Add missing part for last ChangeLog entry:
[official-gcc.git] / gcc / ChangeLog
blob8b663b63a24edfe86a5eadd148ff5cf614c1008a
1 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3         * gcc/config/arc/arc.c (arc_ccfsm_post_advance): Also handle
4         TYPE_UNCOND_BRANCH.
5         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
6         changes statep->state.
8 2013-08-24  Tobias Burnus  <burnus@net-b.de>
10         PR other/33426
11         * tree-cfg.c (replace_loop_annotate): New function.
12         (execute_build_cfg): Call it.
13         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
14         * internal-fn.c (expand_ANNOTATE): New function.
15         * internal-fn.def (ANNOTATE): Define as new internal function.
16         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
17         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
18         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
19         * doc/extend.texi (Pragmas): Document #pragma ivdep.
20         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
22 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
23             Marcus Shawcroft  <marcus.shawcroft@arm.com>
25         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
26         Special case reload SP+C into none GENERAL_REGS.
28 2013-10-24  Michael Matz  <matz@suse.de>
30         * gengtype.c (is_file_equal): Check that files will be same
31         length.
33 2013-10-25  Christian Bruel  <christian.bruel@st.com>
35         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
36         * gcc/config/sh/t-sh (sh-mem.o): New rule.
37         * gcc/config/sh/sh-mem.cc (expand_block_move): Moved here.
38         (sh4_expand_cmpstr): New function.
39         * gcc/config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c
40         * gcc/config/sh/sh-protos.h (sh4_expand_cmpstr): Declare.
41         * gcc/config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
42         (rotlhi3_8): Rename.
44 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
46         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
47         * Makefile.in (CXXDEPMODE): Assign and change users.
48         (CCDEPMODE): Delete.
49         * configure: Regenerate.
51 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
53         * gengtype-parse.c (require_without_advance): New.
54         (type): For GTY-marked types that are not GTY((user)), parse any
55         base classes, requiring them to be single-inheritance, and not
56         be templates.  For non-GTY-marked types and GTY((user)),
57         continue to skip over any C++ inheritance specification.
58         * gengtype-state.c (state_writer::write_state_struct_type):
59         Write base class of type (if any).
60         (read_state_struct_type): Read base class of type (if any).
61         * gengtype.c (new_structure): Add a "base_class" parameter.
62         (create_optional_field_): Update for new parameter to
63         new_structure.
64         (adjust_field_rtx_def): Likewise.
65         (adjust_field_tree_exp): Likewise.
66         * gengtype.h (struct type): Add "base_class" field to the s
67         union field.
68         (new_structure): Add "base" parameter.
70 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
72         PR target/57756
73         * config/i386/i386.c (ix86_option_override_internal):
74         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
75         (ix86_valid_target_attribute_tree):
76         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
77         Change TARGET_SSE to TARGET_SSE_P (opts->...)
79 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
81         * tree-ssa-loop.h: Remove include files.
82         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
83         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
84         * graphite-clast-to-gimple.c: Likewise.
85         * graphite-scop-detection.c: Likewise.
86         * graphite-sese-to-poly.c: Likewise.
87         * ipa-inline-analysis.c: Likewise.
88         * ipa-pure-const.c: Likewise.
89         * loop-init.c: Likewise.
90         * passes.c: Likewise.
91         * predict.c: Likewise.
92         * tree-cfg.c: Likewise.
93         * tree-cfgcleanup.c: Likewise.
94         * tree-chrec.c: Likewise.
95         * tree-data-ref.c: Likewise.
96         * tree-loop-distribution.c: Likewise.
97         * tree-parloops.c: Likewise.
98         * tree-predcom.c: Likewise.
99         * tree-scalar-evolution.c: Likewise.
100         * tree-ssa-address.c: Likewise.
101         * tree-ssa.c: Likewise.
102         * tree-ssa-dce.c: Likewise.
103         * tree-ssa-loop.c: Likewise.
104         * tree-ssa-loop-im.c: Likewise.
105         * tree-ssa-loop-ivcanon.c: Likewise.
106         * tree-ssa-loop-ivopts.c: Likewise.
107         * tree-ssa-loop-manip.c: Likewise.
108         * tree-ssa-loop-niter.c: Likewise.
109         * tree-ssa-loop-prefetch.c: Likewise.
110         * tree-ssa-loop-unswitch.c: Likewise.
111         * tree-ssa-reassoc.c: Likewise.
112         * tree-vect-data-refs.c: Likewise.
113         * tree-vect-loop.c: Likewise.
114         * tree-vect-loop-manip.c: Likewise.
115         * tree-vectorizer.c: Likewise.
116         * tree-vect-stmts.c: Likewise.
117         * tree-vrp.c: Likewise.
119 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
121         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
123 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
125         PR tree-optimization/58775
126         PR tree-optimization/58791
127         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
128         (insert_stmt_after): Rewritten, don't move the stmt, but really
129         insert it.
130         (get_stmt_uid_with_default): Remove.
131         (build_and_add_sum): Use insert_stmt_after and
132         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only
133         labels.
134         (update_range_test): Set uid on stmts added by
135         force_gimple_operand_gsi.  Don't immediately modify statements
136         in inter-bb optimization, just update oe->op values.
137         (optimize_range_tests): Return bool whether any changed have
138         been made.
139         (update_ops): New function.
140         (struct inter_bb_range_test_entry): New type.
141         (maybe_optimize_range_tests): Perform statement changes here.
142         (not_dominated_by, appears_later_in_bb, get_def_stmt,
143         ensure_ops_are_available): Remove.
144         (find_insert_point): Rewritten.
145         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
146         return LHS of the (new resp. old) stmt.  Don't call
147         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
148         instead of last, move new stmt at the right place.
149         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
150         (negate_value): Likewise.  Set uids.
151         (break_up_subtract_bb): Initialize uids.
152         (reassociate_bb): Adjust rewrite_expr_tree caller.
153         (do_reassoc): Don't call renumber_gimple_stmt_uids.
155 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
157         PR target/58838
158         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
159         TARGET_32BIT final condition.
160         (mulsi3_internal2 and splitter): Same.
162 2013-10-23  Jeff Law  <law@redhat.com>
164         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
165         through joiner blocks with abnormal outgoing edges.
167         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
168         Add parameter JOINERS, to allow/disallow threading through joiner
169         blocks.
170         (thread_block): New.  Call thread_block_1.
171         (mark_threaded_blocks): Remove code to filter out certain cases
172         of threading through joiner blocks.
173         (thread_through_all_blocks): Document how we can have a dangling
174         edge AUX field and clear it.
176 2013-10-23  Ian Lance Taylor  <iant@google.com>
178         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
179         (C++ Dialect Options): Likewise.
180         (Optimize Options): Likewise.
182 2013-10-23  Tom de Vries  <tom@codesourcery.com>
184         PR tree-optimization/58805
185         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
187 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
189         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
190         sequence based on get_range_info returned range.
192 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
194         * tree-ssa.h: Remove all #include's
195         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
196         * alias.c: Move required includes from tree-ssa.h.
197         * asan.c: Likewise.
198         * builtins.c: Likewise.
199         * calls.c: Likewise.
200         * cfgexpand.c: Likewise.
201         * cfghooks.c: Likewise.
202         * cfgloop.c: Likewise.
203         * cfgloopmanip.c: Likewise.
204         * cgraph.c: Likewise.
205         * cgraphbuild.c: Likewise.
206         * cgraphclones.c: Likewise.
207         * cgraphunit.c: Likewise.
208         * dse.c: Likewise.
209         * except.c: Likewise.
210         * expr.c: Likewise.
211         * final.c: Likewise.
212         * fold-const.c: Likewise.
213         * ggc-page.c: Likewise.
214         * gimple-builder.c: Likewise.
215         * gimple-fold.c: Likewise.
216         * gimple-iterator.c: Likewise.
217         * gimple-low.c: Likewise.
218         * gimple-pretty-print.c: Likewise.
219         * gimple-ssa-strength-reduction.c: Likewise.
220         * gimple-streamer-in.c: Likewise.
221         * gimple-streamer-out.c: Likewise.
222         * gimplify.c: Likewise.
223         * graphite-blocking.c: Likewise.
224         * graphite-clast-to-gimple.c: Likewise.
225         * graphite-dependences.c: Likewise.
226         * graphite-interchange.c: Likewise.
227         * graphite-optimize-isl.c: Likewise.
228         * graphite-poly.c: Likewise.
229         * graphite-scop-detection.c: Likewise.
230         * graphite-sese-to-poly.c: Likewise.
231         * graphite.c: Likewise.
232         * ipa-cp.c: Likewise.
233         * ipa-inline-analysis.c: Likewise.
234         * ipa-inline-transform.c: Likewise.
235         * ipa-inline.c: Likewise.
236         * ipa-prop.c: Likewise.
237         * ipa-pure-const.c: Likewise.
238         * ipa-reference.c: Likewise.
239         * ipa-split.c: Likewise.
240         * ipa-utils.c: Likewise.
241         * loop-init.c: Likewise.
242         * lto-cgraph.c: Likewise.
243         * lto-section-in.c: Likewise.
244         * lto-section-out.c: Likewise.
245         * lto-streamer-in.c: Likewise.
246         * lto-streamer-out.c: Likewise.
247         * lto-streamer.c: Likewise.
248         * omp-low.c: Likewise.
249         * passes.c: Likewise.
250         * predict.c: Likewise.
251         * print-tree.c: Likewise.
252         * profile.c: Likewise.
253         * sese.c: Likewise.
254         * targhooks.c: Likewise.
255         * tracer.c: Likewise.
256         * trans-mem.c: Likewise.
257         * tree-call-cdce.c: Likewise.
258         * tree-cfg.c: Likewise.
259         * tree-cfgcleanup.c: Likewise.
260         * tree-chrec.c: Likewise.
261         * tree-complex.c: Likewise.
262         * tree-data-ref.c: Likewise.
263         * tree-dfa.c: Likewise.
264         * tree-eh.c: Likewise.
265         * tree-emutls.c: Likewise.
266         * tree-if-conv.c: Likewise.
267         * tree-inline.c: Likewise.
268         * tree-into-ssa.c: Likewise.
269         * tree-loop-distribution.c: Likewise.
270         * tree-mudflap.c: Likewise.
271         * tree-nested.c: Likewise.
272         * tree-nrv.c: Likewise.
273         * tree-object-size.c: Likewise.
274         * tree-outof-ssa.c: Likewise.
275         * tree-parloops.c: Likewise.
276         * tree-phinodes.c: Likewise.
