         a zero subreg_regno_offset when accessing the register in
6929         the requested mode.
6931 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6932             Igor Zamyatin  <igor.zamyatin@intel.com>
6934         Silvermont (SLM) architecture pipeline model, tuning and
6935         insn selection.
6936         * config.gcc: Add slm config options and target.
6938         * config/i386/slm.md: New.
6940         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6942         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
6943         PROCESSOR_SLM.
6944         (ix86_target_macros_internal): Likewise.
6946         * gcc/config/i386/i386.c (slm_cost): New cost.
6947         (m_SLM): New macro flag.
6948         (initial_ix86_tune_features): Set m_SLM.
6949         (x86_accumulate_outgoing_args): Likewise.
6950         (x86_arch_always_fancy_math_387): Likewise.
6951         (processor_target_table): Add slm cost.
6952         (cpu_names): Add slm cpu name.
6953         (x86_option_override_internal): Set SLM ISA.
6954         (ix86_issue_rate): New case PROCESSOR_SLM.
6955         (ia32_multipass_dfa_lookahead): Likewise.
6956         (fold_builtin_cpu): Add slm.
6958         * config/i386/i386.h (TARGET_SLM): New target macro.
6959         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6960         (processor_type): Add PROCESSOR_SLM.
6962         * config/i386/i386.md (cpu): Add new value "slm".
6963         (slm.md): Include slm.md.
6965 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
6966             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6968         * config/arm/arm-protos.h: Add and update function protos.
6969         * config/arm/arm.c (use_simple_return_p): New added.
6970         (thumb2_expand_return): Check simple_return flag.
6971         * config/arm/arm.md: Add simple_return and conditional simple_return.
6972         * config/arm/iterators.md: Add iterator for return and simple_return.
6974 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6976         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6977         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6978         (arm_emit_vfp_multi_reg_pop): Likewise.
6979         (thumb2_emit_ldrd_pop): Likewise.
6980         (arm_expand_epilogue): Add misc REG_CFA notes.
6981         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6983 2013-05-29  Lawrence Crowl  <crowl@google.com>
6985         * config/arm/t-arm: Update for below.
6987         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6988         Change type to hash_table.  Update dependent calls and types.
6990         * config/i386/t-cygming: Update for below.
6992         * config/i386/t-interix: Update for below.
6994         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6995         Change type to hash_table.  Update dependent calls and types.
6996         (i386_find_on_wrapper_list::wrappers): Likewise.
6998         * config/ia64/t-ia64: Update for below.
7000         * config/ia64/ia64.c (bundle_state_table):
7001         Change type to hash_table.  Update dependent calls and types.
7003         * config/mips/mips.c (mips_reorg_process_insns::htab):
7004         Change type to hash_table.  Update dependent calls and types.
7006         * config/sol2.c (solaris_comdat_htab):
7007         Change type to hash_table.  Update dependent calls and types.
7009         * config/t-sol2: Update for above.
7011 2013-05-29  Teresa Johnson  <tejohnson@google.com>
7013         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7014         functions are not yet marked as defined.
7016 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
7017             Pat Haugen <pthaugen@us.ibm.com>
7018             Peter Bergner <bergner@vnet.ibm.com>
7020         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7021         instructions.
7022         (VEC_A): Likewise.
7023         (VEC_C): Likewise.
7024         (vrotl<mode>3): Likewise.
7025         (vashl<mode>3): Likewise.
7026         (vlshr<mode>3): Likewise.
7027         (vashr<mode>3): Likewise.
7029         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7030         support for power8 V2DI builtins.
7032         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7033         power8 V2DI builtins.
7034         (vupkhsw): Likewise.
7035         (vupklsw): Likewise.
7036         (vaddudm): Likewise.
7037         (vminsd): Likewise.
7038         (vmaxsd): Likewise.
7039         (vminud): Likewise.
7040         (vmaxud): Likewise.
7041         (vpkudum): Likewise.
7042         (vpksdss): Likewise.
7043         (vpkudus): Likewise.
7044         (vpksdus): Likewise.
7045         (vrld): Likewise.
7046         (vsld): Likewise.
7047         (vsrd): Likewise.
7048         (vsrad): Likewise.
7049         (vsubudm): Likewise.
7050         (vcmpequd): Likewise.
7051         (vcmpgtsd): Likewise.
7052         (vcmpgtud): Likewise.
7053         (vcmpequd_p): Likewise.
7054         (vcmpgtsd_p): Likewise.
7055         (vcmpgtud_p): Likewise.
7056         (vupkhsw): Likewise.
7057         (vupklsw): Likewise.
7058         (vaddudm): Likewise.
7059         (vmaxsd): Likewise.
7060         (vmaxud): Likewise.
7061         (vminsd): Likewise.
7062         (vminud): Likewise.
7063         (vpksdss): Likewise.
7064         (vpksdus): Likewise.
7065         (vpkudum): Likewise.
7066         (vpkudus): Likewise.
7067         (vrld): Likewise.
7068         (vsld): Likewise.
7069         (vsrad): Likewise.
7070         (vsrd): Likewise.
7071         (vsubudm): Likewise.
7073         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7074         support for power8 V2DI instructions.
7076         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7077         power8 V2DI instructions.  Combine pack and unpack insns to use an
7078         iterator for each mode.  Check whether a particular mode supports
7079         Altivec instructions instead of just checking TARGET_ALTIVEC.
7080         (UNSPEC_VPKUWUM): Likewise.
7081         (UNSPEC_VPKSHSS): Likewise.
7082         (UNSPEC_VPKSWSS): Likewise.
7083         (UNSPEC_VPKUHUS): Likewise.
7084         (UNSPEC_VPKSHUS): Likewise.
7085         (UNSPEC_VPKUWUS): Likewise.
7086         (UNSPEC_VPKSWUS): Likewise.
7087         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7088         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7089         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7090         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7091         (UNSPEC_VUPKHSB): Likewise.
7092         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7093         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7094         (UNSPEC_VUPKHSH): Likewise.
7095         (UNSPEC_VUPKLSB): Likewise.
7096         (UNSPEC_VUPKLSH): Likewise.
7097         (VI2): Likewise.
7098         (VI_char): Likewise.
7099         (VI_scalar): Likewise.
7100         (VI_unit): Likewise.
7101         (VP): Likewise.
7102         (VP_small): Likewise.
7103         (VP_small_lc): Likewise.
7104         (VU_char): Likewise.
7105         (add<mode>3): Likewise.
7106         (altivec_vaddcuw): Likewise.
7107         (altivec_vaddu<VI_char>s): Likewise.
7108         (altivec_vadds<VI_char>s): Likewise.
7109         (sub<mode>3): Likewise.
7110         (altivec_vsubcuw): Likewise.
7111         (altivec_vsubu<VI_char>s): Likewise.
7112         (altivec_vsubs<VI_char>s): Likewise.
7113         (altivec_vavgs<VI_char>): Likewise.
7114         (altivec_vcmpbfp): Likewise.
7115         (altivec_eq<mode>): Likewise.
7116         (altivec_gt<mode>): Likewise.
7117         (altivec_gtu<mode>): Likewise.
7118         (umax<mode>3): Likewise.
7119         (smax<mode>3): Likewise.
7120         (umin<mode>3): Likewise.
7121         (smin<mode>3): Likewise.
7122         (altivec_vpkuhum): Likewise.
7123         (altivec_vpkuwum): Likewise.
7124         (altivec_vpkshss): Likewise.
7125         (altivec_vpkswss): Likewise.
7126         (altivec_vpkuhus): Likewise.
7127         (altivec_vpkshus): Likewise.
7128         (altivec_vpkuwus): Likewise.
7129         (altivec_vpkswus): Likewise.
7130         (altivec_vpks<VI_char>ss): Likewise.
7131         (altivec_vpks<VI_char>us): Likewise.
7132         (altivec_vpku<VI_char>us): Likewise.
7133         (altivec_vpku<VI_char>um): Likewise.
7134         (altivec_vrl<VI_char>): Likewise.
7135         (altivec_vsl<VI_char>): Likewise.
7136         (altivec_vsr<VI_char>): Likewise.
7137         (altivec_vsra<VI_char>): Likewise.
7138         (altivec_vsldoi_<mode>): Likewise.
7139         (altivec_vupkhsb): Likewise.
7140         (altivec_vupkhs<VU_char>): Likewise.
7141         (altivec_vupkls<VU_char>): Likewise.
7142         (altivec_vupkhsh): Likewise.
7143         (altivec_vupklsb): Likewise.
7144         (altivec_vupklsh): Likewise.
7145         (altivec_vcmpequ<VI_char>_p): Likewise.
7146         (altivec_vcmpgts<VI_char>_p): Likewise.
7147         (altivec_vcmpgtu<VI_char>_p): Likewise.
7148         (abs<mode>2): Likewise.
7149         (vec_unpacks_hi_v16qi): Likewise.
7150         (vec_unpacks_hi_v8hi): Likewise.
7151         (vec_unpacks_lo_v16qi): Likewise.
7152         (vec_unpacks_hi_<VP_small_lc>): Likewise.
7153         (vec_unpacks_lo_v8hi): Likewise.
7154         (vec_unpacks_lo_<VP_small_lc>): Likewise.
7155         (vec_pack_trunc_v8h): Likewise.
7156         (vec_pack_trunc_v4si): Likewise.
7157         (vec_pack_trunc_<mode>): Likewise.
7159         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7160         V2DI builtins.
7161         (vec_vmaxsd): Likewise.
7162         (vec_vmaxud): Likewise.
7163         (vec_vminsd): Likewise.
7164         (vec_vminud): Likewise.
7165         (vec_vpksdss): Likewise.
7166         (vec_vpksdus): Likewise.
7167         (vec_vpkudum): Likewise.
7168         (vec_vpkudus): Likewise.
7169         (vec_vrld): Likewise.
7170         (vec_vsld): Likewise.
7171         (vec_vsrad): Likewise.
7172         (vec_vsrd): Likewise.
7173         (vec_vsubudm): Likewise.
7174         (vec_vupkhsw): Likewise.
7175         (vec_vupklsw): Likewise.
7177 2013-05-29  Jan Hubicka  <jh@suse.cz>
7179         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7180         flags; reorder rest of fields in more consistent way.
7181         (varpool_node): Remove analyzed, finalized and alias.
7182         (cgraph_ndoe): Likewise.
7183         (symtab_alias_ultimate_target): New function.
7184         (cgraph_function_node): Move offline.
7185         (cgraph_reset_node): Declare.
7186         (cgraph_comdat_can_be_unshared_p): Remove.
7187         (varpool_remove_initializer): Declare.
7188         (varpool_first_defined_variable, varpool_next_defined_variable
7189         cgraph_first_defined_function, cgraph_next_defined_function): Update.
7190         (cgraph_function_with_gimple_body_p): Update.
7191         (varpool_all_refs_explicit_p): Update.
7192         (symtab_alias_target): New function.
7193         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7194         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7195         (cgraph_function_or_thunk_node): Simplify using
7196         symtab_alias_ultimate_target.
7197         (varpool_variable_node): Likewise.
7198         * cgraph.c (cgraph_create_function_alias): Update.
7199         (cgraph_add_thunk): Update.
7200         (cgraph_remove_node): Update.
7201         (dump_cgraph_node): Do not dump removed flags.
7202         (cgraph_function_body_availability): Update.
7203         (cgraph_propagate_frequency): Update.
7204         (verify_cgraph_node): Check sanity of local flag.
7205         (cgraph_function_node): Move here from cgraph.h; revamp for
7206         cgraph_function_or_thunk_node.
7207         * lto-symtab.c (lto_varpool_replace_node): Update.
7208         (lto_symtab_resolve_can_prevail_p): Update.
7209         (lto_symtab_merge_cgraph_nodes): Update.
7210         * ipa-cp.c (determine_versionability, initialize_node_lattices,
7211         propagate_constants_accross_call, devirtualization_time_bonus,
7212         ipcp_propagate_stage): Update.
7213         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7214         * ipa-inline-transform.c (clone_inlined_nodes,
7215         preserve_function_body_p): Update.
7216         * ipa-reference.c (propagate): Update.
7217         (write_node_summary_p): Update.
7218         * toplev.c (wrapup_global_declaration_2): Update.
7219         * cgraphunit.c (cgraph_analyze_function): Rename to ...
7220         (analyze_function) ... this one.
7221         (cgraph_process_new_functions): Update.
7222         (cgraph_reset_node): Export.
7223         (cgraph_finalize_function): Update.
7224         (cgraph_add_new_function): Update.
7225         (process_function_and_variable_attributes): Update.
7226         (varpool_finalize_decl): Update.
7227         (symbol_finalized): Remove.
7228         (symbol_finalized_and_needed): Rename to ...
7229         (symbol_defined_and_needed): ... update.
7230         (cgraph_analyze_functions): Update.
7231         (handle_alias_pairs): Update.
7232         (mark_functions_to_output): Update.
7233         (assemble_thunk): Update.
7234         (output_in_order): Update.
7235         (output_weakrefs): Update.
7236         (finalize_compilation_unit): Update.
7237         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7238         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7239         input_node, input_varpool_node): Update.
7240         * dbxout.c (dbxout_expand_expr): Update.
7241         * cgraphclones.c (cgraph_clone_node): Update.
7242         (cgraph_copy_node_for_versioning): Update.
7243         (cgraph_materialize_clone): Update.
7244         (cgraph_materialize_all_clones): Update.
7245         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7246         propagate_pure_const, propagate_nothrow): Update.
7247         * lto-streamer-out.c (lto_output, write_symbol): Update.
7248         * ipa-utils.c (ipa_reverse_postorder): Update.
7249         * ipa-inline.c (can_inline_edge_p): Update.
7250         (update_caller_keys, ipa_inline): Update.
7251         * dwarf2out.c (reference_to_unused,
7252         premark_types_used_by_global_vars_helper): Update.
7253         * tree-eh.c (tree_could_trap_p): Update.
7254         * ipa-split.c (consider_split, execute_split_functions): Update.
7255         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7256         has_addr_references_p): Update; move ahead in file for better
7257         readability.
7258         (process_references): Simplify.
7259         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7260         bodies are removed.
7261         (cgraph_comdat_can_be_unshared_p): Make static.
7262         (cgraph_externally_visible_p): Update.
7263         (varpool_externally_visible_p): Update.
7264         (function_and_variable_visibility): Update.
7265         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7266         ipa_tm_mark_force_output_node): Update.
7267         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7268         estimate_edge_devirt_benefit, inline_generate_summary,
7269         inline_write_summary): Update.
7270         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7271         * ipa-prop.c (ipa_compute_jump_functions): Update.
7272         (ipa_print_node_params, ipa_prop_read_section,
7273         ipa_update_after_lto_read, read_replacements_section): Update.
7274         * varasm.c (mark_decl_referenced): Update.
7275         (assemble_alias, dump_tm_clone_pairs): Update.
7276         * tree-inline.c (copy_bb): Update.
7277         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7278         Update.
7279         * symtab.c (dump_symtab_base): Print new flags.
7280         (verify_symtab_base): Verify new flags.
7281         (symtab_alias_ultimate_target): New function.
7282         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7283         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7284         Update.
7285         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7286         Update.
7287         * i386.c (ix86_get_function_versions_dispatcher,
7288         ix86_generate_version_dispatcher_body): Update.
7289         (fold_builtin_cpu): Use varpool_add_new_variable.
7290         * varpool.c (varpool_remove_initializer): Break out from ...
7291         (varpool_remove_node): ... this one.
7292         (dump_varpool_node, varpool_node_for_asm,
7293         cgraph_variable_initializer_availability, varpool_analyze_node,
7294         varpool_assemble_decl, varpool_remove_unreferenced_decls,
7295         varpool_finalize_named_section_flags, varpool_create_variable_alias):
7296         Update.
7298 2013-05-29  Jan Hubicka  <jh@suse.cz>
7300         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7302 2013-05-29  Easwaran Raman  <eraman@google.com>
7304         PR tree-optimization/57442
7305         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7306         when control exits the main loop.
7308 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
7310         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7311         and RX600.
7312         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7313         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7314         * rx/t-rx: Add rx100 under multi library matches option for nofpu
7315         option.
7317 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7319         PR tree-optimization/57441
7320         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7321         Don't limit size of incr_vec to number of candidates.
7323 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
7325         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7326         and mips16 directories.
7327         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7328         (MULTILIB_DIRNAMES): Ditto.
7329         (MULTILIB_EXCEPTIONS): Add new exceptions.
7330         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7331         (MULTILIB_DIRNAMES): Ditto.
7332         (MULTILIB_EXCEPTIONS): Add new exceptions.
7334 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7335             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7337         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7338         SYMBOL_TINY_ABSOLUTE.
7339         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7340         SYMBOL_TINY_ABSOLUTE.
7341         (aarch64_expand_mov_immediate): Likewise.
7342         (aarch64_classify_symbol): Likewise.
7343         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7344         Permit SYMBOL_TINY_ABSOLUTE.
7345         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7347 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7348             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7350         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7351         Refactor if/switch.  Replace gcc_assert with if.
7353 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7355         * config/i386/i386.c (initial_ix86_tune_features): Enable
7356         FP Reassociation for AMD bdver1 and bdver2.
7358 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7360         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7361         and IMAGPART_EXPR do not occur within other handled_components.
7363 2013-05-29  Richard Biener  <rguenther@suse.de>
7365         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7366         access on whether the use is in the BB we currently try to
7367         vectorize.
7368         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7369         vectorize to vect_bb_slp_scalar_cost.
7371 2013-05-29  Richard Biener  <rguenther@suse.de>
7373         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7374         computing scalar cost offsetted by stmts that are kept live
7375         by scalar uses.
7376         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7377         for computation of scalar cost.
7379 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7381         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7383 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7385         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7386         array notation built-in reduction functions.
7387         * doc/passes.texi (Passes): Added documentation about changes done
7388         for Cilk Plus.
7389         * doc/invoke.texi (C Dialect Options): Added documentation about
7390         the -fcilkplus flag.
7391         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7392         (BUILTINS_DEF): Depend on cilkplus.def.
7393         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7394         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7395         * cilkplus.def: New file.
7397 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7399         PR rtl-optimization/57439
7400         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7402 2013-05-28  Easwaran Raman  <eraman@google.com>
7404         PR tree-optimization/57337
7405         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7406         (find_insert_point): Correctly identify the insertion point
7407         when two statements with the same UID is compared.
7409 2013-05-28  Richard Biener  <rguenther@suse.de>
7411         PR tree-optimization/56787
7412         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7413         from the list of data references.
7414         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7415         clobbers.
7416         (vect_analyze_loop_operations): Likewise.
7417         (vect_transform_loop): Remove clobbers.
7419 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7421         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7422         and REALPART_EXPRs have scalar type.
7424 2013-05-28  Richard Biener  <rguenther@suse.de>
7426         PR tree-optimization/57411
7427         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7428         virtual operands.
7429         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7430         virtual operand propagation.
7432 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7434         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7435         destination register for bmasksi_vis.
7436         (vector_init_bshuffle): Likewise.
7437         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7439 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7441         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7442         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7443         mode if the instruction isn't available in the original mode.
7444         * config/sparc/sparc.opt (mfix-ut699): New option.
7445         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7446         (divdf3): Turn into expander.
7447         (divdf3_nofix): New insn.
7448         (divdf3_fix): Likewise.
7449         (divsf3): Disable if -mfix-ut699.
7450         (sqrtdf2): Turn into expander.
7451         (sqrtdf2_nofix): New insn.
7452         (sqrtdf2_fix): Likewise.
7453         (sqrtsf2): Disable if -mfix-ut699.
7455 2013-05-27  Richard Biener  <rguenther@suse.de>
7457         PR middle-end/57412
7458         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7459         block for the new loop.
7461 2013-05-27  Richard Biener  <rguenther@suse.de>
7463         PR tree-optimization/57343
7464         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7465         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7466         (number_of_iterations_cond): Do not build the folded tree.
7468 2013-05-27  Richard Biener  <rguenther@suse.de>
7470         Revert
7471         PR middle-end/57381
7472         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7473         OEP_CONSTANT_ADDRESS_OF retained.
7475         PR tree-optimization/57417
7476         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7477         for unchanged base.
7478         (set_ssa_val_to): Compare addresses using
7479         get_addr_base_and_unit_offset.
7481 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7483         PR rtl-optimization/56833
7484         * postreload.c (move2add_record_mode): New function.
7485         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7486         (move2add_use_add2_insn): Use move2add_record_sym_value.
7487         (move2add_use_add3_insn): Likewise.
7488         (reload_cse_move2add): Use move2add_valid_value_p and
7489         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7490         affected regs by setting reg_mode to VOIDmode.
7491         (move2add_note_store): Don't pretend the inside of a SUBREG is
7492         the actual destination.  Invalidate single/leading registers by
7493         setting reg_mode to VOIDmode.
7494         Use move2add_record_sym_value, move2add_valid_value_p and
7495         move2add_record_mode.
7497 2013-05-27  Richard Biener  <rguenther@suse.de>
7499         PR tree-optimization/57396
7500         * tree-affine.c (double_int_constant_multiple_p): Properly
7501         return false for val == 0 and div != 0.
7503 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7505         * config/mips/mips.h: Use #elif in preprocessor conditions.
7507 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7509         PR target/53916
7510         * config/mips/constraints.md (kl): New constraint.
7511         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7512         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7513         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7514         from LO for MIPS16.
7515         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7517 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7519         PR target/55777
7520         * config/mips/mips.c (mips_can_inline_p): New function.
7521         (TARGET_CAN_INLINE_P): Define.
7523 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7525         * sched-int.h (ds_t, dw_t): Make unsigned int.
7526         Fix documentation that describes how all the ds_t bits are used.
7527         Reserve the last bit for delayed-branch scheduling.
7528         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7529         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7530         (gen_dep_weak_1): Remove prototype.
7531         * sched-deps.c (get_dep_weak_1): Make static.
7532         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7533         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7534         * doc/tm.texi: Regenerate.
7535         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7537 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7539         PR debug/56950
7540         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7542 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7543             Sandra Loosemore  <sandra@codesourcery.com>
7545         * config.gcc (powerpc-*): Allow native for with-cpu.
7547 2013-05-24  Jeff Law  <law@redhat.com>
7549         PR tree-optimization/57124
7550         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7551         conversion feeding a condition if the range has an overflow
7552         if -fstrict-overflow.  Add warnings for when we do make the
7553         transformation.
7555 2013-05-24  Dehao Chen  <dehao@google.com>
7557         * tree-cfg.c (locus_discrim_map): Fix the typo.
7558         (locus_discrim_hasher): Likewise.
7559         (locus_discrim_hasher::hash): Likewise.
7560         (locus_discrim_hasher::equal): Likewise.
7562 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7564         PR tree-optimization/57294
7565         * cgraph.h (ipa_record_stmt_references): Declare.
7566         * cgraphbuild.c (ipa_record_stmt_references): New function.
7567         (build_cgraph_edges): Use ipa_record_stmt_references.
7568         (rebuild_cgraph_edges): Likewise.
7569         (cgraph_rebuild_references): Likewise.
7570         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7571         associated with the old statement and build references from the
7572         newly built statements.
7573         * ipa-ref.c (ipa_remove_stmt_references): New function.
7574         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7576 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7578         * lra-constraints.c (emit_spill_move): Use smaller mode for
7579         mem-mem moves.
7580         (check_and_process_move): Consider mem-reg moves for secondary
7581         too.
7582         (curr_insn_transform): Don't lose insns emitted before for
7583         secondary memory moves.
7584         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7585         reg set up in the current insn.
7587 2013-05-24  Dehao Chen  <dehao@google.com>
7589         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7590         hash function.
7591         (locus_descrim_hasher::equal): Likewise.
7592         (build_gimple_cfg): New discriminator assignment algorithm.
7593         (make_edges): Likewise.
7594         (next_discriminator_for_locus): Likewise.
7595         (same_line_p): Likewise.
7596         (assign_discriminators): Likewise.
7597         (make_cond_expr_edges): Likewise.
7598         (make_gimple_switch_edges): Likewise.
7599         (make_goto_expr_edges): Likewise.
7600         (make_gimple_asm_edges): Likewise.
7602 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
7604         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7605         X format specifier to only display bottom 16 bits.
7606         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7607         immediate to match for operand 2, since it will be masked.
7609 2013-05-24  Richard Biener  <rguenther@suse.de>
7611         PR tree-optimization/57287
7612         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7613         all SSA names that occur in abnormal PHIs.
7615 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7617         PR tree-ssa/57385
7618         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7619         that index is not negative.
7621 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
7623         PR rtl-optimization/55177
7624         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7625         (simplify_byte_swapping_operation): New.
7626         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7627         (simplify_relational_operation_1): Deal with BSWAP.
7629 2013-05-23  Richard Henderson  <rth@redhat.com>
7631         PR target/56742
7632         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7633         (ix86_reorg): Call it.
7635 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
7637         PR target/57379
7638         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7639         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7640         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7642 2013-05-23  Christian Bruel  <christian.bruel@st.com>
7644         PR debug/57351
7645         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7647 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7648             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7650         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7651         * config/aarch64/constraints.md (Usa): Remove.
7652         * doc/md.texi (AArch64 Usa): Remove.
7654 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7655             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7657         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7658         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7659         * config/aarch64/predicates.md (aarch64_const_address): Remove.
7660         (aarch64_mov_operand): Use aarch64_mov_operand_p.
7662 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
7664         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7665         instruction (AdvSIMD).
7666         * config/aarch64/aarch64-builtins.c
7667         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7668         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7670 2013-05-23  Martin Jambor  <mjambor@suse.cz>
7672         PR middle-end/57347
7673         * tree.h (contains_bitfld_component_ref_p): Declare.
7674         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7675         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
7676         caller.
7677         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7678         not access a bit-field.  Assert all final offsets are byte-aligned.
7680 2013-05-23  Richard Biener  <rguenther@suse.de>
7682         PR tree-optimization/57380
7683         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7684         least one invariant or re-used load.
7685         * passes.c (init_optimization_passes): Move pass_phiprop before
7686         pass_forwprop.
7688 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
7690         * config/aarch64/aarch64-simd.md
7691         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7693 2013-05-23  Richard Biener  <rguenther@suse.de>
7695         PR middle-end/57381
7696         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7697         OEP_CONSTANT_ADDRESS_OF retained.
7699 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
7701         PR middle-end/57344
7702         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7703         don't lower unit.  Handle unit not being always BITS_PER_WORD.
7705 2013-05-23  Richard Biener  <rguenther@suse.de>
7707         PR rtl-optimization/57341
7708         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7709         instead of true_dependence.
7711 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
7713         * bb-reorder.c (branch_threshold): Make const.
7714         (exec_threshold): Ditto.
7716 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7717             Pat Haugen <pthaugen@us.ibm.com>
7718             Peter Bergner <bergner@vnet.ibm.com>
7720         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7721         documentation for the power8 crypto builtins.
7723         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7725         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7726         macros for defining power8 builtin functions.
7727         (BU_P8V_AV_2): Likewise.
7728         (BU_P8V_AV_P): Likewise.
7729         (BU_P8V_VSX_1): Likewise.
7730         (BU_P8V_OVERLOAD_1): Likewise.
7731         (BU_P8V_OVERLOAD_2): Likewise.
7732         (BU_CRYPTO_1): Likewise.
7733         (BU_CRYPTO_2): Likewise.
7734         (BU_CRYPTO_3): Likewise.
7735         (BU_CRYPTO_OVERLOAD_1): Likewise.
7736         (BU_CRYPTO_OVERLOAD_2): Likewise.
7737         (XSCVSPDP): Fix typo, point to the correct instruction.
7738         (VCIPHER): Add power8 crypto builtins.
7739         (VCIPHERLAST): Likewise.
7740         (VNCIPHER): Likewise.
7741         (VNCIPHERLAST): Likewise.
7742         (VPMSUMB): Likewise.
7743         (VPMSUMH): Likewise.
7744         (VPMSUMW): Likewise.
7745         (VPERMXOR_V2DI): Likewise.
7746         (VPERMXOR_V4SI: Likewise.
7747         (VPERMXOR_V8HI: Likewise.
7748         (VPERMXOR_V16QI: Likewise.
7749         (VSHASIGMAW): Likewise.
7750         (VSHASIGMAD): Likewise.
7751         (VPMSUM): Likewise.
7752         (VPERMXOR): Likewise.
7753         (VSHASIGMA): Likewise.
7755         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7756         __CRYPTO__ if the crypto instructions are available.
7757         (altivec_overloaded_builtins): Add support for overloaded power8
7758         builtins.
7760         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7761         support for power8 crypto builtins.
7762         (builtin_function_type): Likewise.
7763         (altivec_init_builtins): Add support for builtins that take vector
7764         long long (V2DI) arguments.
7766         * config/rs6000/crypto.md: New file, define power8 crypto
7767         instructions.
7769 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7770             Pat Haugen <pthaugen@us.ibm.com>
7771             Peter Bergner <bergner@vnet.ibm.com>
7773         * doc/invoke.texi (Option Summary): Add power8 options.
7774         (RS/6000 and PowerPC Options): Likewise.
7776         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7777         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
7778         wm, wn, wr documentation.
7780         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7781         registers if direct move instructions are enabled.
7782         (wn): New constraint for no registers.
7783         (wq): New constraint for quad word even GPR registers.
7784         (wr): New constraint if 64-bit instructions are enabled.
7785         (wv): New constraint if power8 vector instructions are enabled.
7786         (wQ): New constraint for quad word memory locations.
7788         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7789         constraint for 0..15 for crypto instructions.
7790         (gpc_reg_operand): If VSX allow registers in VSX registers as well
7791         as GPR and floating point registers.
7792         (int_reg_operand): New predicate to match only GPR registers.
7793         (base_reg_operand): New predicate to match base registers.
7794         (quad_int_reg_operand): New predicate to match even GPR registers
7795         for quad memory operations.
7796         (vsx_reg_or_cint_operand): New predicate to allow vector logical
7797         operations in both GPR and VSX registers.
7798         (quad_memory_operand): New predicate for quad memory operations.
7799         (reg_or_indexed_operand): New predicate for direct move support.
7801         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7802         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7803         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7804         (POWERPC_MASKS): Add power8 options.
7805         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7806         various options.
7808         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7809         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7811         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7812         (-mpower8-fusion): New power8 options.
7813         (-mpower8-fusion-sign): Likewise.
7814         (-mpower8-vector): Likewise.
7815         (-mcrypto): Likewise.
7816         (-mdirect-move): Likewise.
7817         (-mquad-memory): Likewise.
7819         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7820         power8.
7821         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7822         registers.
7823         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7824         (rs6000_debug_vector_unit): Add p8_vector.
7825         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7826         definitions.  Also print fusion state.
7827         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7828         (rs6000_builtin_mask_calculate): Add power8 builtin support.
7829         (rs6000_option_override_internal): Add support for power8.
7830         (rs6000_common_init_builtins): Add debugging for skipped builtins
7831         if -mdebug=builtin.
7832         (rs6000_adjust_cost): Add power8 support.
7833         (rs6000_issue_rate): Likewise.
7834         (insn_must_be_first_in_group): Likewise.
7835         (insn_must_be_last_in_group): Likewise.
7836         (force_new_group): Likewise.
7837         (rs6000_register_move_cost): Likewise.
7838         (rs6000_opt_masks): Likewise.
7840         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7841         power8 capable assembler, default to power7 options.
7842         (TARGET_DIRECT_MOVE): Likewise.
7843         (TARGET_CRYPTO): Likewise.
7844         (TARGET_P8_VECTOR): Likewise.
7845         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7846         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7847         (VECTOR_MEM_P8_VECTOR_P): Likewise.
7848         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7849         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7850         (TARGET_XSCVDPSPN): Likewise.
7851         (TARGET_XSCVSPDPN): Likewsie.
7852         (TARGET_SYNC_HI_QI): Likewise.
7853         (TARGET_SYNC_TI): Likewise.
7854         (MASK_CRYPTO): Likewise.
7855         (MASK_DIRECT_MOVE): Likewise.
7856         (MASK_P8_FUSION): Likewise.
7857         (MASK_P8_VECTOR): Likewise.
7858         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7859         temporary used by some of the direct move instructions to get two FP
7860         temporary registers does not force creation of a stack frame.
7861         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7862         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7863         that any VSX registers are tieable, even if they are also an
7864         Altivec vector mode.
7865         (r6000_reg_class_enum): Add wm, wr, wv constraints.
7866         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7867         (RS6000_BTM_CRYPTO): Likewise.
7868         (RS6000_BTM_COMMON): Likewise.
7870         * config/rs6000/rs6000.md (cpu attribute): Add power8.
7871         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7872         (enum rs6000_vector): Add power8 vector support.
7874 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7876         PR target/19599
7877         PR target/57340
7878         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7879         (any_sibcall_could_use_r3): this and handle indirect calls.
7880         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7882 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7884         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7886 2013-05-22  Richard Biener  <rguenther@suse.de>
7888         PR middle-end/57349
7889         * profile.c (branch_prob): Do not split blocks that are
7890         abnormally receiving from ECF_RETURNS_TWICE functions.
7892 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7894         * recog.c (offsettable_address_addr_space_p): Fix calculation of
7895         address mode.  Move pointer mode initialization to the same place.
7897 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7899         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7900         while it has any effect.
7902 2013-05-21  Easwaran Raman  <eraman@google.com>
7904         PR tree-optimization/57322
7905         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7906         UID of the statement added to the BB to be 1.
7908 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
7910         PR tree-optimization/57331
7911         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7912         of conversion from pointer type to integral type with integer.
7914 2013-05-21  Martin Jambor  <mjambor@suse.cz>
7916         PR lto/57289
7917         * ipa-prop.c (ipa_read_node_info): Process param_used and
7918         controlled_uses in the same order as when writing.
7920 2013-05-21  Magnus Granberg  <baldrick@free.fr>
7922         PR plugins/56754
7923         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7925 2013-05-21  Richard Biener  <rguenther@suse.de>
7927         PR tree-optimization/57318
7928         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7929         estimate stmts with side-effects as likely eliminated.
7931 2013-05-21  Richard Biener  <rguenther@suse.de>
7933         PR tree-optimization/57330
7934         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7935         preserve the call stmts fntype.
7937 2013-05-21  Richard Biener  <rguenther@suse.de>
7939         PR tree-optimization/57303
7940         * tree-ssa-sink.c (statement_sink_location): Improve killing
7941         stmt detection and properly handle self-assignments.
7943 2013-05-21  Christian Bruel  <christian.bruel@st.com>
7945         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7946         spanning registers. LEAF_REG_REMAP is supported only for contiguous
7947         registers. Set register size out of the PARALLEL loop.
7949 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
7951         PR target/56547
7952         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7953         (*fmasf4, *fmasf4_media): New insns.
7955 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7957         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7958         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7959         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7960         (mips_idiv_insns): Update the comments to say that the returned
7961         instruction counts are in units of BASE_INSN_LENGTH.
7962         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7963         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
7964         using 2 rather than 4 as the length of indirect MIPS16 and
7965         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
7966         length of a NOP.  Don't divide MIPS16 lengths by 2.
7967         (mips16_split_long_branches): Assume a branch is long if the
7968         length is greater than 4 rather than 8.
7969         * config/mips/mips.md (length): Give MIPS16 lengths directly,
7970         rather than multiplying them by 2.  Multiply instruction counts
7971         by BASE_INSN_LENGTH rather than 4.
7972         (*jump_mips16, tls_get_tp_mips16_<mode>)
7973         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7975 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7977         * config/mips/mips.md (extended_mips16): Remove branch case.
7978         (length): Remove duplicated extended_mips16 test.
7980 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7982         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7984 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
7986         * recog.h (Recog_data): Rename to...
7987         (recog_data_d): ...this.
7988         (recog_data): Update accordingly.
7989         * recog.c (recog_data): Likewise.
7990         * reload.c (save_recog_data): Likewise.
7991         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7992         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7994 2013-05-17  Julian Brown  <julian@codesourcery.com>
7996         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7997         found in a REG_EQUAL note, invalidate it.
7999 2013-05-17  Easwaran Raman  <eraman@google.com>
8001         * tree-ssa-reassoc.c (find_insert_point): New function.
8002         (insert_stmt_after): Likewise.
8003         (get_def_stmt): Likewise.
8004         (ensure_ops_are_available): Likewise.
8005         (not_dominated_by): Likewise.
8006         (rewrite_expr_tree): Do not move statements beyond what is
8007         necessary. Remove call to swap_ops_for_binary_stmt...
8008         (reassociate_bb): ... and move it here.
8009         (build_and_add_sum): Assign UIDs for new statements.
8010         (linearize_expr): Likewise.
8011         (do_reassoc): Renumber gimple statement UIDs.
8013 2013-05-17  Jan Hubicka  <jh@suse.cz>
8015         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8016         weakrefs.
8017         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8018         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8019         weakrefs.
8020         (output_weakrefs): Update.
8022 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
8023             Martin Jambor  <mjambor@suse.cz>
8025         PR middle-end/57276
8026         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8027         value that corresponds to the given aggval is found in values vector.
8029 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
8031         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8032         sse, sse2, sse3, ssse3 and sse4a flags to options.
8034 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
8036         * gengtype-state.c: (s_expr_writer): New class, to handle
8037         prettifying of output layout of s-expressions.
8038         (state_writer): New class, to write out gtype.state.
8039         (state_written_type_count): Move this variable into member data of
8040         state_writer.
8041         (s_expr_writer::s_expr_writer): New code: constructor for new class
8042         (state_writer::state_writer(): ditto
8043         (s_expr_writer::write_new_line): New function
8044         (s_expr_writer::write_any_indent): ditto
8045         (s_expr_writer::begin_s_expr): ditto
8046         (s_expr_writer::end_s_expr): ditto
8047         (write_state_fileloc): convert to method of state_writer...
8048         (state_writer:: write_state_fileloc): ...and use methods of
8049         s_expr_writer to write indentation into the gtype.state output file
8050         to visually represent the hierarchical structure of the list
8051         structures
8052         (write_state_fields): ditto, renaming to...
8053         (state_writer::write_state_fields)
8054         (write_state_a_string): ditto, renaming to...
8055         (state_writer::write_state_a_string)
8056         (write_state_string_option): ditto, renaming to...
8057         (state_writer::write_state_string_option)
8058         (write_state_type_option): ditto, renaming to...
8059         (state_writer::write_state_type_option)
8060         (write_state_nested_option): ditto, renaming to...
8061         (state_writer::write_state_nested_option)
8062         (write_state_option): ditto, renaming to...
8063         (state_writer::write_state_option)
8064         (write_state_options): ditto, renaming to...
8065         (state_writer::write_state_options)
8066         (write_state_lang_bitmap): ditto, renaming to...
8067         (state_writer::write_state_lang_bitmap)
8068         (write_state_version): ditto, renaming to...
8069         (state_writer::write_state_version)
8070         (write_state_scalar_type): ditto, renaming to...
8071         (state_writer::write_state_scalar_type)
8072         (write_state_string_type): ditto, renaming to...
8073         (state_writer::write_state_string_type)
8074         (write_state_undefined_type): ditto, renaming to...
8075         (state_writer::write_state_undefined_type)
8076         (write_state_struct_union_type): ditto, renaming to...
8077         (state_writer::write_state_struct_union_type)
8078         (write_state_struct_type): ditto, renaming to...
8079         (state_writer::write_state_struct_type)
8080         (write_state_user_struct_type): ditto, renaming to...
8081         (state_writer::write_state_user_struct_type)
8082         (write_state_lang_struct_type): ditto, renaming to...
8083         (state_writer::write_state_lang_struct_type)
8084         (write_state_param_struct_type): ditto, renaming to...
8085         (state_writer::write_state_param_struct_type)
8086         (write_state_pointer_type): ditto, renaming to...
8087         (state_writer::write_state_pointer_type)
8088         (write_state_array_type): ditto, renaming to...
8089         (state_writer::write_state_array_type)
8090         (write_state_gc_used): ditto, renaming to...
8091         (state_writer::write_state_gc_used)
8092         (write_state_common_type_content): ditto, renaming to...
8093         (state_writer::write_state_common_type_content)
8094         (write_state_type): ditto, renaming to...
8095         (state_writer::write_state_type)
8096         (write_state_pair_list): ditto, renaming to...
8097         (state_writer::write_state_pair_list)
8098         (write_state_pair): ditto, renaming to...
8099         (state_writer::write_state_pair)
8100         (write_state_typedefs): ditto, renaming to...
8101         (state_writer::write_state_typedefs)
8102         (write_state_structures): ditto, renaming to...
8103         (state_writer::write_state_structures)
8104         (write_state_param_structs): ditto, renaming to...
8105         (state_writer::write_state_param_structs)
8106         (write_state_variables): ditto, renaming to...
8107         (state_writer::write_state_variables)
8108         (write_state_srcdir): ditto, renaming to...
8109         (state_writer::write_state_srcdir)
8110         (write_state_files_list): ditto, renaming to...
8111         (state_writer::write_state_files_list)
8112         (write_state_languages): ditto, renaming to...
8113         (state_writer::write_state_languages)
8114         (write_state): create a state_writer instance and use it when
8115         writing out the state file
8117 2013-05-17  Mike Stump  <mikestump@comcast.net>
8119         PR rtl-optimization/57304
8120         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8121         accessing DF_REF_REAL_LOC.
8123 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8125         PR rtl-optimization/57281
8126         PR rtl-optimization/57300
8127         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8128         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8129         what the other splitter did if the registers are dead.
8131 2013-05-17  Richard Biener  <rguenther@suse.de>
8133         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8134         MEM_REF offsets.
8136 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8138         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8139         linking.
8141 2013-05-17  Marek Polacek  <polacek@redhat.com>
8143         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8144         length when doing non-zero store of storing '\0' to '\0'.
8146 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8148         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8149         vect_external_def oprnd1 with loop_vinfo, try to emit
8150         optional cast, negation and and stmts on the loop preheader
8151         edge instead of into the pattern def seq.
8153         PR tree-optimization/57051
8154         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8155         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8157 2013-05-16  Nick Clifton  <nickc@redhat.com>
8159         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8160         (rl78_is_naked_func): New function.
8161         (rl78_expand_prologue): Skip prologue generation for naked functions.
8162         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8163         * doc/extend.texi (naked): Add RL78 to the list of processors
8164         that supports this attribute.
8166 2013-05-16  Jeff Law  <law@redhat.com>
8168         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8170 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8172         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8173         cache parameters using detect_caches_amd also for CYRIX,
8174         NSC and TM2 signatures.
8176 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8177             Dzianis Kahanovich  <mahatma@eu.by>
8179         PR target/45359
8180         PR target/46396
8181         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8182         VIA/Centaur processors and determine their cache parameters
8183         using detect_caches_amd.
8185 2013-05-16  Teresa Johnson  <tejohnson@google.com>
8187         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8188         (rtl_verify_edges): New function.
8189         (rtl_verify_bb_insns): Ditto.
8190         (rtl_verify_bb_pointers): Ditto.
8191         (rtl_verify_bb_insn_chain): Ditto.
8192         (rtl_verify_fallthru): Ditto.
8193         (rtl_verify_bb_layout): Ditto.
8194         (rtl_verify_flow_info_1): Outline checks into new functions.
8195         (rtl_verify_flow_info): Ditto.
8197 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
8199         * cfghooks.c (copy_bbs): Add update_dominance argument.
8200         * cfghooks.h (copy_bbs): Update prototype.
8201         * tree-cfg.c (gimple_duplicate_sese_region):
8202         Add update_dominance argument.
8203         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8204         * tree-ssa-loop-ch.c (copy_loop_headers): Update
8205         gimple_duplicate_sese_region call.
8206         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8207         Update copy_bbs call.
8208         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8209         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8211 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8213         * tree-vectorizer.h (NUM_PATTERNS): Increment.
8214         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8215         vect_recog_rotate_pattern.
8216         (vect_recog_rotate_pattern): New function.
8218 2013-05-16  Jason Merrill  <jason@redhat.com>
8220         * Makefile.in (LLINKER): New variable.
8221         (mostlyclean): Remove link mutex.