277         * tree-predcom.c: Likewise.
278         * tree-pretty-print.c: Likewise.
279         * tree-profile.c: Likewise.
280         * tree-scalar-evolution.c: Likewise.
281         * tree-sra.c: Likewise.
282         * tree-ssa-address.c: Likewise.
283         * tree-ssa-alias.c: Likewise.
284         * tree-ssa-ccp.c: Likewise.
285         * tree-ssa-coalesce.c: Likewise.
286         * tree-ssa-copy.c: Likewise.
287         * tree-ssa-copyrename.c: Likewise.
288         * tree-ssa-dce.c: Likewise.
289         * tree-ssa-dom.c: Likewise.
290         * tree-ssa-dse.c: Likewise.
291         * tree-ssa-forwprop.c: Likewise.
292         * tree-ssa-ifcombine.c: Likewise.
293         * tree-ssa-live.c: Likewise.
294         * tree-ssa-loop-ch.c: Likewise.
295         * tree-ssa-loop-im.c: Likewise.
296         * tree-ssa-loop-ivcanon.c: Likewise.
297         * tree-ssa-loop-ivopts.c: Likewise.
298         * tree-ssa-loop-manip.c: Likewise.
299         * tree-ssa-loop-niter.c: Likewise.
300         * tree-ssa-loop-prefetch.c: Likewise.
301         * tree-ssa-loop-unswitch.c: Likewise.
302         * tree-ssa-loop.c: Likewise.
303         * tree-ssa-math-opts.c: Likewise.
304         * tree-ssa-operands.c: Likewise.
305         * tree-ssa-phiopt.c: Likewise.
306         * tree-ssa-phiprop.c: Likewise.
307         * tree-ssa-pre.c: Likewise.
308         * tree-ssa-propagate.c: Likewise.
309         * tree-ssa-reassoc.c: Likewise.
310         * tree-ssa-sccvn.c: Likewise.
311         * tree-ssa-sink.c: Likewise.
312         * tree-ssa-strlen.c: Likewise.
313         * tree-ssa-structalias.c: Likewise.
314         * tree-ssa-tail-merge.c: Likewise.
315         * tree-ssa-ter.c: Likewise.
316         * tree-ssa-threadedge.c: Likewise.
317         * tree-ssa-threadupdate.c: Likewise.
318         * tree-ssa-uncprop.c: Likewise.
319         * tree-ssa-uninit.c: Likewise.
320         * tree-ssa.c: Likewise.
321         * tree-ssanames.c: Likewise.
322         * tree-stdarg.c: Likewise.
323         * tree-streamer-in.c: Likewise.
324         * tree-switch-conversion.c: Likewise.
325         * tree-tailcall.c: Likewise.
326         * tree-vect-data-refs.c: Likewise.
327         * tree-vect-generic.c: Likewise.
328         * tree-vect-loop-manip.c: Likewise.
329         * tree-vect-loop.c: Likewise.
330         * tree-vect-patterns.c: Likewise.
331         * tree-vect-slp.c: Likewise.
332         * tree-vect-stmts.c: Likewise.
333         * tree-vectorizer.c: Likewise.
334         * tree-vrp.c: Likewise.
335         * tree.c: Likewise.
336         * tsan.c: Likewise.
337         * value-prof.c: Likewise.
338         * var-tracking.c: Likewise.
339         * varpool.c: Likewise.
340         * vtable-verify.c: Likewise.
342 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
344         * config/tilegx/tilegx.c: Include "tree.h".
346 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
348         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
349         the info, not after it.
350         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
351         dump_ssaname_info call.
352         (pp_gimple_stmt_1): Adjust caller.
353         (dump_phi_nodes): Likewise.  Don't print "# " here.
355 2013-10-22  Jan Hubicka  <jh@suse.cz>
357         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New tuning flag.
358         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
359         * i386.c (expand_small_movmem_or_setmem): New function.
360         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New function
361         (alg_usable_p): Add support for value ranges; cleanup.
362         (ix86_expand_set_or_movmem): Add support for misaligned moves.
364 2013-10-22  Sterling Augustine  <saugustine@google.com>
366         * doc/invoke.texi: Document -ggnu-pubnames.
367         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
368         logic.
369         * dwarf2out.c: Include gdb/gdb-index.h.
370         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
371         debug_generate_pub_sections.
372         (is_java, output_pubtables, output_pubname): New functions.
373         (include_pubname_in_output): Handle debug_generate_pub_sections at
374         level 2.
375         (size_of_pubnames): Use new local space_for_flags based on
376         debug_generate_pub_sections.
377         (output_pubnames): Unify pubnames and pubtypes output logic.
378         Genericize comments.  Call output_pubname.
379         (dwarf2out_finish): Move logic to output_pubnames and call it.
381 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
383         PR target/58779
384         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
385         Remove CCCmode handling.
386         <case LTU>: Return 'c' suffix for CCCmode.
387         <case GEU>: Return 'nc' suffix for CCCmode.
388         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
389         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
390         (*sub<mode>3_cc_overflow): Ditto.
391         (*subsi3_zext_cc_overflow): Ditto.
393 2013-10-22  Steve Ellcey  <sellcey@mips.com>
395         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
396         (AND (NOT OP1) (NOT OP2)).
398 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
400         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
401         meaning of merge-high and merge-low masks for little endian; avoid
402         use of vector-pack masks for little endian for mismatched modes.
404 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
406         * config/tilepro/tilepro.c: Include "tree.h".
408 2013-10-22  Andreas Schwab  <schwab@suse.de>
410         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
411         with PRE_DEC.
413 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
415         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
416         [Ed Smith-Rowland]): New entries.
417         ([Stephen M. Webb]): Update.
419 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
421         * tree-ssa-ter.h: Remove duplicate copy of file contents.
423 2013-10-21  Marek Polacek  <polacek@redhat.com>
425         PR middle-end/58809
426         * fold-const.c (fold_range_test): Return 0 if the type is not
427         an integral type.
429 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
431         * system.h: Move hwint.h include further down.
432         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
433         gcc_checking_asserts.
434         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
436 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
438         Fix volatile issues in optimize_bit_field_compare.
439         * fold-const.c (optimize_bit_field_compare): Bail out if
440         lvolatilep or rvolatilep.
442 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
444         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
445         and get_inner_reference returning different pmode for non-volatile
446         bit-field members dependent on flag_strict_volatile_bitfields.
447         * stor-layout.c (layout_decl): Remove special handling of
448         flag_strict_volatile_bitfields.
449         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
450         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
452 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
454         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
455         calculations.
457 2013-10-21  Jeff Law  <law@redhat.com>
459         * tree-ssa-threadedge.c (thread_through_normal_block): New
460         argument VISITED.  Remove VISISTED as a local variable.  When we
461         have a threadable jump, verify the destination of the jump has not
462         been visised.
463         (thread_across_edge): Allocate VISITED bitmap once at function
464         scope and use it throughout.  Make sure to set appropriate bits in
465         VISITED for E (start of jump thread path).
466         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
467         through a joiner if any edge on the path has a recorded jump thread.
469 2013-10-21  Ian Lance Taylor  <iant@google.com>
471         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
472         don't imply that attributes can solve all problems.
473         (Directory Options): Fix typo.
475 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
477         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
478         extend and extend_add.
480 2013-10-21  Richard Biener  <rguenther@suse.de>
482         PR tree-optimization/58794
483         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
484         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
486 2013-10-21  Richard Biener  <rguenther@suse.de>
488         PR middle-end/58742
489         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
490         to (T) X for sign-changing conversions (or no conversion).
492 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
494         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
496 2013-10-20  Jan Hubicka  <jh@suse.cz>
498         * config/i386/i386-tune.def: Add comment; organize into categories
500 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
502         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
503         argument.  Update function comment.
504         (expand_set_or_movmem_via_rep): New function combining
505         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
506         (expand_movmem_via_rep_mov): Remove.
507         expand_setmem_via_rep_stos): Remove.
508         (expand_movmem_epilogue): Update calls correspondingly.
509         (expand_setmem_epilogue_via_loop): Likewise.
510         (emit_memset): New.
511         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
512         (expand_set_or_movmem_prologue): New function combining
513         expand_movmem_prologue and expand_setmem_prologue.
514         (expand_movmem_prologue): Remove.
515         (expand_setmem_prologue): Remove.
516         (expand_set_or_movmem_constant_prologue): New function combining
517         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
518         (expand_constant_movmem_prologue): Remove.
519         (expand_constant_setmem_prologue): Remove.
520         (promote_duplicated_reg): Allow vector-const0 value.
521         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
522         and ix86_expand_setmem.
523         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
524         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
526 2013-10-21  Diego Novillo  <dnovillo@google.com>
528         * asan.c: Include tree.h
529         * bb-reorder.c: Likewise.
530         * cfgcleanup.c: Likewise.
531         * cfgloopmanip.c: Likewise.
532         * data-streamer-in.c: Likewise.
533         * data-streamer-out.c: Likewise.
534         * data-streamer.c: Likewise.
535         * dwarf2cfi.c: Likewise.
536         * graphite-blocking.c: Likewise.
537         * graphite-clast-to-gimple.c: Likewise.
538         * graphite-dependences.c: Likewise.
539         * graphite-interchange.c: Likewise.
540         * graphite-optimize-isl.c: Likewise.
541         * graphite-poly.c: Likewise.
542         * graphite-scop-detection.c: Likewise.
543         * graphite-sese-to-poly.c: Likewise.
544         * graphite.c: Likewise.
545         * ipa-devirt.c: Likewise.
546         * ipa-profile.c: Likewise.
547         * ipa.c: Likewise.
548         * ira.c: Likewise.
549         * loop-init.c: Likewise.
550         * loop-unroll.c: Likewise.
551         * lower-subreg.c: Likewise.
552         * lto/lto-object.c: Likewise.
553         * recog.c: Likewise.
554         * reginfo.c: Likewise.
555         * tree-loop-distribution.c: Likewise.
556         * tree-parloops.c: Likewise.
557         * tree-ssa-strlen.c: Likewise.
558         * tree-streamer.c: Likewise.
559         * value-prof.c: Likewise.
560         * target-globals.c: Likewise.
561         * expr.h: Include tree-core.h instead of tree.h.
562         * gimple.h: Likewise.
563         * ipa-prop.h: Likewise.
564         * ipa-utils.h: Likewise.
565         * lto-streamer.h: Likewise.
566         * streamer-hooks.h: Likewise.
567         * ipa-reference.h: Include cgraph.h instead of tree.h.
568         * cgraph.h: Include basic-block.h instead of tree.h.
569         * tree-streamer.h: Do not include tree.h.
570         * genattrtab.c (write_header): Generate inclusion of
571         tree.h.
572         * genautomata.c (main): Likewise.
573         * genemit.c: Likewise.
574         * genopinit.c: Likewise.
575         * genoutput.c (output_prologue): Likewise.
576         * genpeep.c: Likewise.
578 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
580         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
581         little endian.
582         (vec_unpacku_hi_v8hi): Likewise.
583         (vec_unpacku_lo_v16qi): Likewise.
584         (vec_unpacku_lo_v8hi): Likewise.
586 2013-10-20  Jan Hubicka  <jh@suse.cz>
588         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
589         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
590         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
592 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
594         * config/mips/mips.h (ISA_HAS_WSBH): Define.
595         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
596         constants.
597         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
599 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
601         PR target/58603
602         * system.h: Undef m_slot.
604 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
606         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
607         all elements for both endian flavors.
609 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
611         PR target/58792
612         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
613         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
614         and SI_REG for 64bit SYSV ABI targets.
616 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
618         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
619         to multi_reg_return.  Clarify that we are skipping USEs of multiple
620         return registers.  Use bool type where appropriate.
622 2013-10-18  Jan Hubicka  <jh@suse.cz>
624         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
625         for cold functions.
626         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
627         (X86_TUNE_PUSH_MEMORY): Likewise.
628         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
629         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
630         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
631         New.
632         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
633         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
634         Remove.
635         (ix86_option_override_internal): Update to use tune features instead
636         of variables.
638 2013-10-18  Cong Hou  <congh@google.com>
640         PR tree-optimization/58508
641         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
642         statement that contains data refs with zero-step.
644 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
646         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
647         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
648         * gimple-low.c (gimple_check_call_arg,
649         gimple_check_call_matching_types): Move to cgraph.c.
650         * gimple-low.h: Remove prototype.
651         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
652         Relocate from gimple-low.c.
653         * cgraph.h: Add prototype.  Don't include basic-block.h.
654         * gimplify.c: Add gimple-low to include list.
655         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
656         * tree-eh.c: Add gimple-low to include list.
657         * tree-nested.c: Likewise.
658         * cfgexpand.c: Add tree-ssa-address.h to include list.
659         * expr.c: Likewise.
660         * gimple-fold.c: Likewise.
661         * gimple-ssa-strength-reduction.c: Likewise.
662         * trans-mem.c: Likewise.
663         * tree-mudflap.c: Likewise.
664         * tree-ssa-loop-ivopts.c: Likewise.
665         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
666         (degenerate_phi_result): Move to tree-phinodes.c.
667         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
668         * tree-ssa-threadedge.c: Likewise.
669         * tree-vrp.c: Likewise.
670         * tree-phinodes.c (degenerate_phi_result): Relocate here.
671         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
672         * tree-phinodes.h (degenerate_phi_result): Add prototype.
673         * tree-ssa-copy.c: Include tree-ssa-dom.h.
674         * tree-ssa-forwprop.c: Likewise.
675         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
676         pass_data_cleanup_cfg_post_optimizing,
677         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
678         * tree-optimize.c: Delete File.
679         * graphite.c: Include tree-cfgcleanup.h.
680         * passes.c: Likewise.
681         * tree-cfg.c: Likewise.
682         * tree-profile.c: Likewise.
683         * tree-ssa-dse.c: Likewise.
684         * tree-ssa-loop-ivcanon.c: Likewise.
685         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
686         * tree-outof-ssa.c: Include sbitmap.h.
687         * tree-ssa-live.c: Likewise.
688         * tree-ssa-propagate.c: Likewise.
689         * tree-ssa-structalias.c: Likewise.
690         * tree-stdarg.c: Likewise.
691         * Makefile.in (OBJS): Delete tree-optimize.o.
692         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
693         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
694         * varasm.c: Include basic-block.h.
695         * cfgloop.h: Include function.h instead of basic-block.h
696         (bb_loop_depth): Move to cfgloop.c.
697         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
699 2013-10-18  Teresa Johnson  <tejohnson@google.com>
701         * predict.c (probably_never_executed): Compare frequency-based
702         count to number of training runs.
703         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
705 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
707         * config/arm/arm.c (cortexa9_extra_costs): New table.
708         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
710 2013-10-18  Jeff Law  <law@redhat.com>
712         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
714         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
715         (register_jump_thread): Add "registered_jump_thread" debug
716         counter support.
717         * dbgcnt.def (registered_jump_thread): New debug counter.
719 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
721         * config/rs6000/rs6000.c: Include cgraph.h.
723 2013-10-18  Teresa Johnson  <tejohnson@google.com>
725         * tree-ssa-tail-merge.c (replace_block_by): Update edge
726         weights during merging.
728 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
730         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
731         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
732         * tree-ssa-operands.h: Remove prototype.
733         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
734         * gimple.c (virtual_operand_p): Relocate from gimple.c.
735         * gimple.h: Add prototype.
736         * gimple-ssa.h: Include tree-ssa-operands.h.
737         * tree-dump.c: Add tree-cfg.h to include list.
738         * tree-ssa-alias.c: Add ipa-reference.h to include list.
739         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
740         * config/i386/i386.c: Don't include tree-flow.h.
741         * config/rs6000/rs6000.c: Likewise.
743 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
745         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
747 2013-10-18  Richard Biener  <rguenther@suse.de>
749         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
750         or TYPE_UNSIGNED of integral types.
751         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
752         NULL_TREE for zero-precision integral types.
754 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
756         * config/aarch64/arm_neon.h
757         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
759 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
761         PR target/57756
762         * opth-gen.awk: Define target_flags_explicit.
764 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
766         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
767         fields to the reg_addr array that describes the valid addressing
768         mode for any register, general purpose registers, floating point
769         registers, and Altivec registers.
770         (FIRST_RELOAD_REG_CLASS): Likewise.
771         (LAST_RELOAD_REG_CLASS): Likewise.
772         (struct reload_reg_map_type): Likewise.
773         (reload_reg_map_type): Likewise.
774         (RELOAD_REG_VALID): Likewise.
775         (RELOAD_REG_MULTIPLE): Likewise.
776         (RELOAD_REG_INDEXED): Likewise.
777         (RELOAD_REG_OFFSET): Likewise.
778         (RELOAD_REG_PRE_INCDEC): Likewise.
779         (RELOAD_REG_PRE_MODIFY): Likewise.
780         (reg_addr): Likewise.
781         (mode_supports_pre_incdec_p): New helper functions to say whether
782         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
783         (mode_supports_pre_modify_p): Likewise.
784         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
785         print the valid address mode bits for each mode.
786         (rs6000_debug_print_mode): Likewise.
787         (rs6000_debug_reg_global): Likewise.
788         (rs6000_setup_reg_addr_masks): New function to set up the address
789         mask bits for each type.
790         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
791         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
792         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
793         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
794         PRE_MODIFY are supported.
795         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
796         registers, instead of {src,dest}_av_p.
797         (rs6000_print_options_internal): Tweak the debug output slightly.
799 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
801         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
802         isa attribute.
804 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
806         * tree-flow.h (struct omp_region): Move to omp-low.c.
807         Remove omp_ prototypes and variables.
808         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
809         (copy_var_decl): Relocate prototype from tree-flow.h.
810         * gimple.c (copy_var_decl): Relocate from omp-low.c.
811         * tree.h: Move prototype to omp-low.h.
812         * omp-low.h: New File.  Relocate prototypes here.
813         * omp-low.c (struct omp_region): Make local here.
814         (root_omp_region): Make static.
815         (copy_var_decl) Move to gimple.c.
816         (new_omp_region): Make static.
817         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
818         * tree-cfg.c: Include omp-low.h.
819         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
820         * gimplify.c: Include omp-low.h.
821         * tree-parloops.c: Likewise.
823 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
825         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
826         operands are in memory, prefer to force non-matched operand 1 to
827         the register.
829 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
831         PR target/58673
832         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
833         restrict TImode addresses to single indirect registers if both
834         -mquad-memory and -mvsx-timode are used.
835         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
836         we should emit load/store quad.  Remove using %y for quad memory
837         addresses.
839         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
840         constraints to allow load/store quad on machines where TImode is
841         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
842         for TImode to load integer constants.
844 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
846         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
848 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
850         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
851         handling of STACK_REG.
853 2013-10-17  Richard Biener  <rguenther@suse.de>
855         PR tree-optimization/58143
856         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
857         New function.
858         (rewrite_to_defined_overflow): Likewise.
859         (move_computations_dom_walker::before_dom): Rewrite stmts
860         with undefined signed overflow that are not always executed
861         into unsigned arithmetic.
863 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
865         PR target/57756
866         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
867         explicit isa flag to be an options variable, instead of using
868         global_options_set.  Remove define from rs6000.h.
869         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
871         * config/rs6000/rs6000.c (rs6000_option_override_internal):
872         Initialize rs6000_isa_flags_explicit.
873         (rs6000_function_specific_save): Add gcc_options* parameter, so
874         that the powerpc builds after the 2013-10-15 changes.
875         (rs6000_function_specific_restore): Likewise.
877 2013-10-16  DJ Delorie  <dj@redhat.com>
879         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
880         op is a REG before checking REGNO.
881         (rl78_alloc_physical_registers): Verify pattern is a SET before
882         checking SET_SRC.
884 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
886         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
887         endianness.
888         (vec_unpacks_lo_v4sf): Likewise.
889         (vec_unpacks_float_hi_v4si): Likewise.
890         (vec_unpacks_float_lo_v4si): Likewise.
891         (vec_unpacku_float_hi_v4si): Likewise.
892         (vec_unpacku_float_lo_v4si): Likewise.
894 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
896         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
897         (vsx_concat_v2sf): Likewise.
899 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
901         * config/aarch64/aarch64.md
902         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
904 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
906         PR tree-optimization/58697
907         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
908         estimated_loop_iterations_int.
909         (max_stmt_executions_int): Call get_max_loop_iterations_int.
910         (get_max_loop_iterations_int): New.  HWINT version of
911         get_max_loop_iterations.
912         * cfgloop.h: Add prototypes.
913         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
914         * loop-unroll.c (decide_peel_once_rolling): Call
915         get_estimated_loop_iterations_int.
916         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
917         * tree-ssa-loop-niter.h: Tweak prototypes.
919 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
921         * gengtype-parse.c (struct_field_seq): Ignore access-control
922         keywords ("public:" etc).