8222         * configure.ac: Handle --enable-link-mutex.
8223         * lock-and-run.sh: New script.
8225 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8227         PR target/19599
8228         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8229         for NULL decl.
8231 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8233         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8235 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
8237         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8238         * config/arm/arm.c (next_consecutive_mem): New function.
8239         (gen_movmem_ldrd_strd): Likewise.
8240         * config/arm/arm.md (movmemqi): Update condition and code.
8241         (unaligned_loaddi, unaligned_storedi): New patterns.
8243 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8245         * config.gcc: Obsolete *-*-solaris2.9*.
8246         * doc/install.texi (Specific, *-*-solaris2*): Document it.
8248 2013-05-16  Richard Biener  <rguenther@suse.de>
8250         * passes.c (init_optimization_passes): Move pass_parallelize_loops
8251         earlier, after GRAPHITE transforms and IV canonicalization.
8253 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8255         * omp-low.c (extract_omp_for_data): For collapsed loops,
8256         if at least one of the loops is known at compile time to
8257         iterate zero times, set count to 0.
8258         (expand_omp_regimplify_p): New function.
8259         (expand_omp_for_generic): For collapsed loops, if at least
8260         one of the loops isn't known to iterate at least once,
8261         add runtime check with setting count to 0.
8262         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8263         For unsigned types if it isn't known at compile time that
8264         the loop will iterate at least once, add runtime check to bypass
8265         the whole loop if initial condition isn't true.
8267 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
8269         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8271 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
8273         PR middle-end/57286
8274         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8275         transformations to avoid an infinite loop.
8277 2013-05-16  Marek Polacek  <polacek@redhat.com>
8279         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8281 2013-05-15  Leif Ekblad  <leif@rdos.net>
8283         * config/i386/i386.c (ix86_decompose_address): Use
8284         DEFAULT_TLS_SEG_REG to access TLS segment register.
8285         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8286         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8287         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8289 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
8291         PR target/57260
8292         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8293         sibling calls to functions that would normally be lazily bound,
8294         unless $gp is call-clobbered.
8296 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
8298         * config/i386/i386.c (ix86_option_override_internal): Update
8299         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
8300         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8301         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8302         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8303         of TARGET_3DNOW.
8304         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8306 2013-05-15  Andreas Schwab  <schwab@suse.de>
8308         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8309         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
8310         third operand.
8312 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8314         * loop-unroll.c (report_unroll_peel): Check decision before
8315         emitting unroll/peel message.
8317 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8319         * function.h (has_bb_partition): New rtl_data flag.
8320         (bb_reorder_complete): Ditto.
8321         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8322         instead of flag_reorder_blocks_and_partition.
8323         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8324         with some enhancements.
8325         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8326         * bb-reorder.c (connect_traces): Check for has_bb_partition
8327         instead of flag_reorder_blocks_and_partition.
8328         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8329         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8330         verify_hot_cold_block_grouping.
8331         (partition_hot_cold_basic_blocks): Set has_bb_partition.
8333 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8335         PR target/19599
8336         * config/arm/predicates.md (call_insn_operand): New predicate.
8337         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
8338         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8339         if insn is not a tail call.
8340         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8341         registers.
8342         * config/arm/arm.h (enum reg_class): New caller save register class.
8343         (REG_CLASS_NAMES): Likewise.
8344         (REG_CLASS_CONTENTS): Likewise.
8345         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8346         without decls.
8348 2013-05-15  Richard Biener  <rguenther@suse.de>
8350         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8351         of MSG_OPTIMIZED_LOCATIONS.
8352         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8353         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8354         message.
8355         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8356         of MSG_OPTIMIZED_LOCATIONS.
8357         (execute_vect_slp): Likewise.
8358         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8359         (vect_create_cond_for_alias_checks): Likewise.
8360         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8361         (vect_recog_widen_mult_pattern): Likewise.
8362         (vect_recog_widen_sum_pattern): Likewise.
8363         (vect_recog_over_widening_pattern): Likewise.
8364         (vect_recog_widen_shift_pattern): Likewise.
8365         (vect_recog_vector_vector_shift_pattern): Likewise.
8366         (vect_recog_divmod_pattern): Likewise.
8367         (vect_recog_mixed_size_cond_pattern): Likewise.
8368         (vect_recog_bool_pattern): Likewise.
8369         (vect_pattern_recog_1): Likewise.
8371 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8373         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8374         non-functions to builtin_unreachable.
8375         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8376         correct when new direct edges were discovered.
8378 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8380         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8381         header, print symbol order instead of node uid, print more information
8382         about indirect edge targets.
8383         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8384         uids.
8385         (ipa_make_edge_direct_to_target): Likewise.
8386         (remove_described_reference): Likewise.
8387         (propagate_controlled_uses): Likewise.
8388         (ipa_print_node_params): Also print symbol order.
8389         (ipcp_transform_function): Print symbol order instead of node uids.
8390         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8391         (cgraph_get_create_real_symbol_node): Likewise.
8392         * ipa-cp.c (print_lattice): Likewise.
8393         (print_all_lattices): Likewise.
8394         (determine_versionability): Likewise.
8395         (initialize_node_lattices): Likewise.
8396         (estimate_local_effects): Likewise.
8397         (update_profiling_info): Likewise.
8398         (create_specialized_node): Likewise.
8399         (perhaps_add_new_callers): Likewise.
8400         (decide_about_value): Likewise.
8401         (decide_whether_version_node): Likewise.
8402         (identify_dead_nodes): Likewise.
8403         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8404         (dump_inline_summary): Likewise.
8405         (estimate_node_size_and_time): Likewise.
8406         (inline_analyze_function): Likewise.
8407         * ipa-inline.c (report_inline_failed_reason): Likewise.
8408         (want_early_inline_function_p): Likewise.
8409         (edge_badness): Likewise.
8410         (update_edge_key): Likewise.
8411         (inline_small_functions): Likewise.  Add dumping of order to two other
8412         dumps.
8413         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8414         instead of node uids.
8415         (propagate_pure_const): Likewise.
8416         (propagate_pure_const): Likewise.
8417         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8418         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8419         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8420         of node uids.
8421         * tree-pretty-print.c (dump_function_header): Likewise.
8422         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8423         Print symbol order instead of node uids.
8425 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8427         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8428         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8430 2013-05-15  Richard Biener  <rguenther@suse.de>
8432         PR tree-optimization/57275
8433         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8434         return value for fail to do runtime alias checks for gather loads.
8436 2013-05-15  Jan Hubicka  <jh@suse.cz>
8438         PR lto/57038
8439         PR lto/47375
8440         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8441         weakrefs are not external.
8442         (lto_symtab_merge_decls): Fix thinko when dealing with
8443         non-lto_symtab decls.
8444         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8445         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8446         * varpool.c (dump_varpool_node): Dump more flags.
8448 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8450         * config/i386/i386.c (processor_alias_table): Add instruction
8451         FSGSBASE for AMD bdver3 architecture.
8453 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8455         * tree.c (warn_deprecated_use): Print file:line using locus color.
8456         * diagnostic.c (diagnostic_report_current_module): Print file:line
8457         and file:line:column using locus color.
8459 2013-05-14  Mike Stump  <mikestump@comcast.net>
8461         * gdbinit.in: Add __null.
8463 2013-05-14  Mike Stump  <mikestump@comcast.net>
8465         * recog.h: Rename struct recog_data to Recog_data.
8466         * recog.c: Likewise.
8467         * reload.c (can_reload_into): Likewise.
8468         * config/picochip/picochip.c: Likewise.
8470 2013-05-14  Mike Stump  <mikestump@comcast.net>
8472         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8474 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8476         * resource.h (struct resources): Remove unch_memory member.
8477         (CLEAR_RESOURCE): Don't clear unch_memory.
8478         * resource.c (mark_referenced_resources): Don't set it.
8479         (mark_set_resources): Likewise.
8480         (mark_target_live_regs): Don't clear it.
8481         (init_resource_info): Likewise.
8482         * reorg.c (resource_conflicts_p): Don't compare it.
8483         (redundant_insn): Don't set it.
8485         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8486         Remove prototypes.
8487         * emit-rtl.c (next_label): Remove unused function.
8488         (skip_consecutive_labels, link_cc0_insns): Move to ...
8489         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8490         only place where these functions are used, and make them static.
8492 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8494         * fold-const.c (fold_negate_expr): Handle vectors.
8495         (fold_truth_not_expr): Make it static.
8496         (fold_invert_truthvalue): New static function.
8497         (invert_truthvalue_loc): Handle vectors. Do not call
8498         fold_truth_not_expr directly.
8499         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8500         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8501         (fold_comparison): Handle vector constants.
8502         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8503         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8504         * tree.h (fold_truth_not_expr): Remove declaration.
8506 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8508         * config/aarch64/aarch64-simd.md
8509         (aarch64_vcond_internal<mode>): Rename to...
8510         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8511         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8512         float modes. Clarify all iterator modes.
8513         (vcond<mode><mode>): Use new name for vcond expanders.
8514         (vcond<v_cmp_result><mode>): Likewise.
8515         (vcondu<mode><mode>: Likewise.
8516         * config/aarch64/iterators.md (VDQF_COND): New.
8518 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8520         PR bootstrap/57266
8521         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8522         variable for the shift amount. Check that we shift by non-negative
8523         amounts.
8525 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8527         PR target/42017
8528         * config/arm/arm.h (EPILOGUE_USES): Only return true
8529         for LR_REGNUM after epilogue_completed.
8531 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8533         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8534         is error_mark_node.
8536 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8538         PR target/57261
8539         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8540         and Solaris 11+/x86 with gld.
8541         * configure: Regenerate.
8543 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8545         * expmed.c (expand_shift_1): Canonicalize rotates by
8546         constant bitsize / 2 to bitsize - 1.
8547         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8548         case ROTATERT>: Likewise.
8550         Revert:
8551         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8553         * config/i386/i386.md (rotateinv): New code attr.
8554         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8555         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8556         roll $31, %eax, etc.
8558 2013-05-14  Richard Biener  <rguenther@suse.de>
8560         PR middle-end/57235
8561         * tree-eh.c (sink_clobbers): Give up for successors with
8562         multiple predecessors and no virtual uses.
8564 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8566         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8567         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8569 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8571         PR middle-end/57251
8572         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8573         the case when both op0 and op1 have VOIDmode.
8575 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8577         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8578         in multiply-accumulate mode.
8580 2013-05-13  Guozhi Wei  <carrot@google.com>
8582         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8584 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8586         PR target/56975
8587         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8588         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8589         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8590         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8591         (ix86_expand_prologue): Likewise.
8592         (ix86_expand_split_stack_prologue): Likewise.
8593         (legitimate_pic_address_disp_p): Likewise.
8594         (legitimize_pic_address): Likewise.
8595         (legitimize_tls_address): Likewise.
8596         (legitimize_pe_coff_symbol): Likewise.
8597         (output_pic_addr_const): Likewise.
8598         (construct_plt_address): Likewise.
8599         (ix86_expand_call): Likewise.
8600         (x86_output_mi_thunk): Likewise.
8601         (x86_function_profiler): Likewise.
8603 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
8605         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8606         similar switch cases.
8607         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8608         (aarch64_simd_mov_to_<mode>low): Delete.
8609         (aarch64_simd_mov_to_<mode>high): Delete.
8610         (move_lo_quad_<mode>): Add w<-r alternative.
8611         (aarch64_simd_move_hi_quad_<mode>): Likewise.
8612         (aarch64_simd_mov_from_*): Update type attribute.
8613         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8614         statement.
8616 2013-05-13  Jan Hubicka  <jh@suse.cz>
8618         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8619         * config/i386/i386.c (ix86_compute_frame_layout,
8620         ix86_expand_epilogue, emit_i387_cw_initialization,
8621         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8622         ix86_local_alignment): Fix use of size/speed predicates.
8624 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8626         PR tree-optimization/45216
8627         PR tree-optimization/57157
8628         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8629         the (-Y) & (B - 1) variant if OP is |.
8630         * expmed.c (expand_shift_1): For rotations by const0_rtx just
8631         return shifted.  Use (-op1) & (prec - 1) as other_amount
8632         instead of prec - op1.
8634 2013-05-13  Martin Jambor  <mjambor@suse.cz>
8636         PR middle-end/42371
8637         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8638         (ipa_constant_data): New type.
8639         (ipa_jump_func): Use ipa_constant_data to hold information about
8640         constant jump functions.
8641         (ipa_get_jf_constant): Adjust to jump function type changes.
8642         (ipa_get_jf_constant_rdesc): New function.
8643         (ipa_param_descriptor): New field controlled_uses.
8644         (ipa_get_controlled_uses): New function.
8645         (ipa_set_controlled_uses): Likewise.
8646         * ipa-ref.h (ipa_find_reference): Declare.
8647         * ipa-prop.c (ipa_cst_ref_desc): New type.
8648         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8649         changes.
8650         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
8651         New parameter cs.  Adjust all callers.
8652         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8653         (remove_described_reference): New function.
8654         (jfunc_rdesc_usable): Likewise.
8655         (try_make_edge_direct_simple_call): Decrement controlled use count,
8656         attempt to remove reference if it hits zero.
8657         (combine_controlled_uses_counters): New function.
8658         (propagate_controlled_uses): Likewise.
8659         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8660         (ipa_edge_duplication_hook): Duplicate reference descriptions.
8661         (ipa_print_node_params): Print described use counter.
8662         (ipa_write_jump_function): Adjust to jump function type changes.
8663         (ipa_read_jump_function): New parameter CS, pass it to
8664         ipa_set_jf_constant.  Adjust caller.
8665         (ipa_write_node_info): Stream controlled use count
8666         (ipa_read_node_info): Likewise.
8667         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8668         asserting.
8669         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8670         count.  Remove cloning-added reference if it reaches zero.
8671         * ipa-ref.c (ipa_find_reference): New function.
8673 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8675         * config/i386/i386.c (processor_target_table): Modified default
8676         alignment values for AMD BD and BT architectures.
8678 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
8680         * tree-vect-generic.c (uniform_vector_p): Move ...
8681         * tree.c (uniform_vector_p): ... here.
8682         * tree.h (uniform_vector_p): Declare it.
8683         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8684         into a scalar.
8686 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8688         PR tree-optimization/57230
8689         * tree-ssa-strlen.c (handle_char_store): Record length for
8690         array store from STRING_CST.
8692         PR tree-optimization/57230
8693         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8694         check.
8696 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
8698         * config/epiphany/epiphany.c (epiphany_init): Check size of
8699         NUM_MODES_FOR_MODE_SWITCHING.
8700         (epiphany_expand_prologue):
8701         Remove CONFIG_REGNUM initial value handling code.
8702         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8703         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8704         (emit_set_fp_mode, epiphany_mode_after): Likewise.
8705         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8706         Don't return 1 for FP_MODE_NONE.
8707         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8708         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8709         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8710         * config/epiphany/epiphany.md (save_config): New pattern.
8712 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
8714         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8716 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
8718         * config/i386/i386.md (memory): Handle sseishft1.
8719         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8720         (*vec_extractv2di_1): Ditto.
8722 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
8724         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8725         saved registers.
8727 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8729         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8730         Add mthumb/march=armv7-a multilib.
8731         Add mthumb/march=armv7-r multilib.
8732         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8734 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
8736         * config/v850/t-rtems: Add more multilibs.
8738 2013-05-10  Richard Biener  <rguenther@suse.de>
8740         PR tree-optimization/57214
8741         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8742         not propagate from SSA names that occur in abnormal PHI nodes.
8744 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
8746         * stor-layout.c (element_precision): New function.
8747         * machmode.h (element_precision): Declare it.
8748         * tree.c (build_minus_one_cst): New function.
8749         (element_precision): Likewise.
8750         * tree.h (build_minus_one_cst): Declare new function.
8751         (element_precision): Likewise.
8752         * fold-const.c (operand_equal_p): Use element_precision.
8753         (fold_binary_loc): Handle vector types.
8754         * convert.c (convert_to_integer): Use element_precision.
8755         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8756         separately.
8758 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
8760         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8761         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8762         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8763         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8764         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8765         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8766         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8767         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8768         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8769         (Uuw8): New constraints.
8770         (Usb4): Move into alphabetical order.
8771         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8772         (sd8_operand, ub8_operand, uw8_operand): New predicates.
8773         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8774         previously unnamed patterns.
8775         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8776         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8777         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8778         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8779         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
8780         of specific lengths.
8782 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8784         * config/i386/i386.md (rotateinv): New code attr.
8785         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8786         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8787         roll $31, %eax, etc.
8789         PR tree-optimization/45216
8790         PR tree-optimization/57157
8791         * tree-ssa-forwprop.c (simplify_rotate): New function.
8792         (ssa_forward_propagate_and_combine): Call it.
8794 2013-05-10  Richard Biener  <rguenther@suse.de>
8796         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8797         disable peeling when we version for aliasing.
8798         (vector_alignment_reachable_p): Honor explicit user alignment.
8799         (vect_supportable_dr_alignment): Likewise.
8800         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8801         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8802         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8803         then peeling to arrange for the cost-model check to come first.
8805 2013-05-10  Alan Modra  <amodra@gmail.com>
8807         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8808         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8809         * configure: Regenerate.
8811 2013-05-10  Alan Modra  <amodra@gmail.com>
8813         PR target/55033
8814         * varasm.c (default_elf_select_section): Move !DECL_P check..
8815         (get_named_section): ..to here before calling get_section_name.
8816         Adjust assertion.
8817         (default_section_type_flags): Add DECL_P check.
8818         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8819         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8821 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
8823         * config/epiphany/epiphany.c (epiphany_expand_prologue):
8824         When using gen_stack_adjust_str with a register offset, add a
8825         REG_FRAME_RELATED_EXPR note.
8827 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
8829         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8830         (*vec_extractv4si_zext_mem): Ditto.
8831         (*vec_extractv2di): Add 0->x and x->x alternatives.
8832         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8833         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8835 2013-05-09  Jason Merrill  <jason@redhat.com>
8837         N3639 C++1y VLA support
8838         * gimplify.c (gimplify_vla_decl): Don't touch an existing
8839         DECL_VALUE_EXPR.
8841         * tree.c (build_constructor_va): New.
8842         * tree.h: Declare it.
8844 2013-05-09  Martin Jambor  <mjambor@suse.cz>
8846         PR lto/57084
8847         * gimple-fold.c (canonicalize_constructor_val): Call
8848         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8850 2013-05-09  Jan Hubicka  <jh@suse.cz>
8851             Richard Biener  <rguenther@suse.de>
8853         PR lto/54095
8854         * symtab.c (symtab_make_decl_local): Do not add private names.
8856 2013-05-09  Jan Hubicka  <jh@suse.cz>
8858         PR lto/54095
8859         * symtab.c (insert_to_assembler_name_hash): Handle clones.
8860         (unlink_from_assembler_name_hash): Likewise.
8861         (symtab_prevail_in_asm_name_hash, symtab_register_node,
8862         symtab_unregister_node, symtab_initialize_asm_name_hash,
8863         change_decl_assembler_name): Update.
8865 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
8867         * config/aarch64/aarch64.md: New movtf split.
8868         (*movtf_aarch64): Update.
8869         (aarch64_movdi_tilow): Handle TF modes and rename to
8870         aarch64_movdi_<mode>low.
8871         (aarch64_movdi_tihigh): Handle TF modes and rename to
8872         aarch64_movdi_<mode>high
8873         (aarch64_movtihigh_di): Handle TF modes and rename to
8874         aarch64_mov<mode>high_di
8875         (aarch64_movtilow_di): Handle TF modes and rename to
8876         aarch64_mov<mode>low_di
8877         (aarch64_movtilow_tilow): Remove spurious whitespace.
8878         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8879         splits.
8880         (aarch64_print_operand): Update.
8882 2013-05-09  Alan Modra  <amodra@gmail.com>
8884         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8885         powerpc64le.
8886         * configure: Regenerate.
8888 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8890         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8891         splitter preparation statements.
8892         * config/i386/sse.md (*vec_extract* splitters): Ditto.
8893         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8894         adjust_address_nv.
8896 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8898         * gimple-ssa-strength-reduction.c (count_candidates): Change
8899         return value to int.
8900         (analyze_candidates_and_replace): Change type of length to int.
8902 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8904         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8905         (*vec_extract<mode>): Use VI12_128 mode iterator.
8906         (*vec_extract<mode>_mem): Ditto.
8907         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8908         attribute.
8910 2013-05-08  Diego Novillo  <dnovillo@google.com>
8912         PR bootstrap/54659
8914         Revert:
8915         2012-08-17  Diego Novillo  <dnovillo@google.com>
8917         PR bootstrap/54281
8918         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8919         * config.in: Regenerate.
8920         * configure: Regenerate.
8921         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8923 2013-05-08  Jan Hubicka  <jh@suse.cz>
8925         PR lto/54095
8926         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8927         * cgraph.h (symtab_node_base): Add unique_name.
8928         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8929         input_overwrite_node, input_varpool_node): Stream unique_name.
8930         * cgraphclones.c (cgraph_create_virtual_clone,
8931         cgraph_function_versioning): Set unique_name.
8932         * ipa.c (function_and_variable_visibility): Set unique_name.
8934 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8936         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8937         (alloc_cand_and_find_basis): Restrict conditional candidate
8938         processing to CAND_MULTs.
8940 2013-05-08  Jan Hubicka  <jh@suse.cz>
8942         PR lto/54095
8943         lto-symtab.c (lto_symtab_symbol_p): New function.
8944         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8945         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8946         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8947         Skip static symbols.
8949 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
8951         PR tree-optimization/57200
8952         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8953         Only call inform if the preceding warning_at returns true.
8955 2013-05-07  Han Shen  <shenhan@google.com>
8957         * cfgexpand.c (record_or_union_type_has_array_p): New function.
8958         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8959         * common.opt (fstack-protector-strong): New option.
8960         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8961         * doc/invoke.texi (Optimization Options): Document
8962         "-fstack-protector-strong".
8963         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8965 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8967         * config/mips/mips.c (mips_machine_reorg2): Return 0.
8969 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
8971         * ira.c (update_equiv_regs): Add insn having equiv memory even if
8972         it is not lhs of the insn.
8973         (setup_reg_equiv): Remove insn having equiv memory which it is not
8974         lhs of the insn.
8975         * lra-constraints.c (process_address): Try to improve generation
8976         code for address base + disp.
8977         (lra_constraints): Make correct the code for checking insn setting
8978         up backward equivalence.  Remove insn only if it is in the init
8979         insn list.
8980         * lra-eliminations.c (update_reg_eliminate): Change return value.
8981         (lra_eliminate): Use the result.
8983 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
8985         * config/i386/sse.md (ssescalarnummask): New mode attribute.
8986         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8987         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8988         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8989         register target operands.
8990         (*vec_extractv8hi_sse2): New pattern.
8991         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8992         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8993         (*vec_extract<mode>_mem): New insn and split pattern.
8995 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
8997         * config/arm/arm.c (arm_asan_shadow_offset): New function.
8998         (TARGET_ASAN_SHADOW_OFFSET): Define.
8999         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9000         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9002 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9004         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9005         (incr_vec_index): Return -1 if increment not found.
9006         (create_add_on_incoming_edge): Assert if increment not found.
9007         (record_increment): Limit number of increments recorded.
9008         (all_phi_incrs_profitable): Return false if an increment not found.
9009         (replace_profitable_candidates): Don't process increments that were
9010         not recorded.
9011         (analyze_candidates_and_replace): Limit size of incr_vec.
9013 2013-05-07  Richard Biener  <rguenther@suse.de>
9015         * calls.c (special_function_p): setjmp-like functions are leaf.
9016         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9017         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9019 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
9021         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9022         (aarch64_simd_mov<mode>): New expander.
9023         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9024         (aarch64_simd_mov_to_<mode>high): Likewise.
9025         (aarch64_simd_mov_from_<mode>low): Likewise.
9026         (aarch64_simd_mov_from_<mode>high): Likewise.
9027         (aarch64_dup_lane<mode>): Update.
9028         (aarch64_dup_lanedi): New instruction pattern.
9029         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9030         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9032 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9034         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9035         (replace_mult_candidate): Remove unnecessary argument; remove
9036         unnecessary parameter from call to introduce_cast_before_cand.
9037         (replace_unconditional_candidate): Remove unnecessary parameter
9038         from call to replace_mult_candidate.
9039         (replace_conditional_candidate): Likewise.
9040         (insert_initializers): Use make_temp_ssa_name.
9041         (introduce_cast_before_cand): Remove unnecessary argument; use
9042         make_temp_ssa_name.
9043         (replace_one_candidate): Remove unnecessary argument; remove
9044         unnecessary parameter from calls to introduce_cast_before_cand.
9045         (replace_profitable_candidates): Remove unnecessary parameters
9046         from calls to replace_one_candidate.
9048 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9050         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9051         phi def as possibly hiding a basis for a CAND_ADD whose operands
9052         have been commuted in the analysis.
9053         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9055 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
9057         * config/aarch64/aarch64.md
9058         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9059         shift value between 0-4.
9061 2013-05-07  Richard Biener  <rguenther@suse.de>
9063         * double-int.h (rshift): New overload.
9064         * double-int.c (rshift): New function.
9065         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9066         (create_reference_ops_from_ref): Remove.
9067         (vn_reference_insert): Use shared ops for constructing the
9068         reference and copy it.
9070 2013-05-07  Richard Biener  <rguenther@suse.de>
9072         PR middle-end/57190
9073         * tree-eh.c (sink_clobbers): Properly propagate
9074         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9076 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
9078         PR tree-optimization/57149
9079         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9080         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9081         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9082         uninit_undefined_value_p instead of ssa_undefined_value_p.
9084         PR debug/57184
9085         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9086         for modifier == EXPAND_INITIALIZER.
9088 2013-05-07  Anton Blanchard  <anton@samba.org>
9090         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9091         for powerpc64 little endian.
9092         * configure: Regenerate.
9094 2013-05-06  Graham Stott  <grahams@btinternet.com>
9096         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9097         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9098         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9099         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9101 2013-05-06  Graham Stott  <grahams@btinternet.com>
9103         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9104         codes which allow non-lvalues.
9106 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
9108         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9109         components are all 1s.
9110         (integer_minus_onep): New function.
9111         * tree.h (integer_minus_onep): Declare it.
9112         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9113         integer_minus_onep instead of integer_all_onesp.
9115 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9117         PR target/52933
9118         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9119         variations of these patterns.
9121 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9123         * config/i386/i386.md (isa): Add x64_sse4 member.
9124         (enabled): Handle x64_sse4.
9125         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9126         instruction for 64bit SSE4_1 targets.  Update insn attributes.
9127         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9128         instruction for SSE4_1 targets.  Update insn attributes.
9129         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9130         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9131         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9132         const_1 selector.
9133         (*vec_extractv4si): Rename from *sse4_1_pextrd.
9134         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9135         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9137 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9139         PR target/57108
9140         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9142 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
9144         * final.c (do_assembler_dialects): Don't handle curly braces and
9145         vertical bar escaped by % as dialect delimiters.
9146         (output_asm_insn): Print curly braces and vertical bar if escaped
9147         by % and ASSEMBLER_DIALECT defined.
9148         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9149         * doc/tm.texi: Regenerated.
9151 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9153         * config/mips/mips.c: Include tree-pass.h.
9154         (mips_reorg): Split in pre- and post-dbr_schedule parts.
9155         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9156         (pass_mips_machine_reorg2): New machine specific pass.
9157         (insert_pass_mips_machine_reorg2): New pass plugin definition.
9158         (mips_option_override): Register the new pass.
9159         * rtl.h (cleanup_barriers): Remove prototype.
9160         (dbr_schedule): Likewise.
9161         * jump.c (cleanup_barriers): Make static.
9162         * reorg.c (dbr_schedule): Likewise.
9164 2013-05-06  Richard Biener  <rguenther@suse.de>
9166         PR tree-optimization/57185
9167         * tree-parloops.c (add_field_for_reduction): Handle anonymous
9168         SSA names properly.
9170 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9172         PR target/57106
9173         * config/i386/i386.c (add_parameter_dependencies): Add dependence
9174         between "first_arg" and "insn", not "last" and "insn".
9176 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9178         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9179         (find_candidates_in_block): Re-enable slsr_process_phi.
9180         (create_phi_basis): Fix double counting of candidate adjustment.
9182 2013-05-06  Richard Biener  <rguenther@suse.de>
9184         PR middle-end/57147
9185         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9186         the edge is also fallthru, preserve it and just clear the
9187         abnormal flag.
9188         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9189         also complex, preserve that and just clear the fallthru flag.
9190         * tree-inline.c (update_ssa_across_abnormal_edges): Also
9191         update virtual operands.
9193 2013-05-06  Alan Modra  <amodra@gmail.com>
9195         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9196         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9197         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9198         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9199         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9200         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9202 2013-05-06  Alan Modra  <amodra@gmail.com>
9204         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9205         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
9206         (DEFAULT_ASM_ENDIAN): Define.
9207         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9208         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9209         Update -K PIC clause from sysv4.h.
9210         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9211         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9213 2013-05-06  Alan Modra  <amodra@gmail.com>
9215         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9216         twice for little-endian.
9217         (ashrdi3_no_power, ashrdi3): Support little-endian.
9219 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9221         PR target/55303
9222         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9223         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9224         related expanders.
9225         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9226         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9227         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9228         New predicates.
9230 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
9231             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
9233         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9234         * config/pa/pa.opt: Make mbig-switch a no-op.
9235         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9236         (CASE_VECTOR_MODE): Always return SImode.
9237         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9238         for the !TARGET_BIG_SWITCH case.
9239         * config/pa/pa-linux.h: Likewise.
9240         * config/pa/pa-openbsd.h: Likewise.
9241         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9242         * config/pa/pa.md (short_jump): Remove define_insn.
9243         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9244         (casesi0): Remove define_insn.
9245         (type): Remove btable_branch.
9246         (pa_combine_type): Likewise.
9247         (in_nullified_branch_delay): Likewise.
9248         (in_call_delay): Likewise.
9249         (define_delay): Likewise.
9250         (define_insn_reservation "Z3"): Likewise.
9251         (define_insn_reservation "Z4"): Likewise.
9252         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9253         (pa_adjust_insn_length): Remove adjustment for btable branches.
9254         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9255         and mno-big-switch
9257 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
9259         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9260         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9261         Add m->r,x alternatives.
9262         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9263         splitters using SWI48x mode iterator.
9264         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
9265         TARGET_64BIT.  Add m->x alternative.
9266         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9267         Add o->x alternative.  Enable for TARGET_SSE.
9268         (sse_storeq): Remove expander.
9269         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9270         with memory input operand.
9271         (*vec_extractv2di_1 splitter): New.
9272         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9273         * config/i386/i386.md (ssevecmodelower): New mode attribute.
9275 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9277         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
9278         (INT_LOWPART): Delete.
9279         (extract_MB): Adjust.
9280         (extract_ME): Adjust.
9281         (print_operand): Adjust.
9283 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9285         * config/rs6000/predicates.md (reg_or_add_cint_operand,
9286         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9287         (reg_or_logical_cint_operand, easy_fp_constant,
9288         logical_const_operand): Delete "CONST_DOUBLE" case.
9289         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9290         "HOST_BITS_PER_WIDE_INT == 64" test.
9291         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
9292         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9293         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9294         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9295         test.
9296         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9297         CONST_DOUBLE DImode/VOIDmode case.
9298         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9299         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9300         CONST_DOUBLE VOIDmode case.
9301         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9302         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9303         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9304         Delete CONST_DOUBLE case.
9305         (splitters for mov FMOVE64 const_double): Delete
9306         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9307         "HOST_BITS_PER_WIDE_INT >= 64" test.
9308         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9309         case.
9310         (mov DI const_double): Delete.
9312 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
9314         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9315         on op shows all bits zero in mode of a lowpart subreg, return zero.
9317 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
9319         PR target/57150
9320         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9321         to save TFmode registers and DImode to save TImode registers for
9322         caller save operations.
9323         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9324         mark being partially clobbered since they only use the first
9325         double word.
9327         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9328         and TDmode only use the upper 64-bits of each VSX register.
9330 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9332         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9333         (find_candidates_in_block): Disable slsr_process_phi.
9335 2013-05-03  Guozhi Wei  <carrot@google.com>
9337         * coverage.c (coverage_obj_init): Move the construction of gcov
9338         constructor to ...
9339         (build_init_ctor): ... here.
9341 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9343         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9344         (slsr_cand_d): Redefine def_phi.
9345         (stride_status, phi_adjust_status, count_phis_status): New enums.
9346         (find_phi_def): New.
9347         (find_basis_for_base_expr): New.
9348         (find_basis_for_candidate): Handle hidden bases.
9349         (alloc_cand_and_find_basis): Handle phi candidates.
9350         (slsr_process_phi): New.
9351         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9352         (create_mul_imm_cand): Likewise.
9353         (create_add_ssa_cand): Exclude phi base candidates.
9354         (create_add_imm_cand): Likewise.
9355         (slsr_process_cast): Likewise.
9356         (slsr_process_copy): Likewise.
9357         (find_candidates_in_block): Handle phi candidates.
9358         (dump_candidate): Likewise.
9359         (unconditional_cands): Delete.
9360         (unconditional_cands_with_known_stride_p): Delete.
9361         (phi_dependent_cand_p): New.
9362         (cand_increment): Handle phi-dependent candidates.
9363         (replace_dependent): Delete.
9364         (replace_mult_candidate): New.
9365         (replace_unconditional_candidate): New.
9366         (incr_vec_index): Move to avoid forward reference.
9367         (create_add_on_incoming_edge): New.
9368         (create_phi_basis): New.
9369         (replace_dependents): Delete.
9370         (replace_conditional_candidate): New.
9371         (phi_add_costs): New.
9372         (replace_uncond_cands_and_profitable_phis): New.
9373         (record_increment): Handle phi adjustments.
9374         (record_phi_increments): New.
9375         (record_increments): Handle phi adjustments.
9376         (phi_incr_cost): New.
9377         (lowest_cost_path): Handle phis.
9378         (total_savings): Likewise.
9379         (analyze_increments): Likewise.
9380         (ncd_with_phi): New.
9381         (ncd_of_cand_and_phis): New.
9382         (nearest_common_dominator_for_cands): Handle phi increments.
9383         (all_phi_incrs_profitable): New.
9384         (replace_profitable_candidates): Handle phi-dependent candidates.
9385         (analyze_candidates_and_replace): Likewise.
9387 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9389         PR bootstrap/57154
9390         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9391         do not exceed REG_BR_PROB_BASE.
9393 2013-05-03  Jeff Law  <law@redhat.com>
9395         PR tree-optimization/57144
9396         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9397         operand of the condition will bit into the new type when eliminating
9398         a cast feeding a condition.
9400 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9402         PR rtl-optimization/57130
9403         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9404         of COMPARE as in_code to the recursive call if needed.
9406 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9408         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9409         (enabled): Handle new members.
9410         * config/i386/sse.md (*vec_concatv2si): Merge from
9411         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9412         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9414 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9416         PR tree-optimization/57027
9417         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9418         for fnms opportunity, check we got the prerequisite kind
9419         of tree / gimple before using accessor functions.
9421 2013-05-03  Richard Biener  <rguenther@suse.de>
9423         * double-int.h (lshift): New overload without precision
9424         and arith argument.
9425         (operator *=, operator +=, operator -=): Move ...
9426         * double-int.c (operator *=, operator +=, operator -=): ... here
9427         and implement more efficiently.
9428         (mul_double_with_sign): Remove.
9429         (lshift_double): Adjust to take unsinged shift argument, push
9430         dispatching code to callers.
9431         (mul_double_wide_with_sign): Add early out for callers that
9432         are not interested in high parts or overflow.
9433         (lshift): New function.
9434         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9435         dispatch code here.
9436         (lrotate, rrotate): Use logical shifts.
9437         * expr.c (get_inner_reference): Use lshift.
9438         * fixed-value.c (do_fixed_divide): Likewise.
9439         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9440         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9441         (indirect_refs_may_alias_p): Likewise.
9442         (stmt_kills_ref_p_1): Likewise.
9444 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9446         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9448 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9450         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9451         scalar form of FABD instruction.
9453 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9455         * lra-constraints.c (process_alt_operands): Add checking alt
9456         number to choose the best alternative.
9458 2013-05-02  Richard Biener  <rguenther@suse.de>
9460         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9461         bitmap and its handling.
9462         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9464 2013-05-02  Richard Biener  <rguenther@suse.de>
9466         PR middle-end/57140
9467         * tree-inline.c (copy_loops): Properly handle removed loops.
9468         (copy_cfg_body): Mark destination loops for fixup if source
9469         loops needed fixup.
9471 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9473         PR target/56732
9474         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9475         generating simple_return for naked functions.
9477 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9479         PR middle-end/56988
9480         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9481         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9482         flags match.
9483         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9484         ipa_agg_replacement_value structures.
9485         (known_aggs_to_agg_replacement_list): Likewise.
9486         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9487         (read_agg_replacement_chain): Likewise.
9488         (ipcp_transform_function): Also check that by_ref flags match.
9490 2013-05-02  Richard Biener  <rguenther@suse.de>
9492         * graphds.h (struct graph): Add obstack member.
9493         * graphds.c (new_graph): Initialize obstack and allocate
9494         vertices from it.
9495         (add_edge): Allocate edge from the obstack.
9496         (free_graph): Free the obstack instead of all edges and vertices.
9498 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9500         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9501         divides.
9502         * cfg.c (update_bb_profile_for_threading): Ditto.
9503         * tree-inline.c (copy_bb): Ditto.
9504         (copy_edges_for_bb): Ditto.
9505         (initialize_cfun): Ditto.
9506         (copy_cfg_body): Ditto.
9507         (expand_call_inline): Ditto.
9508         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9509         (estimate_node_size_and_time): Ditto.
9510         (inline_merge_summary): Ditto.
9511         * cgraphclones.c (cgraph_clone_edge): Ditto.
9512         (cgraph_clone_node): Ditto.
9513         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9514         (compute_trg_info): Ditto.
9516 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9518         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9519         S reg when fp attribute set.
9520         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9522 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9524         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9525         New pattern.
9526         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9527         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9528         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9530 2013-05-02  Richard Biener  <rguenther@suse.de>
9532         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9533         (struct instantiate_cache_entry): New type.
9534         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9535         (struct instantiate_cache_type): New type.
9536         (set_instantiated_value, get_instantiated_value): Remove.
9537         (get_instantiated_value_entry): New function.
9538         (instantiate_scev_name): Use the new cache and adjust.
9539         (instantiate_scev_poly): Adjust.
9540         (instantiate_scev_binary): Likewise.
9541         (instantiate_array_ref): Likewise.
9542         (instantiate_scev_convert): Likewise.
9543         (instantiate_scev_not): Likewise.
9544         (instantiate_scev_3): Likewise.
9545         (instantiate_scev_2): Likewise.
9546         (instantiate_scev_r): Likewise.
9547         (instantiate_scev): Likewise.
9548         (resolve_mixers): Likewise.
9550 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9552         PR target/57091
9553         * lra-constraints.c (best_small_class_operands_num): Remove.
9554         (process_alt_operands): Remove small_class_operands_num.  Take
9555         small classes operands into losers and only if the operand is not
9556         matched.  Modify debugging output.
9557         (curr_insn_transform): Remove best_small_class_operands_num.
9558         Print insn name.
9560 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9562         * config/aarch64/aarch64-builtins.c
9563         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9564         * config/aarch64/aarch64-simd-builtins.def
9565         (reduc_splus_): Add new modes.
9566         (reduc_uplus_): New.
9567         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9568         (reduc_uplus_v4sf): Likewise.
9569         (reduc_splus_v4sf): Likewise.
9570         (aarch64_addv<mode>): Likewise.
9571         (reduc_uplus_<mode>): Likewise.
9572         (reduc_splus_<mode>): Likewise.
9573         (aarch64_addvv2di): Likewise.
9574         (reduc_uplus_v2di): Likewise.
9575         (reduc_splus_v2di): Likewise.
9576         (aarch64_addvv2si): Likewise.
9577         (reduc_uplus_v2si): Likewise.
9578         (reduc_splus_v2si): Likewise.
9579         (reduc_<sur>plus_<mode>): New.
9580         (reduc_<sur>plus_v2di): Likewise.
9581         (reduc_<sur>plus_v2si): Likewise.
9582         (reduc_<sur>plus_v4sf): Likewise.
9583         (aarch64_addpv4sf): Likewise.
9584         * config/aarch64/arm_neon.h
9585         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9586         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9587         add UNSPEC_SADDV, UNSPEC_UADDV.
9588         (SUADDV): New.
9589         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9591 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9593         * config/aarch64/arm_neon.h
9594         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9596 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9598         * config/aarch64/aarch64-builtins
9599         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9601 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9603         * config/aarch64/aarch64-simd-builtins.def
9604         (reduc_smax_): New.
9605         (reduc_smin_): Likewise.
9606         (reduc_umax_): Likewise.
9607         (reduc_umin_): Likewise.
9608         (reduc_smax_nan_): Likewise.
9609         (reduc_smin_nan_): Likewise.
9610         (fmax): Remove.
9611         (fmin): Likewise.
9612         (smax): Update for V2SF, V4SF and V2DF modes.
9613         (smin): Likewise.
9614         (smax_nan): New.
9615         (smin_nan): Likewise.
9616         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9617         (<su><maxmin><mode>3): ...This, refactor.
9618         (s<maxmin><mode>3): New.
9619         (<maxmin_uns><mode>3): Likewise.
9620         (reduc_<maxmin_uns>_<mode>): Refactor.
9621         (reduc_<maxmin_uns>_v4sf): Likewise.
9622         (reduc_<maxmin_uns>_v2si): Likewise.
9623         (aarch64_<fmaxmin><mode>: Remove.
9624         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9625         new builtin names.
9626         (vmin<q>_f<32,64>): Likewise.
9627         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9628         (FMAXMIN): New.
9629         (su): Add mappings for smax, smin, umax, umin.
9630         (maxmin): New.
9631         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9632         (FMAXMIN): Rename as...
9633         (FMAXMIN_UNS): ...This.
9634         (maxminv): Remove.
9635         (fmaxminv): Likewise.
9636         (fmaxmin): Likewise.
9637         (maxmin_uns): New.
9638         (maxmin_uns_op): Likewise.
9640 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9642         * config/aarch64/arm_neon.h
9643         (vac<ge, gt><sd>_f<32, 64>): Rename to...
9644         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9645         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9647 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9649         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9650         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9652 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9654         * config/aarch64/aarch64-simd.md
9655         (vcond<mode>_internal): Handle special cases for constant masks.
9656         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9657         (vcondu<mode><mode>): Likewise.
9658         (vcond<v_cmp_result><mode>): New.
9660 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9662         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9663         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9664         * config/aarch64/aarch64-simd-builtins.def
9665         (cmeq): Update to BUILTIN_VALLDI.
9666         (cmgt): Likewise.
9667         (cmge): Likewise.
9668         (cmle): Likewise.
9669         (cmlt): Likewise.
9670         * config/aarch64/arm_neon.h
9671         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9672         to builtins or C as appropriate.
9674 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9676         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9677         (cmgeu): ...This.
9678         (cmhi): Rename to...
9679         (cmgtu): ...This.
9680         * config/aarch64/aarch64-simd.md
9681         (simd_mode): Add SF.
9682         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9683         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9684         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9685         (cstore<mode>_neg): ...This.
9686         * config/aarch64/iterators.md
9687         (VALLF): new.
9688         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9689         (COMPARISONS): New.
9690         (UCOMPARISONS): Likewise.
9691         (optab): Add missing comparisons.
9692         (n_optab): New.
9693         (cmp_1): Likewise.
9694         (cmp_2): Likewise.
9695         (CMP): Likewise.
9696         (cmp): Remove.
9697         (VCMP_S): Likewise.
9698         (VCMP_U): Likewise.
9699         (V_cmp_result): Add DF, SF modes.
9700         (v_cmp_result): Likewise.
9701         (v): Likewise.
9702         (vmtype): Likewise.
9703         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9705 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
9707         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9708         define_insn to define_insn_and_split.