924 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
926         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
927         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
929 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
931         * config/arm/arm.opt (mlra): New option.
932         * config/arm/arm.c (arm_lra_p): New function.
933         (TARGET_LRA_P): Define.
935 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
937         * tree-core.h (tree_code_name): Remove.
938         * tree.h (get_tree_code_name): New prototype.
939         * tree.c (tree_code_name): Make static.
940         (get_tree_code_name): New function.
941         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
942         tree_class_check_failed, tree_range_check_failed,
943         tree_not_class_check_failed, omp_clause_check_failed,
944         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
945         wrapper get_tree_code_name instead of calling tree_code_name directly.
946         * tree-vrp.c (dump_asserts_for): Likewise.
947         * tree-dump.c (dequeue_and_dump): Likewise.
948         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
949         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
950         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
951         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
952         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
953         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
954         dump_gimple_omp_for): Likewise.
955         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
956         * tree-ssa-pre.c (print_pre_expr): Likewise.
957         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
958         * print-tree.c (print_node_brief, print_node): Likewise.
959         * gimple.c (gimple_check_failed): Likewise.
960         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
961         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
962         * config/mep/mep.c (mep_validate_vliw): Likewise.
963         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
964         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
966 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
968         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
969         for AMD bdver3.
971 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
973         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
974         (MULTILIB_MATCHES): Add multilib for -march=v8.
976 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
978         PR target/57756
979         * optc-save-gen.awk: Add extra parameter to the save and restore
980         target calls.
981         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
982         * tree.c (build_optimization_node): New parameter.  Add extra parameter
983         to call to cl_optimization_save.
984         (build_target_option_node): New parameter. Add extra parameter
985         to call to cl_target_option_save.
986         * tree.h (build_optimization_node): New parameter.
987         (build_target_option_node): New parameter.
988         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
989         build_optimization_node and build_target_option_node.
990         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
991         (handle_pragma_push_options): Ditto.
992         * toplev.c (process_options): Ditto.
993         * opts.c (init_options_struct): Check for opts_set non-null.
994         * target.def (target_option.save): New parameter.
995         (target_option.restore): New parameter.
996         * tm.texi: Generate.
997         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
998         (ix86_pragma_target_parse): Ditto.
999         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
1000         parameters.
1001         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
1002         to  build_optimization_node and build_target_option_node.
1003         (rs6000_valid_attribute_p): Ditto.
1004         (rs6000_pragma_target_parse): Ditto.
1005         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
1006         data.
1007         * config/i386/i386.h:
1008         TARGET_64BIT_P: New Macro
1009         TARGET_MMX_P: New Macro.
1010         TARGET_3DNOW_P: New Macro.
1011         TARGET_3DNOW_A_P: New Macro.
1012         TARGET_SSE_P: New Macro.
1013         TARGET_SSE2_P: New Macro.
1014         TARGET_SSE3_P: New Macro.
1015         TARGET_SSSE3_P: New Macro.
1016         TARGET_SSE4_1_P: New Macro.
1017         TARGET_SSE4_2_P: New Macro.
1018         TARGET_AVX_P: New Macro.
1019         TARGET_AVX2_P: New Macro.
1020         TARGET_AVX512F_P: New Macro.
1021         TARGET_AVX512PF_P: New Macro.
1022         TARGET_AVX512ER_P: New Macro.
1023         TARGET_AVX512CD_P: New Macro.
1024         TARGET_FMA_P: New Macro.
1025         TARGET_SSE4A_P: New Macro.
1026         TARGET_FMA4_P: New Macro.
1027         TARGET_XOP_P: New Macro.
1028         TARGET_LWP_P: New Macro.
1029         TARGET_ABM_P: New Macro.
1030         TARGET_BMI_P: New Macro.
1031         TARGET_BMI2_P: New Macro.
1032         TARGET_LZCNT_P: New Macro.
1033         TARGET_TBM_P: New Macro.
1034         TARGET_POPCNT_P: New Macro.
1035         TARGET_SAHF_P: New Macro.
1036         TARGET_MOVBE_P: New Macro.
1037         TARGET_CRC32_P: New Macro.
1038         TARGET_AES_P: New Macro.
1039         TARGET_PCLMUL_P: New Macro.
1040         TARGET_CMPXCHG16B_P: New Macro.
1041         TARGET_FSGSBASE_P: New Macro.
1042         TARGET_RDRND_P: New Macro.
1043         TARGET_F16C_P: New Macro.
1044         TARGET_RTM_P: New Macro.
1045         TARGET_HLE_P: New Macro.
1046         TARGET_RDSEED_P: New Macro.
1047         TARGET_PRFCHW_P: New Macro.
1048         TARGET_ADX_P: New Macro.
1049         TARGET_FXSR_P: New Macro.
1050         TARGET_XSAVE_P: New Macro.
1051         TARGET_XSAVEOPT_P: New Macro.
1052         TARGET_LP64_P: New Macro.
1053         TARGET_X32_P: New Macro.
1054         TARGET_FPMATH_DEFAULT_P: New Macro.
1055         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
1056         * config/i386/i386.c (ix86_option_override_internal): New parameters.
1057         opts and opts_set.
1058         Change ix86_tune_string to access opts->x_ix86_tune_string.
1059         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
1060         Change ix86_arch_string to access opts->x_ix86_arch_string.
1061         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
1062         Change ix86_pmode to access opts->x_ix86_pmode.
1063         Change ix86_abi to access opts->x_ix86_abi.
1064         Change ix86_cmodel to access opts->x_ix86_cmodel.
1065         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
1066         Change ix86_isa_flags_explicit to access
1067         opts->x_ix86_isa_flags_explicit.
1068         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
1069         Change ix86_regparm to access opts->x_ix86_regparm.
1070         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
1071         Change ix86_preferred_stack_boundary_arg to access
1072         opts->x_ix86_preferred_stack_boundary_arg.
1073         Change ix86_force_align_arg_pointer to access
1074         opts->x_ix86_force_align_arg_pointer.
1075         Change ix86_incoming_stack_boundar_arg to access
1076         opts->x_ix86_incoming_stack_boundar_arg.
1077         Change ix86_fpmath to access opts->x_ix86_fpmath.
1078         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
1079         Change ix86_recip_name to access opts->x_ix86_recip_name.
1080         Change ix86_stack_protector_guard to access
1081         opts->x_ix86_stack_protector_guard.
1082         Change ix86_tune_memcpy_strategy to access
1083         opts->x_ix86_tune_memcpy_strategy.
1084         Change ix86_tune_memset_strategy to access
1085         opts->x_ix86_tune_memset_strategy.
1086         Change global_options to access opts.
1087         Change global_options_set to access opts_set.
1088         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
1089         Change TARGET_MMX to TARGET_MMX_P (opts->...).
1090         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
1091         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
1092         Change TARGET_SSE to TARGET_SSE_P (opts->...).
1093         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
1094         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
1095         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
1096         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
1097         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
1098         Change TARGET_AVX to TARGET_AVX_P (opts->...).
1099         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
1100         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
1101         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
1102         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
1103         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
1104         Change TARGET_FMA to TARGET_FMA_P (opts->...).
1105         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
1106         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
1107         Change TARGET_XOP to TARGET_XOP_P (opts->...).
1108         Change TARGET_LWP to TARGET_LWP_P (opts->...).
1109         Change TARGET_ABM to TARGET_ABM_P (opts->...).
1110         Change TARGET_BMI to TARGET_BMI_P (opts->...).
1111         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
1112         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
1113         Change TARGET_TBM to TARGET_TBM_P (opts->...).
1114         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
1115         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
1116         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
1117         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
1118         Change TARGET_AES to TARGET_AES_P (opts->...).
1119         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
1120         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
1121         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
1122         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
1123         Change TARGET_F16C to TARGET_F16C_P (opts->...).
1124         Change TARGET_RTM to TARGET_RTM_P (opts->...).
1125         Change TARGET_HLE to TARGET_HLE_P (opts->...).
1126         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
1127         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
1128         Change TARGET_ADX to TARGET_ADX_P (opts->...).
1129         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
1130         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
1131         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
1132         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
1133         Change TARGET_X32 to TARGET_X32_P (opts->...).
1134         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
1135         Change TARGET_FLOAT_RETURNS_IN_80387 to
1136         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
1137         (ix86_function_specific_save): New parameter. Use opts-> fields
1138         to replace global fields.
1139         (ix86_function_specific_restore): Ditto.
1140         (ix86_valid_target_attribute_inner_p): New parameters.
1141         Fix recursive call.
1142         Fix call to ix86_handle_option and set_option.
1143         (ix86_valid_target_attribute_tree): New parameters.
1144         Change global_options to access opts.
1145         Change global_options_set to access opts_set.
1146         Fix call to ix86_valid_target_attribute_inner_p.
1147         Change ix86_tune_string to access opts->x_ix86_tune_string.
1148         Change ix86_arch_string to access opts->x_ix86_arch_string.
1149         Change ix86_fpmath to access opts->x_ix86_fpmath
1150         Fix call to ix86_option_override_internal.
1151         Fix call to ix86_add_new_builtins.
1152         Fix calls to build_optimization_node and build_target_option_node.
1153         (ix86_valid_target_attribute_p): Remove access to global_options.
1154         Use new gcc_options structure func_options.
1155         Fix call to ix86_valid_target_attribute_tree.
1156         Fix call to  build_optimization_node.
1157         (get_builtin_code_for_version): Fix call to
1158         ix86_valid_target_attribute_tree.
1160 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
1162         * Makefile.in (PICFLAG): New.
1163         (enable_host_shared): New.
1164         (INTERNAL_CFLAGS): Use PICFLAG.
1165         (LIBIBERTY): Use pic build of libiberty.a if configured with
1166         --enable-host-shared.
1167         * configure.ac: Add --enable-host-shared, setting up new
1168         PICFLAG variable.
1169         * configure: Regenerate.
1170         * doc/install.texi (--enable-shared): Add note contrasting it with ...
1171         (--enable-host-shared): New option.
1173 2013-10-15  Richard Biener  <rguenther@suse.de>
1175         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
1176         for built-in functions.
1178 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
1180         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
1181         (optimize_range_tests_1): New function,
1182         extracted from optimize_range_tests.
1183         (optimize_range_tests_xor): Similarly.
1184         (optimize_range_tests_diff): New function.
1185         (optimize_range_tests): Use optimize_range_tests_1.
1187 2013-10-15  Cong Hou  <congh@google.com>
1189         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
1190         requirement of the reduction pattern so that one operand of the
1191         reduction operation can come from outside of the loop.
1193 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1195         * config/arm/neon-schedgen.ml: Remove.
1196         * config/arm/cortex-a9-neon.md: Remove comment regarding
1197         neon-schedgen.ml.
1199 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1201         * config/arm/types: Remove old neon types.
1203 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1205         * config/arm/cortex-a7.md
1206         (cortex_a7_neon_type): New.
1207         (cortex_a7_neon_mul): Update for new types.
1208         (cortex_a7_neon_mla): Likewise.
1209         (cortex_a7_neon): Likewise.
1211 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1213         * config/arm/cortex-a15-neon.md
1214         (cortex_a15_neon_type): New,
1216         (cortex_a15_neon_int_1): Remove.
1217         (cortex_a15_neon_int_2): Likewise.
1218         (cortex_a15_neon_int_3): Likewise.
1219         (cortex_a15_neon_int_4): Likewise.
1220         (cortex_a15_neon_int_5): Likewise.
1221         (cortex_a15_neon_vqneg_vqabs): Likewise.
1222         (cortex_a15_neon_vmov): Likewise.
1223         (cortex_a15_neon_vaba): Likewise.
1224         (cortex_a15_neon_vaba_qqq): Likewise.
1225         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1226         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1227         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1228         Likewise.
1229         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1230         (cortex_a15_neon_mla_qqq_8_16): Likewise.
1231         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
1232         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1233         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1234         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
1235         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1236         (cortex_a15_neon_shift_1): Likewise.
1237         (cortex_a15_neon_shift_2): Likewise.
1238         (cortex_a15_neon_shift_3): Likewise.
1239         (cortex_a15_neon_vshl_ddd): Likewise.
1240         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1241         (cortex_a15_neon_vsra_vrsra): Likewise.
1242         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1243         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1244         (cortex_a15_neon_bp_3cycle): Likewise.
1245         (cortex_a15_neon_ldm_2): Likewise.
1246         (cortex_a15_neon_stm_2): Likewise.
1247         (cortex_a15_neon_mcr): Likewise.
1248         (cortex_a15_neon_mrc): Likewise.
1249         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
1250         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
1251         (cortex_a15_neon_fp_vmul_ddd): Likewise.
1252         (cortex_a15_neon_fp_vmul_qqd): Likewise.
1253         (cortex_a15_neon_fp_vmla_ddd): Likewise.
1254         (cortex_a15_neon_fp_vmla_qqq): Likewise.
1255         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1256         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1257         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1258         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1259         (cortex_a15_neon_bp_simple): Likewise.
1260         (cortex_a15_neon_bp_2cycle): Likewise.
1261         (cortex_a15_neon_bp_3cycle): Likewise.
1262         (cortex_a15_neon_vld1_1_2_regs): Likewise.
1263         (cortex_a15_neon_vld1_3_4_regs): Likewise.
1264         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1265         (cortex_a15_neon_vld2_4_regs): Likewise.
1266         (cortex_a15_neon_vld3_vld4): Likewise.
1267         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1268         (cortex_a15_neon_vst1_3_4_regs): Likewise.
1269         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
1270         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
1271         (cortex_a15_neon_vst3_vst4): Rename to...
1272         (cortex_a15_neon_vst4): This, update for new attributes.
1273         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
1274         (cortex_a15_neon_vld3_vld4_lane): Likewise.
1275         (cortex_a15_neon_vst1_vst2_lane): Likewise.
1276         (cortex_a15_neon_vst3_vst4_lane): Likewise.
1277         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
1278         (cortex_a15_neon_ldm_2): Likewise.
1279         (cortex_a15_neon_stm_2): Likewise.
1280         (cortex_a15_neon_mcr): Likewise.
1281         (cortex_a15_neon_mcr_2_mcrr): Likewise.
1282         (cortex_a15_neon_mrc): Likewise.
1283         (cortex_a15_neon_mrrc): Likewise.
1285         (cortex_a15_neon_abd): New.
1286         (cortex_a15_neon_abd_q): Likewise.
1287         (cortex_a15_neon_aba): Likewise.
1288         (cortex_a15_neon_aba_q): Likewise.
1289         (cortex_a15_neon_acc): Likewise.
1290         (cortex_a15_neon_acc_q): Likewise.
1291         (cortex_a15_neon_arith_basic): Likewise.
1292         (cortex_a15_neon_arith_complex): Likewise.
1293         (cortex_a15_neon_multiply): Likewise.
1294         (cortex_a15_neon_multiply_q): Likewise.
1295         (cortex_a15_neon_mla): Likewise.
1296         (cortex_a15_neon_mla_q): Likewise.
1297         (cortex_a15_neon_sat_mla_long): Likewise.
1298         (cortex_a15_neon_shift_acc): Likewise.
1299         (cortex_a15_neon_shift_imm_basic): Likewise.
1300         (cortex_a15_neon_shift_imm_complex): Likewise.
1301         (cortex_a15_neon_shift_reg_basic): Likewise.
1302         (cortex_a15_neon_shift_reg_basic_q): Likewise.
1303         (cortex_a15_neon_shift_reg_complex): Likewise.
1304         (cortex_a15_neon_shift_reg_complex_q): Likewise.
1305         (cortex_a15_neon_fp_negabs): Likewise
1306         (cortex_a15_neon_fp_arith): Likewise
1307         (cortex_a15_neon_fp_arith_q): Likewise
1308         (cortex_a15_neon_fp_cvt_int): Likewise
1309         (cortex_a15_neon_fp_cvt_int_q): Likewise
1310         (cortex_a15_neon_fp_cvt_16): Likewise
1311         (cortex_a15_neon_fp_mul): Likewise
1312         (cortex_a15_neon_fp_mul_q): Likewise
1313         (cortex_a15_neon_fp_mla): Likewise
1314         (cortex_a15_neon_fp_mla_q): Likewise
1315         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
1316         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
1317         (cortex_a15_neon_bitops): Likewise.
1318         (cortex_a15_neon_bitops_q): Likewise.
1319         (cortex_a15_neon_from_gp): Likewise.
1320         (cortex_a15_neon_from_gp_q): Likewise.
1321         (cortex_a15_neon_tbl3_tbl4): Likewise.
1322         (cortex_a15_neon_zip_q): Likewise.
1323         (cortex_a15_neon_to_gp): Likewise.
1324         (cortex_a15_neon_load_a): Likewise.
1325         (cortex_a15_neon_load_b): Likewise.
1326         (cortex_a15_neon_load_c): Likewise.
1327         (cortex_a15_neon_load_d): Likewise.
1328         (cortex_a15_neon_load_e): Likewise.
1329         (cortex_a15_neon_load_f): Likewise.
1330         (cortex_a15_neon_store_a): Likewise.
1331         (cortex_a15_neon_store_b): Likewise.
1332         (cortex_a15_neon_store_c): Likewise.
1333         (cortex_a15_neon_store_d): Likewise.
1334         (cortex_a15_neon_store_e): Likewise.
1335         (cortex_a15_neon_store_f): Likewise.
1336         (cortex_a15_neon_store_g): Likewise.
1337         (cortex_a15_neon_store_h): Likewise.
1338         (cortex_a15_vfp_to_from_gp): Likewise.
1340 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1342         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
1344         (cortex_a9_neon_vshl_ddd): Remove.
1345         (cortex_a9_neon_vst3_vst4): Likewise.
1346         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1348         (cortex_a9_neon_bit_ops_q): New.
1350         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
1351         (cortex_a9_neon_int_2): Likewise.
1352         (cortex_a9_neon_int_3): Likewise.
1353         (cortex_a9_neon_int_4): Likewise.
1354         (cortex_a9_neon_int_5): Likewise.
1355         (cortex_a9_neon_vqneg_vqabs): Likewise.
1356         (cortex_a9_neon_vmov): Likewise.
1357         (cortex_a9_neon_vaba): Likewise.
1358         (cortex_a9_neon_vaba_qqq): Likewise.
1359         (cortex_a9_neon_shift_1): Likewise.
1360         (cortex_a9_neon_shift_2): Likewise.
1361         (cortex_a9_neon_shift_3): Likewise.
1362         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1363         (cortex_a9_neon_vsra_vrsra): Likewise.
1364         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1365         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1366         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1367         Likewise.
1368         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1369         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1370         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1371         Likewise.
1372         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1373         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1374         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1375         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1376         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1377         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1378         (cortex_a9_neon_fp_vsum): Likewise.
1379         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1380         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1381         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1382         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1383         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1384         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1385         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1386         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1387         (cortex_a9_neon_bp_simple): Likewise.
1388         (cortex_a9_neon_bp_2cycle): Likewise.
1389         (cortex_a9_neon_bp_3cycle): Likewise.
1390         (cortex_a9_neon_ldr): Likewise.
1391         (cortex_a9_neon_str): Likewise.
1392         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1393         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1394         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1395         (cortex_a9_neon_vld2_4_regs): Likewise.
1396         (cortex_a9_neon_vld3_vld4): Likewise.
1397         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1398         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1399         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1400         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1401         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1402         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1403         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1404         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1405         (cortex_a9_neon_mcr): Likewise.
1406         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1407         (cortex_a9_neon_mrc): Likewise.
1408         (cortex_a9_neon_mrrc): Likewise.
1410 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1412         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
1414         (cortex_a8_neon_vshl_ddd): Remove.
1415         (cortex_a8_neon_vst3_vst4): Likewise.
1416         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1418         (cortex_a8_neon_bit_ops_q): New.
1420         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
1421         (cortex_a8_neon_int_2): Likewise..
1422         (cortex_a8_neon_int_3): Likewise.
1423         (cortex_a8_neon_int_5): Likewise.
1424         (cortex_a8_neon_vqneg_vqabs): Likewise.
1425         (cortex_a8_neon_int_4): Likewise.
1426         (cortex_a8_neon_vaba): Likewise.
1427         (cortex_a8_neon_vaba_qqq): Likewise.
1428         (cortex_a8_neon_shift_1): Likewise.
1429         (cortex_a8_neon_shift_2): Likewise.
1430         (cortex_a8_neon_shift_3): Likewise.
1431         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1432         (cortex_a8_neon_vsra_vrsra): Likewise.
1433         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1434         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1435         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1436         Likewise.
1437         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1438         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1439         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1440         Likewise.
1441         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1442         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1443         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1444         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1445         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1446         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1447         (cortex_a8_neon_fp_vsum): Likewise.
1448         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1449         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1450         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1451         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1452         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1453         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1454         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1455         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1456         (cortex_a8_neon_bp_simple): Likewise.
1457         (cortex_a8_neon_bp_2cycle): Likewise.