9709         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9710         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9711         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9712         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9713         (thumb2_negscc): Likewise.
9715 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9717         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9719 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9721         * config/arm/thumb2.md: Remove trailing whitespaces.
9723 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9725         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9726         Use gen_int_mode rather than GEN_INT.
9728 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
9730         * value-prof.c (stream_in_histogram_value): Remove the strayed
9731         debug_gimple_stmt.
9733 2013-04-30  Richard Biener  <rguenther@suse.de>
9735         PR middle-end/57122
9736         * cfghooks.c (split_edge): Properly check for the loop latch edge.
9738 2013-04-30  Richard Biener  <rguenther@suse.de>
9740         PR middle-end/57107
9741         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9743 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
9745         PR rtl-optimization/56957
9746         PR rtl-optimization/57105
9747         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9748         variable.  Use just INSN_UID for determining whether an insn
9749         should be only disconnected from the insn stream.
9750         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9752 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
9754         PR tree-optimization/57104
9755         * tsan.c (instrument_expr): Don't instrument accesses to
9756         DECL_HARD_REGISTER VAR_DECLs.
9758 2013-04-30  Richard Biener  <rguenther@suse.de>
9760         * function.h (loops_for_fn): New inline function.
9761         (set_loops_for_fn): Likewise.
9762         * cfgloop.h (place_new_loop): Add struct function parameter.
9763         (get_loop): Likewise.
9764         (get_loops): Likewise.
9765         (number_of_loops): Likewise.
9766         (fel_next): Adjust.
9767         (fel_init): Likewise.
9768         * cfg.c (get_loop_copy): Adjust.
9769         * cfgloop.c (flow_loops_dump): Likewise.
9770         (record_loop_exits): Likewise.
9771         (verify_loop_structure): Likewise.
9772         * cfgloopanal.c (mark_irreducible_loops): Likewise.
9773         (estimate_reg_pressure_cost): Likewise.
9774         (mark_loop_exit_edges): Likewise.
9775         * cfgloopmanip.c (place_new_loop): Likewise.
9776         (add_loop): Likewise.
9777         (duplicate_loop): Likewise.
9778         * graph.c (draw_cfg_nodes): Likewise.
9779         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9780         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9781         (extract_affine_chrec): Likewise.
9782         (build_scop_iteration_domain): Likewise.
9783         * graphite.c (graphite_initialize): Likewise.
9784         * ira-build.c (create_loop_tree_nodes): Likewise.
9785         (more_one_region_p): Likewise.
9786         (rebuild_regno_allocno_maps): Likewise.
9787         (mark_loops_for_removal): Likewise.
9788         (mark_all_loops_for_removal): Likewise.
9789         (remove_unnecessary_regions): Likewise.
9790         (ira_build): Likewise.
9791         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9792         * loop-init.c (fix_loop_structure): Likewise.
9793         (gate_rtl_move_loop_invariants): Likewise.
9794         (gate_rtl_unswitch): Likewise.
9795         (gate_rtl_unroll_and_peel_loops): Likewise.
9796         (rtl_doloop): Likewise.
9797         * lto-streamer-in.c (input_cfg): Likewise.
9798         * lto-streamer-out.c (output_cfg): Likewise.
9799         * modulo-sched.c (sms_schedule): Likewise.
9800         * predict.c (tree_estimate_probability): Likewise.
9801         (tree_estimate_probability_driver): Likewise.
9802         (estimate_loops): Likewise.
9803         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9804         (move_sese_region_to_fn): Likewise.
9805         (debug_loop_num): Likewise.
9806         * tree-chrec.c (chrec_evaluate): Likewise.
9807         (hide_evolution_in_other_loops_than_loop): Likewise.
9808         (chrec_component_in_loop_num): Likewise.
9809         (reset_evolution_in_loop): Likewise.
9810         (evolution_function_is_invariant_rec_p): Likewise.
9811         * tree-if-conv.c (main_tree_if_conversion): Likewise.
9812         * tree-inline.c (copy_loops): Likewise.
9813         (copy_cfg_body): Likewise.
9814         (tree_function_versioning): Likewise.
9815         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9816         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9817         Likewise.
9818         (add_to_evolution_1): Likewise.
9819         (scev_const_prop): Likewise.
9820         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9821         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9822         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9823         (tree_ssa_lim_initialize): Likewise.
9824         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9825         (verify_loop_closed_ssa): Likewise.
9826         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
9827         (tree_ssa_loop_im): Likewise.
9828         (tree_ssa_loop_unswitch): Likewise.
9829         (tree_vectorize): Likewise.
9830         (check_data_deps): Likewise.
9831         (tree_ssa_loop_ivcanon): Likewise.
9832         (tree_ssa_loop_bounds): Likewise.
9833         (tree_complete_unroll): Likewise.
9834         (tree_complete_unroll_inner): Likewise.
9835         (tree_parallelize_loops): Likewise.
9836         (tree_ssa_loop_prefetch): Likewise.
9837         (tree_ssa_loop_ivopts): Likewise.
9838         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9839         * tree-vectorizer.c (vectorize_loops): Likewise.
9841 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
9843         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9844         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9845         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9846         with EABI_LINK_SPEC.
9848 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9850         PR target/44578
9851         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9852         alternative.
9854 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
9856         PR target/57097
9857         * lra-constraints.c (process_alt_operands): Discourage a bit more
9858         using memory for pseudos.  Print cost dump for alternatives.
9859         Modify cost values for conflicts with early clobbers.
9860         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9862 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9864         PR target/57098
9865         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9867 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9869         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9870         from/to S register.
9871         (movdi_aarch64): Support LDR/STR from/to D register.
9873 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9875         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9876         or higher by default.
9878 2013-04-29  Richard Biener  <rguenther@suse.de>
9880         PR middle-end/57075
9881         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9882         even if not adding abnormal edges for calls that can make
9883         abnormal gotos.
9885 2013-04-29  Richard Biener  <rguenther@suse.de>
9887         PR middle-end/57103
9888         * tree-cfg.c (move_stmt_op): Fix condition under which to update
9889         TREE_BLOCK.
9890         (move_stmt_r): Remove redundant checking.
9892 2013-04-29  Teresa Johnson  <tejohnson@google.com>
9894         PR bootstrap/57077
9895         * basic-block.h (apply_scale): New function.
9896         (apply_probability): Use apply_scale.
9897         * gimple-streamer-in.c (input_bb): Ditto.
9898         * lto-streamer-in.c (input_cfg): Ditto.
9899         * lto-cgraph.c (merge_profile_summaries): Ditto.
9900         * tree-optimize.c (execute_fixup_cfg): Ditto.
9901         * tree-inline.c (copy_bb): Update comment to use apply_scale.
9902         (copy_edges_for_bb): Ditto.
9903         (copy_cfg_body): Ditto.
9905 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9907         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9908         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9909         (tail_merge_optimize): Handle current_loops == NULL.
9911 2013-04-26  Jeff Law  <law@redhat.com>
9913         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9914         (simplify_cond_using_ranges): Generalize code to simplify
9915         COND_EXPRs where one argument is a constant and the other
9916         is an SSA_NAME created by an integral type conversion.
9918 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9920         * config/arm/arm.md (store_minmaxsi): Use only when
9921         optimize_insn_for_size_p.
9923 2013-04-29  Christian Bruel  <christian.bruel@st.com>
9925         PR target/57108
9926         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9928 2013-04-29  Richard Biener  <rguenther@suse.de>
9930         PR middle-end/57089
9931         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9932         loop tree make sure to schedule a fixup for the child as well.
9933         (expand_omp_for_generic): Properly add loops.
9934         (expand_omp_for_static_nochunk): Likewise.
9935         (expand_omp_for_static_chunk): Likewise.
9936         (expand_omp_for): For the degenerate case fixup loops.
9937         (expand_omp_sections): Fix default bb placement in loops.
9938         (expand_omp_atomic_pipeline): Properly add loops.
9940 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9942         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9944 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9946         * tree-ssa-tail-merge.c: Update header comment.
9948 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9950         * config/aarch64/arm_neon.h
9951         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9952         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9953         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9954         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9955         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9956         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9957         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9958         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9960 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9962         * config/aarch64/aarch64-simd.md
9963         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9964         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9965         fix_trunc, fixuns_trunc.
9966         (ftrunc<VDQF:mode>2): New.
9967         * config/aarch64/iterators.md (optab): Add fix, fixuns.
9968         (fix_trunc_optab): New.
9970 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9972         * config/aarch64/aarch64-builtins.c
9973         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9974         iceilf, lround, iroundf.
9976 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9978         PR target/54349
9979         * config/i386/i386.h (enum ix86_tune_indices)
9980         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9981         New, split from X86_TUNE_INTER_UNIT_MOVES.
9982         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9983         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9984         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9985         (TARGET_INTER_UNIT_MOVES): Remove.
9986         * config/i386/i386.c (initial_ix86_tune_features): Update.
9987         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9988         (ix86_expand_convert_uns_didf_sse): Use
9989         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9990         (ix86_expand_vector_init_one_nonzero): Ditto.
9991         (ix86_expand_vector_init_interleave): Ditto.
9992         (inline_secondary_memory_needed): Return true for moves from SSE class
9993         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9994         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9995         * config/i386/constraints.md (Yi, Ym): Depend on
9996         TARGET_INTER_UNIT_MOVES_TO_VEC.
9997         (Yj, Yn): New constraints.
9998         * config/i386/i386.md (*movdi_internal): Change constraints of
9999         operand 1 from Yi to Yj and from Ym to Yn.
10000         (*movsi_internal): Ditto.
10001         (*movdf_internal): Ditto.
10002         (*movsf_internal): Ditto.
10003         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10004         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10005         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10006         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10007         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10008         * config/i386/sse.md (movdi_to_sse): Ditto.
10009         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10010         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10011         TARGET_INTER_UNIT_MOVES.
10012         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10013         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10014         instead of TARGET_INTER_UNIT_MOVES.
10015         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10016         operand 1 from Yi to Yj and from Ym to Yn.
10018 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10020         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10021         (float_truncate_hi_): Likewise.
10022         (float_extend_lo_): Likewise.
10023         (float_truncate_lo_): Likewise.
10024         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10025         (aarch64_float_extend_lo_v2df): Likewise.
10026         (vec_unpacks_hi_v4sf): Likewise.
10027         (aarch64_float_truncate_lo_v2sf): Likewise.
10028         (aarch64_float_truncate_hi_v4sf): Likewise.
10029         (vec_pack_trunc_v2df): Likewise.
10030         (vec_pack_trunc_df): Likewise.
10032 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10034         * config/aarch64/aarch64-builtins.c
10035         (aarch64_fold_builtin): Fold float conversions.
10036         * config/aarch64/aarch64-simd-builtins.def
10037         (floatv2si, floatv4si, floatv2di): New.
10038         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10039         * config/aarch64/aarch64-simd.md
10040         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10041         * config/aarch64/iterators.md (FLOATUORS): New.
10042         (optab): Add float, floatuns.
10043         (su_optab): Likewise.
10045 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10047         * config/aarch64/aarch64-builtins.c
10048         (aarch64_builtin_vectorized_function): Use new names for
10049         fcvt builtins.
10050         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10051         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10052         (fcvtzu): Split as...
10053         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10054         (fcvtas): Split as...
10055         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10056         (fcvtau): Split as...
10057         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10058         (fcvtps): Split as...
10059         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10060         (fcvtpu): Split as...
10061         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10062         (fcvtms): Split as...
10063         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10064         (fcvtmu): Split as...
10065         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10066         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10067         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10068         (lfrintnusf, lfrintnudf): Likewise.
10069         * config/aarch64/aarch64-simd.md
10070         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10071         define_insn.
10072         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10073         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10074         (fcvt_pattern): Likewise.
10076 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10078         * config/aarch64/aarch64-simd.md
10079         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10080         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10082 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10084         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10085         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10086         (vrnd<a,m,n,p>_f32): Implement using builtins.
10087         (vrnd<i,x><q>_f<32, 64>): New.
10089 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10091         * config/aarch64/aarch64-builtins.c
10092         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10093         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10094         (frintz): Rename to...
10095         (btrunc): ...this.
10096         (frintp): Rename to...
10097         (ceil): ...this.
10098         (frintm): Rename to...
10099         (floor): ...this.
10100         (frinti): Rename to...
10101         (nearbyint): ...this.
10102         (frintx): Rename to...
10103         (rint): ...this.
10104         (frinta): Rename to...
10105         (round): ...this.
10106         * config/aarch64/aarch64-simd.md
10107         (aarch64_frint<frint_suffix><mode>): Delete.
10108         (<frint_pattern><mode>2): Convert to insn.
10109         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10110         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10111         (frint_pattern): Likewise.
10112         (frint_suffix): Likewise.
10114 2013-04-29  Richard Biener  <rguenther@suse.de>
10116         PR tree-optimization/57081
10117         * loop-init.c: Include tree-flow.h.
10118         (loop_optimizer_finalize): Free number of iteration estimates.
10119         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10121 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
10123         PR tree-optimization/57083
10124         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10125         non-singleton shift count range, zero extend low_bound for uns case.
10127         * config/i386/predicates.md (general_vector_operand): New predicate.
10128         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10129         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10130         if they aren't nonimmediate operands.  If their original values
10131         satisfy const_vector_equal_evenodd_p, don't shift them.
10132         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10133         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
10134         if not nonimmediate_operand.
10135         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10136         instead of register_operand.
10137         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10139 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
10141         * stor-layout.c (finalize_size_functions): Allocate a structure and
10142         reset cfun before dumping the functions.
10144 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
10146         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10148         PR target/56866
10149         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10150         use xop_pmacsdqh if uns_p.
10151         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10152         the immediate rotate count.
10154 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
10156         * rtl.h (struct rtx_def): Add comment for field jump.
10157         (LRA_SUBREG_P): New macro.
10158         * recog.c (register_operand): Check LRA_SUBREG_P.
10159         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10160         stack frame.
10161         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10162         (lra_final_code_change): Skip subreg change for operators.
10163         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10164         if there are no operand changes.
10165         * lra-constraints.c (curr_insn_set): New.
10166         (match_reload): Set LRA_SUBREG_P.
10167         (emit_spill_move): Ditto.
10168         (check_and_process_move): Use curr_insn_set. Process only single
10169         set insns.  Don't initialize sec_mem_p and change_p.
10170         (simplify_operand_subreg): Use LRA_SUBREG_P.
10171         (reg_in_class_p): New function.
10172         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10173         of #ifdef.  Add code to remove cycling.
10174         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10175         non-null disp.  Reload inner instead of disp when base and index
10176         are null.  Try to put lo_sum into register.
10177         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10178         (check_and_process_move): Move code for move cost check to
10179         simple_move_p.  Remove equiv_substitution.
10180         (simple_move_p): New function.
10181         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10182         curr_insn_set.  Call check_and_process_move only for single set
10183         insns.  Use the new function.  Move call of check_and_process_move
10184         after operand equiv substitution and address process.
10186 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10188         PR go/57045
10189         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10190         with nonlocal goto receivers or returns twice calls, ignore
10191         unininitialized values from abnormal edges to nl goto receiver
10192         or returns twice call.
10194 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10196         PR tree-optimization/57051
10197         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10198         and VEC_RSHIFT_EXPR if shift count is a multiple of element
10199         bitsize.
10201 2013-04-26  Richard Biener  <rguenther@suse.de>
10203         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10204         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
10205         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10206         (fixup_loop_arrays_after_move): New function.
10207         (move_sese_region_to_fn): Properly outline the loop tree parts
10208         of the SESE region.
10210 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
10212         * config/i386/i386.md (type, unit): Fix long lines.
10214 2013-04-26  Richard Biener  <rguenther@suse.de>
10216         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10217         (lto-streamer-out.o): Likewise.
10218         * cfgloop.c (init_loops_structure): Export, add struct function
10219         argument and adjust.
10220         (flow_loops_find): Adjust.
10221         * cfgloop.h (enum loop_estimation): Add EST_LAST.
10222         (init_loops_structure): Declare.
10223         * lto-streamer-in.c: Include cfgloop.h.
10224         (input_cfg): Input the loop tree.
10225         * lto-streamer-out.c: Include cfgloop.h.
10226         (output_cfg): Output the loop tree.
10227         (output_struct_function_base): Do not drop PROP_loops.
10229 2013-03-26  Richard Biener  <rguenther@suse.de>
10231         * tree-cfg.c (execute_build_cfg): Build the loop tree.
10232         (pass_build_cfg): Provide PROP_loops.
10233         (move_sese_region_to_fn): Remove loops that are outlined into fn
10234         for now.
10235         * tree-inline.c: Include cfgloop.h.
10236         (initialize_cfun): Do not drop PROP_loops.
10237         (copy_loops): New function.
10238         (copy_cfg_body): Copy loop structure.
10239         (tree_function_versioning): Initialize destination loop tree.
10240         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10241         (pass_parallelize_loops): Do IL verification.
10242         * loop-init.c (loop_optimizer_init): Fixup loops if required.
10243         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10244         the CFG make sure we fixup loops as well.
10245         * tree-ssa-tail-merge.c: Include cfgloop.h.
10246         (replace_block_by): When merging loop latches mark loops for fixup.
10247         * lto-streamer-out.c (output_struct_function_base): Drop
10248         PROP_loops for now.
10249         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10250         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10251         * ipa-split.c: Include cfgloop.h.
10252         (split_function): Add the new return block to the loop tree root.
10253         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10254         whether we have removed the forwarder block.
10255         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10256         * cfgloop.h (place_new_loop): Declare.
10257         * cfgloopmanip.c (place_new_loop): Export.
10258         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10259         (tree-switch-conversion.o): Likewise.
10260         (tree-complex.o): Likewise.
10261         (tree-inline.o): Likewise.
10262         (tree-ssa-tailmerge.o): Likewise.
10263         (ipa-split.o): Likewise.
10264         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10265         (tree-ssa-copy.o): Likewise.
10266         * tree-switch-conversion.c: Include cfgloop.h
10267         (process_switch): If we emit a bit-test cascade, schedule loops
10268         for fixup.
10269         * tree-complex.c: Include cfgloop.h.
10270         (expand_complex_div_wide): Properly add new basic-blocks to loops.
10271         * asan.c: Include cfgloop.h.
10272         (create_cond_insert_point): Properly add new basic-blocks to
10273         loops, schedule loop fixup.
10274         * cfgloop.c (verify_loop_structure): Check that looks are not
10275         marked for fixup.
10276         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10277         to loops.
10278         (expand_omp_for_generic): Likewise.
10279         (expand_omp_sections): Likewise.
10280         (expand_omp_atomic_pipeline): Schedule loops for fixup.
10281         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10282         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10283         is initialized, not when loops are present.
10284         * tree-parloops.c (parallelize_loops): Remove checking here.
10285         * passes.c (init_optimization_passes): Schedule a copy-propagation
10286         pass before complete unrolling of inner loops.
10288 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10290         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10291         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10292         (colorize_init): Add argument to _WIN32 version.
10293         * toplev.c: Include diagnostic-color.h.
10294         (process_options): Default to -fdiagnostics-color=auto if
10295         GCC_COLORS env var is in the environment.
10296         * common.opt (fdiagnostics-color=): Add Var and Init.
10297         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10298         env var is in the environment, the default is auto rather than never.
10300         * diagnostic.h (file_name_as_prefix): Add context argument.
10301         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
10302         the string as locus.
10303         * langhooks.c (lhd_print_error_function): Adjust caller.
10305 2013-04-25  Lawrence Crowl  <crowl@google.com>
10307         * var-tracking.c (shared_hash_def::htab):
10308         Change type to hash_table.  Update dependent calls and types.
10310 2013-04-25  Lawrence Crowl  <crowl@google.com>
10312         * Makefile.in: Update as needed below.
10314         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10315         Move declaration to after the type's method definitons.
10317         * attribs.c (htab_t scoped_attributes::attribute_hash):
10318         Change type to hash_table.  Update dependent calls and types.
10320         * bitmap.c (htab_t bitmap_desc_hash):
10321         Change type to hash_table.  Update dependent calls and types.
10323         * cselib.c (htab_t cselib_hash_table):
10324         Change type to hash_table.  Update dependent calls and types.
10326         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10327         (hash_string_slot_node): Move implementation into lto-streamer.h
10328         struct string_slot_hasher.
10329         (eq_string_slot_node): Likewise.
10331         * data-streamer-out.c: Update output_block::string_hash_table
10332         dependent calls and types.
10334         * dwarf2cfi.c (htab_t trace_index):
10335         Change type to hash_table.  Update dependent calls and types.
10337         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10338         Change type to hash_table.  Update dependent calls and types.
10339         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10340         (htab_t optimize_external_refs::map): Likewise.
10341         (htab_t output_comp_unit::extern_map): Likewise.
10342         (htab_t output_comdat_type_unit::extern_map): Likewise.
10343         (htab_t output_macinfo::macinfo_htab): Likewise.
10344         (htab_t optimize_location_lists::htab): Likewise.
10345         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10347         * except.c (htab_t ehspec_hash_type):
10348         Change type to hash_table.  Update dependent calls and types.
10349         (assign_filter_values::ttypes): Likewise.
10350         (assign_filter_values::ehspec): Likewise.
10351         (sjlj_assign_call_site_values::ar_hash): Likewise.
10352         (convert_to_eh_region_ranges::ar_hash): Likewise.
10354         * gcse.c (htab_t pre_ldst_table):
10355         Change type to hash_table.  Update dependent calls and types.
10357         * ggc-common.c (htab_t saving_htab):
10358         Change type to hash_table.  Update dependent calls and types.
10359         (htab_t loc_hash): Likewise.
10360         (htab_t ptr_hash): Likewise.
10361         (call_count): Rename ggc_call_count.
10362         (call_alloc): Rename ggc_call_alloc.
10363         (loc_descriptor): Rename make_loc_descriptor.
10364         (add_statistics): Rename ggc_add_statistics.
10366         * ggc-common.c (saving_htab):
10367         Change type to hash_table.  Update dependent calls and types.
10369         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10370         (push_gimplify_context): Likewise.
10371         (pop_gimplify_context): Likewise.
10372         (struct gimple_temp_hash_elt): Added.
10373         (struct gimplify_hasher): Likewise.
10374         (struct gimplify_ctx.temp_htab):
10375         Change type to hash_table.  Update dependent calls and types.
10377         * gimple-fold.c: Include gimplify-ctx.h.
10379         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10380         Change type to hash_table.  Update dependent calls and types.
10381         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10382         avoid potential global name collision.
10384         * gimplify.c: Include gimplify-ctx.h.
10385         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10386         (htab_t gimplify_ctx::temp_htab):
10387         Update dependent calls and types for new type hash_table.
10388         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10389         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10391         * gimplify-ctx.h: New.
10392         (struct gimple_temp_hash_elt): Move from gimplify.c.
10393         (class gimplify_hasher): New.
10394         (struct gimplify_ctx): Move from gimple.h.
10395         (htab_t gimplify_ctx::temp_htab):
10396         Change type to hash_table.  Update dependent calls and types.
10398         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10399         (htab_t ivs_params::newivs_index):
10400         Change type to hash_table.  Update dependent calls and types.
10401         (htab_t ivs_params::params_index): Likewise.
10402         (htab_t print_generated_program::params_index): Likewise.
10403         (htab_t gloog::newivs_index): Likewise.
10404         (htab_t gloog::params_index): Likewise.
10406         * graphite.c: Include graphite-htab.h.
10407         4htab_t graphite_transform_loops::bb_pbb_mapping):
10408         Change type to hash_table.  Update dependent calls and types.
10410         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10411         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10412         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10414         * graphite-dependences.c: Include graphite-htab.h.
10415         (loop_is_parallel_p): Change hash table type of parameter.
10417         * graphite-htab.h: New.
10418         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10419         (extern find_pbb_via_hash): Move from graphite-poly.h.
10420         (extern loop_is_parallel_p): Move from graphite-poly.h.
10421         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10423         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10424         (extern loop_is_parallel_p): Move to graphite-htab.h.
10425         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10427         * haifa-sched.c (htab_t delay_htab):
10428         Change type to hash_table.  Update dependent calls and types.
10429         (htab_t delay_htab_i2): Likewise.
10431         * ira-color.c (htab_t allocno_hard_regs_htab):
10432         Change type to hash_table.  Update dependent calls and types.
10434         * ira-costs.c (htab_t cost_classes_htab):
10435         Change type to hash_table.  Update dependent calls and types.
10437         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10438         Change type to hash_table.  Update dependent calls and types.
10440         * loop-iv.c (htab_t bivs):
10441         Change type to hash_table.  Update dependent calls and types.
10443         * loop-unroll.c (htab_t opt_info::insns_to_split):
10444         Change type to hash_table.  Update dependent calls and types.
10445         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10447         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10448         (struct string_slot_hasher): New.
10449         (htab_t output_block::string_hash_table):
10450         Change type to hash_table.  Update dependent calls and types.
10452         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10453         (htab_t file_name_hash_table):
10454         Change type to hash_table.  Update dependent calls and types.
10456         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10457         calls and types.
10459         * lto-streamer.c (htab_t tree_htab):
10460         Change type to hash_table.  Update dependent calls and types.
10462         * omp-low.c: Include gimplify-ctx.h.
10464         * passes.c (htab_t name_to_pass_map):
10465         Change type to hash_table.  Update dependent calls and types.
10466         (pass_traverse): Rename to passes_pass_traverse.
10468         * plugin.c (htab_t event_tab):
10469         Change type to hash_table.  Update dependent calls and types.
10471         * postreload-gcse.c (htab_t expr_table):
10472         Change type to hash_table.  Update dependent calls and types.
10473         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10475         * sese.c (debug_rename_map_1): Make extern.
10476         (htab_t copy_bb_and_scalar_dependences::rename_map):
10477         Change type to hash_table.  Update dependent calls and types.
10479         * sese.h (extern debug_rename_map): Move to .c file.
10481         * store-motion.c (htab_t store_motion_mems_table):
10482         Change type to hash_table.  Update dependent calls and types.
10484         * trans-mem.c (htab_t tm_new_mem_hash):
10485         Change type to hash_table.  Update dependent calls and types.
10487         * tree-browser.c (htab_t TB_up_ht):
10488         Change type to hash_table.  Update dependent calls and types.
10490         * tree-cfg.c (htab_t discriminator_per_locus):
10491         Change type to hash_table.  Update dependent calls and types.
10493         * tree-complex.c: Include tree-hasher.h
10494         (htab_t complex_variable_components):
10495         Change type to hash_table.  Update dependent calls and types.
10497         * tree-eh.c (htab_t finally_tree):
10498         Change type to hash_table.  Update dependent calls and types.
10500         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10501         struct int_tree_hasher.
10502         (extern int_tree_map_eq): Likewise.
10503         (uid_decl_map_hash): Removed.
10504         (extern decl_tree_map_eq): Likewise.
10506         * tree-hasher.h: New.
10507         (struct int_tree_hasher): New.
10508         (typedef int_tree_htab_type): New.
10510         * tree-inline.c: Include gimplify-ctx.h.
10512         * tree-mudflap.c: Include gimplify-ctx.h.
10514         * tree-parloops.c: Include tree-hasher.h.
10515         (htab_t eliminate_local_variables_stmt::decl_address):
10516         Change type to hash_table.  Update dependent calls and types.
10517         (htab_t separate_decls_in_region::decl_copies): Likewise.
10519         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10520         Change type to hash_table.  Update dependent calls and types.
10522         * tree-sra.c (candidates):
10523         Change type to hash_table.  Update dependent calls and types.
10525         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10526         in tree-flow.h.
10527         (int_tree_map_hash): Likewise.
10529         * tree-ssa-dom.c (htab_t avail_exprs):
10530         Change type to hash_table.  Update dependent calls and types.
10532         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10533         Change type to hash_table.  Update dependent calls and types.
10535         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10536         Change type to hash_table.  Update dependent calls and types.
10538         * tree-ssa-phiopt.c (seen_ssa_names):
10539         Change type to hash_table.  Update dependent calls and types.
10541         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10542         Change type to hash_table.  Update dependent calls and types.
10544         * tree-ssa-uncprop.c (equiv):
10545         Change type to hash_table.  Update dependent calls and types.
10547 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10549         PR rtl-optimization/57003
10550         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10551         call note_stores with kill_clobbered_value callback again after
10552         killing regs_invalidated_by_call.
10554 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10556         * config/aarch64/aarch64-simd.md
10557         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10558         (aarch64_simd_bsl<mode>): Likewise.
10559         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10561 2013-04-25  Marek Polacek  <polacek@redhat.com>
10563         PR tree-optimization/57066
10564         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10566 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10568         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10570 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10572         * config/aarch64/aarch64-builtins.c
10573         (aarch64_fold_builtin): New.
10574         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10575         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10576         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10577         * config/aarch64/arm_neon.h
10578         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10580 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10581             Tejas Belagod  <tejas.belagod@arm.com>
10583         * config/aarch64/aarch64-builtins.c
10584         (aarch64_gimple_fold_builtin): New.
10585         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10586         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10587         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10588         (addvv4sf): Update.
10589         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10591 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10593         * config/aarch64/aarch64.md
10594         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10596 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10598         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10599         (*ngcsi_uxtw): New pattern.
10601 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10602             Julian Brown  <julian@codesourcery.com>
10604         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10605         (TB_DREG): Add T_V4HF.
10606         (v4hf_UP): New macro.
10607         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10608         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10609         Handle initialisation of V4HF. Adjust initialisation of reinterpret
10610         built-ins.
10611         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10612         (arm_vector_mode_supported_p): Handle V4HF.
10613         (arm_mangle_map): Handle V4HFmode.
10614         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10615         * config/arm/arm_neon_builtins.def: Add entries for
10616         vcvtv4hfv4sf, vcvtv4sfv4hf.
10617         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10618         (neon_vcvtv4hfv4sf): Likewise.
10619         * config/arm/neon-gen.ml: Handle half-precision floating point
10620         features.
10621         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10622         * config/arm/arm_neon.h: Regenerate.
10623         * config/arm/neon.ml (type elts): Add F16.
10624         (type vectype): Add T_float16x4, T_floatHF.
10625         (type vecmode): Add V4HF.
10626         (type features): Add Requires_FP_bit feature.
10627         (elt_width): Handle F16.
10628         (elt_class): Likewise.
10629         (elt_of_class_width): Likewise.
10630         (mode_of_elt): Refactor.
10631         (type_for_elt): Handle F16, fix error messages.
10632         (vectype_size): Handle T_float16x4.
10633         (vcvt_sh): New function.
10634         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10635         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10636         (string_of_mode): Handle V4HF.
10637         * doc/arm-neon-intrinsics.texi: Regenerate.
10639 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10641         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10642         format specifier in 'X' case.
10644 2013-04-25  Alan Modra  <amodra@gmail.com>
10646         PR target/57052
10647         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10648         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10649         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10650         Repeat for many other rotate/shift and mask patterns using subregs.
10651         Name lshiftrt insns.
10652         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10653         on WORDS_BIG_ENDIAN.
10655 2013-04-25  Alan Modra  <amodra@gmail.com>
10657         * config.gcc: Support little-endian powerpc-linux targets.
10658         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10659         (LINK_OS_LINUX_SPEC): Define.
10660         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10661         Preserve MASK_LITTLE_ENDIAN.
10662         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10663         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10664         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10665         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10666         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10667         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10668         Correct fp word order for little-endian.  Don't shift toc entries
10669         smaller than a word for little-endian.
10670         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10671         (bswapdi2 splits): Correct low-part subreg for little-endian.
10672         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10673         low/high where such is correct only for be.
10674         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10675         little-endian for -mcall-aixdesc.
10677 2013-04-25  Alan Modra  <amodra@gmail.com>
10679         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10680         replace_equiv_address_nv.
10682 2013-04-25  Alan Modra  <amodra@gmail.com>
10684         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10686 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10688         Revert:
10689         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10690         * rtl.h (struct rtx_def): ...
10692 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10694         PR rtl-optimizations/57046
10695         * lra-constraints (split_reg): Set up lra_risky_transformations_p
10696         for multi-reg splits.
10698 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
10700         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10702 2013-04-24  Sterling Augustine  <saugustine@google.com>
10704         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10705         (comp_dir_string, debug_str_dwo_section): New.
10706         (DEBUG_STR_DWO_SECTION): Rename to ...
10707         (DEBUG_DWO_STR_SECTION): ... this.
10708         (DEBUG_NORM_STR_SECTION): Delete.
10709         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10710         (DEBUG_STR_DWO_SECTION_FLAGS): New.
10711         (find_AT_string): Move most logic to ...
10712         (find_AT_string_in_table): ... here.  New.
10713         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10714         add_skeleton_AT_string.  Delete logic.
10715         (output_skeleton_debug_sections): Remove call to
10716         add_top_level_skeleton_die_attrs.
10717         (add_comp_dir_attribute): Move logic to comp_dir_string.
10718         (dwarf2out_init): Initialize debug_str_dwo_section.
10719         (output_indirect_string): Call find_string_form.
10720         (output_indirect_strings): Rewrite.
10721         (prune_unused_types): Empty skeleton_debug_str_hash.
10722         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10723         (dwarf2out_finish):  Call output_indirect_strings.
10725 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10727         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10729 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10731         * rtl.h (struct rtx_def): Add comment for field jump.
10732         (LRA_SUBREG_P): New macro.
10733         * recog.c (register_operand): Check LRA_SUBREG_P.
10734         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10735         stack frame.
10736         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10737         (lra_final_code_change): Skip subreg change for operators.
10738         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10739         if there are no operand changes.
10740         * lra-constraints.c (curr_insn_set): New.
10741         (match_reload): Set LRA_SUBREG_P.
10742         (emit_spill_move): Ditto.
10743         (check_and_process_move): Use curr_insn_set. Process only single
10744         set insns.  Don't initialize sec_mem_p and change_p.
10745         (simplify_operand_subreg): Use LRA_SUBREG_P.
10746         (reg_in_class_p): New function.
10747         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10748         of #ifdef.  Add code to remove cycling.
10749         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10750         non-null disp.  Reload inner instead of disp when base and index
10751         are null.  Try to put lo_sum into register.
10752         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10753         (check_and_process_move): Move code for move cost check to
10754         simple_move_p.  Remove equiv_substitution.
10755         (simple_move_p): New function.
10756         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10757         curr_insn_set.  Call check_and_process_move only for single set
10758         insns.  Use the new function.  Move call of check_and_process_move
10759         after operand equiv substitution and address process.
10761 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
10763         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10764         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10765         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10767 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10769         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10771 2013-04-24  Marek Polacek  <polacek@redhat.com>
10773         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10774         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10775         (select_loops_exit_conditions): Likewise.
10776         (number_of_iterations_for_all_loops): Likewise.
10777         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10778         (scev_analysis): Likewise.
10780 2013-04-02  Catherine Moore  <clm@codesourcery.com>
10781             Chao-ying Fu <fu@mips.com>
10783         * config/mips/micromips.md (jraddiusp): New pattern.
10784         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10785         instruction if possible.
10787 2013-04-24  Alan Modra  <amodra@gmail.com>
10789         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10791 2013-04-24  Julian Brown  <julian@codesourcery.com>
10792             Chung-Lin Tang  <cltang@codesourcery.com>
10794         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10795         dependency behavior in enumeration type DIE generation. Add TODO note
10796         to comments about future DW_FORM_sdata/udata re-work of related code.
10798 2013-04-23  Lawrence Crowl  <crowl@google.com>
10800         * Makefile.in: Update as needed below.
10802         * hash-table.h (class hash_table):
10803         Correct many methods with parameter types compare_type to the correct
10804         value_type.  (Correct code was unlikely to notice the change.)
10805         (hash_table::elements_with_deleted) New.
10806         (class hashtable::iterator): New.
10807         (hashtable::begin()): New.
10808         (hashtable::end()): New.
10809         (FOR_EACH_HASH_TABLE_ELEMENT): New.
10811         * statistics.c (statistics_hashes):
10812         Change type to hash_table.  Update dependent calls and types.
10814         * tree-into-ssa.c (var_infos):
10815         Change type to hash_table.  Update dependent calls and types.
10817         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10818         Change type to hash_table.  Update dependent calls and types.
10820         * tree-ssa-loop-im.c (struct mem_ref.refs):
10821         Change type to hash_table.  Update dependent calls and types.
10823         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10824         Change type to hash_table.  Update dependent calls and types.
10826         * tree-ssa-sccvn.c (vn_tables_s::nary):
10827         Change type to hash_table.  Update dependent calls and types.
10828         (vn_tables_s::phis): Likewise.
10829         (vn_tables_s::references): Likewise.
10831         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10832         (vn_reference_eq): Update parameter and return types.
10834         * tree-ssa-structalias.c (pointer_equiv_class_table):
10835         Change type to hash_table.  Update dependent calls and types.
10836         (location_equiv_class_table): Likewise.
10838         * tree-vect-data-refs.c: Consequential changes for making
10839         peeling a hash_table.
10841         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10842         (destroy_loop_vec_info): Dependent hash_table update.
10844         * tree-vectorizer.h (peeling_htab):
10845         Change type to hash_table.  Update dependent calls and types.
10847 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
10849         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10850         to check the register content is equal or not.
10851         * lra-constraints.c (match_reload): Use lra_assign_reg_val
10852         to assign register content record.
10853         * lra-eliminations.c (update_reg_eliminate): Use
10854         lra_update_reg_val_offset to update register content offset.
10855         * lra-int.h (struct lra_reg): Add offset member.
10856         (lra_reg_val_equal_p): New static inline function.
10857         (lra_update_reg_val_offset): New static inline function.
10858         (lra_assign_reg_val): New static inline function.
10859         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10860         to assign register content record.
10861         (initialize_lra_reg_info_element): Initial offset to zero.
10863 2013-04-23  Catherine Moore  <clm@codesourcery.com>
10865         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10866         operands.  Record compression.
10868 2013-04-23  Xinliang David Li  <davidxl@google.com>
10870         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10872 2013-04-23  Richard Biener  <rguenther@suse.de>
10874         PR middle-end/57036
10875         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10876         parameter, only add abnormal goto edges from the copied body
10877         if the call could perform abnormal gotos.
10878         (copy_cfg_body): Adjust.
10880 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
10882         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10884 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
10886         * coretypes.h (gimple_stmt_iterator): Add struct to make
10887         compatible with C.
10889 2013-04-23  Richard Biener  <rguenther@suse.de>
10891         PR tree-optimization/57026
10892         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10893         from SSA names occuring in abnormal PHI nodes.
10895 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10897         * lto/lto.c (print_lto_report_1): Fix LTO report names.
10899 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10901         * lto/lto.c (print_lto_report_1): Declare early.
10902         (read_cgraph_and_symbols): Call print_lto_report_1 early.
10904 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10906         * common.opt (-flto-report-wpa): Add.
10907         * doc/invoke.texi (-flto-report-wpa): Add.
10908         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10909         (lto_main): dito.
10911 2013-04-22  Xinliang David Li  <davidxl@google.com>
10913         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10914         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10915         * Makefile.in: New dependency
10917             David Daney <ddaney.cavm@gmail.com>
10919         * configure.ac (gcc_cv_as_micromips_support):  Use the
10920         --fatal-warnings option.
10921         * configure: Regenerate.
10923 2013-04-22  Marek Polacek  <polacek@redhat.com>
10925         PR sanitizer/56990
10926         * tsan.c (instrument_expr): Don't instrument expression
10927         in case its size is zero.
10929 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
10931         PR target/57032
10932         Revert:
10933         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
10935         * config/alpha/alpha.c (TARGET_LRA_P): New define.
10937 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10939         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10940         (gimple_stmt_iterator): New typedef.
10941         * gimple.h (gimple_stmt_iterator): Rename to...
10942         (gimple_stmt_iterator_d): ... This.
10943         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10944         trees be valid for GIMPLE and GENERIC.
10945         (TARGET_GIMPLE_FOLD_BUILTIN): New.
10946         * gimple-fold.c (gimple_fold_call): Call target hook
10947         gimple_fold_builtin.
10948         * hooks.c (hook_bool_gsiptr_false): New.
10949         * hooks.h (hook_bool_gsiptr_false): New.
10950         * target.def (fold_stmt): New.
10951         * doc/tm.texi: Regenerate.
10953 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
10955         PR target/57018
10956         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10957         a set sp if no stack realignment.
10959 2013-04-22  Nick Clifton  <nickc@redhat.com>
10961         * config.gcc (tilegx-linux): Extend extra_objs rather than
10962         overwriting it.
10963         (tilepro-linux): Likewise.
10965 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10967         * config/aarch64/aarch64-builtins.c
10968         (CF): Remove.
10969         (CF0, CF1, CF2, CF3, CF4, CF10): New.
10970         (VAR<1-12>): Add MAP parameter.
10971         (BUILTIN_*): Likewise.
10972         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10973         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10974         (aarch64_ushl_n<mode>): Likewise.
10975         (aarch64_sshr_n<mode>): Likewise.
10976         (aarch64_ushr_n<mode>): Likewise.
10977         (aarch64_<maxmin><mode>): Likewise.
10978         (aarch64_sqrt<mode>): Likewise.
10979         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10980         (vshr<q>_n_*): Likewise.
10982 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10984         * config/aarch64/aarch64-builtins.c
10985         (aarch64_simd_builtin_type_mode): Handle SF types.
10986         (sf_UP): Define.
10987         (BUILTIN_GPF): Define.
10988         (aarch64_init_simd_builtins): Handle SF types.
10989         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10990         (frecps): Likewise.
10991         (frecpx): Likewise.
10992         * config/aarch64/aarch64-simd.md
10993         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10994         (aarch64_frecpe<mode>): New.
10995         (aarch64_frecps<mode>): Likewise.
10996         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10997         (v8type): Add frecp<esx>.
10998         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10999         (aarch64_frecps<mode>): Likewise.
11000         * config/aarch64/iterators.md (FRECP): New.
11001         (frecp_suffix): Likewise.
11002         * config/aarch64/arm_neon.h
11003         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11005 2013-04-22  Christian Bruel  <christian.bruel@st.com>
11007         PR target/56995
11008         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11009         (REG_CLASS_NAMES): Idem.
11010         (REG_CLASS_CONTENTS): Idem.
11011         (REGCLASS_HAS_FP_REG): Idem.
11012         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11013         (sh_conditional_register_usage): Idem.
11015 2013-04-21  Jeff Law  <law@redhat.com>
11017         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11018         (ssa_forward_propagate_and_combine): Use it.
11020 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11022         * lra.c: Update the flow chart diagram.
11024 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11026         PR rtl-optimization/56847
11027         * lra-constraints.c (process_alt_operands): Discourage alternative
11028         with non-matche doffsettable memory constraint fro memory with
11029         known offset.
11031 2013-04-19  Richard Biener  <rguenther@suse.de>
11033         PR tree-optimization/56982
11034         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11035         function.
11036         * gimplify.c (gimplify_call_expr): Notice special calls.
11037         (gimplify_modify_expr): Likewise.
11038         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11039         abnormal control flow receivers.
11040         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11041         in the same way as cfun->has_nonlocal_labels.
11042         (gimple_purge_dead_abnormal_call_edges): Likewise.
11043         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11044         receivers start a basic-block.
11046 2013-04-19  Richard Biener  <rguenther@suse.de>
11048         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11049         member ...
11050         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
11051         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11052         (SLP_TREE_LOAD_PERMUTATION): Add.
11053         (vect_transform_slp_perm_load): Adjust prototype.
11054         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11055         (vect_free_slp_instance): Likewise.
11056         (vect_create_new_slp_node): Likewise.
11057         (vect_supported_slp_permutation_p): Remove.
11058         (vect_slp_rearrange_stmts): Adjust.
11059         (vect_supported_load_permutation_p): Likewise.  Inline
11060         vect_supported_slp_permutation_p here.
11061         (vect_analyze_slp_instance): Compute load permutations per
11062         slp node instead of per instance.
11063         (vect_get_slp_defs): Adjust.
11064         (vect_transform_slp_perm_load): Likewise.
11065         (vect_schedule_slp_instance): Remove redundant code.