1458         (cortex_a8_neon_bp_3cycle): Likewise.
1459         (cortex_a8_neon_ldr): Likewise.
1460         (cortex_a8_neon_str): Likewise.
1461         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1462         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1463         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1464         (cortex_a8_neon_vld2_4_regs): Likewise.
1465         (cortex_a8_neon_vld3_vld4): Likewise.
1466         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1467         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1468         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1469         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1470         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1471         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1472         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1473         (cortex_a8_neon_mcr): Likewise.
1474         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1475         (cortex_a8_neon_mrc): Likewise.
1476         (cortex_a8_neon_mrrc): Likewise.
1478 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1480         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1481         (fp): New.
1482         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1483         (aarch64_simd_dup<mode>): Add "type" attribute.
1484         (aarch64_dup_lane<mode>): Likewise.
1485         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1486         (*aarch64_simd_mov<mode>): Likewise.
1487         (aarch64_simd_mov_from_<mode>low): Likewise.
1488         (aarch64_simd_mov_from_<mode>high): Likewise.
1489         (orn<mode>3): Likewise.
1490         (bic<mode>3): Likewise.
1491         (add<mode>3): Likewise.
1492         (sub<mode>3): Likewise.
1493         (mul<mode>3): Likewise.
1494         (*aarch64_mul3_elt<mode>): Likewise.
1495         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1496         (*aarch64_mul3_elt_to_128df): Likewise.
1497         (*aarch64_mul3_elt_to_64v2df): Likewise.
1498         (neg<mode>2): Likewise.
1499         (abs<mode>2): Likewise.
1500         (abd<mode>_3): Likewise.
1501         (aba<mode>_3): Likewise.
1502         (fabd<mode>_3): Likewise.
1503         (*fabd_scalar<mode>3): Likewise.
1504         (and<mode>3): Likewise.
1505         (ior<mode>3): Likewise.
1506         (xor<mode>3): Likewise.
1507         (one_cmpl<mode>2): Likewise.
1508         (aarch64_simd_vec_set<mode>): Likewise.
1509         (aarch64_simd_lshr<mode>): Likewise.
1510         (aarch64_simd_ashr<mode>): Likewise.
1511         (aarch64_simd_imm_shl<mode>): Likewise.
1512         (aarch64_simd_reg_sshl<mode): Likewise.
1513         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1514         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1515         (aarch64_simd_vec_setv2di): Likewise.
1516         (aarch64_simd_vec_set<mode>): Likewise.
1517         (aarch64_mla<mode>): Likewise.
1518         (*aarch64_mla_elt<mode>): Likewise.
1519         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1520         (aarch64_mls<mode>): Likewise.
1521         (*aarch64_mls_elt<mode>): Likewise.
1522         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1523         (<su><maxmin><mode>3): Likewise.
1524         (move_lo_quad_<mode>): Likewise.
1525         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1526         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1527         (vec_pack_trunc_<mode>): Likewise.
1528         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1529         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1530         (*aarch64_<su>mlal_lo<mode>): Likewise.
1531         (*aarch64_<su>mlal_hi<mode>): Likewise.
1532         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1533         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1534         (*aarch64_<su>mlal<mode>): Likewise.
1535         (*aarch64_<su>mlsl<mode>): Likewise.
1536         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1537         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1538         (add<mode>3): Likewise.
1539         (sub<mode>3): Likewise.
1540         (mul<mode>3): Likewise.
1541         (div<mode>3): Likewise.
1542         (neg<mode>2): Likewise.
1543         (abs<mode>2): Likewise.
1544         (fma<mode>4): Likewise.
1545         (*aarch64_fma4_elt<mode>): Likewise.
1546         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1547         (*aarch64_fma4_elt_to_128df): Likewise.
1548         (*aarch64_fma4_elt_to_64v2df): Likewise.
1549         (fnma<mode>4): Likewise.
1550         (*aarch64_fnma4_elt<mode>): Likewise.
1551         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1552         (*aarch64_fnma4_elt_to_128df): Likewise.
1553         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1554         (<frint_pattern><mode>2): Likewise.
1555         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1556         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1557         (vec_unpacks_lo_v4sf): Likewise.
1558         (aarch64_float_extend_lo_v2df): Likewise.
1559         (vec_unpacks_hi_v4sf): Likewise.
1560         (aarch64_float_truncate_lo_v2sf): Likewise.
1561         (aarch64_float_truncate_hi_v4sf): Likewise.
1562         (aarch64_vmls<mode>): Likewise.
1563         (<su><maxmin><mode>3): Likewise.
1564         (<maxmin_uns><mode>3): Likewise.
1565         (reduc_<sur>plus_<mode>): Likewise.
1566         (reduc_<sur>plus_v2di): Likewise.
1567         (reduc_<sur>plus_v2si): Likewise.
1568         (reduc_<sur>plus_<mode>): Likewise.
1569         (aarch64_addpv4sf): Likewise.
1570         (clz<mode>2): Likewise.
1571         (reduc_<maxmin_uns>_<mode>): Likewise.
1572         (reduc_<maxmin_uns>_v2di): Likewise.
1573         (reduc_<maxmin_uns>_v2si): Likewise.
1574         (reduc_<maxmin_uns>_<mode>): Likewise.
1575         (reduc_<maxmin_uns>_v4sf): Likewise.
1576         (aarch64_simd_bsl<mode>_internal): Likewise.
1577         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1578         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1579         (aarch64_get_lane<mode>): Likewise.
1580         (*aarch64_combinez<mode>): Likewise.
1581         (aarch64_combine<mode>): Likewise.
1582         (aarch64_simd_combine<mode>): Likewise.
1583         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1584         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1585         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1586         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1587         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1588         (aarch64_<sur>h<addsub><mode>): Likewise.
1589         (aarch64_<sur><addsub>hn<mode>): Likewise.
1590         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1591         (aarch64_pmul<mode>): Likewise.
1592         (aarch64_<su_optab><optab><mode>): Likewise.
1593         (aarch64_<sur>qadd<mode>): Likewise.
1594         (aarch64_sqmovun<mode>): Likewise.
1595         (aarch64_<sur>qmovn<mode>): Likewise.
1596         (aarch64_s<optab><mode>): Likewise.
1597         (aarch64_sq<r>dmulh<mode>): Likewise.
1598         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1599         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1600         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1601         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1602         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1603         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1604         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1605         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1606         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1607         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1608         (aarch64_sqdmull<mode>): Likewise.
1609         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1610         (aarch64_sqdmull_n<mode>): Likewise.
1611         (aarch64_sqdmull2<mode>_internal): Likewise.
1612         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1613         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1614         (aarch64_<sur>shl<mode>): Likewise.
1615         (aarch64_<sur>q<r>shl<mode>
1616         (aarch64_<sur>shll_n<mode>): Likewise.
1617         (aarch64_<sur>shll2_n<mode>): Likewise.
1618         (aarch64_<sur>shr_n<mode>): Likewise.
1619         (aarch64_<sur>sra_n<mode>): Likewise.
1620         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1621         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1622         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1623         (aarch64_cm<optab><mode>): Likewise.
1624         (aarch64_cm<optab>di): Likewise.
1625         (aarch64_cm<optab><mode>): Likewise.
1626         (aarch64_cm<optab>di): Likewise.
1627         (aarch64_cmtst<mode>): Likewise.
1628         (aarch64_cmtstdi): Likewise.
1629         (aarch64_cm<optab><mode>): Likewise.
1630         (*aarch64_fac<optab><mode>): Likewise.
1631         (aarch64_addp<mode>): Likewise.
1632         (aarch64_addpdi): Likewise.
1633         (sqrt<mode>2): Likewise.
1634         (vec_load_lanesoi<mode>): Likewise.
1635         (vec_store_lanesoi<mode>): Likewise.
1636         (vec_load_lanesci<mode>): Likewise.
1637         (vec_store_lanesci<mode>): Likewise.
1638         (vec_load_lanesxi<mode>): Likewise.
1639         (vec_store_lanesxi<mode>): Likewise.
1640         (*aarch64_mov<mode>): Likewise.
1641         (aarch64_ld2<mode>_dreg): Likewise.
1642         (aarch64_ld2<mode>_dreg): Likewise.
1643         (aarch64_ld3<mode>_dreg): Likewise.
1644         (aarch64_ld3<mode>_dreg): Likewise.
1645         (aarch64_ld4<mode>_dreg): Likewise.
1646         (aarch64_ld4<mode>_dreg): Likewise.
1647         (aarch64_tbl1<mode>): Likewise.
1648         (aarch64_tbl2v16qi): Likewise.
1649         (aarch64_combinev16qi): Likewise.
1650         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1651         (aarch64_st2<mode>_dreg): Likewise.
1652         (aarch64_st2<mode>_dreg): Likewise.
1653         (aarch64_st3<mode>_dreg): Likewise.
1654         (aarch64_st3<mode>_dreg): Likewise.
1655         (aarch64_st4<mode>_dreg): Likewise.
1656         (aarch64_st4<mode>_dreg): Likewise.
1657         (*aarch64_simd_ld1r<mode>): Likewise.
1658         (aarch64_frecpe<mode>): Likewise.
1659         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1660         (aarch64_frecps<mode>): Likewise.
1662 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1664         * config/arm/iterators.md (V_elem_ch): New.
1665         (q): Likewise.
1666         (VQH_type): Likewise.
1667         * config/arm/arm.md (is_neon_type): New.
1668         (conds): Use is_neon_type.
1669         (anddi3_insn): Update type attribute.
1670         (xordi3_insn): Likewise.
1671         (one_cmpldi2): Likewise.
1672         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1673         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1674         (*movmisalign<mode>_neon_store): Likewise.
1675         (*movmisalign<mode>_neon_load): Likewise.
1676         (vec_set<mode>_internal): Likewise.
1677         (vec_set<mode>_internal): Likewise.
1678         (vec_setv2di_internal): Likewise.
1679         (vec_extract<mode>): Likewise.
1680         (vec_extract<mode>): Likewise.
1681         (vec_extractv2di): Likewise.
1682         (*add<mode>3_neon): Likewise.
1683         (adddi3_neon): Likewise.
1684         (*sub<mode>3_neon): Likewise.
1685         (subdi3_neon): Likewise.
1686         (fma<VCVTF:mode>4): Likewise.
1687         (fma<VCVTF:mode>4_intrinsic): Likewise.
1688         (*fmsub<VCVTF:mode>4): Likewise.
1689         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1690         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1691         (ior<mode>3): Likewise.
1692         (and<mode>3): Likewise.
1693         (orn<mode>3_neon): Likewise.
1694         (orndi3_neon): Likewise.
1695         (bic<mode>3_neon): Likewise.
1696         (bicdi3_neon): Likewise.
1697         (xor<mode>3): Likewise.
1698         (one_cmpl<mode>2): Likewise.
1699         (abs<mode>2): Likewise.
1700         (neg<mode>2): Likewise.
1701         (negdi2_neon): Likewise.
1702         (*umin<mode>3_neon): Likewise.
1703         (*umax<mode>3_neon): Likewise.
1704         (*smin<mode>3_neon): Likewise.
1705         (*smax<mode>3_neon): Likewise.
1706         (vashl<mode>3): Likewise.
1707         (vashr<mode>3_imm): Likewise.
1708         (vlshr<mode>3_imm): Likewise.
1709         (ashl<mode>3_signed): Likewise.
1710         (ashl<mode>3_unsigned): Likewise.
1711         (neon_load_count): Likewise.
1712         (ashldi3_neon_noclobber): Likewise.
1713         (ashldi3_neon): Likewise.
1714         (signed_shift_di3_neon): Likewise.
1715         (unsigned_shift_di3_neon): Likewise.
1716         (ashrdi3_neon_imm_noclobber): Likewise.
1717         (lshrdi3_neon_imm_noclobber): Likewise.
1718         (<shift>di3_neon): Likewise.
1719         (widen_ssum<mode>3): Likewise.
1720         (widen_usum<mode>3): Likewise.
1721         (quad_halves_<code>v4si): Likewise.
1722         (quad_halves_<code>v4sf): Likewise.
1723         (quad_halves_<code>v8hi): Likewise.
1724         (quad_halves_<code>v16qi): Likewise.
1725         (reduc_splus_v2di): Likewise.
1726         (neon_vpadd_internal<mode>): Likewise.
1727         (neon_vpsmin<mode>): Likewise.
1728         (neon_vpsmax<mode>): Likewise.
1729         (neon_vpumin<mode>): Likewise.
1730         (neon_vpumax<mode>): Likewise.
1731         (*ss_add<mode>_neon): Likewise.
1732         (*us_add<mode>_neon): Likewise.
1733         (*ss_sub<mode>_neon): Likewise.
1734         (*us_sub<mode>_neon): Likewise.
1735         (neon_vadd<mode>_unspec): Likewise.
1736         (neon_vaddl<mode>): Likewise.
1737         (neon_vaddw<mode>): Likewise.
1738         (neon_vhadd<mode>): Likewise.
1739         (neon_vqadd<mode>): Likewise.
1740         (neon_vaddhn<mode>): Likewise.
1741         (neon_vmul<mode>): Likewise.
1742         (neon_vfms<VCVTF:mode>): Likewise.
1743         (neon_vmlal<mode>): Likewise.
1744         (neon_vmls<mode>): Likewise.
1745         (neon_vmlsl<mode>): Likewise.
1746         (neon_vqdmulh<mode>): Likewise.
1747         (neon_vqdmlal<mode>): Likewise.
1748         (neon_vqdmlsl<mode>): Likewise.
1749         (neon_vmull<mode>): Likewise.
1750         (neon_vqdmull<mode>): Likewise.
1751         (neon_vsub<mode>_unspec): Likewise.
1752         (neon_vsubl<mode>): Likewise.
1753         (neon_vsubw<mode>): Likewise.
1754         (neon_vqsub<mode>): Likewise.
1755         (neon_vhsub<mode>): Likewise.
1756         (neon_vsubhn<mode>): Likewise.
1757         (neon_vceq<mode>): Likewise.
1758         (neon_vcge<mode>): Likewise.
1759         (neon_vcgeu<mode>): Likewise.
1760         (neon_vcgt<mode>): Likewise.
1761         (neon_vcgtu<mode>): Likewise.
1762         (neon_vcle<mode>): Likewise.
1763         (neon_vclt<mode>): Likewise.
1764         (neon_vcage<mode>): Likewise.
1765         (neon_vcagt<mode>): Likewise.
1766         (neon_vtst<mode>): Likewise.
1767         (neon_vabd<mode>): Likewise.
1768         (neon_vabdl<mode>): Likewise.
1769         (neon_vaba<mode>): Likewise.
1770         (neon_vabal<mode>): Likewise.
1771         (neon_vmax<mode>): Likewise.
1772         (neon_vmin<mode>): Likewise.
1773         (neon_vpaddl<mode>): Likewise.
1774         (neon_vpadal<mode>): Likewise.
1775         (neon_vpmax<mode>): Likewise.
1776         (neon_vpmin<mode>): Likewise.
1777         (neon_vrecps<mode>): Likewise.
1778         (neon_vrsqrts<mode>): Likewise.
1779         (neon_vqabs<mode>): Likewise.
1780         (neon_vqneg<mode>): Likewise.
1781         (neon_vcls<mode>): Likewise.
1782         (clz<mode>2): Likewise.
1783         (popcount<mode>2): Likewise.
1784         (neon_vrecpe<mode>): Likewise.
1785         (neon_vrsqrte<mode>): Likewise.
1786         (neon_vget_lane<mode>_sext_internal): Likewise.
1787         (neon_vget_lane<mode>_zext_internal): Likewise.
1788         (neon_vdup_n<mode>): Likewise.
1789         (neon_vdup_n<mode>): Likewise.
1790         (neon_vdup_nv2di): Likewise.
1791         (neon_vdup_lane<mode>_interal): Likewise.
1792         (*neon_vswp<mode>): Likewise.
1793         (neon_vcombine<mode>): Likewise.
1794         (float<mode><V_cvtto>2): Likewise.
1795         (floatuns<mode><V_cvtto>2): Likewise.
1796         (fix_trunc<mode><V_cvtto>2): Likewise.
1797         (fixuns_trunc<mode><V_cvtto>2
1798         (neon_vcvt<mode>): Likewise.
1799         (neon_vcvt<mode>): Likewise.
1800         (neon_vcvtv4sfv4hf): Likewise.
1801         (neon_vcvtv4hfv4sf): Likewise.
1802         (neon_vcvt_n<mode>): Likewise.
1803         (neon_vcvt_n<mode>): Likewise.
1804         (neon_vmovn<mode>): Likewise.
1805         (neon_vqmovn<mode>): Likewise.
1806         (neon_vqmovun<mode>): Likewise.
1807         (neon_vmovl<mode>): Likewise.
1808         (neon_vmul_lane<mode>): Likewise.
1809         (neon_vmul_lane<mode>): Likewise.
1810         (neon_vmull_lane<mode>): Likewise.
1811         (neon_vqdmull_lane<mode>): Likewise.
1812         (neon_vqdmulh_lane<mode>): Likewise.
1813         (neon_vqdmulh_lane<mode>): Likewise.
1814         (neon_vmla_lane<mode>): Likewise.
1815         (neon_vmla_lane<mode>): Likewise.
1816         (neon_vmlal_lane<mode>): Likewise.
1817         (neon_vqdmlal_lane<mode>): Likewise.
1818         (neon_vmls_lane<mode>): Likewise.
1819         (neon_vmls_lane<mode>): Likewise.
1820         (neon_vmlsl_lane<mode>): Likewise.
1821         (neon_vqdmlsl_lane<mode>): Likewise.
1822         (neon_vext<mode>): Likewise.
1823         (neon_vrev64<mode>): Likewise.
1824         (neon_vrev32<mode>): Likewise.
1825         (neon_vrev16<mode>): Likewise.
1826         (neon_vbsl<mode>_internal): Likewise.
1827         (neon_vshl<mode>): Likewise.
1828         (neon_vqshl<mode>): Likewise.
1829         (neon_vshr_n<mode>): Likewise.
1830         (neon_vshrn_n<mode>): Likewise.
1831         (neon_vqshrn_n<mode>): Likewise.
1832         (neon_vqshrun_n<mode>): Likewise.
1833         (neon_vshl_n<mode>): Likewise.
1834         (neon_vqshl_n<mode>): Likewise.
1835         (neon_vqshlu_n<mode>): Likewise.
1836         (neon_vshll_n<mode>): Likewise.
1837         (neon_vsra_n<mode>): Likewise.
1838         (neon_vsri_n<mode>): Likewise.
1839         (neon_vsli_n<mode>): Likewise.
1840         (neon_vtbl1v8qi): Likewise.
1841         (neon_vtbl2v8qi): Likewise.
1842         (neon_vtbl3v8qi): Likewise.
1843         (neon_vtbl4v8qi): Likewise.
1844         (neon_vtbl1v16qi): Likewise.
1845         (neon_vtbl2v16qi): Likewise.
1846         (neon_vcombinev16qi): Likewise.
1847         (neon_vtbx1v8qi): Likewise.
1848         (neon_vtbx2v8qi): Likewise.
1849         (neon_vtbx3v8qi): Likewise.
1850         (neon_vtbx4v8qi): Likewise.
1851         (*neon_vtrn<mode>_insn): Likewise.
1852         (*neon_vzip<mode>_insn): Likewise.
1853         (*neon_vuzp<mode>_insn): Likewise.
1854         (neon_vld1<mode>): Likewise.
1855         (neon_vld1_lane<mode>): Likewise.
1856         (neon_vld1_lane<mode>): Likewise.
1857         (neon_vld1_dup<mode>): Likewise.
1858         (neon_vld1_dup<mode>): Likewise.
1859         (neon_vld1_dupv2di): Likewise.
1860         (neon_vst1<mode>): Likewise.
1861         (neon_vst1_lane<mode>): Likewise.
1862         (neon_vst1_lane<mode>): Likewise.
1863         (neon_vld2<mode>): Likewise.
1864         (neon_vld2<mode>): Likewise.
1865         (neon_vld2_lane<mode>): Likewise.
1866         (neon_vld2_lane<mode>): Likewise.
1867         (neon_vld2_dup<mode>): Likewise.
1868         (neon_vst2<mode>): Likewise.
1869         (neon_vst2<mode>): Likewise.
1870         (neon_vst2_lane<mode>): Likewise.
1871         (neon_vst2_lane<mode>): Likewise.
1872         (neon_vld3<mode>): Likewise.
1873         (neon_vld3qa<mode>): Likewise.
1874         (neon_vld3qb<mode>): Likewise.
1875         (neon_vld3_lane<mode>): Likewise.
1876         (neon_vld3_lane<mode>): Likewise.
1877         (neon_vld3_dup<mode>): Likewise.
1878         (neon_vst3<mode>): Likewise.