11066         (vect_schedule_slp): Remove hack for PR56270, add it ...
11067         * tree-vect-stmts.c (vectorizable_load): ... here, do not
11068         CSE loads for SLP.  Adjust.
11070 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11072         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11073         spelling in two comments.
11075 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11077         PR target/56797
11078         * config/arm/arm.c (load_multiple_sequence): Require SP
11079         as base register for loads if SP is in the register list.
11081 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11083         PR tree-optimization/56718
11084         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11085         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11086         and made public.  Adjusted all callers.
11087         (ipa_intraprocedural_devirtualization): New function.
11088         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11089         (ipa_intraprocedural_devirtualization): Likewise.
11090         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11092 2013-04-19  Richard Biener  <rguenther@suse.de>
11094         PR tree-optimization/57000
11095         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11097 2013-04-19  Terry Guo  <terry.guo@arm.com>
11099         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11100         Replace with ...
11101         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
11102         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11103         (cortex_m4_fmacs): Use new reservations.
11104         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11106 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
11108         PR rtl-optimization/56999
11109         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11110         related code.
11111         (lra_coalesce): Remove split_origin_bitmap and related code.
11112         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
11113         ranges if necessary.
11115 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
11117         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11118         New array.
11119         (ix86_expand_call): Remove clobbered_registers array and use
11120         x86_64_ms_sysv_extra_clobbered_registers instead.
11121         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11122         Declare here.
11123         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11124         predicate.
11125         * config/i386/i386.md (*call_rex64_ms_sysv): Use
11126         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
11127         (*call_value_rex64_ms_sysv): Ditto.
11129 2013-04-18  Cary Coutant  <ccoutant@google.com>
11131         * dwarf2out.c (output_pubnames): Check die_perennial_p of
11132         parent instead of die_mark.
11134 2013-04-18  Diego Novillo  <dnovillo@google.com>
11136         * gimple.c (create_gimple_tmp): New.
11137         (get_expr_type): New.
11138         (build_assign): New.
11139         (build_type_cast): New.
11140         * gimple.h (enum ssa_mode): Define.
11141         (gimple_seq_set_location): New.
11142         * asan.c (build_check_stmt): Change some gimple_build_* calls
11143         to use build_assign and build_type_cast.
11145 2013-04-18  Richard Biener  <rguenther@suse.de>
11147         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11148         handle negative step.  Remove redundant checks.
11149         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11150         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11151         for negative step and grouped loads fail to vectorize.
11153 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11155         * emit-rtl.c (reset_insn_used_flags): New function.
11156         (reset_all_used_flags): Use it.
11157         (verify_insn_sharing): New function.
11158         (verify_rtl_sharing): Fix verification for SEQUENCEs.
11160 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
11162         PR tree-optimization/56984
11163         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11164         and (x >> M) >= N don't register any assertion if N << M is the
11165         minimum value.
11167 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11169         * lower-subreg.c (resolve_simple_move): If called self-recursive,
11170         do not delete_insn insns that have not yet been emitted, only
11171         unlink them with remove_insn.
11172         * df-scan.c (df_insn_delete): Revert r197492.
11174 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
11176         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11177         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11179 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11181         * config/arm/arm.md (movsicc_insn): Convert define_insn into
11182         define_insn_and_split.
11183         (and_scc,ior_scc,negscc): Likewise.
11184         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11186 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11188         * config/arm/arm.c (use_return_insn): Return 0 for targets that
11189         can benefit from using a sequence of LDRD instructions in epilogue
11190         instead of a single LDM instruction.
11192 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11194         PR 45688
11195         * doc/extend.texi: Fix typo.
11197 2013-04-17  Richard Biener  <rguenther@suse.de>
11199         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11200         (vect_build_slp_tree): ... here.
11201         (vect_build_slp_tree_1): Compute which stmts of the SLP group
11202         match.  Remove special-casing of mismatched complex loads.
11203         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11204         re-try the match with swapped commutative operands.
11205         (vect_supported_load_permutation_p): Remove special-casing of
11206         mismatched complex loads.
11207         (vect_analyze_slp_instance): Adjust.
11209 2013-04-17  Richard Biener  <rguenther@suse.de>
11211         PR rtl-optimization/56921
11212         * cfgloop.h (struct loop): Add simple_loop_desc member.
11213         (struct niter_desc): Mark with GTY(()).
11214         (simple_loop_desc): Do not use aux field but simple_loop_desc.
11215         * loop-iv.c (get_simple_loop_desc): Likewise.
11216         (free_simple_loop_desc): Likewise.
11218         Revert
11219         2013-04-16  Richard Biener  <rguenther@suse.de>
11221         PR rtl-optimization/56921
11222         * loop-init.c (pass_rtl_move_loop_invariants): Add
11223         TODO_do_not_ggc_collect to todo_flags_finish.
11224         (pass_rtl_unswitch): Same.
11225         (pass_rtl_unroll_and_peel_loops): Same.
11226         (pass_rtl_doloop): Same.
11228 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
11230         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11231         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11232         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11233         references.
11234         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11235         * tree-streamer.c (record_common_node): Adjust reference in comment.
11237 2013-04-17  Terry Guo  <terry.guo@arm.com>
11239         * config/arm/cortex-m4.md: Add a new bypass.
11241 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11243         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11244         New pattern.
11245         (*subs_<optab><mode>_multp2): New pattern.
11246         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11247         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11249 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11251         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11252         (*subs_mul_imm_<mode>): New pattern.
11254 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
11256         PR target/56948
11257         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11258         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
11259         (vsx_movti_32bit): Change j->wa to O->wa.
11261 2013-04-16  Richard Biener  <rguenther@suse.de>
11263         PR rtl-optimization/56921
11264         * loop-init.c (pass_rtl_move_loop_invariants): Add
11265         TODO_do_not_ggc_collect to todo_flags_finish.
11266         (pass_rtl_unswitch): Same.
11267         (pass_rtl_unroll_and_peel_loops): Same.
11268         (pass_rtl_doloop): Same.
11270 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11272         * config/arm/arm.c (emit_multi_reg_push): New declaration
11273         for an existing function.
11274         (arm_emit_strd_push): New function.
11275         (arm_expand_prologue): Used here.
11276         (arm_emit_ldrd_pop): New function.
11277         (arm_expand_epilogue): Used here.
11278         (arm_get_frame_offsets): Update condition.
11279         (arm_emit_multi_reg_pop): Add a special case for load of a single
11280         register with writeback.
11282 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
11284         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11285         description.
11287 2013-04-16  Richard Biener  <rguenther@suse.de>
11289         PR tree-optimization/56756
11290         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11291         (first_mem_ref_loc): New.
11292         (execute_sm): Place the load temporarily before a previous
11293         access instead of in the latch edge to ensure its SSA dependencies
11294         are defined at points dominating the load.
11296 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
11298         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11299         correct fix by moving header and footer insn to the footer of
11300         the merged basic block.  Clear BB_END of the merged-away block.
11302         PR middle-end/43631
11303         * emit-rtl.c (make_note_raw): New function.
11304         (link_insn_into_chain): New static inline function.
11305         (add_insn): Use it.
11306         (add_insn_before, add_insn_after): Factor insn chain linking code...
11307         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11308         using link_insn_into_chain.
11309         (note_outside_basic_block_p): New helper function for emit_note_after
11310         and emit_note_before.
11311         (emit_note_after): Use nobb variant of add_insn_after if the note
11312         should not be contained in a basic block.
11313         (emit_note_before): Use nobb variant of add_insn_before if the note
11314         should not be contained in a basic block.
11315         (emit_note_copy): Use make_note_raw.
11316         (emit_note): Likewise.
11317         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11318         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11319         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11320         the moved barrier the tail of the basic block it follows.
11321         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11323 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
11325         PR tree-optimization/56962
11326         * gimple-ssa-strength-reduction.c (record_increment): Only set
11327         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11328         either rhs1 or rhs2 is equal to c->base_expr.
11330 2013-04-15  Richard Biener  <rguenther@suse.de>
11332         PR tree-optimization/56933
11333         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11334         member.
11335         (GROUP_READ_WRITE_DEPENDENCE): Remove.
11336         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11337         * tree-vect-data-refs.c (vect_analyze_group_access): Move
11338         dependence check ...
11339         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11340         ... here.
11341         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11342         GROUP_READ_WRITE_DEPENDENCE.
11344 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11346         * emit-rtl.c (reset_all_used_flags): New function.
11347         (verify_rtl_sharing): Call reset_all_used_flags before and after
11348         performing the checks.
11350 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11352         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11353         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11354         * config/arm/constraints.md (De): New constraint.
11355         * config/arm/neon.md (anddi3_neon): Delete.
11356         (neon_vand<mode>): Expand to standard anddi3 pattern.
11357         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11358         Move earlier in the file.
11359         (neon_inv_logic_op2): Likewise.
11360         (arm_anddi_operand_neon): New predicate.
11362 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11364         * configure.ac (gcc_cv_ld_as_needed): Set
11365         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11366         Use -z ignore, -z record on *-*-solaris2*.
11367         (HAVE_LD_AS_NEEDED): Update comment.
11368         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11369         * configure: Regenerate.
11370         * config.in: Regenerate.
11371         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11372         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11373         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11374         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11375         equivalents.  Fix markup.
11376         * doc/tm.texi: Regenerate.
11378 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11380         * config/i386/i386.opt: New option mstack-protector-guard=.
11381         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11382         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11383         TARGET_SSP_TLS_GUARD.
11384         * config/i386/i386.c (ix86_option_override_internal): Set
11385         ix86_stack_protector_guard.
11386         * config/i386/i386.md (stack_protect_set): Enable for
11387         TARGET_SSP_TLS_GUARD only.
11388         (stack_protect_set_<mode>): Ditto.
11389         (stack_protect_test): Ditto.
11390         (stack_protect_test_<mode>): Ditto.
11391         * doc/invoke.texi (i386 Option): Document.
11393 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11395         PR target/56890
11396         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11397         (S_MODES): Set H_MODE bit.
11398         (SF_MODES): Set only S_MODE and SF_MODE bits.
11399         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11400         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11401         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11402         <MODE_FLOAT>: Likewise.
11404 2013-04-15  Joey Ye  <joey.ye@arm.com>
11406         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11408 2013-04-15  Joey Ye  <joey.ye@arm.com>
11410         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11411         for real far jump.
11412         (thumb_far_jump_used_p): Count instruction size and set
11413         far_jump_used.
11415 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11417         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11418         * resource.c (mark_target_live_regs): Reformat conditional block.
11420 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11422         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11423         notes, they are emitted only just before final.
11424         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11426 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11428         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11429         * cfgrtl.c (delete_insn): Call it here instead.
11430         * lra-spills.c (lra_final_code_change): Use delete_insn.
11431         * haifa-sched.c (sched_remove_insn): Likewise.
11432         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11433         returning to the nop pool.
11434         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11435         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11437 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11439         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11440         * doc/tm.texi: Regenerated.
11442 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11444         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11445         QImode checks.
11447 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11449         * df-core.c (df_find_def): Compare register numbers.
11450         (df_find_use): Likewise.
11452 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11454         PR target/56903
11455         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11456         lra_in_progress for return.
11458 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11460         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11461         define_insn into define_insn_and_split and emit movsicc patterns.
11463 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11465         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11467 2013-04-12  Richard Biener  <rguenther@suse.de>
11469         * tree-pass.h (TODO_do_not_ggc_collect): New.
11470         * passes.c (execute_one_ipa_transform_pass): Honor
11471         TODO_do_not_ggc_collect.
11472         (execute_one_pass): Likewise.
11474         Revert
11475         2013-04-10  Richard Biener  <rguenther@suse.de>
11477         * passes.c (init_optimization_passes): Remove reload pass.
11478         * ira.c (do_reload): Merge into ...
11479         (ira): ... this.
11480         (rest_of_handle_reload): Remove.
11481         (pass_reload): Likewise.
11482         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11483         of reload for vzeroupper pass placement.
11485 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11487         PR tree-optimization/56918
11488         PR tree-optimization/56920
11489         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11490         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11491         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11492         use wide_mul_with_sign method.
11494 2013-04-12  Richard Biener  <rguenther@suse.de>
11496         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11497         not be considered a gimple constant.
11499 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11501         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11502         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11503         operations.
11505 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11506             Jakub Jelinek  <jakub@redhat.com>
11508         * opts.c: Include diagnostic-color.h.
11509         (common_handle_option): Handle OPT_fdiagnostics_color_.
11510         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11511         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11512         (diagnostic-color.o): New.
11513         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11514         (diagnostic_color_rule): New enum.
11515         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11516         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11517         the location string.
11518         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11519         either NULL, or color kind.
11520         * diagnostic-color.c: New file.
11521         * diagnostic-color.h: New file.
11522         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11523         arguments.
11524         * doc/invoke.texi (-fdiagnostics-color): Document.
11525         * pretty-print.h (pp_show_color): Define.
11526         (struct pretty_print_info): Add show_color field.
11527         * diagnostic.c: Include diagnostic-color.h.
11528         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11529         macros.  Colorize error:, warning: etc. strings and also the location
11530         string.
11531         (diagnostic_show_locus): Colorize the caret line.
11532         * pretty-print.c: Include diagnostic-color.h.
11533         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11534         inside of %< %> quotes or quoted through q format modifier.
11536 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11538         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11540 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11542         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11543         code in CC_NZ mode.
11544         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11545         pattern.
11547 2013-04-11  Marek Polacek  <polacek@redhat.com>
11549         PR tree-optimization/48184
11550         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11552 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11554         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11555         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11556         (skip_simple_arithmetic): Tidy up.
11557         * tree.h (skip_simple_constant_arithmetic): Declare.
11559 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11561         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11563 2013-04-11  Richard Biener  <rguenther@suse.de>
11565         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11566         generate vector constants.
11568 2013-04-11  Richard Biener  <rguenther@suse.de>
11570         PR tree-optimization/56878
11571         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11572         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11573         New function.
11574         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11575         Prefer to align the DR with the most invariant base address.
11577 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11579         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11580         comment.
11582 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11584         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11585         floating-point vector comparisons against 0.
11587 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11589         PR tree-optimization/56899
11590         * fold-const.c (extract_muldiv_1): Apply distributive law
11591         only if TYPE_OVERFLOW_WRAPS (ctype).
11593 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11595         PR target/56124
11596         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11597         loading has side effect.
11599 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
11601         * config/sparc/sparc.c: Include tree-pass.h.
11602         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11603         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
11604         head of file.  Change return type.  Split off gate function.
11605         (sparc_gate_work_around_errata): New function.
11606         (pass_work_around_errata): New pass definition.
11607         (insert_pass_work_around_errata) New pass insert definition to
11608         insert pass_work_around_errata just after delayed-branch scheduling.
11609         (sparc_option_override): Insert the pass.
11610         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11612 2013-04-10  David S. Miller  <davem@davemloft.net>
11614         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11615         or -mcpu=hypersparc.
11617         * target.def (cstore_mode): New hook.
11618         * target.h: Include insn-codes.h
11619         * targhooks.c: Likewise.
11620         (default_cstore_mode): New function.
11621         * targhooks.h: Declare it.
11622         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11623         * doc/tm.texi: Rebuild.
11624         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11625         target hook, rather than inspecting the insn_data.
11626         * config/sparc/sparc.c (sparc_cstore_mode): New function.
11627         (TARGET_CSTORE_MODE): Redefine.
11628         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11629         result patterns.
11630         * config/sparc/predicates.md (cstore_result_operand): New special
11631         predicate.
11632         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11633         Use it for operand 0.
11634         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11635         (*snesi_special): Likewise.
11636         (*snesi_zero): Likewise.
11637         (*seqsi_zero): Likewise.
11638         (*sltu_insn): Likewise.
11639         (*sgeu_insn): Likewise.
11640         (*seqdi_special): Make operand 0 and comparison operation be of
11641         DImode.
11642         (*snedi_special): Likewise.
11643         (*snedi_special_vis3): Likewise.
11644         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11645         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11646         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11647         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11648         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11649         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11650         (*sltu_extend_sp64): Likewise.
11651         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11652         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11653         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11654         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11655         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11657 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
11659         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11660         (aarch64_start_file): Use the new function.
11662 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11663             Jason Merrill  <jason@redhat.com>
11665         * common.opt: Add -gdwarf.
11666         * opts.c (common_handle_option): Handle it.
11667         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11669 2013-04-10  Richard Biener  <rguenther@suse.de>
11671         * passes.c (execute_todo): Do not call ggc_collect conditional here.
11672         (execute_one_ipa_transform_pass): But unconditionally here.
11673         (execute_one_pass): And here.
11674         (init_optimization_passes): Remove reload pass.
11675         * tree-pass.h (TODO_ggc_collect): Remove.
11676         (pass_reload): Likewise.
11677         * ira.c (do_reload): Merge into ...
11678         (ira): ... this.
11679         (rest_of_handle_reload): Remove.
11680         (pass_reload): Likewise.
11681         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11682         of reload for vzeroupper pass placement.
11683         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11684         and todo_flags_finish of all passes.
11686 2013-04-10  Richard Biener  <rguenther@suse.de>
11688         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11689         first_const_oprnd field, rename first_def_type to first_op_type.
11690         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11691         (vect_get_and_check_slp_defs): Always use the type of the
11692         operand.  Allow mixed vect_external_def, vect_constant_def types.
11693         (vect_get_constant_vectors): Handle mixed vect_external_def,
11694         vect_constant_def types.
11696 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
11698         PR tree-optimization/55524
11699         * tree-ssa-math-opts.c
11700         (convert_mult_to_fma): Don't use an fms construct
11701         when we don't have an fms operation, but fnma, and it looks
11702         likely that we'll be able to use the latter.
11704 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11706         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11707         function.
11708         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11709         inline fail caused by overwritable functions.
11711 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
11713         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11714         unnecessary bits in the constant power of two case.
11716 2013-04-10  Richard Biener  <rguenther@suse.de>
11718         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11719         broken code swapping operands.
11720         (vect_build_slp_tree): Do not compute load permutations here.
11721         (vect_analyze_slp_instance): Compute load permutations here,
11722         after building the SLP tree.
11724 2013-04-09  Christian Bruel  <christian.bruel@st.com>
11726         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11727         of next/prev_real_insn.
11729 2013-04-09  Jan Hubicka  <jh@suse.cz>
11731         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11732         Drop aliased parameter.
11733         (function_and_variable_visibility): Do not handle alias pairs.
11734         * cgraph.c (varpool_externally_visible_p): Update prototype.
11735         * varpool.c (varpool_add_new_variable): Update.
11737 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11739         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11741 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
11743         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11745         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11747 2013-04-09  Marek Polacek  <polacek@redhat.com>
11749         PR tree-optimization/48762
11750         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11752 2013-04-09  Richard Biener  <rguenther@suse.de>
11754         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11755         dealing with cost.
11756         (vect_build_slp_tree): Likewise.
11757         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11758         calculating the cost of a SLP instance.
11759         (vect_analyze_slp_instance): Use it from here, after building
11760         the SLP tree.
11762 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11764         PR middle-end/56883
11765         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11766         expand_omp_for_static_chunk): Use simple_p = true in
11767         force_gimple_operand_gsi calls when assigning to addressable decls.
11769 2013-04-09  Jeff Law  <law@redhat.com>
11771         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11772         when the boolean was created by converting a wider object which
11773         had a boolean range.
11775 2013-04-09  Richard Biener  <rguenther@suse.de>
11777         * tree-vectorizer.h (slp_void_p): Remove.
11778         (slp_tree): Typedef before _slp_tree declaration.
11779         (struct _slp_tree): Use a vector of slp_tree as children.
11780         (vect_get_place_in_interleaving_chain): Remove.
11781         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11782         Move ...
11783         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11784         and make static.
11785         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11786         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11787         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11788         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11789         Use slp_node instead of slp_void_p and adjust.
11791 2013-04-09  Richard Biener  <rguenther@suse.de>
11793         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11794         work that is not necessary.
11796 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11798         PR tree-optimization/56854
11799         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11800         forward into clobber stmts if it would change MEM_REF lhs into
11801         non-MEM_REF.
11803 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
11805         * tree.c (type_hash_lookup, type_hash_add): Make static.
11806         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11808 2013-04-09  Richard Biener  <rguenther@suse.de>
11810         * tree.h (unsave_expr_now): Remove.
11811         * tree-inline.c (mark_local_for_remap_r): Remove.
11812         (unsave_expr_1): Likewise.
11813         (unsave_r): Likewise.
11814         (unsave_expr_now): Likewise.
11815         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11816         (propagate_tree_value): Likewise.
11818 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
11820         * doc/rtl.texi (sequence): Rewrite documentation to match the
11821         current use of SEQUENCE rtl objects.
11822         * rtl.def (SEQUENCE): Likewise.
11824         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11825         Update documentation.
11826         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11827         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11829         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11831 2013-04-08  Teresa Johnson  <tejohnson@google.com>
11833         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11834         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11835         methods.
11836         (estimate_edge_size_and_time): Add comment to suggest using rounding
11837         methods.
11838         (estimate_node_size_and_time): Ditto.
11839         (remap_edge_change_prob): Use helper rounding divide methods.
11840         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11841         (gimple_mod_pow2_value_transform): Ditto.
11842         (gimple_mod_subtract_transform): Ditto.
11843         (gimple_ic_transform): Ditto.
11844         (gimple_stringops_transform): Ditto.
11845         * stmt.c (conditional_probability): Ditto.
11846         (emit_case_dispatch_table): Ditto.
11847         * lto-cgraph.c (merge_profile_summaries): Ditto.
11848         * tree-optimize.c (execute_fixup_cfg): Ditto.
11849         * cfgcleanup.c (try_forward_edges): Ditto.
11850         * cfgloopmanip.c (scale_loop_profile): Ditto.
11851         (loopify): Ditto.
11852         (duplicate_loop_to_header_edge): Ditto.
11853         (lv_adjust_loop_entry_edge): Ditto.
11854         * tree-vect-loop.c (vect_transform_loop): Ditto.
11855         * profile.c (compute_branch_probabilities): Ditto.
11856         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11857         * lto-streamer-in.c (input_cfg): Ditto.
11858         * gimple-streamer-in.c (input_bb): Ditto.
11859         * ipa-cp.c (update_profiling_info): Ditto.
11860         (update_specialized_profile): Ditto.
11861         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11862         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11863         rounding methods.
11864         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11865         (compute_trg_info): Ditto.
11866         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11867         (purge_dead_edges): Ditto.
11868         * loop-unswitch.c (unswitch_loop): Ditto.
11869         * cgraphclones.c (cgraph_clone_edge): Ditto.
11870         (cgraph_clone_node): Ditto.
11871         * tree-inline.c (copy_bb): Ditto.
11872         (copy_edges_for_bb): Ditto.
11873         (initialize_cfun): Ditto.
11874         (copy_cfg_body): Ditto.
11875         (expand_call_inline): Ditto.
11877 2013-04-08  Kai Tietz  <ktietz@redhat.com>
11879         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11880         TARGET_CYGWIN64 by TARGET_64BIT.
11882 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11884         * config/epiphany/epiphany.md (GPR_1): New constant.
11885         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11886         * config/epiphany/epiphany.c (gen_compare_reg):
11887         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11888         is already in place.
11889         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11890         Don't require being called during rtl expansion; If y operlaps r0,
11891         return 0.
11892         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11893         (epiphany_expand_epilogue): Likewise.
11895         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11896         Don't use CC_FPmode for ORDERED / UNORDERED.
11897         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11899         * config/epiphany/constraints.md (CnL): New constraint.
11900         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11901         * config/epiphany/predicates.md (add_operand): Allow 1024.
11903         * config/epiphany/epiphany.md (logical_op): New code iterator.
11904         (op_mnc): New code attribute.
11905         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11906         (mov_f+1, mov_f+2): New peephole2 patterns.
11908         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11909         (cstoresi4): Also allow re-use of zero result when doing a NE
11910         comparison to a non-zero operand.
11911         Use (clobber (scratch)) for first insn if the gpr output is not needed.
11913         * config/epiphany/epiphany.md (<insn_opname>v2si3):
11914         Use gen_addsi3_i / gen_subsi3_i.
11916 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11918         PR c++/34949
11919         PR c++/50243
11920         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11921         contain anything but clobbers, at most one __builtin_stack_restore,
11922         optionally debug stmts and final resx, and if it has at least one
11923         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
11924         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11925         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11926         which isn't defaut definition, remove them.
11927         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11928         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11929         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11930         with MEM_REF LHS with SSA_NAME address.
11932 2013-04-08  Jeff Law  <law@redhat.com>
11934         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11936 2013-04-08  Richard Biener  <rguenther@suse.de>
11938         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11939         extra newline.
11940         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11941         determined vector type.
11942         (vect_analyze_data_refs): Likewise.
11943         (vect_get_new_vect_var): Adjust.
11944         (vect_create_destination_var): Preserve SSA name versions.
11945         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11946         not dump anything here.
11948 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11950         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11951         Add member lr_slot_known.
11952         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11953         if necessary.
11954         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11955         Remove code that sets lr_slot_offset according to what a previous
11956         version of epiphany_emit_save_restore used to do.
11957         (epiphany_emit_save_restore): When doing an lr save or restore,
11958         set/verify lr_slot_known and lr_slot_offset.
11960 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
11962         PR target/54338
11963         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11964         in ALL_REGS.
11966 2013-04-08  Richard Biener  <rguenther@suse.de>
11968         * alias.c (find_base_term): Fix thinko in previous change.
11970 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11972         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11973         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11974         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
11975         if possible to compute val.
11976         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11977         For QImode integers don't require anything about precision.  Use
11978         const_with_all_bytes_same to find out if the constant doesn't have
11979         repeated bytes in it.
11981 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11983         * config/s390/s390.c (s390_expand_insv): Only accept insertions
11984         within mode size.
11986 2013-04-08  Marek Polacek  <polacek@redhat.com>
11988         PR rtl-optimization/48182
11989         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11990         value to 1.
11992 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11994         PR target/55487
11995         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11996         nuses, make sure we have a label.
11998 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12000         PR target/56843
12001         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12002         (rs6000_emit_swdiv_low_precision): Remove.
12003         (rs6000_emit_swdiv): Rewrite to handle between one and four
12004         iterations of Newton-Raphson generally; modify required number of
12005         iterations for some cases.
12006         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12008 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
12010         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12011         set-but-unused variable.
12013         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12014         basic blocks of released function bodies garbage-collectable.
12016         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12017         (struct rtl_opt_pass): Add TODO_df_finish.
12019         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12021 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12023         * config/arm/constraints.md (q): New constraint.
12024         * config/arm/ldrdstrd.md: New file.
12025         * config/arm/arm.md (ldrdstrd.md) New include.
12026         (arm_movdi): Use "q" instead of "r" constraint
12027         for double-word memory access.
12028         (movdf_soft_insn): Likewise.
12029         * config/arm/vfp.md (movdi_vfp): Likewise.
12030         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12031         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12032         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12033         (mem_ok_for_ldrd_strd): Likewise.
12034         (output_move_double): Update assertion.
12036 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12038         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12040 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12042         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12043         define_insn_and_split.
12044         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12046 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12048         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12049         define_insn_and_split.
12050         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12051         (shiftsi3_compare): New pattern.
12052         (rrx): New pattern.
12053         * config/arm/unspecs.md (UNSPEC_RRX): New.
12055 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12057         * config/arm/arm.md (negdi_extendsidi): New pattern.
12058         (negdi_zero_extendsidi): Likewise.
12060 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12062         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12063         define_insn_and_split.
12064         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12065         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12067 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12069         * config/arm/arm.md (arm_subdi3): Convert define_insn into
12070         define_insn_and_split.
12071         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12072         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12074 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12076         * config/arm/arm.md (subsi3_carryin): New pattern.
12077         (subsi3_carryin_const): Likewise.
12078         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12079         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12081 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12083         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12085 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12087         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12088         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12090 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12092         * config/arm/arm.c (arm_expand_builtin): Change fcode
12093         type to unsigned int.
12095 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12097         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12099 2013-04-04  Ian Lance Taylor  <iant@google.com>
12101         * doc/standards.texi (Standards): The Go frontend supports the Go 1
12102         language standard.
12104 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
12106         PR middle-end/56729
12107         * df-scan.c (df_insn_delete): Disable failing assert.
12109 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12111         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12112         New function prototype.
12113         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12114         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12115         (arm_builtin_vectorized_function): New function.
12117 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12119         * config/arm/arm_neon_builtins.def: New file.
12120         * config/arm/arm.c (neon_builtin_data): Move contents to
12121         arm_neon_builtins.def.
12122         (enum arm_builtins): Include neon builtin definitions.
12123         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12124         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12126 2013-04-04  Marek Polacek  <polacek@redhat.com>
12128         PR tree-optimization/48186
12129         * predict.c (maybe_hot_frequency_p): Return false if
12130         HOT_BB_FREQUENCY_FRACTION is 0.
12131         (cgraph_maybe_hot_edge_p): Likewise.
12133 2013-04-04  Richard Biener  <rguenther@suse.de>
12135         PR tree-optimization/56826
12136         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12137         more accurately.
12139 2013-04-04  Richard Biener  <rguenther@suse.de>
12141         PR tree-optimization/56213
12142         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12143         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12145 2013-04-04  Richard Biener  <rguenther@suse.de>
12147         PR tree-optimization/56837
12148         * tree-loop-distribution.c (classify_partition): For non-zero
12149         values require that the value has the same precision as its
12150         mode to be useful as memset value.
12152 2013-04-03  Nick Clifton  <nickc@redhat.com>
12154         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12155         (fmssf4): Use fmsf.s on E3V5 architectures.
12156         (fnmasf4): Use fnmaf.s on E3V5 architectures.
12157         (fnmssf4): Use fnmsf.s on E3V5 architectures.
12159 2013-04-03  Jeff Law  <law@redhat.com>
12161         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12162         (lra-eliminations.o): Likewise.
12164 2013-04-03  Teresa Johnson  <tejohnson@google.com>
12166         * gcov-io.c (compute_working_sets): Moved most of body of old
12167         compute_working_sets here from profile.c.
12168         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12169         (gcov_working_set_t): Moved typedef here from basic-block.h
12170         (compute_working_set): Declare.
12171         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12172         (get_working_sets): Renamed from compute_working_set,
12173         replace most of body with call to new compute_working_sets.
12174         (get_exec_counts): Replace call to compute_working_sets
12175         to get_working_sets.
12176         * profile.h (get_working_sets): Renamed from compute_working_set.
12177         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12178         to get_working_sets.
12179         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12180         * gcov-dump.c (dump_working_sets): New function.
12182 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
12184         * hwint.c (sext_hwi, zext_hwi): New functions.
12185         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12186         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12187         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12188         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12189         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12190         (sext_hwi, zext_hwi): New functions.
12192 2013-04-03  Jeff Law  <law@redhat.com>
12194         PR tree-optimization/56799
12195         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12196         back test for widening conversion erroneously dropped in prior change.
12198 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12200         PR target/56809
12201         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12202         instead of next_real_insn.
12204 2013-04-03  Marek Polacek  <polacek@redhat.com>
12206         PR sanitizer/55702
12207         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12209 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12211         PR target/56809
12212         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12213         next_real_insn.
12214         (thumb1_output_casesi): Likewise.
12215         (thumb2_output_casesi): Likewise.
12217 2013-04-03  Richard Biener  <rguenther@suse.de>
12219         PR tree-optimization/56817
12220         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12221         Split out ...
12222         (tree_unroll_loops_completely_1): ... new function to manually
12223         walk the loop tree, properly defering outer loops of unrolled
12224         loops to later iterations.
12226 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12228         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12229         (vectorizable_load): Likewise.
12230         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12231         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12233 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12235         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12236         BIT_FIELD_REF.
12238 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12240         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12242 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
12244         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12246 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12248         PR tree-optimization/56790
12249         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12250         folding.
12252 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12254         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12255         Handle VEC_MERGE.
12256         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12257         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12258         equal arguments.
12260 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
12262         PR c/19449
12263         * tree.h (force_folding_builtin_constant_p): New decl.
12264         * builtins.c (force_folding_builtin_constant_p): New variable.
12265         (fold_builtin_constant_p): Fold immediately also if
12266         force_folding_builtin_constant_p.
12268 2013-04-03  Richard Biener  <rguenther@suse.de>
12270         PR tree-optimization/56812
12271         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12272         DRs of the same interleaving chain are independent.
12274 2013-04-02  Jason Merrill  <jason@redhat.com>
12276         * gdbinit.in (pbb): Use debug fn.
12278 2013-04-02  Lawrence Crowl  <crowl@google.com>
12280         * sese.h (struct ivtype_map_elt_s): Remove unused.
12281         (extern debug_ivtype_map): Remove unused.
12282         (extern eq_ivtype_map_elts): Remove unused.
12283         * sese.c (debug_ivtype_map): Removed unused.
12284         (debug_ivtype_map_1): Removed unused.
12285         (debug_ivtype_elt): Remove unused.
12286         (eq_ivtype_map_elts): Remove unused.
12289 2013-04-02  Kai Tietz  <ktietz@redhat.com>
12291         PR target/52790
12292         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12293         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12294         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12295         function.
12296         (legitimize_pe_coff_symbol): Likewise.
12297         (is_imported_p): New helper-function.
12298         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12299         for Windows x64 targets.
12300         (ix86_expand_prologue): Optimize for pe-coff targets.
12301         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12302         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12303         medium/large code-model.
12304         (legitimize_pic_address): Likewise.
12305         (legitimize_tls_address): Likewise.
12306         (ix86_expand_call): Likewise.
12307         (x86_output_mi_thunk): Likewise.
12308         (get_dllimport_decl): Add new beimport argument.
12309         (construct_plt_address): Don't assert for x64 pe-coff targets.
12310         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12311         targets.
12312         (SYMBOL_FLAG_STUBVAR): New macro.
12313         (SYMBOL_REF_STUBVAR_P): Likewise.
12314         * config/i386/winnt.c (stub_list): New structure.
12315         (stub_head): New local variable.
12316         (i386_pe_record_stub): New function.
12317         (i386_pe_file_end): Emit refptr-stubs.
12319 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
12321         PR rtl-optimization/56745
12322         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12323         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12325         PR c++/34949
12326         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12327         and both of them are MEM_REFs, just compare first argument for
12328         equality and attempt to deal even with differing offsets.
12330         PR c++/34949
12331         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12332         of gimple_clobber_p to be MEM_REF.
12333         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12334         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
12335         after gimplification.
12336         * asan.c (get_mem_ref_of_assignment): Don't instrument
12337         gimple_clobber_p stmts.
12338         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12339         gimple_clobber_p stmt if they have MEM_REF lhs and
12340         are dead because of another gimple_clobber_p stmt.
12341         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12342         gimple_clobber_p stmts like debug stmts.
12343         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12344         that refer to unused VAR_DECLs or uninitialized values.
12345         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12346         gimple_clobber_p stmts if they refer to removed parameters.
12347         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12348         formatting.
12350 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12352         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12353         using SWI48 mode attribute.
12355 2013-04-02  Wei Mi  <wmi@google.com>
12357         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12358         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12359         *<rotate_insn><mode>3_mask in i386.md.
12361 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12363         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12365 2013-04-02  Richard Biener  <rguenther@suse.de>
12367         PR tree-optimization/56778
12368         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12369         Runtime alias tests are not supported for gather loads.
12370         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12371         stmts referenced from SSA operands before updating SSA form.
12373 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12374             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12376         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12377         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12378         * config/arm/cortex-a53.md: New file.
12379         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12380         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12381         * config/arm/arm.c (arm_issue_rate): Likewise.
12382         * config/arm/arm-tune.md: Regenerate
12383         * config/arm/arm-tables.opt: Regenerate.
12384         * config/arm/arm-cores.def: Add cortex-a53.
12386 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12388         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12389         non-static link.
12391 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12393         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12394         scalar load/store operations using B/H registers.
12395         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12397 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12399         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12400         scalar move.
12401         * config/aarch64/aarch64.c
12402         (aarch64_simd_scalar_immediate_valid_for_move): New.
12403         * config/aarch64/aarch64-protos.h
12404         (aarch64_simd_scalar_immediate_valid_for_move): New.
12405         * config/aarch64/constraints.md (Dh, Dq): New.
12406         * config/aarch64/iterators.md (hq): New.
12408 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12410         * reorg.c (get_branch_condition): Deal with conditional returns.
12411         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12413 2013-04-01  Wei Mi  <wmi@google.com>
12415         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12416         Truncate operand 2 using %b asm operand modifier.
12417         (*<shift_insn><mode>3_mask): Ditto.
12418         (*<rotate_insn><mode>3_mask): Ditto.
12420 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12422         PR middle-end/56798
12423         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12425 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12427         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12428         of next_real_insn.
12429         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12431 2013-03-30  Lawrence Crowl  <crowl@google.com>
12433         * dse.c (clear_alias_sets): Remove never set.
12434         (disqualified_clear_alias_sets): Remove never set.
12435         (clear_alias_mode_pool): Remove never set.
12436         (dse_step0): Remove condition that is never true.
12437         (canon_address): Remove condition that is never true.
12438         (dse_step7): Remove condition that is never true.
12439         (rest_of_handle_dse): Remove condition that is never true.
12440         (rest_of_handle_dse::did_global): Remove never read from above.
12441         (dse_step2_spill): Remove never called from above.
12442         (dse_step5_spill): Remove never called from above.
12444 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12446         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12447         JUMP_TABLE_DATA changes.
12448         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12449         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12450         (Insns) <jump_table_data>: New entry.
12451         * doc/tm.texi: Regenerate.
12453         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12455         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12456         for table jump at the end of a basic block using tablejump_p.
12457         * targhooks.c (default_invalid_within_doloop): Likewise.
12458         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12459         target hook implementation that is identical to the default hook.
12460         (rs6000_invalid_within_doloop): Remove.
12462         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12463         unused variable from tablejump_p call.
12465         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12466         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12467         (INSN_DELETED_P): Likewise.
12468         (emit_jump_table_data): New prototype.
12469         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12470         after 4th as unused.
12471         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12472         * sched-vis.c (print_insn): Likewise.
12473         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12474         insn for compatibility with back ends that use next_active_insn to
12475         identify jump table data.
12476         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12477         (remove_insn): Likewise.
12478         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12479         to be emitted.
12480         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12481         (emit_jump_table_data): New function.
12483         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12484         basic block, a JUMP_TABLE_DATA never is.
12485         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12486         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12487         off from code handling real insns.
12488         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12489         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12490         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12491         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12492         is not a NONDEBUG_INSN_P.
12493         * ira-costs.c (scan_one_insn): Likewise.
12494         * jump.c (mark_all_labels): Likewise.
12495         (mark_jump_label_1): Likewise.
12496         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12497         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12498         a basic block.
12499         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12500         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12501         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12502         JUMP_TABLE_DATA_P insns.
12503         (calculate_elim_costs_all_insns): Likewise.
12504         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12505         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12506         (delete_output_reload): Code style fixups.
12507         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12508         insn flags on this non-insn.
12509         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12510         as scheduling barriers, for pre-change compatibility.
12511         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12512         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12514         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12515         redundant JUMP_TABLE_DATA_P test.
12516         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12517         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12518         (frv_for_each_packet): Likewise.
12519         * config/i386/i386.c (min_insn_size): Likewise.
12520         (ix86_avoid_jump_mispredicts): Likewise.
12521         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12522         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12523         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12524         (mips16_insn_length): Robustify.
12525         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12526         (mips16_split_long_branches): Likewise.
12527         * config/pa/pa.c (pa_combine_instructions): Likewise.
12528         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12529         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12530         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12531         as contributing to pool range lengths.
12532         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12533         Remove redundant JUMP_TABLE_DATA_P test.
12534         (sh_loop_align): Likewise.
12535         (split_branches): Likewise.
12536         (sh_insn_length_adjustment): Likewise.
12537         * config/spu/spu.c (get_branch_target): Likewise.
12539 2013-03-29  Jan Hubicka  <jh@suse.cz>
12541         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12542         gcov streaming; stream hot bb threshold to ltrans.
12543         * predict.c (get_hot_bb_threshold): Break out from ....
12544         (maybe_hot_count_p): ... here.
12545         (set_hot_bb_threshold): New function.
12546         * lto-section-in.c (lto_section_name): Add profile.
12547         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12548         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12549         and data-streamer.h
12550         (histogram_entry): New structure.
12551         (histogram, histogram_pool): New global vars.
12552         (histogram_hash): New structure.
12553         (histogram_hash::hash): New method.
12554         (histogram_hash::equal): Likewise.
12555         (account_time_size): New function.
12556         (cmp_counts): New function.
12557         (dump_histogram): New function.
12558         (ipa_profile_generate_summary): New function.
12559         (ipa_profile_write_summary): New function.
12560         (ipa_profile_read_summary): New function.
12561         (ipa_profile): Decide on threshold.
12562         (pass_ipa_profile): Add ipa_profile_write_summary and
12563         ipa_profile_read_summary.
12564         * Makefile.in (ipa.o): Update dependencies.
12565         * lto-streamer.h (LTO_section_ipa_profile): New section.
12567 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12569         * tree.h (VAR_P): New.
12571 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12573         PR lto/56777
12574         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12576 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12578         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12579         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12580         (control_flow_insn_p): Likewise.
12581         * cfgrtl.c (duplicate_insn_chain): Likewise.
12582         * final.c (get_attr_length_1): Likewise.
12583         (shorten_branches): Likewise.
12584         (final_scan_insn): Likewise.
12585         * function.c (instantiate_virtual_regs): Likewise.
12586         * gcse.c (insert_insn_end_basic_block): Likewise.
12587         * ira-costs.c (scan_one_insn): Likewise.
12588         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12589         * lra.c (check_rtl): Likewise.
12590         * reload1.c (elimination_costs_in_insn): Likewise.
12591         * reorg.c (follow_jumps): Likewise.
12593         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12594         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12595         (thumb_far_jump_used_p): Likewise.
12596         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12597         (workaround_speculation): Likewise.
12598         (add_sched_insns_for_speculation): Likewise.
12599         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12600         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12601         (frv_for_each_packet): Likewise.
12602         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12603         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12604         (final_emit_insn_group_barriers): Likewise.
12605         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12606         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12607         (mips16_insn_length): Likewise.
12608         * config/pa/pa.c (pa_reorg): Likewise.
12609         (pa_combine_instructions): Likewise.
12610         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12611         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12612         (sh_reorg): Likewise.
12613         (split_branches): Likewise.
12614         * config/spu/spu.c (get_branch_target): Likewise.
12616         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12617         JUMP_TABLE_DATA_P.
12619 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
12621         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12622         Fix declaration name.
12624 2013-03-28  Lawrence Crowl  <crowl@google.com>
12626         * graphds.h (struct graph.indicies): Remove unused.
12627         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12628         (SCOP_ORIGINAL_PDDRS): Remove unused.
12629         * sese.h (extern insert_loop_close_phis): Removed unused.
12630         (extern insert_guard_phis): Removed unused.
12631         (extern ivtype_map_elt_info): Removed unused.
12632         (new_ivtype_map_elt): Removed unused.
12633         * sese.c (ivtype_map_elt_info): Removed unused.
12635 2013-03-28  Lawrence Crowl  <crowl@google.com>
12637         * Makefile.in: Add several missing include dependences.
12638         (DUMPFILE_H): New.
12639         (test-dump.o): New.  This object is not added to any executable,
12640         but is present for ad-hoc testing.
12641         * bitmap.c
12642         (debug (const bitmap_head_def &)): New.
12643         (debug (const bitmap_head_def *)): New.
12644         * bitmap.h
12645         (extern debug (const bitmap_head_def &)): New.
12646         (extern debug (const bitmap_head_def *)): New.
12647         * cfg.c
12648         (debug (edge_def &)): New.
12649         (debug (edge_def *)): New.
12650         * cfghooks.c
12651         (debug (basic_block_def &)): New.
12652         (debug (basic_block_def *)): New.
12653         * dumpfile.h
12654         (dump_node (const_tree, int, FILE *)): Correct source file.
12655         * dwarf2out.c
12656         (debug (die_struct &)): New.
12657         (debug (die_struct *)): New.
12658         * dwarf2out.h
12659         (extern debug (die_struct &)): New.
12660         (extern debug (die_struct *)): New.
12661         * gimple-pretty-print.c
12662         (debug (gimple_statement_d &)): New.
12663         (debug (gimple_statement_d *)): New.
12664         * gimple-pretty-print.h
12665         (extern debug (gimple_statement_d &)): New.
12666         (extern debug (gimple_statement_d *)): New.
12667         * ira-build.c
12668         (debug (ira_allocno_copy &)): New.
12669         (debug (ira_allocno_copy *)): New.
12670         (debug (ira_allocno &)): New.
12671         (debug (ira_allocno *)): New.
12672         * ira-int.h
12673         (extern debug (ira_allocno_copy &)): New.
12674         (extern debug (ira_allocno_copy *)): New.
12675         (extern debug (ira_allocno &)): New.
12676         (extern debug (ira_allocno *)): New.
12677         * ira-lives.c
12678         (debug (live_range &)): New.
12679         (debug (live_range *)): New.
12680         * lra-int.h
12681         (debug (lra_live_range &)): New.
12682         (debug (lra_live_range *)): New.
12683         * lra-lives.c
12684         (debug (lra_live_range &)): New.
12685         (debug (lra_live_range *)): New.
12686         * omega.c
12687         (debug (omega_pb_d &)): New.
12688         (debug (omega_pb_d *)): New.
12689         * omega.h
12690         (extern debug (omega_pb_d &)): New.
12691         (extern debug (omega_pb_d *)): New.
12692         * print-rtl.c
12693         (debug (const rtx_def &)): New.
12694         (debug (const rtx_def *)): New.
12695         * print-tree.c
12696         (debug_tree (tree): Move within file.
12697         (debug_raw (const tree_node &)): New.
12698         (debug_raw (const tree_node *)): New.
12699         (dump_tree_via_hooks (const tree_node *, int)): New.
12700         (debug (const tree_node &)): New.
12701         (debug (const tree_node *)): New.
12702         (debug_verbose (const tree_node &)): New.
12703         (debug_verbose (const tree_node *)): New.
12704         (debug_head (const tree_node &)): New.
12705         (debug_head (const tree_node *)): New.
12706         (debug_body (const tree_node &)): New.
12707         (debug_body (const tree_node *)): New.
12708         (debug_vec_tree (tree): Move and reimplement in terms of dump.
12709         (debug (vec<tree, va_gc> &)): New.
12710         (debug (vec<tree, va_gc> *)): New.
12711         * rtl.h
12712         (extern debug (const rtx_def &)): New.
12713         (extern debug (const rtx_def *)): New.
12714         * sbitmap.c
12715         (debug_raw (simple_bitmap_def &)): New.
12716         (debug_raw (simple_bitmap_def *)): New.
12717         (debug (simple_bitmap_def &)): New.
12718         (debug (simple_bitmap_def *)): New.
12719         * sbitmap.h
12720         (extern debug (simple_bitmap_def &)): New.
12721         (extern debug (simple_bitmap_def *)): New.
12722         (extern debug_raw (simple_bitmap_def &)): New.
12723         (extern debug_raw (simple_bitmap_def *)): New.
12724         * sel-sched-dump.c
12725         (debug (vinsn_def &)): New.
12726         (debug (vinsn_def *)): New.
12727         (debug_verbose (vinsn_def &)): New.
12728         (debug_verbose (vinsn_def *)): New.
12729         (debug (expr_def &)): New.
12730         (debug (expr_def *)): New.
12731         (debug_verbose (expr_def &)): New.
12732         (debug_verbose (expr_def *)): New.
12733         (debug (vec<rtx> &)): New.
12734         (debug (vec<rtx> *)): New.
12735         * sel-sched-dump.h
12736         (extern debug (vinsn_def &)): New.
12737         (extern debug (vinsn_def *)): New.
12738         (extern debug_verbose (vinsn_def &)): New.
12739         (extern debug_verbose (vinsn_def *)): New.
12740         (extern debug (expr_def &)): New.
12741         (extern debug (expr_def *)): New.
12742         (extern debug_verbose (expr_def &)): New.
12743         (extern debug_verbose (expr_def *)): New.
12744         (extern debug (vec<rtx> &)): New.
12745         (extern debug (vec<rtx> *)): New.
12746         * sel-sched-ir.h
12747         (_list_iter_cond_expr): Make inline instead of static.
12748         * sreal.c
12749         (debug (sreal &)): New.
12750         (debug (sreal *)): New.
12751         * sreal.h
12752         (extern debug (sreal &)): New.
12753         (extern debug (sreal *)): New.
12754         * tree.h
12755         (extern debug_raw (const tree_node &)): New.
12756         (extern debug_raw (const tree_node *)): New.
12757         (extern debug (const tree_node &)): New.
12758         (extern debug (const tree_node *)): New.
12759         (extern debug_verbose (const tree_node &)): New.
12760         (extern debug_verbose (const tree_node *)): New.
12761         (extern debug_head (const tree_node &)): New.
12762         (extern debug_head (const tree_node *)): New.
12763         (extern debug_body (const tree_node &)): New.
12764         (extern debug_body (const tree_node *)): New.
12765         (extern debug (vec<tree, va_gc> &)): New.
12766         (extern debug (vec<tree, va_gc> *)): New.
12767         * tree-cfg.c
12768         (debug (struct loop &)): New.
12769         (debug (struct loop *)): New.
12770         (debug_verbose (struct loop &)): New.
12771         (debug_verbose (struct loop *)): New.
12772         * tree-dump.c: Add header dependence.
12773         * tree-flow.h
12774         (extern debug (struct loop &)): New.
12775         (extern debug (struct loop *)): New.
12776         (extern debug_verbose (struct loop &)): New.
12777         (extern debug_verbose (struct loop *)): New.
12778         * tree-data-ref.c
12779         (debug (data_reference &)): New.
12780         (debug (data_reference *)): New.
12781         (debug (vec<data_reference_p> &)): New.
12782         (debug (vec<data_reference_p> *)): New.
12783         (debug (vec<ddr_p> &)): New.
12784         (debug (vec<ddr_p> *)): New.
12785         * tree-data-ref.h
12786         (extern debug (data_reference &)): New.
12787         (extern debug (data_reference *)): New.
12788         (extern debug (vec<data_reference_p> &)): New.
12789         (extern debug (vec<data_reference_p> *)): New.
12790         (extern debug (vec<ddr_p> &)): New.
12791         (extern debug (vec<ddr_p> *)): New.
12792         * tree-ssa-alias.c
12793         (debug (pt_solution &)): New.
12794         (debug (pt_solution *)): New.
12795         * tree-ssa-alias.h
12796         (extern debug (pt_solution &)): New.
12797         (extern debug (pt_solution *)): New.
12798         * tree-ssa-alias.c
12799         (debug (_var_map &)): New.
12800         (debug (_var_map *)): New.
12801         (debug (tree_live_info_d &)): New.
12802         (debug (tree_live_info_d *)): New.
12803         * tree-ssa-alias.h
12804         (extern debug (_var_map &)): New.
12805         (extern debug (_var_map *)): New.
12806         (extern debug (tree_live_info_d &)): New.
12807         (extern debug (tree_live_info_d *)): New.
12809 2013-03-28  Jan Hubicka  <jh@suse.cz>
12811         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12813 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
12815         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12816         record only when desired or required.
12818 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
12820         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12821         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
12823 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12825         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12826         (*andsi3_compare0_uxtw): New pattern.
12827         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12828         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12830 2013-03-28  Jan Hubicka  <jh@suse.cz>
12832         * data-streamer-in.c (streamer_read_gcov_count): New function.
12833         * gimple-streamer-out.c: Include value-prof.h.
12834         (output_gimple_stmt): Output histogram.
12835         (output_bb): Use streamer_write_gcov_count.
12836         * value-prof.c: Include data-streamer.h
12837         (dump_histogram_value): Add HIST_TYPE_MAX.
12838         (stream_out_histogram_value): New function.
12839         (stream_in_histogram_value): New function.
12840         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12841         (stream_out_histogram_value, stream_in_histogram_value): Declare.
12842         * data-streamer-out.c (streamer_write_gcov_count): New function.
12843         (streamer_write_gcov_count_stream): New function.
12844         * lto-cgraph.c (lto_output_edge): Update counter streaming.
12845         (lto_output_node): Likewise.
12846         (input_node, input_edge): Likewise.
12847         * lto-streamer-out.c (output_cfg): Update streaming.
12848         * lto-streamer-in.c (input_cfg): Likewise.
12849         * data-streamer.h (streamer_write_gcov_count,
12850         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12851         * gimple-streamer-in.c: Include value-prof.h
12852         (input_gimple_stmt): Input histograms.
12853         (input_bb): Update profile streaming.
12855 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12857         * genmodes.c (emit_max_int): New function.
12858         (emit_insn_modes_h): Added call to emit_max_function.
12859         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12860         Added doc.
12861         * machmode.def: Fixed comment.
12863 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12865         * combine.c (try_combine): Removed useless assert.
12866         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12868 2013-03-28  Marek Polacek  <polacek@redhat.com>
12869             Richard Biener  <rguenther@suse.de>
12871         PR tree-optimization/56695
12872         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12873         build signed result of a vector comparison.
12874         * tree-cfg.c (verify_gimple_comparison): Check that a result
12875         of a vector comparison has signed type.
12877 2013-03-28  Richard Biener  <rguenther@suse.de>
12879         PR tree-optimization/37021
12880         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12881         do not restrict gaps between groups.
12882         * tree-vect-stmts.c (vectorizable_load): Properly account for
12883         a gap between groups.
12885 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
12887         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12888         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12889         is not enabled.
12891 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
12893         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12894         * doc/extend.texi (Named Address Spaces): Ditto.
12895         (Variable Attributes): Ditto.
12897 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12899         * config.build: Add support for cygwin x64 target.
12900         * config.gcc: Likewise.
12901         * config.host: Likewise.
12902         * configure.ac: Likewise
12903         * configure: Regenerated.
12905 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12907         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12908         * config/i386/t-cygwin-w64: New file.
12909         * config/i386/cygwin-w64.h: New file.
12910         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12911         and add support for x64-cygwin target.
12912         (CPP_SPEC): Likewise.
12913         (CXX_WRAP_SPEC_LIST): Undefine before define.
12914         (LIBGCJ_SONAME): Use 15 as version.
12916 2013-03-27  Richard Biener  <rguenther@suse.de>
12918         PR tree-optimization/56716
12919         * tree-ssa-structalias.c (perform_var_substitution): Adjust
12920         dumping for ref nodes.
12922 2013-03-27  Martin Jambor  <mjambor@suse.cz>
12924         PR tree-optimization/55334
12925         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12926         restricted pointers to arrays.
12928 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12930         * Makefile.in (.SUFFIXES): Add .cc.
12931         (.c.o): Apply same recipe for implicit rule .cc.o.
12933 2013-03-27  Richard Biener  <rguenther@suse.de>
12935         PR tree-optimization/37021
12936         * tree-vect-data-refs.c (vect_check_strided_load): Allow
12937         REALPART/IMAGPART_EXPRs around the supported refs.
12938         * tree-ssa-structalias.c (find_func_aliases): Assume that
12939         floating-point values are not used to transfer pointers.
12941 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12943         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
12944         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12945         * gcc/doc/tm.texi: Regenerate.
12946         * gcc/targhooks.h (default_has_ifunc_p): New.
12947         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
12948         * gcc/config/linux-protos.h: New file.
12949         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12950         this hook for linux which disables support of indirect functions in
12951         android.
12952         * gcc/config/linux-android.c: New file.
12953         * gcc/config/t-linux-android.c: Ditto.
12954         * gcc/config.gcc: Added new object file linux-android.o.
12955         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
12956         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12957         * gcc/varasm.c (do_assemble_alias): Likewise.
12958         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12959         doesn't support indirect functions.
12960         * configure: Regenerate.
12962 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
12964         PR target/56102
12965         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12966         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12967         mult-word mode.
12969 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12971         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12973 2013-03-27  Terry Guo  <terry.guo@arm.com>
12975         * config/arm/arm-cores.def: Added core cortex-r7.
12976         * config/arm/arm-tune.md: Regenerated.
12977         * config/arm/arm-tables.opt: Regenerated.
12978         * doc/invoke.texi: Added entry for core cortex-r7.
12980 2013-03-27  Walter Lee  <walt@tilera.com>
12982         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12983         double-decrement of next_scratch_regno.
12985 2013-03-27  Walter Lee  <walt@tilera.com>
12987         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12988         input operands.
12989         (insn_v1mulus): Ditto.
12990         (insn_v2muls): Ditto.
12992 2013-03-27  Walter Lee  <walt@tilera.com>
12994         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12995         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12997 2013-03-27  Walter Lee  <walt@tilera.com>
12999         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13000         (*sibcall_value): Ditto.
13002 2013-03-27  Walter Lee  <walt@tilera.com>
13004         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13005         (insn_mnz_v8qi): ... this ...
13006         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
13007         vector equivalent.
13008         (insn_v<n>mnz): Replaced by ...
13009         (insn_v1mnz): ... this ...
13010         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
13011         equivalent.
13012         (insn_mz_<mode>): Replaced by ...
13013         (insn_mz_v8qi): ... this ...
13014         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
13015         vector equivalent.
13016         (insn_v<n>mz): Replaced by ...
13017         (insn_v1mz): ... this ...
13018         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
13019         equivalent.
13021 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
13023         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13025 2013-03-26  Roland McGrath  <mcgrathr@google.com>
13027         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13028         than fprintf with a non-constant, non-format string.
13030 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13032         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13033         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
13034         operand 0 predicate.
13035         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13036         attribute.  Use general_x64nomem_operand as operand 1 predicate.
13037         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13038         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13039         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13040         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13041         (mov<mode>_insv_1): Remove expander.  Merge insn with
13042         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13043         Use general_x64nomem_operand as operand 1 predicate.
13044         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13045         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13046         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13047         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13048         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
13049         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13050         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13051         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13052         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
13053         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13054         (general_x64nomem_operand): Ditto.
13056 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13058         * config/rtems.opt: Add -pthread option.
13060 2013-03-26  Richard Biener  <rguenther@suse.de>
13062         * alias.c (find_base_term): Avoid redundant and not used recursion.
13063         (base_alias_check): Get the initial base term from the caller.
13064         (true_dependence_1): Compute and pass base terms to base_alias_check.
13065         (write_dependence_p): Likewise.
13066         (may_alias_p): Likewise.
13068 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
13070         * config/aarch64/aarch64.c (aarch64_classify_address): Support
13071         PC-relative load in SI modes and above only.
13073 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
13075         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13076         * config/arm/iwmmxt.md (WCGR0): Update.
13077         (WCGR1, WCGR2, WCGR3): Likewise.
13079 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13081         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13082         Use x64 and nox64 isa attributes.
13084 2013-03-26  Richard Biener  <rguenther@suse.de>
13086         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13087         alignment computations and rely on get_object_alignment_1
13088         for the !TYPE_P case.
13089         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13091 2013-03-26  Walter Lee  <walt@tilera.com>
13093         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13094         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13096 2013-03-25  Jeff Law  <law@redhat.com>
13098         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13099         check for INTEGRAL_TYPE_P that was missing due to checking in
13100         wrong version of prior patch.
13102 2013-03-25  Walter Lee  <walt@tilera.com>
13104         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13105         TILEGX_INSN_SHUFFLEBYTES1.
13106         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13107         shufflebytes1.
13108         (tilegx_builtins): Ditto.
13109         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13111 2013-03-25  Walter Lee  <walt@tilera.com>
13113         * config/tilegx/tilegx.md (floatsisf2): New pattern.
13114         (floatunssisf2): New pattern.
13115         (floatsidf2): New pattern.
13116         (floatunssidf2): New pattern.
13118 2013-03-25  Walter Lee  <walt@tilera.com>
13120         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13121         tests for constraint J, K, N, P.
13123 2013-03-25  Walter Lee  <walt@tilera.com>
13125         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13126         Use indirect/pcrel encoding.
13127         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13128         Ditto.
13130 2013-03-25  Steve Ellcey  <sellcey@mips.com>
13132         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13133         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13134         * config/mips/mips.c (mips_option_override): Set IMADD default.
13135         * config/mips/mips.h (PTF_AVOID_IMADD): New.
13136         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13137         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13138         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13139         * doc/invoke.texi (-mimadd/-mno-imadd): New.
13141 2013-03-25  Jeff Law  <law@redhat.com>
13143         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13144         slightly to avoid creating and folding useless trees.  Simplify
13145         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13147 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
13149         * config/i386/i386.md (*zero_extendsidi2): Merge with
13150         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
13151         * config/i386/predicates.md (x86_64_zext_operand): Rename from
13152         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
13153         targets.  Clarify comment.
13155 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13157         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13158         pass-through jump functions differently.
13159         (ipa_read_jump_function): Likewise.  Also use setter functions to set
13160         up jump functions.
13162 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13164         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13165         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13166         process it.
13167         (ipa_get_indirect_edge_target): New function.
13168         (devirtualization_time_bonus): New parameter known_aggs, pass it to
13169         ipa_get_indirect_edge_target.  Update all callers.
13170         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
13171         ipa_get_indirect_edge_target_1 instead of calling
13172         ipa_get_indirect_edge_target.
13173         (create_specialized_node): Pass aggvlas to
13174         ipcp_discover_new_direct_edges.
13176 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13178         * config/arm/arm.md (f_sels, f_seld): New types.
13179         (*cmov<mode>): New pattern.
13180         * config/arm/predicates.md (arm_vsel_comparison_operator): New
13181         predicate.
13183 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13185         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13186         POSIX-printf for mingw-hosted  builds.
13188 2013-03-25  Richard Biener  <rguenther@suse.de>
13190         PR middle-end/56694
13191         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13192         must-not-throw stmt location.
13194 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13196         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13197         Emit load-acquire versions when acq is true.
13198         (arm_emit_store_exclusive): Add rel parameter.
13199         Emit store-release versions when rel is true.
13200         (arm_split_compare_and_swap): Use acquire-release instructions
13201         instead.
13202         of barriers when appropriate.
13203         (arm_split_atomic_op): Likewise.
13204         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13205         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13206         (VUNSPEC_SLX): Likewise.
13207         (VUNSPEC_LDA): Likewise.
13208         (VUNSPEC_STL): Likewise.
13209         * config/arm/sync.md (atomic_load<mode>): New pattern.
13210         (atomic_store<mode>): Likewise.
13211         (arm_load_acquire_exclusive<mode>): Likewise.
13212         (arm_load_acquire_exclusivesi): Likewise.
13213         (arm_load_acquire_exclusivedi): Likewise.
13214         (arm_store_release_exclusive<mode>): Likewise.
13216 2013-03-25  Catherine Moore  <clm@codesourcery.com>
13218         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13219         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13220         * config/mip/predicates.md (lwsp_swsp_operand,
13221         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13222         sb16_operand, db4_operand, db7_operand, ib3_operand,
13223         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13224         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13225         andi16_operand): New predicates.
13226         * config/mips/mips.md (compression): New attribute.
13227         (enabled): New attribute.
13228         (length): Consider compression in computing length.
13229         (shift_compression): New code attribute.
13230         (*add<mode>3): New operands. Record compression.
13231         (sub<mode>3): Likewise.
13232         (one_cmpl<mode>2): Likewise.
13233         (*and<mode>3): Likewise.
13234         (*ior<mode>3): Likewise.
13235         (unnamed pattern for xor): Likewise.
13236         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13237         (*<optab><mode>3): Likewise.
13238         (*mov<mode>_internal: Likewise.
13239         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13240         (mips_unsigned_immediate_p): New.
13241         (umips_lwsp_swsp_address_p): New.
13242         (m16_based_address_p): New.
13243         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13244         (mips_unsigned_immediate_p): New prototype.
13245         (lwsp_swsp_address_p): New prototype.
13246         (m16_based_address_p): New prototype.
13247         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13248         (mips_signed_immediate_p): New function.
13249         (m16_based_address_p): New function.
13250         (lwsp_swsp_address_p): New function.
13251         (mips_print_operand_punctuation): Recognize short delay slot insns
13252         for microMIPS.add<mode>3"
13254 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13256         PR target/56720
13257         * config/arm/iterators.md (v_cmp_result): New mode attribute.
13258         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13260 2013-03-25  Richard Biener  <rguenther@suse.de>
13262         PR tree-optimization/56689
13263         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13264         any edge.
13266 2013-03-25  Richard Biener  <rguenther@suse.de>
13268         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13269         of bitmap.
13270         (memory_references): Likewise.
13271         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13272         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13273         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13274         (gather_mem_refs_in_loops): Fold into ...
13275         (analyze_memory_references): ... this.  Move initialization
13276         to tree_ssa_lim_initialize.
13277         (fill_always_executed_in): Rename to ...
13278         (fill_always_executed_in_1): ... this.
13279         (fill_always_executed_in): Move contains_call computation to
13280         this new function from ...
13281         (tree_ssa_lim_initialize): ... here.
13282         (tree_ssa_lim): Call fill_always_executed_in.
13284 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13286         * postreload.c (reload_combine): Fix code detecting returns.
13288 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13290         * function.c (emit_use_return_register_into_block): On cc0 targets,
13291         do not emit the sequence between cc0 setter and user.
13293 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13295         * config/i386/predicates.md (local_symbolic_operand): Interpret
13296         dll-imported symbols as none-local.
13298 2013-03-25  Richard Biener  <rguenther@suse.de>
13300         * tree-ssa-loop-im.c (struct depend): Remove.
13301         (struct lim_aux_data): Make depends a vec of gimples.
13302         (free_lim_aux_data): Adjust.
13303         (add_dependency): Likewise.
13304         (set_level): Likewise.
13306 2013-03-25  Richard Biener  <rguenther@suse.de>
13308         PR middle-end/56434
13309         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13310         the pointer returned by calls with ECF_MALLOC set.
13312 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13314         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13316 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13318         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13319         using MMXMODE mode iterator.
13320         (*move<mode>_internal): Merge with *movv2sf_internal and
13321         *movv2sf_internal_rex64 using MMXMODE mode iterator.
13323 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13325         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13326         (record_last_mem_set_info): Likewise.
13328         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13329         of XNEWVEC followed by memset.
13330         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13332 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13334         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13335         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13336         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13337         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13338         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13339         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13340         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13341         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13342         BARRIER_P instead of GET_CODE.
13344 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13346         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13347         inaccuracy in the probing code.
13349         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13350         (ctrapdi4): Likewise.
13352 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13354         * calls.c (expand_call): Add missing guard to code handling return
13355         of non-BLKmode structures in MSB.
13356         * function.c (expand_function_end): Likewise.
13358 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13360         * combine.c (try_combine): Adjust comment.  Do not add the set of
13361         insn #0 if the destination indirectly is set or dies in insn #2.
13362         Tidy up code to distribute a new note.
13364 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13366         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13367         also for alternatives 16 and 17.
13369 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13371         * config/i386/sse.md (*mov<mode>_internal): Merge with
13372         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13373         Emit insn template depending on type attribute.  Use
13374         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13375         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13376         attribute calculation.  Remove unit attribute calculation.
13377         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13378         Set prefix_data16 attribute for DImode ssemov types.
13379         Use Ym instead of y for SSE-MMX conversion alternatives.
13380         Reorder operand constraints.
13382 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13384         * df.h (df_insn_delete): Adjust prototype.
13385         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13386         and let it decide whether mark the basic block dirty.
13387         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13388         * df-scan.c (df_insn_info_delete): New helper function, split
13389         off from df_insn_delete.
13390         (df_scan_free_bb_info): Use it.
13391         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13392         Likewise.
13393         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13394         that the insn is actually an insn and it has a non-NULL basic block.
13395         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13397 2013-03-22  Richard Biener  <rguenther@suse.de>
13399         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13400         dep_ref members.
13401         (mem_ref_alloc): Do not allocate them.
13402         (refs_independent_p): Do not query or maintain a cache.
13404 2013-03-22  Richard Biener  <rguenther@suse.de>
13406         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13407         (gather_mem_refs_in_loops): Do not compute it.
13408         (analyze_memory_references): Do not allocate it.
13409         (tree_ssa_lim_finalize): Do not free it.
13410         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13412 2013-03-22  Richard Biener  <rguenther@suse.de>
13414         * is-a.h (as_a): Use gcc_checking_assert.
13416 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13418         * config/aarch64/aarch64.c (aarch64_print_operand): New
13419         format specifier for printing a constant in hex.
13420         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13421         format specifier for printing second operand.
13423 2013-03-22  Richard Biener  <rguenther@suse.de>
13425         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13426         bitmaps.
13427         (gather_mem_refs_in_loops): Perform store accumulation here.
13428         (create_vop_ref_mapping_loop): Remove.
13429         (create_vop_ref_mapping): Likewise.
13430         (analyze_memory_references): Initialize refs_stored_in_loop.
13431         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13432         (record_indep_loop): Remove.
13433         (record_dep_loop): New function.
13434         (ref_indep_loop_p_1): Adjust to only walk over references
13435         in the loop, not its subloops.
13436         (ref_indep_loop_p): Rename to ...
13437         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13438         maintaining a more fine-grained cache.
13439         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13440         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13442 2013-03-22  Richard Biener  <rguenther@suse.de>
13444         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13445         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13446         aggregate mem_ref_loc.
13447         (free_mem_ref_locs): Inline into ...
13448         (memref_free): ... this and adjust.
13449         (mem_ref_alloc): Adjust.
13450         (mem_ref_locs_alloc): Remove.
13451         (record_mem_ref_loc): Adjust.
13452         (get_all_locs_in_loop): Rewrite into ...
13453         (for_all_locs_in_loop): ... this iterator.
13454         (rewrite_mem_ref_loc): New functor.
13455         (rewrite_mem_refs): Use for_all_locs_in_loop.
13456         (sm_set_flag_if_changed): New functor.
13457         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13458         (ref_always_accessed): New functor.
13459         (ref_always_accessed_p): Use for_all_locs_in_loop.
13461 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13463         * tree-pass.h (PROP_gimple_lvec): New.
13464         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13465         (init_optimization_passes): Move pass_lower_vector.
13466         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13467         PROP_gimple_lvec.
13468         (pass_lower_vector): Provide PROP_gimple_lvec.
13469         (pass_lower_vector_ssa): Likewise.
13470         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13472 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13474         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13476 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13478         * config/i386/i386.md (*movdi_internal): Disparage slightly
13479         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13480         conversion alternatives.
13482 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13484         PR middle-end/48087
13485         * diagnostic.def (DK_WERROR): New kind.
13486         * diagnostic.h (werrorcount): Define.
13487         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13488         promoted to DK_ERROR, increment DK_WERROR counter instead of
13489         DK_ERROR counter.
13490         * toplev.c (toplev_main): Call print_ignored_options even if
13491         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13492         even if just werrorcount is non-zero.
13494         PR debug/55608
13495         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13496         on failure.
13497         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13498         (string_cst_pool_decl): New function.
13499         (optimize_one_addr_into_implicit_ptr): New function.
13500         (resolve_addr_in_expr): Optimize DWARF location expression
13501         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13502         which doesn't live in memory, but has DW_AT_location or
13503         DW_AT_const_value, or refers to a string literal, into
13504         DW_OP_GNU_implicit_pointer.
13505         (optimize_location_into_implicit_ptr): New function.
13506         (resolve_addr): If removing DW_AT_location of a variable because
13507         it was DW_OP_addr of address of the variable, but the variable doesn't
13508         live in memory, try to emit const value attribute for the initializer.
13510 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13512         * tree.h (VECTOR_TYPE_P): New macro.
13513         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13514         TYPE_MODE): Use it.
13515         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13516         VEC_COND_EXPR cannot be lvalues.
13517         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13519 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13521         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13522         Restrict the transformation to equal modes.
13524 2013-03-21  Richard Biener  <rguenther@suse.de>
13526         PR tree-optimization/39326
13527         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13528         (MEM_ANALYZABLE): Adjust.
13529         (record_mem_ref_loc): Move bitmap ops ...
13530         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13531         unanalyzable refs, do not record locations for it.
13532         (analyze_memory_references): Allocate ref zero as shared
13533         unanalyzable ref.
13534         (refs_independent_p): Do not test for unanalyzed mems here.
13535         (ref_indep_loop_p_1): Special-case disambiguation against
13536         the unanalyzed ref.
13537         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13539 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13541         * config/arm/arm-protos.h (tune_params): Add
13542         prefer_neon_for_64bits field.
13543         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13544         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13545         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13546         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13547         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13548         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13549         (arm_option_override): Handle -mneon-for-64bits new option.
13550         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13551         (prefer_neon_for_64bits): Declare new variable.
13552         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13553         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13554         (arch_enabled): Handle new arch types. Remove support for onlya8
13555         and nota8.
13556         (one_cmpldi2): Use new arch names.
13557         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13558         * config/arm/arm.opt (mneon-for-64bits): Add option.
13559         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13560         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13561         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13562         of onlya8.
13563         * doc/invoke.texi (-mneon-for-64bits): Document.
13565 2013-03-21  Richard Biener  <rguenther@suse.de>
13567         PR tree-optimization/39326
13568         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13569         (sort_bbs_in_loop_postorder_cmp): New function.
13570         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13572 2013-03-21  Richard Biener  <rguenther@suse.de>
13574         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13575         (vect_insert_into_interleaving_chain): Likewise.
13576         (vect_drs_dependent_in_basic_block): Inline ...
13577         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13578         split out from ...
13579         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13580         (vect_check_interleaving): Simplify.
13581         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13582         (vect_slp_analyze_data_ref_dependences): ... this new function.
13583         (dr_group_sort_cmp): New function.
13584         (vect_analyze_data_ref_accesses): Compute data-reference groups
13585         here instead of in vect_analyze_data_ref_dependence.  Use
13586         a more efficient algorithm.
13587         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13588         vect_slp_analyze_data_ref_dependences.  Call
13589         vect_analyze_data_ref_accesses earlier.
13590         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13591         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13592         (vect_slp_analyze_data_ref_dependences): New prototype.
13594 2013-03-21  Richard Biener  <rguenther@suse.de>
13596         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13597         ref is stored in the loop.
13598         (find_refs_for_sm): Walk only over all stores.
13599         (store_motion_loop): Allocate from lim_bitmap_obstack.
13600         (store_motion): Likewise.
13602 2013-03-21  Richard Biener  <rguenther@suse.de>
13604         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13605         Update virtual SSA form.
13607 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13609         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13610         * configure: Regenerate.
13611         * config.in: Regenerate.
13612         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13613         if !HAVE_LD_EH_FRAME_CIEV3.
13615 2013-03-21  Richard Biener  <rguenther@suse.de>
13617         * tree-cfg.c (verify_expr_no_block): New function.
13618         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13619         nor DECL_VALUE_EXPR have locations with associated blocks.
13620         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13621         (clear_unused_block_pointer): Remove code dealing with
13622         blocks in DECL_DEBUG_EXPR locations.
13624 2013-03-21  Richard Biener  <rguenther@suse.de>
13626         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13627         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
13628         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13629         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13630         instead of DECL_DEBUG_EXPR_IS_FROM.
13631         * gimplify.c (gimplify_modify_expr): Likewise.
13632         * tree-cfg.c (verify_expr_location_1): Likewise.
13633         * tree-complex.c (create_one_component_var): Likewise.
13634         * tree-sra.c (create_access_replacement): Likewise.
13635         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13636         (clear_unused_block_pointer): Likewise.
13637         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13638         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13639         * var-tracking.c (var_debug_decl): Likewise.
13640         (track_expr_p): Likewise.
13641         * tree-inline.c (add_local_variables): Likewise.  Set
13642         DECL_HAS_DEBUG_EXPR_P after copying it.
13643         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13644         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
13646 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13648         PR bootstrap/56656
13649         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13650         * configure: Regenerate.
13651         * config.in: Regenerate.
13652         * config/i386/i386.md (*movdf_internal): Use
13653         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13654         movd instead of movq mnemonic for interunit moves.
13655         (*movdi_internal): Ditto.
13657 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13659         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13660         (abd<mode>_3): New pattern.
13661         (aba<mode>_3): New pattern.
13662         (fabd<mode>_3): New pattern.
13664 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13666         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13667         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13668         occurrence of REGISTER_PREFIX as its empty string.
13670 2013-03-20  Jeff Law  <law@redhat.com>
13672         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13673         addititional equivalences for equality comparisons between an SSA_NAME
13674         and a constant where the SSA_NAME was set from a widening conversion.
13676 2013-03-20  Walter Lee  <walt@tilera.com>
13678         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13680 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13682         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13683         depending on type attribute.
13684         (*movti_internal): Ditto.
13685         (*movtf_internal): Ditto.
13686         (*movxf_internal): Ditto.
13687         (*movdf_internal): Ditto.
13688         (*movsf_internal): Ditto.
13690 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13692         * config/i386/i386.md (*movti_internal): Set prefix attribute to
13693         maybe_vex for sselog1 and ssemov types.
13694         (*movdi_internal): Reorder operand constraints.
13695         (*movsi_internal): Ditto.  Set prefix attribute to
13696         maybe_vex for sselog1 and ssemov types.
13697         (*movtf_internal): Set prefix attribute to maybe_vex
13698         for sselog1 and ssemov types.
13699         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
13700         DImode ssemov types.  Reorder operand constraints.
13701         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
13702         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
13703         attribute for SImode ssemov types.  Reorder operand constraints.
13705 2013-03-20  Martin Jambor  <mjambor@suse.cz>
13707         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13708         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13710 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
13712         * config/rs6000/predicates.md (indexed_address, update_address_mem
13713         update_indexed_address_mem): New predicates.
13714         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13715         attribute for load/store instructions.
13716         * config/rs6000/dfp.md (movsd_store): Likewise.
13717         (movsd_load): Likewise.
13718         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13719         (unnamed HI->DI extend define_insn): Likewise.
13720         (unnamed SI->DI extend define_insn): Likewise.
13721         (unnamed QI->SI extend define_insn): Likewise.
13722         (unnamed QI->HI extend define_insn): Likewise.
13723         (unnamed HI->SI extend define_insn): Likewise.
13724         (unnamed HI->SI extend define_insn): Likewise.
13725         (extendsfdf2_fpr): Likewise.
13726         (movsi_internal1): Likewise.
13727         (movsi_internal1_single): Likewise.
13728         (movhi_internal): Likewise.
13729         (movqi_internal): Likewise.
13730         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13731         attribute for load/store instructions.
13732         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13733         instructions.
13734         (mov<mode>_softfloat): Likewise.
13735         (mov<mode>_hardfloat32): Likewise.
13736         (mov<mode>_hardfloat64): Likewise.
13737         (mov<mode>_softfloat64): Likewise.
13738         (movdi_internal32): Likewise.
13739         (movdi_internal64): Likewise.
13740         (probe_stack_<mode>): Likewise.
13742 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
13744         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13745         floating point, and decimal floating point to reload iterator.
13747         * config/rs6000/constraints.md (wl constraint): New constraints to
13748         return FLOAT_REGS if certain options are used to reduce the number
13749         of separate patterns that exist in the file.
13750         (wx constraint): Likewise.
13751         (wz constraint): Likewise.
13753         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13754         -mdebug=reg, print wg, wl, wx, and wz constraints.
13755         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13756         Initialize the reload functions for 64-bit binary/decimal floating
13757         point types.
13758         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13759         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13760         create the buffer on the stack to overcome not having a 32-bit
13761         load and store.
13762         (rs6000_emit_move): Likewise.
13763         (rs6000_secondary_memory_needed_rtx): Likewise.
13764         (rs6000_alloc_sdmode_stack_slot): Likewise.
13765         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13766         via xxlxor, just like DFmode 0.0.
13768         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13769         define as 1 if we are running on a power7 or newer.
13770         (enum r6000_reg_class_enum): Add new constraints.
13772         * config/rs6000/dfp.md (movsd): Delete, combine with binary
13773         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13774         with other moves by using conditional constraits (wg).  Use LFIWZX
13775         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
13776         (movsd splitter): Likewise.
13777         (movsd_hardfloat): Likewise.
13778         (movsd_softfloat): Likewise.
13780         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13781         binary and decimal floating point moves.
13782         (fmove_ok): New attributes to combine binary and decimal floating
13783         point moves, and to combine power6x (mfpgpr) moves along normal
13784         floating moves.
13785         (real_value_to_target): Likewise.
13786         (f32_lr): Likewise.
13787         (f32_lm): Likewise.
13788         (f32_li): Likewise.
13789         (f32_sr): Likewise.
13790         (f32_sm): Likewise.
13791         (f32_si): Likewise.
13792         (movsf): Combine binary and decimal floating point moves.  Combine
13793         power6x (mfpgpr) moves with other moves by using conditional
13794         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
13795         (mov<mode> for SFmode/SDmode); Likewise.
13796         (SFmode/SDmode splitters): Likewise.
13797         (movsf_hardfloat): Likewise.
13798         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13799         (movsf_softfloat): Likewise.
13800         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13802         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13803         wx and wz constraints.
13805         * config/rs6000/constraints.md (wg constraint): New constraint to
13806         return FLOAT_REGS if -mmfpgpr (power6x) was used.
13808         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13809         constraint.
13811         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13812         -mdebug=reg, print wg, wl, wx, and wz constraints.
13813         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13814         Initialize the reload functions for 64-bit binary/decimal floating
13815         point types.
13816         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13817         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13818         create the buffer on the stack to overcome not having a 32-bit
13819         load and store.
13820         (rs6000_emit_move): Likewise.
13821         (rs6000_secondary_memory_needed_rtx): Likewise.
13822         (rs6000_alloc_sdmode_stack_slot): Likewise.
13823         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13824         via xxlxor, just like DFmode 0.0.
13826         * config/rs6000/dfp.md (movdd): Delete, combine with binary
13827         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13828         with other moves by using conditional constraits (wg).  Use LFIWZX
13829         and STFIWX for loading SDmode on power7.
13830         (movdd splitters): Likewise.
13831         (movdd_hardfloat32): Likewise.
13832         (movdd_softfloat32): Likewise.
13833         (movdd_hardfloat64_mfpgpr): Likewise.
13834         (movdd_hardfloat64): Likewise.
13835         (movdd_softfloat64): Likewise.
13837         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13838         64-bit binary and decimal floating point moves.
13839         (FMOVE64X): Likewise.
13840         (movdf): Combine 64-bit binary and decimal floating point moves.
13841         Combine power6x (mfpgpr) moves with other moves by using
13842         conditional constraits (wg).
13843         (mov<mode> for DFmode/DDmode): Likewise.
13844         (DFmode/DDmode splitters): Likewise.
13845         (movdf_hardfloat32): Likewise.
13846         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13847         (movdf_softfloat32): Likewise.
13848         (movdf_hardfloat64_mfpgpr): Likewise.
13849         (movdf_hardfloat64): Likewise.
13850         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13851         (movdf_softfloat64): Likewise.
13852         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13853         (reload_<mode>_load): Move to later in the file so they aren't in
13854         the middle of the floating point move insns.
13855         (reload_<mode>_store): Likewise.
13857         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13858         constraint.
13860         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13861         constraint if -mdebug=reg.
13862         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13863         Enable using dd reload support if needed.
13865         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13866         binary and decimal floating point moves in rs6000.md.
13867         (movtd_internal): Likewise.
13869         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13870         decimal floating point moves.
13871         (movtf): Likewise.
13872         (movtf_internal): Likewise.
13873         (mov<mode>_internal, TDmode/TFmode): Likewise.
13874         (movtf_softfloat): Likewise.
13875         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13877         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13878         movdi_internal64, using wg constraint for move direct operations.
13879         (movdi_internal64): Likewise.
13881         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13882         MODES_TIEABLE_P for selected modes.  Print the numerical value of
13883         the various virtual registers. Use GPR/FPR first/last values,
13884         instead of hard coding the register numbers.  Print which modes
13885         have reload functions registered.
13886         (rs6000_option_override_internal): If -mdebug=reg, trace the options
13887         settings before/after setting cpu, target and subtarget settings.
13888         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13889         and for secondary reload failures in rs6000_secondary_reload_inner.
13890         (rs6000_secondary_reload_fail): Likewise.
13891         (rs6000_secondary_reload_inner): Likewise.
13893         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13894         macros for first/last GPR and FPR registers.
13895         (LAST_GPR_REGNO): Likewise.
13896         (FIRST_FPR_REGNO): Likewise.
13897         (LAST_FPR_REGNO): Likewise.
13899         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13900         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13901         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13902         (vcond<mode><mode>): Likewise.
13903         (vcondu<mode><mode>): Likewise.
13904         (vector_gtu<mode>): Likewise.
13905         (vector_gte<mode>): Likewise.
13906         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13907         to prevent the compiler from converting DImode operations to TImode.
13908         (ior<mode>3): Likewise.
13909         (and<mode>3): Likewise.
13910         (one_cmpl<mode>2): Likewise.
13911         (nor<mode>3): Likewise.
13912         (andc<mode>3): Likewise.
13914         * config/rs6000/constraints.md (wt constraint): New constraint
13915         that returns VSX_REGS if TImode is allowed in VSX registers.
13917         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13918         constant under VSX.
13920         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13921         similar to TImode, but it is restricted to being in the GPRs.
13923         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13924         TImode to occupy a single VSX register.
13926         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13927         -mvsx-timode for power7/power8.
13928         (power7 cpu): Likewise.
13929         (power8 cpu): Likewise.
13931         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13932         sure that TFmode/TDmode take up two registers if they are ever
13933         allowed in the upper VSX registers.
13934         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13935         registers.
13936         (rs6000_init_hard_regno_mode_ok): Likewise.
13937         (rs6000_debug_reg_global): Add debugging for PTImode and wt
13938         constraint.  Print if LRA is turned on.
13939         (rs6000_option_override_internal): Give an error if -mvsx-timode
13940         and VSX is not enabled.
13941         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
13942         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13943         to reg+offset addressing.  Use PTImode when checking offset
13944         addresses for validity.
13945         (reg_offset_addressing_ok_p): Likewise.
13946         (rs6000_legitimate_offset_address_p): Likewise.
13947         (rs6000_legitimize_address): Likewise.
13948         (rs6000_legitimize_reload_address): Likewise.
13949         (rs6000_legitimate_address_p): Likewise.
13950         (rs6000_eliminate_indexed_memrefs): Likewise.
13951         (rs6000_emit_move): Likewise.
13952         (rs6000_secondary_reload): Likewise.
13953         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
13954         reloads to fpr registers to continue to use reg+offset addressing,
13955         but 64-bit reloads to altivec registers need reg+reg addressing.
13956         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13957         it.  Treat LO_SUM like a PLUS operation.
13958         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13959         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13960         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13961         registers to share a register with a smaller sized type, since VSX
13962         puts scalars in the upper 64-bits.
13963         (print_operand): Add support for PTImode.
13964         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13965         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13966         registers, but don't have arithmetic support.
13967         (rs6000_memory_move_cost): Add test for VSX.
13968         (rs6000_opt_masks): Add -mvsx-timode.
13970         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13971         for TImode.
13972         (VSs): Likewise.
13973         (VSr): Use wt constraint for TImode.
13974         (VSv): Drop TImode support.
13975         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13976         (vsx_movti_64bit): Likewise.
13977         (vsx_movti_32bit): Likewise.
13978         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13979         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13980         one '?' on the appropriate output constraint.  Do not allow TImode
13981         logical operations on 32-bit systems.
13982         (vsx_ior<mode>3): Likewise.
13983         (vsx_xor<mode>3): Likewise.
13984         (vsx_one_cmpl<mode>2): Likewise.
13985         (vsx_nor<mode>3): Likewise.
13986         (vsx_andc<mode>3): Likewise.
13987         (vsx_concat_<mode>): Likewise.
13988         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13990         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13991         OPTION_MASK_VSX_TIMODE.
13992         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13993         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13995         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13996         (TI2 iterator): New iterator for TImode, PTImode.
13997         (wd mode attribute): Add values for vector types.
13998         (movti_string): Replace TI move operations with operations for TImode
13999         and PTImode.  Add support for TImode being allowed in VSX registers.
14000         (mov<mode>_string, TImode/PTImode): Likewise.
14001         (movti_ppc64): Likewise.
14002         (mov<mode>_ppc64, TImode/PTImode): Likewise.
14003         (TI mode splitters): Likewise.
14005         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14006         constraint.
14008 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
14010         PR tree-optimization/56355
14011         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14012         Also handle integers with undefined overflow.
14014 2013-03-20  Catherine Moore  <clm@codesourcery.com>
14015             Maciej W. Rozycki  <macro@codesourcery.com>
14016             Tom de Vries  <tom@codesourcery.com>
14017             Nathan Sidwell <nathan@codesourcery.com>
14018             Iain Sandoe  <iain@codesourcery.com>
14019             Nathan Froyd  <froydnj@codesourcery.com>
14020             Chao-ying Fu <fu@mips.com>
14022         * doc/extend.texi: (micromips, nomicromips, nocompression):
14023         Document new function attributes.
14024         * doc/invoke.texi (minterlink-compressed, mmicromips,
14025         m14k, m14ke, m14kec): Document new options.
14026         (minterlink-mips16): Update documentation.
14027         * doc/md.texi (ZC, ZD): Document new constraints.
14028         * configure.ac (gcc_cv_as_micromips): Check if linker
14029         supports the .set micromips directive.
14030         * configure: Regenerate.
14031         * config.in: Regenerate.
14032         * config/mips/mips-tables.opt: Regenerate.
14033         * config/mips/micromips.md: New file.
14034         * constraints.md (ZC, ZD): New constraints.
14035         * config/mips/predicates.md (movep_src_register): New predicate.
14036         (movep_src_operand): New predicate.
14037         (non_volatile_mem_operand): New predicate.
14038         * config/mips/mips.md (multimem): New type.
14039         (length): Differentiate between 17-bit and 18-bit branch offsets.
14040         (MOVEP1, MOVEP2): New mode iterator.
14041         (mov_<load>l): Use ZC constraint.
14042         (mov_<load>r): Likewise.
14043         (mov_<store>l): Likewise.
14044         (mov_<store>r): Likewise.
14045         (*branch_equality<mode>_inverted): Add microMIPS support.
14046         (*branch_equality<mode>): Likewise.
14047         (*jump_absolute): Likewise.
14048         (indirect_jump_<mode>): Likewise.
14049         (tablejump_<mode>): Likewise.
14050         (<optab>_internal): Likewise.
14051         (sibcall_internal): Likewise.
14052         (sibcall_value_internal): Likewise.
14053         (prefetch): Use constraint ZD.
14054         * config/mips/mips.opt (minterlink-compressed): New option.
14055         (minterlink-mips16): Now an alias for minterlink-compressed.
14056         (mmicromips): New option.
14057         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14058         (compare_and_swap_12): Likewise.
14059         (sync_add<mode>): Likewise.
14060         (sync_<optab>_12): Likewise.
14061         (sync_old_<optab>_12): Likewise.
14062         (sync_new_<optab>_12): Likewise.
14063         (sync_nand_12): Likewise.
14064         (sync_old_nand_12): Likewise.
14065         (sync_new_nand_12): Likewise.
14066         (sync_sub<mode>): Likewise.
14067         (sync_old_add<mode>): Likewise.
14068         (sync_old_sub<mode>): Likewise.
14069         (sync_new_add<mode>): Likewise.
14070         (sync_new_sub<mode>): Likewise.
14071         (sync_<optab><mode>): Likewise.
14072         (sync_old_<optab><mode>): Likewise.
14073         (sync_new_<optab><mode>): Likewise.
14074         (sync_nand<mode>): Likewise.
14075         (sync_old_nand<mode>): Likewise.
14076         (sync_new_nand<mode>): Likewise.
14077         (sync_lock_test_and_set<mode>): Likewise.
14078         (test_and_set_12): Likewise.
14079         (atomic_compare_and_swap<mode>): Likewise.
14080         (atomic_exchange<mode>_llsc): Likewise.
14081         (atomic_fetch_add<mode>_llsc): Likewise.
14082         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14083         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14084         (umips_save_restore_pattern_p): Likewise.
14085         (umips_load_store_pair_p): Likewise.
14086         (umips_output_load_store_pair): Likewise.
14087         (umips_movep_target_p): Likewise.
14088         (umips_12bit_offset_address_p): Likewise.
14089         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14090         (mips_base_mips16): Rename this...
14091         (mips_base_compression_flags): ...to this. Update all uses.
14092         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14093         (mips_mips16_decl_p): Delete.
14094         (mips_nomips16_decl_p): Delete.
14095         (mips_get_compress_on_flags): New function.
14096         (mips_get_compress_off_flags): New function.
14097         (mips_get_compress_mode): New function.
14098         (mips_get_compress_on_name): New function.
14099         (mips_get_compress_off_name): New function.
14100         (mips_insert_attributes): Support multiple compression types.
14101         (mips_merge_decl_attributes): Likewise.
14102         (umips_12bit_offset_address_p): New function.
14103         (mips_start_function_definition): Emit .set micromips directive.
14104         (mips_call_may_need_jalx_p): New function.
14105         (mips_function_ok_for_sibcall): Add microMIPS support.
14106         (mips_print_operand_punctuation): Support short delay slots and
14107         compact jumps.
14108         (umips_swm_mask, umips_swm_encoding): New.
14109         (umips_build_save_restore): New function.
14110         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14111         (was_mips16_p): Remove.
14112         (old_compression_mode): New.
14113         (mips_set_compression_mode): New function.
14114         (mips_set_current_function): Add microMIPS support.
14115         (mips_option_override): Likewise.
14116         (umips_save_restore_pattern_p): New function.
14117         (umips_output_save_restore): New function.
14118         (umips_load_store_pair_p_1): New function.
14119         (umips_load_store_pair_p): New function.
14120         (umips_output_load_store_pair_1): New function.
14121         (umips_output_load_store_pair): New function.
14122         (umips_movep_target_p) New function.
14123         (mips_prepare_pch_save): Add microMIPS support.
14124         * config/mips/mips.h (TARGET_COMPRESSION): New.
14125         (TARGET_CPU_CPP_BUILTINS): Update macro
14126         to use new compression flags and to support microMIPS.
14127         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14128         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14129         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14130         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14131         (ASM_SPEC): Support mmicromips and mno-micromips.
14132         (M16STORE_REG_P): New macro.
14133         (MIPS_CALL): Support TARGET_MICROMIPS.
14134         (MICROMIPS_J): New macro.
14135         (mips_base_mips16): Rename this...
14136         (mips_base_compression_flags): ...to this.
14137         (UMIPS_12BIT_OFFSET_P): New macro.
14138         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14139         (MULTILIB_DIRNAMES): Likewise.
14140 2013-03-20  Richard Biener  <rguenther@suse.de>
14142         PR tree-optimization/56661
14143         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14144         the result does not have to be distinct.
14146 2013-03-20  Richard Biener  <rguenther@suse.de>
14148         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14149         remap_gimple_op_r.
14151 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14152             Steven Bosscher <steven@gcc.gnu.org>
14154         PR rtl-optimization/56605
14155         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14157 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14159         PR bootstrap/56656
14160         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14161         that require movd instead of movq.
14163 2013-03-20  Richard Biener  <rguenther@suse.de>
14165         * tree-ssa-structalias.c (struct variable_info): Add pointer
14166         to the first field of an aggregate with sub-vars.  Make
14167         this and the pointer to the next subfield its ID.
14168         (vi_next): New function.
14169         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14170         storedanything_id, integer_id): Increment by one.
14171         (new_var_info, get_call_vi, lookup_call_clobber_vi,
14172         get_call_clobber_vi): Adjust.
14173         (solution_set_expand): Simplify and speedup.
14174         (solution_set_add): Inline into ...
14175         (set_union_with_increment): ... this.  Adjust accordingly.
14176         (do_sd_constraint): Likewise.
14177         (do_ds_constraint): Likewise.
14178         (do_complex_constraint): Simplify.
14179         (build_pred_graph): Adjust.
14180         (solve_graph): Likewise.  Simplify and speedup.
14181         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14182         get_constraint_for_component_ref, get_constraint_for_1,
14183         first_vi_for_offset, first_or_preceding_vi_for_offset,
14184         create_function_info_for, create_variable_info_for_1,
14185         create_variable_info_for, intra_create_variable_infos): Adjust.
14186         (init_base_vars): Push NULL for ID zero.
14187         (compute_points_to_sets): Adjust.
14189 2013-03-20  Richard Biener  <rguenther@suse.de>
14191         * cfgloop.c (verify_loop_structure): Streamline and avoid
14192         ICEing on corrupt loop tree.
14193         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14194         loop tree.
14196 2013-03-20  Richard Biener  <rguenther@suse.de>
14198         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14199         check whether an SSA update is needed.
14201 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
14203         * config/mips/constraints.md (T): Rename to...
14204         (Yf): ...this.
14205         (U): Rename to...
14206         (Yd): ...this.
14207         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14208         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14210 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14212         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14213         (*subsi3_carryin_uxtw): Likewise.
14215 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14217         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14218         (*rorsi3_insn_uxtw): Likewise.
14220 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14222         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14223         (*extrsi5_insn_uxtw): Likewise.
14225 2013-03-19  Richard Biener  <rguenther@suse.de>
14227         PR tree-optimization/56273
14228         * passes.c (init_optimization_passes): Move second VRP after DOM.
14230 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
14232         * config/i386/i386.md (*movti_internal): Merge from
14233         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
14234         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
14235         nox64 isa attributes.
14237 2013-03-18  Richard Biener  <rguenther@suse.de>
14239         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14240         (unite): Likewise.
14241         (merge_node_constraints): Likewise.
14242         (build_succ_graph): Likewise.
14243         (valid_graph_edge): Inline into single caller.
14244         (unify_nodes): Likewise.  Use bitmap_set_bit return value
14245         and cache varinfo.
14246         (scc_visit): Fix formatting and variable use.
14247         (do_sd_constraint): Use gcc_checking_assert.
14248         (do_ds_constraint): Likewise.
14249         (do_complex_constraint): Likewise.
14250         (condense_visit): Likewise.  Cleanup.
14251         (dump_pred_graph): New function.
14252         (perform_var_substitution): Dump the pred-graph before
14253         variable substitution.
14254         (find_equivalent_node): Use gcc_checking_assert.
14255         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14257 2013-03-18  Richard Biener  <rguenther@suse.de>
14259         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14260         Remove cond_expr_stmt_list argument and do not gimplify the
14261         built expression.
14262         (vect_loop_versioning): Adjust.
14263         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14264         Cleanup to use less temporaries.
14265         (vect_create_data_ref_ptr): Cleanup.
14267 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14269         PR tree-optimization/56635
14270         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14271         require types_compatible_p types.
14273 2013-03-18  Nick Clifton  <nickc@redhat.com>
14275         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14276         spurious backslash.
14278         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14279         Add missing line to comment describing function.
14281 2013-03-18  Richard Biener  <rguenther@suse.de>
14283         PR tree-optimization/56210
14284         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14285         Handle string / character search functions.
14286         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14288 2013-03-18  Richard Biener  <rguenther@suse.de>
14290         PR middle-end/56483
14291         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14292         and implement properly.
14293         * gimple.h (gimple_cond_single_var_p): Remove.
14295 2013-03-18  Richard Biener  <rguenther@suse.de>
14297         * tree-data-ref.h (find_data_references_in_loop): Declare.
14298         * tree-data-ref.c (get_references_in_stmt): Use a stack
14299         vector pre-allocated in the callers.
14300         (find_data_references_in_stmt): Adjust.
14301         (graphite_find_data_references_in_stmt): Likewise.
14302         (create_rdg_vertices): Likewise.
14303         (find_data_references_in_loop): Export.
14304         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14305         Compute dependences here...
14306         (vect_analyze_data_refs): ...not here.  When we encounter
14307         a non-vectorizable data reference in basic-block vectorization
14308         truncate the data reference vector.  Do not bother to
14309         fixup data-dependence information for gather loads.
14310         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14311         of data references, as reported.
14313 2013-03-18  Richard Biener  <rguenther@suse.de>
14315         PR tree-optimization/3713
14316         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
14317         has_constants and expr.
14318         (stmt_has_constants): Properly valueize SSA names when deciding
14319         whether the stmt has constants.
14321 2013-03-18  Richard Biener  <rguenther@suse.de>
14323         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14324         whole function when there is nothing to do.
14325         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14326         * tree-vectorizer.c (vectorize_loops): Update virtual and
14327         loop-closed SSA once.
14328         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14330 2013-03-18  Richard Biener  <rguenther@suse.de>
14332         PR middle-end/56113
14333         * domwalk.c (bb_postorder): New global static.
14334         (cmp_bb_postorder): New function.
14335         (walk_dominator_tree): Replace scheme imposing an order for
14336         visiting dominator sons by one sorting them at the time they
14337         are pushed on the stack.
14339 2013-03-18  Richard Biener  <rguenther@suse.de>
14341         PR tree-optimization/39326
14342         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14343         (struct mem_ref): Replace mem member with ao_ref typed member.
14344         (MEM_ANALYZABLE): Adjust.
14345         (memref_eq): Likewise.
14346         (mem_ref_alloc): Likewise.
14347         (gather_mem_refs_stmt): Likewise.
14348         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14349         (execute_sm_if_changed_flag_set): Adjust.
14350         (execute_sm): Likewise.
14351         (ref_always_accessed_p): Likewise.
14352         (refs_independent_p): Likewise.
14353         (can_sm_ref_p): Likewise.
14355 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14357         PR c/56566
14358         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14359         return 1 even for !unsignedp.
14361 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14363         * config/i386/i386.md (isa): Add x64 and nox64.
14364         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14365         (*pushtf): Enable *roF alternative for x64 isa only.
14366         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14367         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14368         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14369         (*movtf_internal): Merge from *movtf_internal_rex64 and
14370         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14371         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14372         nox64 isa attributes.
14373         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14374         nox64 isa attributes.
14375         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14377 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14379         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14381 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14383         PR target/56640
14384         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14385         class names.  Remove trailing comma after "ALL_REGS".
14387 2013-03-16  Jan Hubicka  <jh@suse.cz>
14389         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14390         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14391         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14392         of cgraph_get_create_node.
14393         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14395 2013-03-16  Jason Merrill  <jason@redhat.com>
14397         PR debug/49090
14398         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14399         with DW_AT_default_value.
14401 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14403         * BASE-VER: Set to 4.9.0.
14405 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14407         PR target/56619
14408         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14409         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14410         Document _x* TSX intrinsics.
14412 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14413             David Holsgrove  <david.holsgrove@xilinx.com>
14415         * configure.ac: Add MicroBlaze TLS support detection.
14416         * configure: Regenerate.
14417         * config/microblaze/microblaze-protos.h
14418         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14419         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14420         * config/microblaze/microblaze.c (microblaze_address_type): Add
14421         ADDRESS_TLS and tls_reloc address types.
14422         (microblaze_address_info): Add tls_reloc.
14423         (TARGET_HAVE_TLS): Define.
14424         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14425          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14426          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14427         load_tls_operand,  microblaze_call_tls_get_addr,
14428         microblaze_legitimize_tls_address): New functions.
14429         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14430         (get_base_reg): Use microblaze_tls_symbol_p.
14431         (microblaze_classify_address): Handle TLS.
14432         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14433         label_mentioned_p and microblaze_tls_referenced_p.
14434         (microblaze_legitimize_address): Handle TLS.
14435         (microblaze_address_insns): Handle ADDRESS_TLS.
14436         (pic_address_needs_scratch): Handle TLS.
14437         (print_operand_address): Handle TLS.
14438         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14439         (microblaze_expand_move): Handle TLS.
14440         (microblaze_legitimate_constant_p): Check
14441         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14442         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14443         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14444         (PIC_OFFSET_TABLE_REGNUM): Set.
14445         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14446         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14447         (addsi3, movsi_internal2, movdf_internal): Update constraints
14448         * config/microblaze/predicates.md (arith_plus_operand): Define
14449         (move_operand): Redefine as move_src_operand,
14450         check microblaze_tls_referenced_p.
14452 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14454         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14455         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14457 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14459         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14460         CC mode for AND.
14462 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14464         PR tree-optimization/53265
14465         * common.opt (Waggressive-loop-optimizations): New option.
14466         * tree-ssa-loop-niter.c: Include tree-pass.h.
14467         (do_warn_aggressive_loop_optimizations): New function.
14468         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14469         if number_of_latch_executions returned constant.
14470         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14471         early.  If number_of_latch_executions returned constant, set
14472         nb_iterations_upper_bound back to it.
14473         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14474         field.
14475         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14476         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14478         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14479         (MULTILIB_OSDIRNAMES): Set.
14480         * genmultilib: If defaultosdirname doesn't start with :: , set
14481         defaultosdirname2 instead, clear it and emit two . multilib_raw
14482         entries instead of just one.
14484 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14486         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14487         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14488         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14489         (SUBTARGET_OVERRIDE_OPTIONS): New.
14491 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14493         PR target/49880
14494         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14495         (musermode): Convert to Var(TARGET_USERMODE).
14496         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14497         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14498         * config/sh/sh.c (sh_option_override): Use
14499         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14500         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14501         condition.
14502         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14503         TARGET_SH4.
14504         (udivsi3_i4_single, divsi3_i4_single): Use
14505         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14507 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14509         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14510         default setting.
14512 2013-03-13  Richard Biener  <rguenther@suse.de>
14514         PR tree-optimization/56608
14515         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14516         calls when vectorizing basic-blocks.
14518 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14520         PR plugins/45078
14521         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14522         tm_file.
14524 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14526         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14528 2013-03-11  Jan Hubicka  <jh@suse.cz>
14530         PR lto/56557
14531         * lto-streamer-out.c (output_symbol_p): Skip references from
14532         constructors of external variables.
14534 2013-03-11  Jan Hubicka  <jh@suse.cz>
14536         PR middle-end/56571
14537         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14538         from pseudos.
14539         * emit-rtl.c (verify_rtx_sharing): Likewise.
14540         (copy_insn_1): Likewise.
14541         * rtl.c (copy_rtx): Likewise.
14543 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14545         PR target/56591
14546         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14547         output_operand_lossage message.
14549 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14551         PR target/56470
14552         * arm.c (shift_op): Validate RTL pattern on the fly.
14553         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14554         the RTL.
14556 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14558         PR target/56347
14559         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14560         new call patterns that clobber %fr12.
14561         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14562         split and postreload patterns.
14563         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14564         registers %fr12 and %fr12R as call used.
14566 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14568         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14569         (canon_address, record_store, replace_read, check_mem_read_rtx,
14570         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14571         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14572         rest_of_handle_dse): Likewise.
14574 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14576         PR middle-end/56524
14577         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14578         Add base_optabs.
14579         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14580         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14581         (save_optabs_if_changed): Replace with...
14582         (init_tree_optimization_optabs): ...this.
14583         * optabs.c (save_optabs_if_changed): Rename to...
14584         (init_tree_optimization_optabs): ...this.  Take the optimization node
14585         as argument.  Do nothing if the base optabs are already correct.
14586         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14587         to recompute optabs.
14588         * function.h (function): Remove optabs field.
14589         * function.c (invoke_set_current_function_hook): Call
14590         init_tree_optimization_optabs.  Use the result to initialize
14591         this_fn_optabs.
14593 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14595         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14596         if GTMA_HAS_NO_INSTRUMENTATION.
14597         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14598         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14599         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14600         * gimple-pretty-print.c (dump_gimple_transaction): Handle
14601         GTMA_HAS_NO_INSTRUMENTATION.
14603 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
14605         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14606         libasan_preinit.o.
14608 2013-03-08  Marek Polacek  <polacek@redhat.com>
14609             Jakub Jelinek  <jakub@redhat.com>
14611         PR tree-optimization/56478
14612         * predict.c (is_comparison_with_loop_invariant_p): Change the
14613         type of loop_step to tree.
14614         (predict_loops): Adjust.
14615         (predict_iv_comparison): Perform the computations on double_ints.
14617 2013-03-08  Richard Biener  <rguenther@suse.de>
14619         PR tree-optimization/56570
14620         * tree-cfg.c (verify_expr_location_1): Verify locations for
14621         DECL_DEBUG_EXPR.
14622         * tree-sra.c (create_access_replacement): Strip locations
14623         from DECL_DEBUG_EXPRs.
14625 2013-03-08  Richard Biener  <rguenther@suse.de>
14627         * tree-inline.c (expand_call_inline): Do not associate
14628         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14629         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14631 2013-03-08  Richard Biener  <rguenther@suse.de>
14633         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14634         or block changes with -Og.  Fix for location / block encoding
14635         changes and PHI arguments with locations.
14637 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
14639         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14640         for all counters.
14641         (struct output_info): Likewise.
14642         (register_overhead): Remove bad gcc_assert.
14643         (bitmap_find_bit): If there is only a single bitmap element, do not
14644         count a miss as a search.
14645         (print_statistics): Update for counter type changes.
14646         (dump_bitmap_statistics): Likewise.  Print headers such that they
14647         are properly lined up with the printed counters.
14649 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
14651         PR tree-optimization/56559
14652         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14653         check that it has only a single use.
14655 2013-03-07  Richard Biener  <rguenther@suse.de>
14657         * doc/invoke.texi (fwhole-program): Discourage use in combination
14658         with -flto.
14660 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
14662         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14664         PR tree-optimization/56539
14665         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14666         instead of GSI_CONTINUE_LINKING as last argument to
14667         force_gimple_operand_gsi.  Adjust function comment.
14669         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14670         aarch64-cores.def.
14672         PR middle-end/56548
14673         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14674         promoted mode, convert the result back to the original mode.
14676 2013-03-06  Richard Biener  <rguenther@suse.de>
14678         PR middle-end/56294
14679         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14680         (insert_updated_phi_nodes_compare_uids): New function.
14681         (update_ssa): Sort symbols_to_rename after UID before
14682         traversing it to insert PHI nodes.
14684 2013-03-06  Richard Biener  <rguenther@suse.de>
14686         PR middle-end/50494
14687         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14688         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14690         Revert
14691         2013-02-13  Richard Biener  <rguenther@suse.de>
14693         PR lto/50494
14694         * varasm.c (output_constant_def_1): Get the decl representing
14695         the constant as argument.
14696         (output_constant_def): Wrap output_constant_def_1.
14697         (make_decl_rtl): Use output_constant_def_1 with the decl
14698         representing the constant.
14699         (build_constant_desc): Optionally re-use a decl already
14700         representing the constant.
14701         (tree_output_constant_def): Adjust.
14703 2013-03-06  Joey Ye  <joey.ye@arm.com>
14705         PR lto/50293
14706         * gcc.c (convert_white_space): New function.
14707         (main): Handles white space in function name.
14709 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
14711         PR target/56529
14712         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14713         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
14714         to SH_DIV_CALL_TABLE for TARGET_SH2.
14715         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14716         list.
14717         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14718         call-table options.
14720 2013-03-05  Sterling Augustine  <saugustine@google.com>
14721             Cary Coutant  <ccoutant@google.com>
14723         PR debug/55364
14724         * dwarf2out.c (resolve_addr): Don't call
14725         remove_loc_list_addr_table_entries a second time for the same
14726         expression.
14728 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14730         PR debug/56510
14731         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14732         (avoid_complex_debug_insns): New function.
14733         (expand_debug_locations): Call it.
14735         PR rtl-optimization/56484
14736         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14737         lifetimes of hard registers on small register class machines.
14739 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14741         * config/microblaze/microblaze-protos.h: Rename
14742         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14743         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14744         fast_interrupt.
14745         (microblaze_fast_interrupt_function_p): New function.
14746         (microblaze_is_interrupt_handler): Rename to
14747         microblaze_is_interrupt_variant and add fast_interrupt check.
14748         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14749         (save_restore_insns): Likewise.
14750         (compute_frame_size): Likewise.
14751         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14752         (microblaze_globalize_label): Likewise.
14753         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14754         * config/microblaze/microblaze.md: Use wrapper
14755         microblaze_is_interrupt_variant.
14757 2013-03-05  Kai Tietz  <ktietz@redhat.com>
14759         * sdbout.c (sdbout_one_type): Switch to current function's section
14760         supporting cold/hot.
14762 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14764         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14765         -mxl-reorder.
14767 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14769         PR middle-end/56461
14770         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14771         if VALGRIND_GET_VBITS is defined, temporarily make object
14772         memory all defined, and restore previous valgrind addressability
14773         and definability afterwards.  Free this_object at the end.
14775         PR middle-end/56461
14776         * lra.c (lra): Call lra_clear_live_ranges if live_p,
14777         right before calling lra_create_live_ranges, also call it
14778         when clearing live_p.  Only call lra_clear_live_ranges
14779         at the end if live_p.
14781         PR middle-end/56461
14782         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14784 2013-03-05  Richard Biener  <rguenther@suse.de>
14786         PR tree-optimization/56521
14787         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14788         value-id.
14790 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
14792         PR c++/55135
14793         * except.h (remove_unreachable_eh_regions): New prototype.
14794         * except.c (remove_eh_handler_splicer): New function, split out
14795         of remove_eh_handler.
14796         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
14797         warning about running it on many EH regions one at a time.
14798         (remove_unreachable_eh_regions_worker): New function, walk the
14799         EH tree in depth-first order and remove non-marked regions.
14800         (remove_unreachable_eh_regions): New function.
14801         * tree-eh.c (mark_reachable_handlers): New function, split out
14802         from remove_unreachable_handlers.
14803         (remove_unreachable_handlers): Use mark_reachable_handlers and
14804         remove_unreachable_eh_regions.
14805         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14806         and remove_unreachable_eh_regions.
14808 2013-03-05  Richard Biener  <rguenther@suse.de>
14810         PR middle-end/56525
14811         * loop-init.c (fix_loop_structure): Remove loops in two stages,
14812         not freeing them until the end.
14814 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14816         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14818 2013-03-05  Richard Biener  <rguenther@suse.de>
14820         PR tree-optimization/56270
14821         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14822         of loads after scheduling an SLP instance.
14824 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14826         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14827         tic6x.exp.
14828         (check_gcc_parallelize): Run guality.exp as a separate job from
14829         vect.exp with unsorted.exp and $(dg_target_exps) separately from
14830         struct-layout-1.exp with stackalign.exp.
14832         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14834         PR middle-end/56461
14835         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14836         load_index sbitmap even if some bit in it isn't set.
14838         PR middle-end/56461
14839         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14840         (discover_iteration_bound_by_body_walk): Change queues to
14841         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
14842         spelling in comment.  Call safe_push on queues[bound_index] directly.
14843         Release queues[queue_index] in every iteration unconditionally.
14844         Release bounds vector.
14846         PR middle-end/56461
14847         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14848         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14849         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14850         inner_phis vector.
14852 2013-03-05  Richard Biener  <rguenther@suse.de>
14854         PR lto/56515
14855         * tree-inline.c (remap_blocks_to_null): New function.
14856         (expand_call_inline): When expanding a call stmt without
14857         an associated block inline remap all callee blocks to NULL.
14859 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14861         PR rtl-optimization/56494
14862         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14863         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14864         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14866         PR middle-end/56461
14867         * sel-sched-ir.c (free_sched_pools): Release
14868         succs_info_pool.stack[succs_info_pool.max_top] vectors too
14869         if succs_info_pool.max_top isn't -1.
14871         PR bootstrap/56509
14872         * opts.c (opts_obstack, opts_concat): Moved to...
14873         * opts-common.c (opts_obstack, opts_concat): ... here.
14875 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14877         PR middle-end/56461
14878         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14880 2013-03-04  Martin Jambor  <mjambor@suse.cz>
14882         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14883         all appropriate places.
14885 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
14887         PR tree-optimization/56424
14888         * ipa-split.c (split_function): Do not set the RSO flag if result is
14889         not by reference and its type is a register type.
14891 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
14893         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14894         (microblaze_legitimate_pic_operand): Likewise
14895         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14896         new function microblaze_legitimate_pic_operand
14897         * config/microblaze/microblaze-protos.h
14898         (microblaze_legitimate_pic_operand): Declare.
14900 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14902         * config/microblaze/predicates.md (call_insn_simple_operand):
14903         New predicate for supported rtx code types.
14904         * config/microblaze/microblaze.md (call_internal1): Use
14905         call_insn_simple_operand predicate.
14907 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14909         PR middle-end/56461
14910         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14911         partitions.ordered_remove.
14913         PR middle-end/56461
14914         * tree-vect-stmts.c (vectorizable_conversion): Don't call
14915         vec_oprnds0.create (1) for modifier == NONE.
14917         PR middle-end/56461
14918         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14919         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14920         vec_oprnds1 right before pushing anything to it for
14921         scalar_shift_arg.
14923         PR middle-end/56461
14924         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14925         set nbbs to 0 instead of having separate code path.
14926         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14927         instead of false as last argument if returning NULL.
14929 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
14931         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14932         the attribute is now called "target" instead of "option".
14933         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14934         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
14935         attribute/pragma name for TARGET_OPTION_VALID_P and
14936         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
14937         * doc/tm.texi: Regenerated.
14939 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
14941         * config/microblaze/microblaze.c:
14942         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14943         * config/microblaze/microblaze.h: Add -mxl-reorder to
14944         DRIVER_SELF_SPECS.
14945         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14946         instructions emitted if TARGET_REORDER.
14947         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14948         or 0 for -m/-mno case, but initialises as 2 to detect default use case
14949         separately.
14951 2013-03-01  Xinliang David Li  <davidxl@google.com>
14953         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14954         walk length.
14956 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
14958         PR middle-end/56461
14959         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14960         vector even when returning true.  Fix up function comment formatting.
14962         PR middle-end/56461
14963         * ira-build.c (ira_loop_nodes_count): New variable.
14964         (create_loop_tree_nodes): Initialize it.
14965         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14967         PR middle-end/56461
14968         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14969         method on dr_chain and result_chain.
14970         * tree-vect-stmts.c (vectorizable_store): Only call
14971         result_chain.create if j == 0.
14973         PR middle-end/56461
14974         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14975         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14976         before overwriting it.
14978 2013-03-01  Tobias Burnus  <burnus@net-b.de>
14980         * doc/extended.texi (C Extensions): Change order in @menu
14981         to match @node.
14982         (Other MIPS Built-in Functions): Move last MIPS entry before
14983         "picoChip Built-in Functions".
14984         (SH Built-in Functions): Move after RX Built-in Functions.
14985         * doc/gcc.texi (Introduction): Change order in @menu
14986         to match @node.
14987         * doc/md.texi (Constraints): Ditto.
14988         * gty.texi (Type Information): Ditto.
14989         (User-provided marking routines for template types): Make
14990         subsection.
14991         * doc/invoke.texi (AArch64 Options): Move before
14992         "Adapteva Epiphany Options".
14994 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
14995             Jakub Jelinek  <jakub@redhat.com>
14997         PR sanitizer/56454
14998         * asan.c (gate_asan): Lookup no_sanitize_address instead of
14999         no_address_safety_analysis attribute.
15000         * doc/extend.texi (no_address_safety_attribute): Rename to
15001         no_sanitize_address attribute, mention no_address_safety_analysis
15002         attribute as deprecated alias.
15004 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
15006         PR middle-end/56461
15007         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15008         type to vec<vec<tree> > *.
15009         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
15010         to be vec<tree> instead of vec<tree> *, set vec_defs
15011         to vNULL and call vec_defs.create (number_of_vects), adjust other
15012         uses of vec_defs.
15013         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15014         vectorizable_condition): Adjust vect_get_slp_defs callers.
15016 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15018         * config/aarch64/aarch64.c
15019         (aarch64_float_const_representable): Remove unused variable.
15021 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15023         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15025 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15027         * config/aarch64/aarch64-builtins.c
15028         (aarch64_init_simd_builtins): Make static.
15030 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15032         * config/aarch64/aarch64.c
15033         (aarch64_simd_make_constant): Make static.
15035 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15037         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15038         with no initialization to the RHS of debug statements.
15040 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15042         PR tree-optimization/56294
15043         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15044         Adjust dumping.
15045         (get_access_replacement): Do not call create_access_replacement.
15046         Assert a replacement exists.
15047         (get_repl_default_def_ssa_name): Create the replacement declaration
15048         itself.
15050 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15052         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15053         final_end_function.
15055 2013-02-28  Marek Polacek  <polacek@redhat.com>
15057         PR rtl-optimization/56466
15058         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15059         if we're changing a loop.
15060         (peel_loops_completely): Likewise.
15062 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
15064         PR c++/55813
15065         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15067 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15069         PR target/56445
15070         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15071         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15072         INTX_FTYPE_FX, FX_FTYPE_INTX.
15073         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15075 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15077         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15078         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15079         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15080         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15081         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15082         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15083         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15084         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15085         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15086         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15087         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15088         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15089         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15090         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15091         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15092         (avrxmega6): Increase max flash segments from 5 to 6.
15093         * config/avr/t-multilib: Regenerate.
15094         * config/avr/avr-tables.opt: Regenerate.
15095         * doc/avr-mmcu.texi: Regenerate.
15097 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15099         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15100         (avr_device_to_arch): Rename to avr_device_to_ld.
15101         (avr_device_to_as): New prototype.
15102         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15103         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15104         * config/avr/driver-avr.c (avr_device_to_as): New.
15105         (avr_device_to_arch): Rename to avr_device_to_ld.
15107 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15109         PR middle-end/56461
15110         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15111         method on dr_chain and result_chain.
15113         PR middle-end/56461
15114         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15115         pointer_set_destroy on not_executed_last_iteration.
15117         PR middle-end/56461
15118         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15120         PR middle-end/56461
15121         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15122         FOR_EACH_DEFINED_FUNCTION when freeing state.
15124         PR middle-end/56461
15125         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15126         pool_free.
15127         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15128         overwriting it.
15130         PR middle-end/56461
15131         * ipa-cp.c (decide_whether_version_node): Call vec_free on
15132         known_aggs[i].items and release known_aggs vector.
15134         PR middle-end/56461
15135         * ipa-reference.c (propagate): Free node_info even for alias nodes.
15137 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15139         * config/microblaze/microblaze.c (microblaze_emit_compare):
15140         Use xor for EQ/NE comparisions.
15141         * config/microblaze/microblaze.md (cstoresf4): Add constraints
15142         (cbranchsf4): Adjust operator to comparison_operator.
15144 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15146         PR middle-end/56461
15147         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15148         vector.
15149         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15150         vec_safe_push, always update *slot.
15151         (redirect_edge_var_map_clear): Use vec_free.
15152         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15153         (free_var_map_entry): Use vec_free.
15154         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15155         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15157 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
15159         PR middle-end/45472
15160         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15161         when the may_trap_p bit of the exprs being merged differs.
15162         Reorder tests for speculativeness in the logical and operator.
15164 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15166         * incpath.c (add_standard_paths): Use reconcat instead of concat
15167         where appropriate and avoid leaking memory.
15169         * opts.h: Include obstack.h.
15170         (opts_concat): New prototype.
15171         (opts_obstack): New declaration.
15172         * opts.c (opts_concat): New function.
15173         (opts_obstack): New variable.
15174         (init_options_struct): Call gcc_init_obstack on opts_obstack.
15175         (finish_options): Use opts_concat instead of concat
15176         and XOBNEWVEC instead of XNEWVEC.
15177         * opts-common.c (generate_canonical_option, decode_cmdline_option,
15178         generate_option): Likewise.
15179         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15180         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15182         PR target/56455
15183         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15184         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15186 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15188         PR middle-end/56461
15189         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15191 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15193         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15194         (arm_block_move_unaligned_straight): Likewise.
15195         (arm_adjust_block_mem): Likewise.
15197 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15199         PR target/48901
15200         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15201         temp, cond and label.
15202         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15204         PR target/52500
15205         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15206         * config/c6x/c6x.h (dbx_register_map): Update declaration.
15208         PR target/52501
15209         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15210         of prologue/epilogue functions.
15212         PR target/52550
15213         * config/tilegx/tilegx.c (tilegx_expand_prologue):
15214         Remove unused variable cfa_offset.
15215         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15217         PR target/54639
15218         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15219         type promotion to unsigned.
15221         PR target/54640
15222         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15223         for HOST_WIDE_INT of 32 bit / same size as int.
15224         (arm_block_move_unaligned_straight): Likewise.
15225         (arm_adjust_block_mem): Likewise.
15227         PR target/54662
15228         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15229         ALL_CFLAGS.
15231 2013-02-26  Marek Polacek  <polacek@redhat.com>
15233         PR tree-optimization/56426
15234         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15236 2013-02-26  Richard Biener  <rguenther@suse.de>
15238         PR target/56444
15239         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15240         unused variable loops.
15242 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15244         PR tree-optimization/56448
15245         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15246         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15247         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15248         later operands of the references, or even first operand for
15249         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15251         PR tree-optimization/56443
15252         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15253         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15254         to type_for_mode langhook.
15256 2013-02-25  Matt Turner  <mattst88@gmail.com>
15258         * doc/invoke.texi: Document r4700.
15260 2013-02-25  Richard Biener  <rguenther@suse.de>
15262         PR tree-optimization/56175
15263         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15264         split out from ...
15265         (simplify_bitwise_binary): ... here.  Also guard the conversion
15266         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15268 2013-02-25  Catherine Moore  <clm@codesourcery.com>
15270         Revert:
15271         2013-02-24  Catherine Moore  <clm@codesourcery.com>
15272             Maciej W. Rozycki  <macro@codesourcery.com>
15273             Tom de Vries  <tom@codesourcery.com>
15274             Nathan Sidwell  <nathan@codesourcery.com>
15275             Iain Sandoe  <iain@codesourcery.com>
15276             Nathan Froyd  <froydnj@codesourcery.com>
15277             Chao-ying Fu  <fu@mips.com>
15279         * doc/extend.texi: (micromips, nomicromips, nocompression):
15280         Document new function attributes.
15281         * doc/invoke.texi (minterlink-compressed, mmicromips,
15282         m14k, m14ke, m14kec): Document new options.
15283         (minterlink-mips16): Update documentation.
15284         * doc/md.texi (ZC, ZD): Document new constraints.
15285         * configure.ac (gcc_cv_as_micromips): Check if linker
15286         supports the .set micromips directive.
15287         * configure: Regenerate.
15288         * config.in: Regenerate.
15289         * config/mips/mips-tables.opt: Regenerate.
15290         * config/mips/micromips.md: New file.
15291         * constraints.md (ZC, AD): New constraints.
15292         * config/mips/predicates.md (movep_src_register): New predicate.
15293         (movep_src_operand): New predicate.
15294         (non_volatile_mem_operand): New predicate.
15295         * config/mips/mips.md (multimem): New type.
15296         (length): Differentiate between 17-bit and 18-bit branch offsets.
15297         (MOVEP1, MOVEP2): New mode iterator.
15298         (mov_<load>l): Use ZC constraint.
15299         (mov_<load>r): Likewise.
15300         (mov_<store>l): Likewise.
15301         (mov_<store>r): Likewise.
15302         (*branch_equality<mode>_inverted): Add microMIPS support.
15303         (*branch_equality<mode>): Likewise.
15304         (*jump_absolute): Likewise.
15305         (indirect_jump_<mode>): Likewise.
15306         (tablejump_<mode>): Likewise.
15307         (<optab>_internal): Likewise.
15308         (sibcall_internal): Likewise.
15309         (sibcall_value_internal): Likewise.
15310         (prefetch): Use constraint ZD.
15311         * config/mips/mips.opt (minterlink-compressed): New option.
15312         (minterlink-mips16): Now an alias for minterlink-compressed.
15313         (mmicromips): New option.
15314         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15315         (compare_and_swap_12): Likewise.
15316         (sync_add<mode>): Likewise.
15317         (sync_<optab>_12): Likewise.
15318         (sync_old_<optab>_12): Likewise.
15319         (sync_new_<optab>_12): Likewise.
15320         (sync_nand_12): Likewise.
15321         (sync_old_nand_12): Likewise.
15322         (sync_new_nand_12): Likewise.
15323         (sync_sub<mode>): Likewise.
15324         (sync_old_add<mode>): Likewise.
15325         (sync_old_sub<mode>): Likewise.
15326         (sync_new_add<mode>): Likewise.
15327         (sync_new_sub<mode>): Likewise.
15328         (sync_<optab><mode>): Likewise.
15329         (sync_old_<optab><mode>): Likewise.
15330         (sync_new_<optab><mode>): Likewise.
15331         (sync_nand<mode>): Likewise.
15332         (sync_old_nand<mode>): Likewise.
15333         (sync_new_nand<mode>): Likewise.
15334         (sync_lock_test_and_set<mode>): Likewise.
15335         (test_and_set_12): Likewise.
15336         (atomic_compare_and_swap<mode>): Likewise.
15337         (atomic_exchange<mode>_llsc): Likewise.
15338         (atomic_fetch_add<mode>_llsc): Likewise.
15339         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15340         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15341         (umips_save_restore_pattern_p): Likewise.
15342         (umips_load_store_pair_p): Likewise.
15343         (umips_output_load_store_pair): Likewise.
15344         (umips_movep_target_p): Likewise.
15345         (umips_12bit_offset_address_p): Likewise.
15346         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15347         (mips_base_mips16): Rename this...
15348         (mips_base_compression_flags): ...to this. Update all uses.
15349         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15350         (mips_mips16_decl_p): Delete.
15351         (mips_nomips16_decl_p): Delete.
15352         (mips_get_compress_on_flags): New function.
15353         (mips_get_compress_off_flags): New function.
15354         (mips_get_compress_mode): New function.
15355         (mips_get_compress_on_name): New function.
15356         (mips_get_compress_off_name): New function.
15357         (mips_insert_attributes): Support multiple compression types.
15358         (mips_merge_decl_attributes): Likewise.
15359         (umips_12bit_offset_address_p): New function.
15360         (mips_start_function_definition): Emit .set micromips directive.
15361         (mips_call_may_need_jalx_p): New function.
15362         (mips_function_ok_for_sibcall): Add microMIPS support.
15363         (mips_print_operand_punctuation): Support short delay slots and
15364         compact jumps.
15365         (umips_swm_mask, umips_swm_encoding): New.
15366         (umips_build_save_restore): New function.
15367         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15368         (was_mips16_p): Remove.
15369         (old_compression_mode): New.
15370         (mips_set_compression_mode): New function.
15371         (mips_set_current_function): Add microMIPS support.
15372         (mips_option_override): Likewise.
15373         (umips_save_restore_pattern_p): New function.
15374         (umips_output_save_restore): New function.
15375         (umips_load_store_pair_p_1): New function.
15376         (umips_load_store_pair_p): New function.
15377         (umips_output_load_store_pair_1): New function.
15378         (umips_output_load_store_pair): New function.
15379         (umips_movep_target_p) New function.
15380         (mips_prepare_pch_save): Add microMIPS support.
15381         * config/mips/mips.h (TARGET_COMPRESSION): New.
15382         (TARGET_CPU_CPP_BUILTINS): Update macro
15383         to use new compression flags and to support microMIPS.
15384         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15385         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15386         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15387         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15388         (ASM_SPEC): Support mmicromips and mno-micromips.
15389         (M16STORE_REG_P): New macro.
15390         (MIPS_CALL): Support TARGET_MICROMIPS.
15391         (MICROMIPS_J): New macro.
15392         (mips_base_mips16): Rename this...
15393         (mips_base_compression_flags): ...to this.
15394         (UMIPS_12BIT_OFFSET_P): New macro.
15395         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15396         (MULTILIB_DIRNAMES): Likewise.
15398 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15400         PR rtl-optimization/56131
15401         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15402         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15403         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15405 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15407         * doc/invoke.texi (-fsanitize=): Move from optimization
15408         to debugging options.
15410 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15412         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15414 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15415             Alexander Monakov  <amonakov@ispras.ru>
15417         PR middle-end/56077
15418         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15419         flush pending lists also on non-jumps.  Adjust comment.
15421 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15422             Maciej W. Rozycki  <macro@codesourcery.com>
15423             Tom de Vries  <tom@codesourcery.com>
15424             Nathan Sidwell  <nathan@codesourcery.com>
15425             Iain Sandoe  <iain@codesourcery.com>
15426             Nathan Froyd  <froydnj@codesourcery.com>
15427             Chao-ying Fu  <fu@mips.com>
15429         * doc/extend.texi: (micromips, nomicromips, nocompression):
15430         Document new function attributes.
15431         * doc/invoke.texi (minterlink-compressed, mmicromips,
15432         m14k, m14ke, m14kec): Document new options.
15433         (minterlink-mips16): Update documentation.
15434         * doc/md.texi (ZC, ZD): Document new constraints.
15435         * configure.ac (gcc_cv_as_micromips): Check if linker
15436         supports the .set micromips directive.
15437         * configure: Regenerate.
15438         * config.in: Regenerate.
15439         * config/mips/mips-tables.opt: Regenerate.
15440         * config/mips/micromips.md: New file.
15441         * constraints.md (ZC, AD): New constraints.
15442         * config/mips/predicates.md (movep_src_register): New predicate.
15443         (movep_src_operand): New predicate.
15444         (non_volatile_mem_operand): New predicate.
15445         * config/mips/mips.md (multimem): New type.
15446         (length): Differentiate between 17-bit and 18-bit branch offsets.
15447         (MOVEP1, MOVEP2): New mode iterator.
15448         (mov_<load>l): Use ZC constraint.
15449         (mov_<load>r): Likewise.
15450         (mov_<store>l): Likewise.
15451         (mov_<store>r): Likewise.
15452         (*branch_equality<mode>_inverted): Add microMIPS support.
15453         (*branch_equality<mode>): Likewise.
15454         (*jump_absolute): Likewise.
15455         (indirect_jump_<mode>): Likewise.
15456         (tablejump_<mode>): Likewise.
15457         (<optab>_internal): Likewise.
15458         (sibcall_internal): Likewise.
15459         (sibcall_value_internal): Likewise.
15460         (prefetch): Use constraint ZD.
15461         * config/mips/mips.opt (minterlink-compressed): New option.
15462         (minterlink-mips16): Now an alias for minterlink-compressed.
15463         (mmicromips): New option.
15464         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15465         (compare_and_swap_12): Likewise.
15466         (sync_add<mode>): Likewise.
15467         (sync_<optab>_12): Likewise.
15468         (sync_old_<optab>_12): Likewise.
15469         (sync_new_<optab>_12): Likewise.
15470         (sync_nand_12): Likewise.
15471         (sync_old_nand_12): Likewise.
15472         (sync_new_nand_12): Likewise.
15473         (sync_sub<mode>): Likewise.
15474         (sync_old_add<mode>): Likewise.
15475         (sync_old_sub<mode>): Likewise.
15476         (sync_new_add<mode>): Likewise.
15477         (sync_new_sub<mode>): Likewise.
15478         (sync_<optab><mode>): Likewise.
15479         (sync_old_<optab><mode>): Likewise.
15480         (sync_new_<optab><mode>): Likewise.
15481         (sync_nand<mode>): Likewise.
15482         (sync_old_nand<mode>): Likewise.
15483         (sync_new_nand<mode>): Likewise.
15484         (sync_lock_test_and_set<mode>): Likewise.
15485         (test_and_set_12): Likewise.
15486         (atomic_compare_and_swap<mode>): Likewise.
15487         (atomic_exchange<mode>_llsc): Likewise.
15488         (atomic_fetch_add<mode>_llsc): Likewise.
15489         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15490         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15491         (umips_save_restore_pattern_p): Likewise.
15492         (umips_load_store_pair_p): Likewise.
15493         (umips_output_load_store_pair): Likewise.
15494         (umips_movep_target_p): Likewise.
15495         (umips_12bit_offset_address_p): Likewise.
15496         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15497         (mips_base_mips16): Rename this...
15498         (mips_base_compression_flags): ...to this. Update all uses.
15499         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15500         (mips_mips16_decl_p): Delete.
15501         (mips_nomips16_decl_p): Delete.
15502         (mips_get_compress_on_flags): New function.
15503         (mips_get_compress_off_flags): New function.
15504         (mips_get_compress_mode): New function.
15505         (mips_get_compress_on_name): New function.
15506         (mips_get_compress_off_name): New function.
15507         (mips_insert_attributes): Support multiple compression types.
15508         (mips_merge_decl_attributes): Likewise.
15509         (umips_12bit_offset_address_p): New function.
15510         (mips_start_function_definition): Emit .set micromips directive.
15511         (mips_call_may_need_jalx_p): New function.
15512         (mips_function_ok_for_sibcall): Add microMIPS support.
15513         (mips_print_operand_punctuation): Support short delay slots and
15514         compact jumps.
15515         (umips_swm_mask, umips_swm_encoding): New.
15516         (umips_build_save_restore): New function.
15517         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15518         (was_mips16_p): Remove.
15519         (old_compression_mode): New.
15520         (mips_set_compression_mode): New function.
15521         (mips_set_current_function): Add microMIPS support.
15522         (mips_option_override): Likewise.
15523         (umips_save_restore_pattern_p): New function.
15524         (umips_output_save_restore): New function.
15525         (umips_load_store_pair_p_1): New function.
15526         (umips_load_store_pair_p): New function.
15527         (umips_output_load_store_pair_1): New function.
15528         (umips_output_load_store_pair): New function.
15529         (umips_movep_target_p) New function.
15530         (mips_prepare_pch_save): Add microMIPS support.
15531         * config/mips/mips.h (TARGET_COMPRESSION): New.
15532         (TARGET_CPU_CPP_BUILTINS): Update macro
15533         to use new compression flags and to support microMIPS.
15534         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15535         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15536         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15537         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15538         (ASM_SPEC): Support mmicromips and mno-micromips.
15539         (M16STORE_REG_P): New macro.
15540         (MIPS_CALL): Support TARGET_MICROMIPS.
15541         (MICROMIPS_J): New macro.
15542         (mips_base_mips16): Rename this...
15543         (mips_base_compression_flags): ...to this.
15544         (UMIPS_12BIT_OFFSET_P): New macro.
15545         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15546         (MULTILIB_DIRNAMES): Likewise.
15548 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15550         PR target/52555
15551         * target-globals.c (save_target_globals): For init_reg_sets and
15552         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15554 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15556         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15557         * config/aarch64/t-aarch64
15558         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15560 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15562         PR inline-asm/56148
15563         * lra-constraints.c (process_alt_operands): Reload operand
15564         conflicting with earlier clobber only if no more other conflicting
15565         operands.
15567 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15569         PR sanitizer/56393
15570         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15571         if not linking a shared library.
15573 2013-02-22  Seth LaForge  <sethml@google.com>
15575         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15577 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15579         * config/arm/arm.md (split for extendsidi): Update condition.
15580         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15581         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15582         (qhs_zextenddi_cstr): Likewise.
15584 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15586         PR middle-end/56420
15587         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15588         avoid signed wrapping.
15589         (expand_mult): Handle properly multiplication by
15590         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15591         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15592         in the compiler if coeff is HOST_WIDE_INT_MIN.
15593         (expand_divmod): Don't make ext_op1 static, change it's type to
15594         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15596         PR rtl-optimization/50339
15597         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15598         field.
15599         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15600         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15601         into splitting_ashiftrt field.
15602         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15603         ASHIFTRT.
15604         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15605         choices.
15607 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
15609         PR middle-end/56108
15610         * trans-mem.c (execute_tm_mark): Do not expand transactions that
15611         are sure to go irrevocable.
15613 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
15615         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15616         scalars are valid operands.
15618 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15620         PR tree-optimization/56310
15621         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15622         only matching indices and non-negative final offsets.
15623         (intersect_aggregates_with_edge): Pass src_idx to
15624         agg_replacements_to_vector.  Pass src_idx insstead of index to
15625         intersect_with_agg_replacements.
15627 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15629         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15630         instead of hard-wired defaults.
15632 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
15634         * doc/invoke.texi (MIPS Options): Update documentation of the
15635         floating-point multiply-accumulate instruction restrictions.
15637 2013-02-21  Kostya Serebryany  <kcc@google.com>
15639         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15640         asan_shadow_offset on x86_64 linux.
15642 2013-02-21  Richard Biener  <rguenther@suse.de>
15644         PR tree-optimization/56415
15645         Revert
15646         2013-02-11  Richard Biener  <rguenther@suse.de>
15648         PR tree-optimization/56273
15649         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15650         first VRP run.
15652 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15654         PR bootstrap/56258
15655         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15656         instead of @itemx.
15658         PR inline-asm/56405
15659         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15660         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15662 2013-02-20  Jan Hubicka  <jh@suse.cz>
15664         PR tree-optimization/56265
15665         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15666         when target is referenced for first time.
15668 2013-02-20  Richard Biener  <rguenther@suse.de>
15670         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15671         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15672         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15673         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15674         not return anything.
15675         (rename_ssa_copies): Do not remove unused locals.
15676         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15677         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15678         * passes.c (execute_function_todo): Do not schedule unused locals
15679         removal if cleanup_tree_cfg did something.
15680         * tree-ssa-live.c (remove_unused_locals): Dump statistics
15681         about the number of removed locals.
15683 2013-02-20  Richard Biener  <rguenther@suse.de>
15685         PR tree-optimization/56398
15686         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15688 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15690         PR tree-optimization/55334
15691         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15692         restricted pointers to arrays.
15694 2013-02-20  Richard Biener  <rguenther@suse.de>
15695             Jakub Jelinek  <jakub@redhat.com>
15697         PR tree-optimization/56396
15698         * tree-ssa-ccp.c (n_const_val): New static variable.
15699         (get_value): Return NULL for SSA names we don't have a lattice
15700         entry for.
15701         (ccp_initialize): Initialize n_const_val.
15702         * tree-ssa-copy.c (n_copy_of): New static variable.
15703         (init_copy_prop): Initialize n_copy_of.
15704         (get_value): Return NULL_TREE for SSA names we don't have a
15705         lattice entry for.
15707 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15709         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15711 2013-02-20  Richard Biener  <rguenther@suse.de>
15713         * genpreds.c (write_lookup_constraint): Do not compare first
15714         letter of the constraint again.
15716 2013-02-20  Richard Biener  <rguenther@suse.de>
15718         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15719         and ceil_log2.
15720         (get_use_iv_cost): Terminate hashtable walk when coming across
15721         an empty entry.
15723 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
15725         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15726         reassociation for avx2 targets.
15728 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15730         * config/microblaze/microblaze.c: microblaze_has_clz = 0
15731         Add version check for v8.10.a to enable microblaze_has_clz
15732         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15733         version and TARGET_PATTERN_COMPARE check
15734         * config/microblaze/microblaze.md: New clzsi2 instruction
15736 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15738         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15739         function before branching.
15741 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15743         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15744         DUMP_INSN_RTX_UID.
15745         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15747 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15749         PR middle-end/55889
15750         * sel-sched.c: Include ira.h.
15751         (implicit_clobber_conflict_p): New function.
15752         (moveup_expr): Use it.
15753         * Makefile.in (sel-sched.o): Depend on ira.h.
15755 2013-02-19  Richard Biener  <rguenther@suse.de>
15757         PR tree-optimization/56384
15758         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15759         (vn_hash_type): Split out from ...
15760         (vn_hash_constant_with_type): ... here.
15761         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15762         (vn_phi_eq): Compare types from vn_phi_s structure.
15763         (vn_phi_lookup): Populate vn_phi_s type.
15764         (vn_phi_insert): Likewise.
15766 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
15768         PR tree-optimization/56350
15769         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15770         if haven't found reduction or nested cycle operand, rather than
15771         asserting we must find it.
15773         PR tree-optimization/56381
15774         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15775         to fold_build3.
15777 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
15778             Jakub Jelinek  <jakub@redhat.com>
15780         PR target/52555
15781         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15782         (swap_optab_enable): Same.
15783         (init_all_optabs): Use argument instead of global.
15784         * tree.h (struct tree_optimization_option): New field target_optabs.
15785         * expr.h (init_all_optabs): Add argument to prototype.
15786         (TREE_OPTIMIZATION_OPTABS): New.
15787         (save_optabs_if_changed): Protoize.
15788         * optabs.h: Declare this_fn_optabs.
15789         * optabs.c (save_optabs_if_changed): New.
15790         Declare this_fn_optabs.
15791         (init_optabs): Add argument to init_all_optabs() call.
15792         * function.c (invoke_set_current_function_hook): Handle per
15793         function optabs.
15794         * function.h (struct function): New field optabs.
15795         * config/mips/mips.c (mips_set_mips16_mode): Handle when
15796         optimization_current_node has changed.
15797         * target-globals.h (save_target_globals_default_opts): Protoize.
15798         * target-globals.c (save_target_globals_default_opts): New.
15800 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15802         PR target/56347
15803         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15804         registers %fr12 and %fr12R as call used.
15806         PR target/56214
15807         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15808         and HImode, require all displacements to be an integer multiple of
15809         their mode size.
15810         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15811         only allow QImode and HImode when reload is in progress and strict is
15812         true.  Likewise for symbolic addresses.  Use base14_operand to check
15813         displacements in REG+BASE addresses.
15815 2013-02-18  Richard Biener  <rguenther@suse.de>
15817         PR tree-optimization/56366
15818         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15819         handle sign-conversion of outer-loop initial induction value.
15821 2013-02-18  Richard Biener  <rguenther@suse.de>
15823         PR middle-end/56349
15824         * cfghooks.c (merge_blocks): If we merge a latch into another
15825         block adjust references to it.
15826         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15827         (verify_loop_structure): Verify that a recorded latch is in fact
15828         a latch.
15830 2013-02-18  Richard Biener  <rguenther@suse.de>
15832         PR tree-optimization/56321
15833         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15834         order SSA name release and virtual operand unlinking.
15836 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15838         * config/microblaze/microblaze.md (save_stack_block): Define.
15839         (restore_stack_block): Likewise.
15841 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15843         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15844         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15845         * config/microblaze/microblaze.c (microblaze_option_override):
15846         Bail out early for PIC modes when target does not support PIC.
15848 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15850         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15851         Replace with a microblaze version.
15852         (microblaze_trampoline_init): Adapt for microblaze.
15853         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15854         microblaze.
15856 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
15857             Dodji Seketeli  <dodji@redhat.com>
15859         PR asan/56330
15860         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15861         (instrument_mem_region_access): Do not forget to always put
15862         instrumentation of the of 'base' and 'base + len' in a "if (len !=
15863         0) statement, even for cases where either 'base' or 'base + len'
15864         are not instrumented -- because they have been previously
15865         instrumented.  Simplify the logic by putting all the statements
15866         instrument 'base + len' inside a sequence, and then insert that
15867         sequence right before the current insertion point.  Then, to
15868         instrument 'base + len', just get an iterator on that statement.
15869         And do not forget to update the pointer to iterator the function
15870         received as argument.
15872 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
15874         PR rtl-optimization/56348
15875         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15877 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
15879         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15880         (clean_graph_dump_file): Pass base to start_graph_dump.
15882 2013-02-14  Richard Henderson  <rth@redhat.com>
15884         PR target/55941
15885         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15887 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
15889         * collect2-aix.h: Define F_LOADONLY.
15891 2013-02-14  Richard Biener  <rguenther@suse.de>
15893         PR lto/50494
15894         * varasm.c (output_constant_def_1): Get the decl representing
15895         the constant as argument.
15896         (output_constant_def): Wrap output_constant_def_1.
15897         (make_decl_rtl): Use output_constant_def_1 with the decl
15898         representing the constant.
15899         (build_constant_desc): Optionally re-use a decl already
15900         representing the constant.
15901         (tree_output_constant_def): Adjust.
15903 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
15905         Fix an asan crash
15906         * asan.c (instrument_builtin_call):  Really put the length of the
15907         second source argument into src1_len.
15909 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15911         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15912         argument.  If it is false, don't create edge from then_bb to
15913         fallthru_bb.
15914         (insert_if_then_before_iter): Pass true to it.
15915         (build_check_stmt): Pass false to it.
15916         (transform_statements): Flush hash table only on extended basic
15917         block boundaries, rather than at the beginning of every bb.
15918         Don't flush hash table on nonfreeing_call_p calls.
15919         * tree-flow.h (nonfreeing_call_p): New prototype.
15920         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15922 2013-02-13  David S. Miller  <davem@davemloft.net>
15924         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15926 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
15928         PR target/56184
15929         * ira.c (max_regno_before_ira): Move from ...
15930         (ira): ... here.
15931         (fix_reg_equiv_init): Use max_regno_before_ira instead of
15932         vec_safe_length.
15934 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15936         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15938 2013-02-13  Richard Biener  <rguenther@suse.de>
15940         PR lto/56295
15941         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15942         globals in MEM_REFs.
15944 2013-02-13  Richard Biener  <rguenther@suse.de>
15946         * loop-init.c (loop_optimizer_init): Clear loop state when
15947         re-initializing preserved loops.
15948         * loop-unswitch.c (unswitch_single_loop): Return whether
15949         we unswitched the loop.  Do not verify loop state here.
15950         (unswitch_loops): When we unswitched a loop discover new loops.
15952 2013-02-13  Kostya Serebryany  <kcc@google.com>
15954         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15955         on x86_64 linux.
15956         * sanitizer.def: Rename __asan_init to __asan_init_v1.
15958 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
15960         Avoid instrumenting duplicated memory access in the same basic block
15961         * Makefile.in (asan.o): Add new dependency on hash-table.h
15962         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15963         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15964         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15965         (free_mem_ref_resources, has_mem_ref_been_instrumented)
15966         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15967         (get_mem_ref_of_assignment): New functions.
15968         (get_mem_refs_of_builtin_call): Extract from
15969         instrument_builtin_call and tweak a little bit to make it fit with
15970         the new signature.
15971         (instrument_builtin_call): Use the new
15972         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
15973         of is_gimple_builtin_call.
15974         (instrument_derefs, instrument_mem_region_access): Insert the
15975         instrumented memory reference into the hash table.
15976         (maybe_instrument_assignment): Renamed instrument_assignment into
15977         this, and change it to advance the iterator when instrumentation
15978         actually happened and return true in that case.  This makes it
15979         homogeneous with maybe_instrument_assignment, and thus give a
15980         chance to callers to be more 'regular'.
15981         (transform_statements): Clear the memory reference hash table
15982         whenever we enter a new BB, when we cross a function call, or when
15983         we are done transforming statements.  Use
15984         maybe_instrument_assignment instead of instrumentation.  No more
15985         need to special case maybe_instrument_assignment and advance the
15986         iterator after calling it; it's now handled just like
15987         maybe_instrument_call.  Update comment.
15989 2013-02-13  Richard Biener  <rguenther@suse.de>
15991         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15992         Fix loop discovery code.
15994 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
15996         PR inline-asm/56148
15997         * lra-constraints.c (process_alt_operands): Match early clobber
15998         operand with itself.  Check conflicts with earlyclobber only if
15999         the operand is not reloaded.  Prefer to reload conflicting operand
16000         if earlyclobber and matching operands are the same.
16002 2013-02-12  Richard Biener  <rguenther@suse.de>
16004         PR lto/56297
16005         * lto-streamer-out.c (write_symbol): Do not output symbols
16006         for hard register variables.
16008 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
16010         PR target/54222
16011         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16012         (umulsidi3_insn, mulsidi3_insn): New insns.
16014 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
16016         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16017         (struct tune_params): Add vec_costs field.
16018         * config/arm/arm.c (arm_builtin_vectorization_cost)
16019         (arm_add_stmt_cost): New functions.
16020         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16021         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16022         (arm_default_vec_cost): New struct of type cpu_vec_costs.
16023         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16024         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16025         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16026         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16028 2013-02-12  Richard Biener  <rguenther@suse.de>
16030         PR lto/56295
16031         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16032         decls again if possible.
16034 2013-02-12  Richard Biener  <rguenther@suse.de>
16036         PR middle-end/56288
16037         * tree-ssa.c (verify_ssa_name): Fix check, move
16038         SSA_NAME_IN_FREE_LIST check up.
16040 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
16041             Steven Bosscher   <steven@gcc.gnu.org>
16043         PR rtl-optimization/56151
16044         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16045         equal to op0 or op1, and last_insn pattern is CODE operation
16046         with MEM dest and one of the operands matches that MEM.
16048 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
16050         * doc/extend.texi: Document Function Multiversioning and "default"
16051         parameter string to target attribute.
16052         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16053         target attribute parameter is "default".
16054         (ix86_compare_version_priority): Remove checks for target attribute.
16055         (ix86_mangle_function_version_assembler_name): Change error to sorry.
16056         Remove check for target attribute equal to NULL. Add assert.
16057         (ix86_generate_version_dispatcher_body): Change error to sorry.
16059 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
16060             Jack Howarth  <howarth@bromo.med.uc.edu>
16061             Patrick Marlier  <patrick.marlier@gmail.com>
16063         PR libitm/55693
16064         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16065         define ENDFILE_SPEC as TM_DESTRUCTOR.
16066         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16068 2013-02-11  Alexander Potapenko  <glider@google.com>
16069             Jack Howarth  <howarth@bromo.med.uc.edu>
16070             Jakub Jelinek  <jakub@redhat.com>
16072         PR sanitizer/55617
16073         * config/darwin.c (cdtor_record): Rename ctor_record.
16074         (sort_cdtor_records): Rename sort_ctor_records.
16075         (finalize_dtors): New routine to sort destructors by
16076         priority before use in assemble_integer.
16077         (machopic_asm_out_destructor): Use finalize_dtors if needed.
16079 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16081         PR rtl-optimization/56275
16082         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16083         offset is non-negative and less than cmode size before
16084         calling simplify_subreg.
16086 2013-02-11  Richard Biener  <rguenther@suse.de>
16088         PR tree-optimization/56264
16089         * cfgloop.h (fix_loop_structure): Adjust prototype.
16090         * loop-init.c (fix_loop_structure): Return the number of
16091         newly discovered loops.
16092         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16093         are discovered, do a full loop-closed SSA rewrite.
16095 2013-02-11  Richard Biener  <rguenther@suse.de>
16097         PR tree-optimization/56273
16098         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16099         first VRP run.
16100         (check_array_ref): Fix missing newline in dumps.
16101         (search_for_addr_array): Likewise.
16103 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
16105         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16107 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
16109         PR target/56256
16110         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16112 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
16114         PR rtl-optimization/56246
16115         * lra-constraints.c (simplify_operand_subreg): Try to reuse
16116         reload pseudo.
16117         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16118         constraints are satisfied.
16120 2013-02-08  Jeff Law  <law@redhat.com>
16122         PR debug/53948
16123         * emit-rtl.c (reg_is_parm_p): New function.
16124         * regs.h (reg_is_parm_p): New prototype.
16125         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16126         callee-clobbered registers.
16128 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
16130         PR target/56043
16131         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16132         If there is no implicit builtin declaration, just return NULL.
16134 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16136         * config/i386/sse.md (FMAMODEM): New mode iterator.
16137         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16138         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16140 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16142         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16143         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16144         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16146 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16148         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16149         (microblaze*-*-elf): Likewise.
16150         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16151         LINK_SPEC.
16152         * config/microblaze/microblaze-c.c: Add builtin defines for
16153         _LITTLE_ENDIAN and _BIG_ENDIAN.
16154         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16155         add to TARGET_DEFAULT flags.
16156         Expand ASM_SPEC and LINK_SPEC.
16157         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16158         * config/microblaze/microblaze.md: Update extendsidi2 and
16159         movdi_internal instructions to use low-order / high-order reg
16160         print_operands.
16161         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16162         options and inversemask / mask of LITTLE_ENDIAN.
16163         * config/microblaze/t-microblaze: Expand multilib options to
16164         include mlittle-endian (le) and update exceptions patterns.
16166 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
16168         PR rtl-optimization/56195
16169         * lra-constraints.c (get_reload_reg): Don't reuse regs
16170         if they have smaller mode than requested, if they have
16171         wider mode than requested, try to return a SUBREG.
16173         PR tree-optimization/56250
16174         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16175         if type is unsigned and code isn't MULT_EXPR.
16177 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16179         PR tree-optimization/56064
16180         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16181         bits according to mode.
16182         * fixed-value.h (fixed_from_double_int)
16183         (const_fixed_from_double_int): Adjust comments.
16185 2013-02-08  Richard Biener  <rguenther@suse.de>
16187         PR lto/56231
16188         * lto-streamer.h (struct data_in): Remove current_file, current_line
16189         and current_col members.
16190         * lto-streamer-out.c (lto_output_location): Stream changed bits
16191         en-block for efficiency.
16192         * lto-streamer-in.c (clear_line_info): Remove.
16193         (lto_input_location): Cache current file, line and column
16194         globally via local statics.  Read changed bits en-block.
16195         (input_function): Do not call clear_line_info.
16196         (lto_read_body): Likewise.
16197         (lto_input_toplevel_asms): Likewise.
16199 2013-02-08  Michael Matz  <matz@suse.de>
16201         PR tree-optimization/52448
16202         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16203         (nt_call_phase): New static.
16204         (add_or_mark_expr): Only mark accesses with newer phase than any
16205         call seen.
16206         (nonfreeing_call_p): New.
16207         (nt_init_block): Update nt_call_phase, mark blocks as visited.
16208         (nt_fini_block): Keep blocks marked as visited.
16209         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16211 2013-02-08  Richard Biener  <rguenther@suse.de>
16213         * ira.c (ira): Free broken dominator information.
16215 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16217         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16219 2013-02-08  Marek Polacek  <polacek@redhat.com>
16221         * cfgloop.c (verify_loop_structure): Add more checking of headers.
16223 2013-02-08  Richard Biener  <rguenther@suse.de>
16225         PR middle-end/56181
16226         * cfgloop.h (flow_loops_find): Adjust.
16227         (bb_loop_header_p): Declare.
16228         * cfgloop.c (bb_loop_header_p): New function split out from ...
16229         (flow_loops_find): ... here.  Adjust function signature,
16230         support incremental loop structure update.
16231         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
16232         * cfgloopmanip.c (fix_loop_structure): Move ...
16233         * loop-init.c (fix_loop_structure): ... here.
16234         (apply_loop_flags): Split out from ...
16235         (loop_optimizer_init): ... here.
16236         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
16237         in incremental mode, only remove dead loops here.
16239 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16241         PR target/54222
16242         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16243         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16244         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16245         (*round<mode>3.libgcc): New insns for fixed-modes.
16246         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16247         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16248         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16249         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16250         implementations.  Define to __builtin_avr_absFX,
16251         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16252         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16253         __builtin_avr_countlsFX, respectively.
16254         * config/avr/avr-c.c (target.h): Include it.
16255         (enum avr_builtin_id): New enum.
16256         (avr_resolve_overloaded_builtin): New static function.
16257         (avr_register_target_pragmas): Use it to set
16258         targetm.resolve_overloaded_builtin.
16259         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16260         tree nodes used by DEF_BUILTIN.
16261         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16262         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16263         <AVR_BUILTIN_xxBITS>: Same.
16265 2013-02-08  Richard Biener  <rguenther@suse.de>
16267         * cfgloop.c (verify_loop_structure): Properly handle
16268         a loop exiting to another loop header.
16269         * ira-int.h (ira_loops): Remove.
16270         * ira.c (ira_loops): Remove.
16271         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16272         (do_reload): Use loop_optimizer_finalize.
16273         * ira-build.c (create_loop_tree_nodes): Use get_loops and
16274         number_of_loops to access the loop tree.
16275         (more_one_region_p): Likewise.
16276         (finish_loop_tree_nodes): Likewise.
16277         (rebuild_regno_allocno_maps): Likewise.
16278         (mark_loops_for_removal): Likewise.
16279         (mark_all_loops_for_removal): Likewise.
16280         (remove_unnecessary_regions): Likewise.
16281         (ira_build): Likewise.
16282         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16284 2013-02-08  Richard Biener  <rguenther@suse.de>
16286         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16287         * ipa-pure-const.c (analyze_function): Avoid calling
16288         mark_irreducible_loops twice.
16289         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16291 2013-02-07  David S. Miller  <davem@davemloft.net>
16293         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16294         on 'reg'.
16295         * var-tracking.c (vt_add_function_parameter): Test the presence of
16296         HAVE_window_save properly and do not remap argument registers when
16297         we have a leaf function.
16299 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
16301         PR bootstrap/56227
16302         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16303         instead of "ll".
16304         * config/i386/i386.c (ix86_print_operand): Ditto.
16306 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16308         * lra-constraints.c (process_alt_operands): Fix recently added comment.
16310 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16312         PR rtl-optimization/56225
16313         * lra-constraints.c (process_alt_operands): Check that reload hard
16314         reg can hold value for strict_low_part.
16316 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16318         PR debug/56154
16319         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16320         dwarf2out_end_function.
16321         (in_first_function_p, maybe_at_text_label_p,
16322         first_loclabel_num_not_at_text_label): New variables.
16323         (dwarf2out_var_location): In the first function find out
16324         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16325         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16326         functions.
16328 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
16330         PR rtl-optimization/56178
16331         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16332         SUBREG of a register.  Tidy up related block of code.
16333         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16334         note if the source is a register or a SUBREG of a register.
16336 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16338         PR target/56228
16339         * config/rs6000/rs6000.md (ptrm): New mode attr.
16340         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16341         call_value_indirect_aix<pttrsize>,
16342         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16343         m in constraints.
16345 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16347         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16348         if -bnortl. Convert to strcmp and strncmp.
16350 2013-02-07  Alan Modra  <amodra@gmail.com>
16352         PR target/54009
16353         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16354         addresses won't wrap when offsetting.
16355         (rs6000_secondary_reload): Provide secondary reloads needed for
16356         wrapping LO_SUM addresses.
16358 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16360         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16361         MACH, just __MACH__.
16363 2013-02-06  Richard Biener  <rguenther@suse.de>
16365         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16366         instead of calling fix_loop_structure.
16368 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16370         PR middle-end/56217
16371         * omp-low.c (use_pointer_for_field): Return false if
16372         lower_send_shared_vars doesn't generate any copy-out code.
16374 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16376         PR rtl-optimization/56131
16377         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16378         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16379         of the label is NULL.  Add comment.
16381 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16383         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16385         PR sanitizer/55374
16386         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16387         (STATIC_LIBTSAN_LIBS): Likewise.
16388         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16389         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16390         is defined, don't add anything else beyond that.
16391         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16392         (LINK_COMMAND_SPEC): Use them.
16394         PR tree-optimization/56205
16395         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16396         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16397         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16399 2013-02-05  Richard Biener  <rguenther@suse.de>
16401         PR tree-optimization/53342
16402         PR tree-optimization/53185
16403         * tree-vectorizer.h (vect_check_strided_load): Remove.
16404         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16405         not disallow peeling for vectorized strided loads.
16406         (vect_check_strided_load): Make static and simplify.
16407         (vect_analyze_data_refs): Adjust.
16408         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16409         correctly when vectorizing strided loads.
16411 2013-02-05  Richard Biener  <rguenther@suse.de>
16413         * doc/install.texi: Refer to ISL, not PPL.
16415 2013-02-05  Jan Hubicka  <jh@suse.cz>
16417         PR tree-optimization/55789
16418         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16420 2013-02-05  Jan Hubicka  <jh@suse.cz>
16422         PR tree-optimization/55789
16423         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16424         the dead call anyway.
16426 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16428         PR sanitizer/55374
16429         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16431 2013-02-04  Alexander Potapenko  <glider@google.com>
16432             Jack Howarth  <howarth@bromo.med.uc.edu>
16433             Jakub Jelinek  <jakub@redhat.com>
16435         PR sanitizer/55617
16436         * config/darwin.c (sort_ctor_records): Stabilized qsort
16437         on constructor priority by using original position.
16438         (finalize_ctors): New routine to sort constructors by
16439         priority before use in assemble_integer.
16440         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16442 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16444         PR libstdc++/54314
16445         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16446         about visibility on artificial decls.
16447         * config/sol2.c (solaris_assemble_visibility): Likewise.
16449 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16451         PR target/56186
16452         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16453         argument and improve checking of return-argument types for 16-byte
16454         modes.
16455         (ix86_function_value_1): Add additional valtype argument on call
16456         of function_value_64.
16457         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16458         handling infunction_value_64 function.
16460 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16462         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16464 2013-02-04  Richard Biener  <rguenther@suse.de>
16466         PR tree-optimization/56188
16467         * tree-ssa-structalias.c (label_visit): Consider case with
16468         initially non-empty points-to set.
16469         (perform_var_substitution): Dump node mapping and clean up.
16471 2013-02-04  Richard Guenther  <rguenther@suse.de>
16473         PR lto/56168
16474         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16475         node prevail as last resort.
16476         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16477         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16479 2013-02-04  Richard Biener  <rguenther@suse.de>
16481         PR tree-optimization/56113
16482         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16483         Merge into ...
16484         (equiv_class_lookup_or_add): ... this.
16485         (label_visit): Adjust and fix error in previous patch.
16486         (perform_var_substitution): Adjust.
16488 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16490         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16491         * config/sh/sh4-300.md: Likewise.
16492         * config/sh/sh4a.md: Likewise.
16493         * config/sh/constraints.md: Likewise.
16494         * config/sh/sh.md: Likewise.
16495         * config/sh/netbsd-elf.h: Likewise.
16496         * config/sh/predicates.md: Likewise.
16497         * config/sh/sh-protos.h: Likewise.
16498         * config/sh/ushmedia.h: Likewise.
16499         * config/sh/linux.h: Likewise.
16500         * config/sh/sh.c: Likewise.
16501         * config/sh/superh.h: Likewise.
16502         * config/sh/elf.h: Likewise.
16503         * config/sh/sh4.md: Likewise.
16504         * config/sh/sh.h: Likewise.
16506 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16508         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16509         constraint to match_test floating_point_store_memory_operand().
16510         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16511         (base14_operand): New.
16512         (floating_point_store_memory_operand): New.
16513         (integer_store_memory_operand): Revise to use base14_operand and
16514         reg_plus_base_memory_operand.
16515         (move_dest_operand): Allow symbolic_memory_operands.
16516         (symbolic_memory_operand): Check for LO_SOM.
16517         (symbolic_operand): Change default case to break.
16518         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16519         CONST_DOUBLE values to be reloaded by putting them into memory when
16520         the destination is a floating point register.
16521         (movdf): Remove code to handle CONST_DOUBLE.
16522         (movsf): Likewise.
16523         (reload_indf_r1): New.
16524         (reload_insf_r1): New.
16525         Consistently use "Q" and "T" constraints with integer and floating
16526         point move instructions, respectively.
16527         (movdi): Remove FAIL.
16528         Change predicate for source operand unamed DImode move from
16529         general_operand to move_src_operand.
16530         (umulsidi3): Change predicate for destination operand to
16531         register_operand.
16532         Likewise for similar unamed patterns.
16533         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16534         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16535         (hppa_legitimize_address): Simplify mask calculation.
16536         (pa_emit_move_sequence): Revised handling of secondary reloads from
16537         REG+D addresses for floating point loads and stores.  Directly handle
16538         loading CONST0_RTX (mode) to a floating point register.
16539         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16540         to floating point registers.  Don't restrict secondary reloads to
16541         floating point registers to integer modes.  Revise some comments and
16542         cleanup some code.
16543         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16544         (pa_legitimate_address_p): New.
16545         (pa_legitimize_reload_address): New.
16546         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16547         (STRICT_REG_OK_FOR_BASE_P): New.
16548         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16549         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16551 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16552             Andrew Dixie  <andrewd@gentrack.com>
16554         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16555         flag set.
16557 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16559         * expmed.c (extract_bit_field_1): Pass the full width of the
16560         structure to get_best_reg_extraction_insn.
16562 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16564         PR target/54601
16565         * configure.ac (use_cxa_atexit): Add AIX.
16566         * configure: Regenerate.
16568         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16570 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16572         PR debug/54793
16573         * final.c (need_profile_function): New variable.
16574         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16575         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16576         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16577         notes, targetm.asm_out.function_prologue doesn't emit anything,
16578         HAVE_prologue and profiler should be emitted before prologue,
16579         set need_profile_function instead of emitting it.
16580         (final_scan_insn): If need_profile_function, emit
16581         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16582         NOTE_INSN_FUNCTION_BEG note.
16584 2013-02-01  Richard Henderson  <rth@redhat.com>
16586         * config/rs6000/rs6000.md (smulditi3): New.
16587         (umulditi3): New.
16589         * config/alpha/alpha.md (umulditi3): New.
16591 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16593         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16594         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16596 2013-02-01  Richard Biener  <rguenther@suse.de>
16598         PR tree-optimization/56113
16599         * tree-ssa-structalias.c (label_visit): Reduce work for
16600         single-predecessor nodes.
16602 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
16604         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16605         range isn't testing for zero.
16607 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
16609         PR middle-end/56113
16610         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16612 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
16613             Nick Clifton  <nickc@redhat.com>
16615         * config/v850/constraints.md (Q): Define as a memory constraint.
16616         * config/v850/predicates.md (label_ref_operand): New predicate.
16617         (e3v5_shift_operand): New predicate.
16618         (ior_operator): New predicate.
16619         * config/v850/t-v850: Add e3v5 multilib.
16620         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16621         (v850_gen_movdi): Prototype.
16622         * config/v850/v850.c: Add support for e3v5 architecture.
16623         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16624         TARGET_V850E_UP.
16625         (construct_save_jarl): Add e3v5 long JARL support.
16626         (v850_adjust_insn_length): New function.  Adjust length of call
16627         insns when using e3v5 instructions.
16628         (v850_gen_movdi): New function: Generate instructions to move a
16629         DImode value.
16630         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16631         (CPP_SPEC): Define __v850e3v5__ as appropriate.
16632         (TARGET_USE_FPU): Enable for e3v5.
16633         (CONST_OK_FOR_W): New macro.
16634         (ADJUST_INSN_LENGTH): Define.
16635         * config/v850/v850.md (UNSPEC_LOOP): Define.
16636         (attr cpu): Add v850e3v5.
16637         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16638         (movdi): New pattern.
16639         (movdi_internal): New pattern.
16640         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16641         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16642         (cstoresf4): Likewise.
16643         (cstoredf4): Likewise.
16644         (insv): New pattern.
16645         (rotlso3_a): New pattern.
16646         (rotlsi3_b): New pattern
16647         (rotlsi3_v850e3v5): New pattern.
16648         (doloop_begin): New pattern.
16649         (fix_loop_counter): New pattern.
16650         (doloop_end): New pattern.
16651         (branch_normal): Add e3v5 long branch support.
16652         (branch_invert): Likewise.
16653         (branch_z_normal): Likewise.
16654         (branch_z_invert): Likewise.
16655         (branch_nz_normal): Likewise.
16656         (branch_nz_invert): Likewise.
16657         (call_internal_short): Add e3v5 register-indirect JARL support.
16658         (call_internal_long): Likewise.
16659         (call_value_internal_short): Likewise.
16660         (call_value_internal_long): Likewise.
16661         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16662         (mloop): New option.
16663         * config.gcc: Add support for configuring v840e3v5 target.
16664         * doc/invoke.texi: Document new v850 specific command line options.
16666 2013-01-31  Paul Koning  <ni1d@arrl.net>
16668         PR debug/55059
16669         PR debug/54508
16670         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16671         children if parent is a class.
16672         (prune_unused_types_prune): Don't add DW_AT_declaration.
16674 2013-01-31  Richard Biener  <rguenther@suse.de>
16676         PR tree-optimization/56157
16677         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16678         match up operand with SLP child.
16680 2013-01-31  Jason Merrill  <jason@redhat.com>
16682         PR debug/54410
16683         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16684         parameters the first time.
16685         (gen_scheduled_generic_parms_dies): Check completeness here.
16687 2013-01-31  Richard Biener  <rguenther@suse.de>
16689         PR middle-end/53073
16690         * common.opt (faggressive-loop-optimizations): New flag,
16691         enabled by default.
16692         * doc/invoke.texi (faggressive-loop-optimizations): Document.
16693         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16694         infer_loop_bounds_from_undefined by it.
16696 2013-01-31  Richard Biener  <rguenther@suse.de>
16698         PR tree-optimization/56150
16699         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16700         visit virtual operands.
16701         (find_uses_to_rename_bb): Likewise.
16703 2013-01-31  Richard Biener  <rguenther@suse.de>
16705         PR tree-optimization/56150
16706         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16707         mixed store non-store stmts.
16709 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
16711         PR sanitizer/55374
16712         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16713         LIBASAN_EARLY_SPEC is defined.
16714         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16715         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16716         before %o.
16717         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16719         PR c++/55742
16720         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16721         invalid args instead of ICEing on it.
16722         (ix86_valid_target_attribute_tree): Return error_mark_node if
16723         ix86_valid_target_attribute_inner_p failed.
16724         (ix86_valid_target_attribute_p): Return false only if
16725         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
16726         target("default") attribute.
16727         (sorted_attr_string): Change argument from const char * to tree,
16728         merge in all target attribute arguments rather than just one.
16729         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
16730         instead of free.  Avoid using strcat.
16731         (ix86_mangle_function_version_assembler_name): Mangle
16732         target("default") as if no target attribute is present.  Adjust
16733         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
16734         instead of xmalloc and XDELETEVEC instead of free.
16735         (ix86_function_versions): Don't return true if one of the decls
16736         doesn't have target attribute.  If they don't and one of the decls
16737         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
16738         sorted_attr_string caller.  Use XDELETEVEC instead of free.
16739         (ix86_supports_function_versions): Remove.
16740         (make_name): Fix up formatting.
16741         (make_dispatcher_decl): Remove resolver_name and its initialization.
16742         Avoid leaking memory.
16743         (is_function_default_version): Return true if there is
16744         target("default") attribute rather than no target attribute at all.
16745         (make_resolver_func): Avoid leaking memory.
16746         (ix86_generate_version_dispatcher_body): Likewise.
16747         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16748         * target.def (supports_function_versions): Remove.
16749         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16750         * doc/tm.texi: Regenerated.
16752 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
16754         PR rtl-optimization/56144
16755         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16756         for values with side effects.
16758 2013-01-30  Richard Biener  <rguenther@suse.de>
16760         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16761         (sparseset_pop): Likewise.
16762         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
16763         to be able to use quick_push in the worker loop.
16765 2013-01-30  Marek Polacek  <polacek@redhat.com>
16767         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16769 2013-01-30  Richard Biener  <rguenther@suse.de>
16771         PR lto/56147
16772         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16774 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
16776         PR tree-optimization/56064
16777         * fixed-value.c (fixed_from_double_int): New function.
16778         * fixed-value.h (fixed_from_double_int): New prototype.
16779         (const_fixed_from_double_int): New static inline function.
16780         * fold-const.c (native_interpret_fixed): New static function.
16781         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16782         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16783         (native_encode_fixed): New static function.
16784         (native_encode_expr) <FIXED_CST>: Use it.
16785         (native_interpret_int): Move double_int worker code to...
16786         * double-int.c (double_int::from_buffer): ...this new static method.
16787         * double-int.h (double_int::from_buffer): Prototype it.
16789 2013-01-30  Richard Biener  <rguenther@suse.de>
16791         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16792         New pointer-map and obstack.
16793         (init_alias_vars): Allocate pointer-map and obstack.
16794         (delete_points_to_sets): Free them.
16795         (find_what_var_points_to): Cache result.
16796         (find_what_p_points_to): Adjust for changed interface of
16797         find_what_var_points_to.
16798         (compute_points_to_sets): Likewise.
16799         (ipa_pta_execute): Likewise.
16801 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16803         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16804         * configure: Regenerate.
16805         * config.in: Regenerate.
16806         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16807         #nobits/#progbits if supported.
16809 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
16811         PR target/56121
16812         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16813         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16814         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16816 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16818         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16819         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16821 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16823         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16824         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16826 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16828         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16829         declaration.
16830         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16831         * config/arm/cortex-a7.md: New bypasses using
16832         arm_mac_accumulator_is_result.
16834 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16836         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
16837         (cortex_a7_neon_mla): Likewise.
16838         (cortex_a7_fpfmad): New reservation.
16839         (cortex_a7_fpmacs): Use ffmas and update required units.
16840         (cortex_a7_fpmuld): Update required units and latency.
16841         (cortex_a7_fpmacd): Likewise.
16842         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16843         (cortex_a7_neon). Likewise.
16844         (bypass) Update participating units.
16846 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16848         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16849         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16850         from fmac to ffma.
16851         * config/arm/vfp11.md (vfp_farith): Use ffmas.
16852         (vfp_fmul): Use ffmad.
16853         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16854         (cortex_r4_fmacd): Use ffmad.
16855         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16856         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
16857         (cortex_a9_fmacd): Use ffmad.
16858         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16859         (cortex_a8_vfp_macd): Use ffmad.
16860         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16861         (cortex_a5_fpmacd): Use ffmad.
16862         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16863         (cortex_a15_vfp_macd): Use ffmad.
16864         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16866 2013-01-29  Jason Merrill  <jason@redhat.com>
16868         PR libstdc++/54314
16869         * varasm.c (default_assemble_visibility): Don't warn about
16870         visibility on artificial decls.
16872 2013-01-29  Richard Biener  <rguenther@suse.de>
16874         PR tree-optimization/56113
16875         * tree-ssa-structalias.c (equiv_class_lookup): Also return
16876         the bitmap leader.
16877         (label_visit): Free duplicate bitmaps and record the leader instead.
16878         (perform_var_substitution): Adjust.
16880 2013-01-29  Richard Biener  <rguenther@suse.de>
16882         PR tree-optimization/55270
16883         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16884         the CFG, schedule loops for fixup.
16886 2013-01-29  Nick Clifton  <nickc@redhat.com>
16888         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16889         SP_REG.
16891 2013-01-28  Leif Ekblad  <leif@rdos.net>
16893         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16894         * config/i386/i386.h (TARGET_RDOS): New macro.
16895         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16896         * config/i386/i386.c (ix86_option_override_internal): For 64bit
16897         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16898         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16899         DEFAULT_LARGE_SECTION_THRESHOLD.
16900         * config/i386/i386.md (R14_REG, R15_REG): New constants.
16901         * config/i386/rdos.h: New file.
16902         * config/i386/rdos64.h: New file.
16904 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
16906         PR other/54814
16907         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16908         TEST_HARD_REG_BIT.
16910 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16912         PR rtl-optimization/56117
16913         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16914         call cselib_lookup_from_insn on the MEM before calling
16915         add_insn_mem_dependence.
16917 2013-01-28  Richard Biener  <rguenther@suse.de>
16919         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16920         to a stmt that didn't have one.
16921         (copy_phis_for_bb): Likewise for PHI arguments.
16922         (copy_debug_stmt): Likewise for debug stmts.
16924 2013-01-28  Richard Biener  <rguenther@suse.de>
16926         PR tree-optimization/56034
16927         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16928         (partition_builtin_p): Adjust.
16929         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
16930         it is the last partition.
16931         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16932         up the vertex for the definition.
16933         (classify_partition): Classify whether a partition is a
16934         PKIND_REDUCTION, thus has uses outside of the loop.
16935         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16936         Merge all PKIND_REDUCTION partitions into the last partition.
16937         (tree_loop_distribution): Seed partitions from reductions as well.
16939 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16941         PR tree-optimization/56125
16942         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16943         pow(x,c) into sqrt(x) * powi(x, n/2) or
16944         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16945         optimizing for size.
16946         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16947         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16948         integer.
16950         PR tree-optimization/56094
16951         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16952         to UNKNOWN_LOCATION while gimplifying expr.
16954 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
16956         PR target/56114
16957         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16958         operand 0 in movabs insn template for -masm=intel asm alternative.
16959         (*movabs<mode>_2): Ditto for operand 1.
16961 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
16963         PR target/54663
16964         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16965         of microblaze-c.o
16967 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16969         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16970         tm_file.
16972 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16974         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16975         Undef to avoid warning.
16977 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16979         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16980         * configure: Regenerate.
16982 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
16984         PR tree-optimization/56098
16985         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16986         for stmts with volatile ops.
16987         (cond_store_replacement): Don't optimize if assign has volatile ops.
16988         (cond_if_else_store_replacement_1): Don't optimize if either
16989         then_assign or else_assign have volatile ops.
16990         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16991         volatile ops.
16993 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16995         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16997 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16999         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17000         missing ':' in asm example.
17002 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
17004         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17005         entries into lane and laneq entries.
17006         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17007         Remove AdvSIMD scalar modes.
17008         (aarch64_sq<r>dmulh_laneq<mode>): New.
17009         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17010         modes.
17011         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17012         builtin implementations to relfect changes in RTL in aarch64-simd.md.
17013         * config/aarch64/iterators.md (VCOND): New.
17014         (VCONQ): New.
17016 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17018         PR target/54222
17019         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17020         Add NULL LIBNAME argument to existing definitions.
17021         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17022         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17023         * config/avr/avr.c (DEF_BUILTIN): Same.
17024         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17025         (avr_expand_builtin): Expand to a vanilla call if a libgcc
17026         implementation is available (DECL_ASSEMBLER_NAME is set).
17027         (avr_fold_absfx): New static function.
17028         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17029         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17030         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17031         AVR_BUILTIN_ABSLLK.
17032         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17033         (abshk, absk, abslk, absllk): Provide as static inline functions.
17035 2013-01-25  Marek Polacek  <polacek@redhat.com>
17037         PR tree-optimization/56035
17038         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17040 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
17042         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17043         (*movtf_internal_rex64): Add (!o,C) alternative
17044         (*movxf_internal_rex64): Ditto.
17045         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17047 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
17049         * doc/invoke.texi: fix typo.
17050         * doc/objc.texi: fix typo.
17052 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
17054         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17055         for the first two alternatives.
17057 2013-01-24  Diego Novillo  <dnovillo@google.com>
17059         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
17060         (ggc-zone.o): Remove.
17061         * configure.ac: Remove option --with-gc.
17062         * configure: Re-generate.
17063         * doc/install.texi: Remove documentation for --with-gc.
17064         * gengtype.c (write_enum_defn): Remove.  Update all users.
17065         (write_Types_process_field): Remove generation of gt_e_* argument.
17066         (output_type_enum): Remove.  Update all users.
17067         (write_enum_defn): Remove.  Update all users.
17068         (enum alloc_zone): Remove.  Update all users.
17069         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17070         * ggc-common.c (ggc_splay_alloc): Remove first argument.
17071         Update all callers.
17072         (struct ptr_data): Remove field TYPE.  Update all users.
17073         (gt_pch_note_object): Remove argument TYPE.  Update all users.
17074         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17075         Update all users.
17076         * ggc-none.c (ggc_alloc_typed_stat): Remove.
17077         (struct alloc_zone): Remove.
17078         (ggc_internal_alloc_zone_stat): Remove.
17079         (ggc_internal_cleared_alloc_zone_stat): Remove.
17080         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17081         (ggc_pch_count_object): Remove last argument.  Update all users.
17082         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17083         (struct alloc_zone): Remove.
17084         * ggc-zone.c: Remove.
17085         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17086         (struct alloc_zone): Remove.
17087         (ggc_alloc_typed_stat): Remove.
17088         (ggc_alloc_typed): Remove.
17089         (ggc_splay_alloc): Remove first argument.
17090         (rtl_zone): Remove.  Update all users.
17091         (tree_zone): Remove.  Update all users.
17092         (tree_id_zone): Remove.  Update all users.
17093         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17094         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17095         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17096         * tree-ssanames.c: Remove references to zone allocator in comments.
17098 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
17100         * config/avr/avr.c (avr_out_fract): Make register numbers that
17101         might be outside of source operand signed.
17103 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
17105         * config/i386/constraints.md (Yf): New constraint.
17106         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17107         of f constraint to conditionaly disable x87 register preferences.
17108         (*movdf_internal): Ditto.
17109         (*movsf_internal): Ditto.
17111 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
17113         PR inline-asm/55934
17114         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17115         that have operands with impossible constraints.
17116         Add a FIXME for a speed-up opportunity.
17117         * lra-constraints.c (process_alt_operands): Verify that a class
17118         selected from constraints on asms is valid for the operand mode.
17119         (curr_insn_transform): Remove incorrect comment.
17121 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
17123         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17124         TOC operand is a valid symbol ref in the constant pool.
17126 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17128         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17130 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
17132         PR target/54222
17133         * config/avr/stdfix.h: New file.
17134         * t-avr (stdfix-gcc.h): New rule to build it.
17135         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17137 2013-01-23  Kostya Serebryany  <kcc@google.com>
17139         * config/darwin.h: remove dependency on
17140         CoreFoundation (asan on Mac OS).
17142 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
17144         PR target/49069
17145         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17146         instead of cmpdi_operand for first comparison operand.
17147         Don't assert that comparison operands aren't both constants.
17149 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17151         * doc/install.texi (Downloading the Source): Update references to
17152         downloading separate components.
17154 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17156         * doc/extend.texi (__int128): Improve grammar.
17158 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
17160         PR target/56028
17161         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17162         alternative to (o,r).
17163         (*movdi_internal_rex64): Remove (!o,n) alternative.
17164         (DImode immediate->memory splitter): Remove.
17165         (DImode immediate->memory peephole2): Remove.
17166         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17167         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17168         alternative to (!o,*r).
17169         (*movtf_internal_sse): New pattern.
17170         (*movxf_internal_rex64): New pattern.
17171         (*movxf_internal): Disable for TARGET_64BIT.
17172         (*movdf_internal_rex64): Remove (!o,F) alternative.
17174 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
17176         PR middle-end/56074
17177         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17178         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17179         * tree-vect-loop-manip.c (find_loop_location): Also ignore
17180         stmt locations where LOCATION_LOCUS of the stmt location is
17181         UNKNOWN_LOCATION or BUILTINS_LOCATION.
17183         PR target/55686
17184         * config/i386/i386.md (UNSPEC_STOS): New.
17185         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17186         *strsetqi_1): Add UNSPEC_STOS.
17188 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
17190         PR c++/56067
17191         * doc/invoke.texi: Remove left over -Wsynth example.
17193 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17195         PR tree-optimization/56051
17196         * fold-const.c (fold_binary_loc): Don't fold
17197         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17198         a narrowing conversion, or widening conversion from signed
17199         to unsigned.
17201 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17203         PR rtl-optimization/56023
17204         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17205         dependent on debug instruction.
17207 2013-01-21  Martin Jambor  <mjambor@suse.cz>
17209         PR middle-end/56022
17210         * function.c (allocate_struct_function): Call
17211         invoke_set_current_function_hook earlier.
17213 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17215         * reload1.c (init_reload): Only initialize reload_obstack
17216         during the first call.
17218 2013-01-21  Marek Polacek  <polacek@redhat.com>
17220         * cfgloop.c (verify_loop_structure): Fix up grammar.
17222 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
17224         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17225         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17227 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17229         PR target/56058
17230         * config/arm/marvell-pj4.md: Update copyright year.
17231         Fix up use of alu to alu_reg and simple_alu_imm.
17233 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17235         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17237 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
17239         PR target/55433
17240         * lra-constraints.c (curr_insn_transform): Don't reuse original
17241         insn for secondary memory move when memory mode should be different.
17243 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17245         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17246         atomic_storedi_1): New patterns.
17248 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17250         btver2 pipeline descriptions.
17251         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17252         descriptions.
17253         * config/i386/i386.md (btver2_decode): New type attributes.
17254         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17255         type attributes.
17256         * config/i386/btver2.md: New file describing btver2 pipelines.
17258 2013-01-19  Andrew Pinski  <apinski@cavium.com>
17260         PR tree-optimization/52631
17261         * tree-ssa-sccvn (visit_use): Before looking up the original
17262         statement, try looking up the simplified expression.
17264 2013-01-19  Anthony Green  <green@moxielogic.com>
17266         * config/moxie/moxie.c (moxie_expand_prologue): Set
17267         current_function_static_stack_size.
17269 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17271         PR tree-optimization/56029
17272         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17273         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17275 2013-01-18  Sharad Singhai  <singhai@google.com>
17277         PR tree-optimization/55995
17278         * dumpfile.c (dump_loc): Print location only if available.
17279         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17281 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
17283         PR target/55433
17284         * lra-constraints.c (curr_insn_transform): Reuse original insn for
17285         secondary memory move.
17286         (inherit_reload_reg): Use rclass instead of cl for
17287         check_secondary_memory_needed_p.
17289 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17291         PR middle-end/56015
17292         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17293         the case where writing real complex part of target modifies op1.
17295 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
17297         * config/aarch64/aarch64-simd.md
17298         (aarch64_vcond_internal<mode>): Handle unordered cases.
17299         * config/aarch64/iterators.md (v_cmp_result): New.
17301 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
17302             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17304         * config/arm/marvell-pj4.md: New file.
17305         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17306         * config/arm/arm.md (generic_sched): Add marvell_pj4.
17307         (generic_vfp): Likewise.
17308         * config/arm/arm-cores.def: Add marvell-pj4.
17309         * config/arm/arm-tune.md: Regenerate.
17310         * config/arm/arm-tables.opt: Regenerate.
17311         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17312         * doc/invoke.texi: Document marvell-pj4.
17314 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
17316         * config/aarch64/arm_neon.h: Map scalar types to standard types.
17318 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17320         PR debug/54114
17321         PR debug/54402
17322         PR debug/49888
17323         * var-tracking.c (negative_power_of_two_p): New.
17324         (global_get_addr_cache, local_get_addr_cache): New.
17325         (get_addr_from_global_cache, get_addr_from_local_cache): New.
17326         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
17327         heading comment.
17328         (vt_stack_offset_p): Remove.
17329         (vt_canon_true_dep): Always canonicalize loc's address.
17330         (clobber_overlapping_mems): Make sure we have a MEM.
17331         (local_get_addr_clear_given_value): New.
17332         (val_reset): Clear local cached entries.
17333         (compute_bb_dataflow): Create and release the local cache.
17334         Disable duplicate MEMs clobbering.
17335         (emit_notes_in_bb): Clobber MEMs likewise.
17336         (vt_emit_notes): Create and release the local cache.
17337         (vt_initialize, vt_finalize): Create and release the global
17338         cache, respectively.
17339         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17341 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17343         PR libmudflap/53359
17344         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17345         not found in the symtab.
17347 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17349         PR debug/56006
17350         PR rtl-optimization/55547
17351         PR rtl-optimization/53827
17352         PR debug/53671
17353         PR debug/49888
17354         * alias.c (offset_overlap_p): New, factored out of...
17355         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17356         the conservative special case for symbolic constants.  Don't
17357         adjust zero sizes on alignment.
17359 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17361         PR rtl-optimization/52573
17362         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17363         REG_UNUSED for the same register.
17365 2013-01-17  Richard Biener  <rguenther@suse.de>
17366             Marek Polacek  <polacek@redhat.com>
17368         PR rtl-optimization/55833
17369         * loop-unswitch.c (unswitch_loops): Move loop verification...
17370         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17371         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17372         Set it to true when we're removing a loop from hierarchy tree in
17373         an irreducible region.
17374         (fix_bb_placements): Adjust caller.
17375         (fix_loop_placements): Likewise.
17377 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17379         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17380         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17381         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17382         Remove ID.  Adjust comments.
17383         * config/avr/avr-c.c (avr_builtin_name): Remove.
17384         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17385         * config/avr/avr.c (avr_tolower): New static function.
17386         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17387         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17388         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17389         default expansion.
17391 2013-01-17  Jan Hubicka  <jh@suse.cz>
17393         PR tree-optimization/55273
17394         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17396 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17398         PR target/55981
17399         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17400         store through atomic_store<mode>_1.
17401         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17403 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17405         PR tree-optimizations/55264
17406         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17407         for virtual methods.
17408         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17409         virtual methods before inlining is over.
17410         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17411         virtual functions.
17412         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17413         non-virtual.
17415 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17417         PR rtl-optimization/56005
17418         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17419         pending reads for prefetch.
17421 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17423         * config/aarch64/aarch64.md
17424         (*cstoresi_neg_uxtw): New pattern.
17425         (*cmovsi_insn_uxtw): New pattern.
17426         (*<optab>si3_uxtw): New pattern.
17427         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17428         (*<optab>si3_insn_uxtw): New pattern.
17429         (*bswapsi2_uxtw): New pattern.
17431 2013-01-16  Richard Biener  <rguenther@suse.de>
17433         * tree-inline.c (tree_function_versioning): Remove set but
17434         never used variable.
17436 2013-01-16  Richard Biener  <rguenther@suse.de>
17438         PR tree-optimization/55964
17439         * tree-flow.h (rename_variables_in_loop): Remove.
17440         (rename_variables_in_bb): Likewise.
17441         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17442         (copy_loop_before): Adjust and delete update-ssa status.
17443         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17444         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17445         (rename_variables_in_loop): Remove.
17446         (slpeel_update_phis_for_duplicate_loop): Likewise.
17447         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17448         use available cfg machinery instead of duplicating it.
17449         Update PHI nodes and perform poor-mans SSA update here.
17450         (slpeel_tree_peel_loop_to_edge): Adjust.
17452 2013-01-16  Richard Biener  <rguenther@suse.de>
17454         PR tree-optimization/54767
17455         PR tree-optimization/53465
17456         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17457         (vrp_visit_phi_node): For PHI arguments coming via backedges
17458         drop all symbolical range information.
17459         (execute_vrp): Compute backedges.
17461 2013-01-16  Richard Biener  <rguenther@suse.de>
17463         * doc/install.texi: Update CLooG and ISL requirements to
17464         0.18.0 and 0.11.1.
17466 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17468         PR target/55301
17469         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17470         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17471         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17473 2013-01-16  DJ Delorie  <dj@redhat.com>
17475         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17476         (UNSPECV_SP_SWITCH_E): New.
17477         (sp_switch_1): Change to an unspec.
17478         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17479         replace $r15.
17481 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17483         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17484         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17485         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17486         (expand_mem_thread_fence): Ditto.
17487         (expand_mem_signal_fence): Ditto.
17488         (expand_atomic_load): Ditto.
17489         (expand_atomic_store): Ditto.
17491 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17493         PR rtl-optimization/55547
17494         PR rtl-optimization/53827
17495         PR debug/53671
17496         PR debug/49888
17497         * alias.c (memrefs_conflict_p): Set sizes to negative after
17498         AND adjustments.
17500 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17502         PR target/55940
17503         * function.c (thread_prologue_and_epilogue_insns): Always
17504         add crtl->drap_reg to set_up_by_prologue.set, even if
17505         stack_realign_drap is false.
17507 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17509         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17510         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17511         *call): Fix indention.
17513 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17515         PR target/55876
17516         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17517         Update comment.
17519 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17521         PR rtl-optimization/55153
17522         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17524 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17526         PR tree-optimization/55920
17527         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17528         accesses as grp_to_be_debug_replaced.
17530 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17532         PR tree-optimization/55920
17533         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17534         there is non-useless type conversion needed from debug rhs to lhs,
17535         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17537 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17538             Mikael Pettersson  <mikpe@it.uu.se>
17540         PR target/43961
17541         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17542         Thumb.
17543         (ASM_OUTPUT_CASE_LABEL): Remove.
17544         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17545         * final.c (shorten_branches): Update alignment of labels before
17546         jump tables if CASE_VECTOR_SHORTEN_MODE.
17548 2013-01-15  Richard Biener  <rguenther@suse.de>
17550         PR bootstrap/55961
17551         * system.h: Do not include gmp.h for building host tools.
17553 2013-01-15  Richard Biener  <rguenther@suse.de>
17555         PR middle-end/55882
17556         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17557         account for bitpos when computing alignment.
17559 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17561         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17562         (ix86_target_macros_internal): Likewise.
17564         * config/i386/i386.c (m_CORE2I7): Removed.
17565         (m_CORE_HASWELL): New macro.
17566         (m_CORE_ALL): Likewise.
17567         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17568         (initial_ix86_arch_features): Likewise.
17569         (processor_target_table): Initializations for Core avx2.
17570         (cpu_names): New names "core-avx2".
17571         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17572         PROCESSOR_CORE_HASWELL.
17573         (ix86_issue_rate): New case.
17574         (ia32_multipass_dfa_lookahead): Likewise.
17575         (ix86_sched_init_global): Likewise.
17577         * config/i386/i386.h (TARGET_HASWELL): New macro.
17578         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17579         (processor_type): New PROCESSOR_HASWELL.
17581 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17583         PR tree-optimization/55955
17584         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17585         *SHIFT_EXPR and *ROTATE_EXPR codes.
17587         PR tree-optimization/48766
17588         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17589         -ftrapv disable -fwrapv.
17591 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17593         PR target/55974
17594         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17595         etc. to 1 and not to __flash.
17596         Use LL suffix for __INT24_MAX__ with -mint8.
17597         Use ULL suffix for __UINT24_MAX__ with -mint8.
17599 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17601         * config/avr/avr-arch.h
17602         (struct base_arch_s): Use typedef avr_arch_t instead.
17603         (struct arch_info_s): Use typedef avr_arch_info_t instead.
17604         (struct mcu_type_s): Use typedef avr_mcu_t instead.
17605         * config/avr/avr.c: Same.
17606         * config/avr/avr-devices.c: Same.
17607         * config/avr/driver-avr.c: Same.
17608         * config/avr/gen-avr-mmcu-texi.c: Same.
17609         * config/avr/avr-mcus.def: Adjust comment.
17611 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
17613         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17614         * config/aarch64/iterators.md (VALLDI): New.
17616 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
17617             Andi Kleen  <ak@linux.intel.com>
17619         PR target/55948
17620         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17621         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17622         memmodel flag.
17624 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17626         * config/avr/avr-stdint.h: Remove trailing blanks.
17627         * config/avr/avr-log.h: Same.
17628         * config/avr/avr-arch.h: Same.
17629         * config/avr/avr-devices.c: Same.
17630         * config/avr/avr-dimode.md: Same.
17631         * config/avr/predicates.md: Same.
17632         * config/avr/avr-c.c: Same.  And fix typo.
17634         * config/avr/avr-protos.h: Same.  And:
17635         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17636         (init_cumulative_args): Rename to avr_init_cumulative_args.
17637         (expand_prologue): Rename to avr_expand_prologue.
17638         (expand_epilogue): Rename to avr_expand_epilogue.
17639         (adjust_insn_length): Rename to avr_adjust_insn_length.
17640         (notice_update_cc): Rename to avr_notice_update_cc.
17641         (final_prescan_insn): Rename to avr_final_prescan_insn.
17642         * config/avr/avr.c: Same.
17643         * config/avr/avr.h: Same.
17644         * config/avr/avr.md: Remove trailing blanks.
17645         (prologue): Use avr_expand_prologue.
17646         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17648 2013-01-14  Richard Biener  <rguenther@suse.de>
17650         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17651         verify_location, collect_subblocks): New functions.
17652         (verify_gimple_in_cfg): Verify that locations only reference
17653         BLOCKs in the functions BLOCK tree.
17655 2013-01-14  Richard Biener  <rguenther@suse.de>
17657         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17658         PHI argument.
17659         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17660         unshare reference.
17661         (insert_out_of_ssa_copy_on_edge): Likewise.
17662         (rewrite_close_phi_out_of_ssa): Likewise.
17663         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17664         debug expressions.
17665         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17666         propagated constants.
17667         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17668         can not be shared.
17670 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17672         * config/avr/avr-modes.def: Add GPL copyright notice.
17674 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
17676         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17677         MEMMODEL_MASK to determine memory model.
17678         (atomic_store<mode>): Ditto from operands[2].
17679         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17681 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
17683         PR fortran/55935
17684         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17685         (fold_gimple_assign): Don't call unshare_expr here.
17686         (fold_ctor_reference): Call unshare_expr.
17688 2013-01-13  Terry Guo  <terry.guo@arm.com>
17690         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17691         * doc/fragments.texi: Document MULTILIB_REUSE.
17692         * gcc.c (multilib_reuse): New internal spec.
17693         (set_multilib_dir): Also search multilib from multilib_reuse.
17694         * genmultilib (tmpmultilib3): Refactor code.
17695         (tmpmultilib4): Ditto.
17696         (multilib_reuse): New multilib argument.
17698 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
17700         * Makefile.in: Update copyright.
17702 2013-01-12  Tom de Vries  <tom@codesourcery.com>
17704         PR middle-end/55890
17705         * calls.c (expand_call): Check if arg_nr is valid.
17707 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17709         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17710         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17711         documentation.  Add missing '__' in front of
17712         __builtin_ia32_packssdw256.
17714 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17716         PR target/55719
17717         * config/s390/s390.c (s390_preferred_reload_class): Do not return
17718         NO_REGS for larl operands.
17719         (s390_reload_larl_operand): Use s390_load_address instead of
17720         emit_move_insn.
17722 2013-01-11  Richard Biener  <rguenther@suse.de>
17724         * tree-cfg.c (verify_node_sharing_1): Split out from ...
17725         (verify_node_sharing): ... here.
17726         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17728 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
17730         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17731         Substitute TREECHECKING.
17732         * configure: Regenerate.
17733         * Makefile.in (TREECHECKING): New.
17735 2013-01-11  Richard Guenther  <rguenther@suse.de>
17737         PR tree-optimization/44061
17738         * tree-vrp.c (extract_range_basic): Compute zero as
17739         value-range for __builtin_constant_p of function parameters.
17741 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
17743         Update copyright years.
17745 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
17747         PR rtl-optimization/55672
17748         * lra-eliminations.c (mark_not_eliminable): Permit addition with
17749         const to be eliminable.
17751 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
17753         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17754         * configure: Regenerate.
17756 2013-01-10  Richard Biener  <rguenther@suse.de>
17758         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17760 2013-01-10  Richard Biener  <rguenther@suse.de>
17762         PR bootstrap/55792
17763         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17764         locations for virtual PHI arguments.
17765         (rewrite_update_phi_arguments): Likewise.
17767 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
17769         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17770         on to assembler.
17772 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
17774         PR tree-optimization/55921
17775         * tree-complex.c (expand_complex_asm): New function.
17776         (expand_complex_operations_1): Call it for GIMPLE_ASM.
17778 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17780         PR target/55718
17781         * config/s390/s390.c (s390_symref_operand_p)
17782         (s390_loadrelative_operand_p): Merge the two functions.
17783         (s390_check_qrst_address, print_operand_address): Add parameters
17784         to s390_loadrelative_operand_p invokation.
17785         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17786         (s390_reload_larl_operand, s390_secondary_reload): Use
17787         s390_loadrelative_operand_p instead of s390_symref_operand_p.
17788         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17790 2013-01-09  Mike Stump  <mikestump@comcast.net>
17792         * dse.c (record_store): Remove unnecessary assert.
17794 2013-01-09  Jan Hubicka  <jh@suse.cz>
17796         PR tree-optimization/55569
17797         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17798         * cfgloop.h (scale_loop_profile): Likewise.
17800 2013-01-09  Jan Hubicka  <jh@suse.cz>
17802         PR lto/45375
17803         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17804         functions.
17805         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17807 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
17809         PR middle-end/55114
17810         * expr.h (maybe_emit_group_store): Declare.
17811         * expr.c (maybe_emit_group_store): New function.
17812         * builtins.c (expand_builtin_int_roundingfn): Call it.
17813         (expand_builtin_int_roundingfn_2): Likewise.
17815 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
17817         PR rtl-optimization/55829
17818         * lra-constraints.c (match_reload): Add code for absent output.
17819         (curr_insn_transform): Add code for reloads of matched inputs
17820         without output.
17822 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
17824         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17825         attribute of movddup insn to DF.
17826         (*vec_interleave_lowv2df): Ditto.
17827         (vec_dupv2df): Ditto.
17829 2013-01-09  Jan Hubicka  <jh@suse.cz>
17831         PR tree-optimiation/55875
17832         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17833         EVERY_ITERATION parameter.
17834         (number_of_iterations_exit): Check if exit is executed every iteration.
17835         (idx_infer_loop_bounds): Similarly here.
17836         (n_of_executions_at_most): Simplify
17837         to only test for cases where statement is dominated by the
17838         particular bound; handle correctly the "postdominance" test.
17839         (scev_probably_wraps_p): Use max loop iterations info
17840         as a global bound first.
17842 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
17843             Nick Clifton  <nickc@redhat.com>
17845         * config/v850/v850.md (cbranchsf4): New pattern.
17846         (cstoresf4): New pattern.
17847         (cbranchdf4): New pattern.
17848         (cstoredf4): New pattern.
17849         (movsicc): Disallow floating point comparisons.
17850         (cmpsf_le_insn): Fix order of operators.
17851         (cmpsf_lt_insn): Likewise.
17852         (cmpsf_eq_insn): Likewise.
17853         (cmpdf_le_insn): Likewise.
17854         (cmpdf_lt_insn): Likewise.
17855         (cmpdf_eq_insn): Likewise.
17856         (cmpsf_ge_insn): Use LE comparison.
17857         (cmpdf_ge_insn): Likewise.
17858         (cmpsf_gt_insn): Use LT comparison.
17859         (cmpdf_gt_insn): Likewise.
17860         (cmpsf_ne_insn): Delete pattern.
17861         (cmpdf_ne_insn): Delete pattern.
17862         * config/v850/v850.c (v850_gen_float_compare): Use
17863         gen_cmpdf_eq_insn for NE comparison.
17864         (v850_float_z_comparison_operator)
17865         (v850_float_nz_comparison_operator): Move from here ...
17866         * config/v850/predicates.md: ... to here.  Move GT and GE
17867         comparisons into v850_float_z_comparison_operator.
17868         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17869         Delete prototype.
17870         (v850_float_nz_comparison_operator): Likewise.
17872 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17874         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17875         with calls to gen_insvsi/gen_insvdi.
17877 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17879         * config/i386/i386.c (initial_ix86_tune_features): Set up
17880         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17882 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
17883             Jakub Jelinek  <jakub@redhat.com>
17885         PR tree-optimization/48189
17886         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17887         If nitercst is 0, don't predict the exit edge.
17889 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17891         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17892         in asm_fprintf with reg_names.
17893         (aarch64_print_operand_address): Likewise.
17894         (aarch64_return_addr): Likewise.
17895         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17897 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17899         * config/pa/pa.h (VAL_U6_BITS_P): Define.
17900         (INT_U6_BITS): Likewise.
17901         * config/pa/predicates.md (uint6_operand): New predicate.
17902         (shift5_operand, shift6_operand): Likewise.
17903         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17904         arith32_operand.
17905         (lshrdi3): Use shift6_operand.
17906         (shrpsi4, shrpdi4): New insn patterns.
17907         (extzv): Delete expander.
17908         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
17909         predicates in unamed zero extract patterns.  Tighten common constraint.
17910         (extv): Delete expander.
17911         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
17912         predicates in unamed sign extract patterns.  Tighten common constraint.
17913         (insv): Delete expander.
17914         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
17915         predicates in unamed insert patterns.  Tighten common constraint.
17916         Change uint32_operand predicate to uint6_operand predicate in unamed
17917         DImode pattern to insert constant values of type 1...1xxxx.
17919 2013-01-04  Jan Hubicka  <jh@suse.cz>
17921         PR tree-optimization/55823
17922         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17923         issue.
17925 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17926             Uros Bizjak  <ubizjak@gmail.com>
17928         PR rtl-optimization/55845
17929         * df-problems.c (can_move_insns_across): Stop scanning at
17930         volatile_insn_p source instruction or give up if
17931         across_from .. across_to range contains any volatile_insn_p
17932         instructions.
17934 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17936         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17937         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17938         Declare.
17939         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17940         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17942 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17944         PR fortran/55341
17945         * asan.c (asan_clear_shadow): New function.
17946         (asan_emit_stack_protection): Use it.
17948 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17950         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17951         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17952         with tab instead of space.
17954 2013-01-08  Nick Clifton  <nickc@redhat.com>
17956         * config/rl78/rl78.c (rl78_expand_prologue): Always select
17957         register bank 0 at the start of an interrupt handler.
17958         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17959         MDBH registers.
17961 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17963         * config/aarch64/aarch64-simd.md
17964         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17965         (aarch64_simd_bsl): Likewise.
17966         (aarch64_vcond_internal<mode>): Likewise.
17967         (vcond<mode><mode>): Likewise.
17968         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17969         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17971 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17973         * config/aarch64/aarch64-builtins.c
17974         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17976 2013-01-08  Martin Jambor  <mjambor@suse.cz>
17978         PR debug/55579
17979         * tree-sra.c (analyze_access_subtree): Return true also after
17980         potentially creating a debug-only replacement.
17982 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17984         PR middle-end/55890
17985         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17987         PR tree-optimization/54120
17988         * tree-vrp.c (range_fits_type_p): Don't allow
17989         src_precision < precision from signed vr to unsigned_p
17990         if vr->min or vr->max is negative.
17991         (simplify_float_conversion_using_ranges): Test can_float_p
17992         against CODE_FOR_nothing.
17994 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17995             Richard Biener  <rguenther@suse.de>
17997         PR middle-end/55851
17998         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17999         types instead of just INTEGER_TYPE types.
18001 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
18003         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18004         TF_SIZE): Define.
18006 2013-01-07  Steve Ellcey  <sellcey@mips.com>
18008         PR target/42661
18009         * config/mips/mips.opt: Change mad to mmad to match documentation.
18011 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18013         PR target/55897
18014         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18015         .progmemx.data now.
18017 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18019         PR target/55897
18020         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18021         (avr_addrspace_t): Add .section_name field.
18022         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18023         array size.
18024         (avr_addrspace): Same.  Initialize .section_name.  Remove last
18025         NULL entry.  Put __memx into .progmemx.data.
18026         (progmem_section_prefix): Remove.
18027         (avr_asm_init_sections): No need to initialize progmem_section.
18028         (avr_asm_named_section): Use avr_addrspace[].section_name to get
18029         section name prefix.
18030         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
18031         retrieve the progmem section.
18032         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18033         boundary to run over avr_addrspace[].
18034         (avr_register_target_pragmas): Ditto.
18036 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
18038         * varasm.c (output_constant_def_contents): For asan_protect_global
18039         protected strings, adjust DECL_ALIGN if needed, before testing for
18040         anchored symbols.
18041         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18042         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
18043         normal decls.
18044         (output_object_block): For asan protected decls, emit asan padding
18045         after their contents.
18046         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18047         (asan_finish_file): Test it here instead.
18049 2013-01-07  Nick Clifton  <nickc@redhat.com>
18050             Matthias Klose  <doko@debian.org>
18051             Doug Kwan  <dougkwan@google.com>
18052             H.J. Lu  <hongjiu.lu@intel.com>
18054         PR driver/55470
18055         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18057         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18059         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18061         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18063         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18065 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18067         PR target/54461
18068         * doc/install.texi (Cross-Compiler-Specific Options): Document
18069         --with-avrlibc.
18071 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
18073         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18074         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18075         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18076         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18077         vqmovun_high_s64): Fix source operand number and update copyright.
18079 2013-01-07  Richard Biener  <rguenther@suse.de>
18081         PR middle-end/55890
18082         * gimple.h (gimple_call_builtin_p): New overload.
18083         * gimple.c (validate_call): New function.
18084         (gimple_call_builtin_p): Likewise.
18085         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18086         Use gimple_call_builtin_p.
18087         (find_func_clobbers): Likewise.
18088         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18089         (strlen_optimize_stmt): Likewise.
18091 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18093         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18094         (vld1q_dup_*): Likewise.
18095         (vld1_*): Likewise.
18096         (vld1q_*): Likewise.
18097         (vld1_lane_*): Likewise.
18098         (vld1q_lane_*): Likewise.
18100 2013-01-07  Richard Biener  <rguenther@suse.de>
18102         * lto-streamer.h (LTO_minor_version): Bump to 2.
18104 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18106         * config/aarch64/aarch64-protos.h
18107         (aarch64_const_double_zero_rtx_p): Rename to...
18108         (aarch64_float_const_zero_rtx_p): ...this.
18109         (aarch64_float_const_representable_p): New.
18110         (aarch64_output_simd_mov_immediate): Likewise.
18111         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18112         move immediate case.
18113         * config/aarch64/aarch64.c
18114         (aarch64_const_double_zero_rtx_p): Rename to...
18115         (aarch64_float_const_zero_rtx_p): ...this.
18116         (aarch64_print_operand): Allow printing of new constants.
18117         (aarch64_valid_floating_const): New.
18118         (aarch64_legitimate_constant_p): Check for valid floating-point
18119         constants.
18120         (aarch64_simd_valid_immediate): Likewise.
18121         (aarch64_vect_float_const_representable_p): New.
18122         (aarch64_float_const_representable_p): Likewise.
18123         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18124         (aarch64_output_simd_mov_immediate): New.
18125         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18126         (*movdf_aarch64): Likewise.
18127         * config/aarch64/constraints.md (Ufc): New.
18128         (Y): call aarch64_float_const_zero_rtx.
18129         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18131 2013-01-07  Richard Biener  <rguenther@suse.de>
18133         PR tree-optimization/55888
18134         PR tree-optimization/55862
18135         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18136         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18137         not if it is contained therein.
18139 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18141         * config/avr/t-avr: Typo.
18143 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18145         PR55243
18146         * config/avr/t-avr: Don't automatically rebuild
18147         $(srcdir)/config/avr/t-multilib
18148         $(srcdir)/config/avr/avr-tables.opt
18149         $(srcdir)/doc/avr-mmcu.texi
18150         (avr-mcus): New phony target to build them on request.
18151         (s-avr-mlib, s-avr-mmcu-texi): Remove.
18152         * avr/avr-mcus.def: Adjust comments.
18154 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
18156         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18158 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18160         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18162 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18164         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18166 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
18168         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18169         to generate profiling.
18170         * config/rs6000/aix64.h (LIB_SPEC): Same.
18172 2013-01-04  Andrew Pinski  <apinski@cavium.com>
18174         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18175         New function.
18176         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18178 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
18180         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18181         unconditionally.
18182         (ix86_expand_move): Ditto.
18183         (ix86_zero_extend_to_Pmode): Ditto.
18184         (ix86_expand_call): Ditto.
18185         (ix86_expand_special_args_builtin): Ditto.
18186         (ix86_expand_builtin): Ditto.
18188 2013-01-04  Richard Biener  <rguenther@suse.de>
18190         PR tree-optimization/55862
18191         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18192         translating them through PHI nodes.
18194 2013-01-04  Martin Jambor  <mjambor@suse.cz>
18196         PR tree-optimization/55755
18197         * tree-sra.c (sra_modify_assign): Do not check that an access has no
18198         children when trying to avoid producing a VIEW_CONVERT_EXPR.
18200 2013-01-04  Marek Polacek  <polacek@redhat.com>
18202         PR middle-end/55859
18203         * opts.c (default_options_optimization): Clarify error message.
18205 2013-01-04  Richard Biener  <rguenther@suse.de>
18207         PR middle-end/55863
18208         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18209         reassociation.
18211 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18213         PR target/53789
18214         * config/pa/pa.md (movsi): Revert previous change.
18215         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18216         references.
18218 2013-01-03  Richard Henderson  <rth@redhat.com>
18220         * config/i386/i386.c (ix86_expand_move): Always assign to op1
18221         after eliminating TLS symbols.
18223 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
18225         PR bootstrap/50167
18226         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18227         * graphite-poly.c (debug_gmp_value): Likewise.
18229 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
18231         PR target/55712
18232         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18233         selected code model, define __code_mode_small__, __code_model_medium__,
18234         __code_model_large__, __code_model_32__ or __code_model_kernel__.
18235         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18236         xchg temporary register with %k.  Declare temporary register as
18237         early clobbered.
18238         [__x86_64__]: For medium and large code models, preserve %rbx register.
18240 2013-01-03  Richard Biener  <rguenther@suse.de>
18242         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18243         (dump_subscript): Adjust.
18244         (finalize_ddr_dependent): Do not dump redundant info.
18245         (analyze_siv_subscript): Adjust.
18246         (subscript_dependence_tester): Likewise.
18247         (compute_affine_dependence): Likewise.
18249 2013-01-03  Richard Biener  <rguenther@suse.de>
18251         Revert
18252         2013-01-03  Richard Biener  <rguenther@suse.de>
18254         PR tree-optimization/55857
18255         * tree-vect-stmts.c (vectorizable_load): Do not setup
18256         re-alignment for invariant loads.
18258         2013-01-02  Richard Biener  <rguenther@suse.de>
18260         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18261         invariant load do not generate a vector load from the scalar location.
18263 2013-01-03  Richard Biener  <rguenther@suse.de>
18265         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18266         for not vectorizing.
18267         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18268         not build INDIRECT_REFs, call get_name once only.
18269         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
18270         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18272 2013-01-03  Richard Biener  <rguenther@suse.de>
18274         PR tree-optimization/55857
18275         * tree-vect-stmts.c (vectorizable_load): Do not setup
18276         re-alignment for invariant loads.
18278 2013-01-03  Richard Biener  <rguenther@suse.de>
18280         PR lto/55848
18281         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18282         prefer a built-in decl.
18284 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18286         * gcc.c (process_command): Update copyright notice dates.
18287         * gcov.c (print_version): Likewise.
18288         * gcov-dump.c (print_version): Likewise.
18290         PR rtl-optimization/55838
18291         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18292         iv0.step, iv1.step and step.
18294 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18295             Marc Glisse  <marc.glisse@inria.fr>
18297         PR tree-optimization/55832
18298         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18299         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18300         integer_{one,zero}_node.
18302 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18304         PR debug/54402
18305         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18306         * var-tracking.c (reverse_op): Don't add reverse ops to
18307         VALUEs that have already
18308         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18310 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
18312         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18314 2013-01-02  Teresa Johnson  <tejohnson@google.com>
18316         * dumpfile.c (dump_loc): Print filename with location.
18317         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18318         new location_t parameter to emit complete unroll message with
18319         new dump framework.
18320         (canonicalize_loop_induction_variables): Compute loops location
18321         and pass to try_unroll_loop_completely.
18322         * loop-unroll.c (report_unroll_peel): New function.
18323         (peel_loops_completely): Use new dump format with location
18324         for main dumpfile message, and invoke report_unroll_peel on success.
18325         (decide_unrolling_and_peeling): Ditto.
18326         (decide_peel_once_rolling): Remove old dumpfile message subsumed
18327         by report_unroll_peel.
18328         (decide_peel_completely): Ditto.
18329         (decide_unroll_constant_iterations): Ditto.
18330         (decide_unroll_runtime_iterations): Ditto.
18331         (decide_peel_simple): Ditto.
18332         (decide_unroll_stupid): Ditto.
18333         * cfgloop.c (get_loop_location): New function.
18334         * cfgloop.h (get_loop_location): Declare.
18336 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18338         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18339         NULL.
18341 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18343         PR middle-end/55198
18344         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18345         BLKmode objects when EXPAND_MEMORY is specified.
18347 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18349         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18350         in loop predicate.
18351         (fold_builtin_cpu): Do not share cpu model decls across statements.
18353 2013-01-02  Jason Merrill  <jason@redhat.com>
18355         PR c++/55804
18356         * tree.c (build_array_type_1): Revert earlier change.
18358 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18360         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18361         "cortex-a57".
18362         * config/aarch64/aarch64-tune.md: Re-generate.
18364 2013-01-02  Richard Biener  <rguenther@suse.de>
18366         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18367         invariant load do not generate a vector load from the scalar location.
18369 2013-01-02  Richard Biener  <rguenther@suse.de>
18371         PR bootstrap/55784
18372         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18373         * configure: Regenerate.
18375 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18377         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18378         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18379         (expand_builtin_int_roundingfn_2): Keep the original target around
18380         for the fallback case.
18382 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18384         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18385         to be clear for sign changes.
18387 2013-01-01  Jan Hubicka  <jh@suse.cz>
18389         * ipa-inline-analysis.c: Fix formatting.
18391 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18393         PR tree-optimization/55831
18394         * tree-vect-loop.c (get_initial_def_for_induction): Use
18395         gsi_after_labels instead of gsi_start_bb.
18397 Copyright (C) 2013 Free Software Foundation, Inc.
18399 Copying and distribution of this file, with or without modification,
18400 are permitted in any medium without royalty provided the copyright
18401 notice and this notice are preserved.