1879         (neon_vst3qa<mode>): Likewise.
1880         (neon_vst3qb<mode>): Likewise.
1881         (neon_vst3_lane<mode>): Likewise.
1882         (neon_vst3_lane<mode>): Likewise.
1883         (neon_vld4<mode>): Likewise.
1884         (neon_vld4qa<mode>): Likewise.
1885         (neon_vld4qb<mode>): Likewise.
1886         (neon_vld4_lane<mode>): Likewise.
1887         (neon_vld4_lane<mode>): Likewise.
1888         (neon_vld4_dup<mode>): Likewise.
1889         (neon_vst4<mode>): Likewise.
1890         (neon_vst4qa<mode>): Likewise.
1891         (neon_vst4qb<mode>): Likewise.
1892         (neon_vst4_lane<mode>): Likewise.
1893         (neon_vst4_lane<mode>): Likewise.
1894         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1895         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1896         (neon_vec_<US>mult_lo_<mode>): Likewise.
1897         (neon_vec_<US>mult_hi_<mode>): Likewise.
1898         (neon_vec_<US>shiftl_<mode>): Likewise.
1899         (neon_unpack<US>_<mode>): Likewise.
1900         (neon_vec_<US>mult_<mode>): Likewise.
1901         (vec_pack_trunc_<mode>): Likewise.
1902         (neon_vec_pack_trunc_<mode>): Likewise.
1903         (neon_vabd<mode>_2): Likewise.
1904         (neon_vabd<mode>_3): Likewise.
1906 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1908         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1909         (load_pair): Update type attribute.
1910         (store_pair): Update type attribute.
1911         * config/aarch64/iterators.md (q): New.
1913 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1915         * config/arm/types.md: Add new types for Neon insns.
1917 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1918             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1919             Sergey Lega  <sergey.s.lega@intel.com>
1920             Anna Tikhonova  <anna.tikhonova@intel.com>
1921             Ilya Tocar  <ilya.tocar@intel.com>
1922             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1923             Ilya Verbin  <ilya.verbin@intel.com>
1924             Kirill Yukhin  <kirill.yukhin@intel.com>
1925             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1927         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1928         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1929         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1930         (rcp14<mode>): New.
1931         (srcp14<mode>): Ditto.
1932         (rsqrt14<mode>): Ditto.
1933         (rsqrt14<mode>): Ditto.
1934         (avx512f_vmscalef<mode>): Ditto.
1935         (avx512f_scalef<mode>): Ditto.
1936         (avx512f_getexp<mode>): Ditto.
1937         (avx512f_sgetexp<mode>): Ditto.
1938         (avx512f_fixupimm<mode>): Ditto.
1939         (avx512f_sfixupimm<mode>): Ditto.
1940         (avx512f_rndscale<mode>): Ditto.
1941         (*avx512er_exp2<mode>): Ditto.
1942         (*avx512er_rcp28<mode>): Ditto.
1943         (avx512er_rsqrt28<mode>): Ditto.
1944         (avx512f_getmant<mode>): Ditto.
1945         (avx512f_getmant<mode>): Ditto.
1946         (avx512f_rndscale<mode>): Fix formatting.
1948 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1950         * ipa-utils.h (ipa_edge_within_scc): Declare.
1951         * ipa-cp.c (edge_within_scc): Moved...
1952         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1954 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1955             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1956             Sergey Lega  <sergey.s.lega@intel.com>
1957             Anna Tikhonova  <anna.tikhonova@intel.com>
1958             Ilya Tocar  <ilya.tocar@intel.com>
1959             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1960             Ilya Verbin  <ilya.verbin@intel.com>
1961             Kirill Yukhin  <kirill.yukhin@intel.com>
1962             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1964         * config/i386/predicates.md (const_8_to_15_operand): New.
1965         (const_16_to_31_operand): Ditto.
1966         * config/i386/sse.md (V8FI): New.
1967         (V16FI): Ditto.
1968         (reduc_splus_v8df): Ditto.
1969         (reduc_splus_v16sf): Ditto.
1970         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1971         (vec_extract_hi_<mode>): Ditto.
1972         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1973         (vec_set_lo_<mode>): Ditto.
1974         (vec_set_hi_<mode>): Ditto.
1975         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1976         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1977         (avx512f_pshufd_1): Ditto.
1978         (avx512f_broadcast<mode>): Ditto.
1979         (avx512f_broadcast<mode>): Ditto.
1980         (define_split): Split vec_extract_lo into move.
1981         (ssequartermode): Ditto.
1982         (ssedoublemode): Extened with wider modes.
1983         (vec_extract_lo_<mode>): Ditto.
1985 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1986             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1987             Sergey Lega  <sergey.s.lega@intel.com>
1988             Anna Tikhonova  <anna.tikhonova@intel.com>
1989             Ilya Tocar  <ilya.tocar@intel.com>
1990             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1991             Ilya Verbin  <ilya.verbin@intel.com>
1992             Kirill Yukhin  <kirill.yukhin@intel.com>
1993             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1995         * config/i386/predicates.md (register_or_constm1_operand): New.
1996         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1997         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1998         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1999         UNSPEC_SCATTER_PREFETCH
2000         (VI48_512): New.
2001         (avx512f_ucmp<mode>3): Ditto.
2002         (avx512f_vternlog<mode>): Ditto.
2003         (avx512f_align<mode>): Ditto.
2004         (<shift_insn><mode>3): Ditto.
2005         (avx512f_<rotate>v<mode>): Ditto.
2006         (avx512f_<rotate><mode>): Ditto.
2007         (avx512f_eq<mode>3): Ditto.
2008         (avx512f_eq<mode>3_1): Ditto.
2009         (avx512f_gt<mode>3): Ditto.
2010         (avx512f_testm<mode>3): Ditto.
2011         (avx512f_testnm<mode>3): Ditto.
2012         (avx512pf_gatherpf<mode>): Ditto.
2013         (*avx512pf_gatherpf<mode>_mask): Ditto.
2014         (*avx512pf_gatherpf<mode>): Ditto.
2015         (avx512pf_scatterpf<mode>): Ditto.
2016         (*avx512pf_scatterpf<mode>_mask): Ditto.
2017         (*avx512pf_scatterpf<mode>): Ditto.
2018         (avx512f_vec_dup_gpr<mode>): Ditto.
2019         (clz<mode>2): Ditto.
2020         (conflict<mode>): Ditto.
2021         (REDUC_SMINMAX_MODE): Extened with wider modes.
2022         (reduc_<code>_<mode>): Ditto.
2023         (vlshr<mode>3): Ditto.
2024         (vashl<mode>3): Ditto.
2026 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2027             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2028             Sergey Lega  <sergey.s.lega@intel.com>
2029             Anna Tikhonova  <anna.tikhonova@intel.com>
2030             Ilya Tocar  <ilya.tocar@intel.com>
2031             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2032             Ilya Verbin  <ilya.verbin@intel.com>
2033             Kirill Yukhin  <kirill.yukhin@intel.com>
2034             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2036         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
2037         UNSPEC_SCATTER.
2038         (VI48F_512): New.
2039         (avx512fmaskmode): Ditto.
2040         (bcstscalarsuff): Ditto.
2041         (avx512f_blendm<mode>): Ditto.
2042         (cmp_imm_predicate): Ditto.
2043         (avx512f_cmp<mode>3): Ditto.
2044         (avx512f_vec_dup<mode>): Ditto.
2045         (avx512f_vec_dup_mem<mode>): Ditto.
2046         (avx512f_vpermi2var<mode>3): Ditto.
2047         (avx512f_vpermt2var<mode>3): Ditto.
2048         (vec_init<mode>): Ditto.
2049         (avx512f_gathersi<mode>): Ditto.
2050         (*avx512f_gathersi<mode>): Ditto.
2051         (*avx512f_gathersi<mode>_2): Ditto.
2052         (avx512f_gatherdi<mode>): Ditto.
2053         (*avx512f_gatherdi<mode>): Ditto.
2054         (*avx512f_gatherdi<mode>_2): Ditto.
2055         (avx512f_scattersi<mode>): Ditto.
2056         (*avx512f_scattersi<mode>): Ditto.
2057         (avx512f_scatterdi<mode>): Ditto.
2058         (*avx512f_scatterdi<mode>): Ditto.
2059         (sseintprefix): Extened with wider modes.
2060         (VEC_GATHER_IDXSI): Ditto.
2061         (VEC_GATHER_IDXDI): Ditto.
2062         (VEC_GATHER_SRCDI): Ditto.
2064 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
2065             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2067         * config/arm/t-aprofile: New file.
2068         * config.gcc: Handle --with-multilib-list option.
2070 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
2072         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
2073         the input for r2.
2075 2013-10-15  Richard Biener  <rguenther@suse.de>
2077         * tree-loop-distribution.c (build_empty_rdg): Inline into
2078         single user.
2079         (rdg_flag_vertex): Inline into single user.
2080         (rdg_flag_vertex_and_dependent): Likewise.
2081         (build_rdg_partition_for_vertex): Remove processed bitmap.
2082         (rdg_build_partitions): Simplify.
2084 2013-10-15  Richard Biener  <rguenther@suse.de>
2086         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
2087         Restructure forwarding through conversions and copies to
2088         avoid performing copy-propagation the wrong way.  Adjust
2089         recursion invocations.
2090         (forward_propagate_addr_expr): Add argument stating if we
2091         are recursing from a single-use.
2092         (ssa_forward_propagate_and_combine): Adjust.
2094 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
2096         * dumpfile.h (gcc::dump_manager): New class, to hold state
2097         relating to dumpfile management.
2098         (get_dump_file_name): Remove in favor of method of dump_manager.
2099         (dump_initialized_p): Likewise.
2100         (dump_start): Likewise.
2101         (dump_finish): Likewise.
2102         (dump_switch_p): Likewise.
2103         (dump_register): Likewise.
2104         (get_dump_file_info): Likewise.
2105         * context.c (gcc::context::context): Construct the dump_manager
2106         instance.
2107         * context.h (gcc::context::get_dumps): New.
2108         (gcc::context::m_dumps): New.
2109         * coverage.c (coverage_init): Port to dump_manager API.
2110         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
2111         (extra_dump_files_in_use): Likewise.
2112         (extra_dump_files_alloced): Likewise.
2113         (gcc::dump_manager::dump_manager): New.
2114         (dump_register): Convert to...
2115         (gcc::dump_manager::dump_register): ...method, replacing
2116         function-static next_dump with m_next_dump field.
2117         (get_dump_file_info): Convert to...
2118         (gcc::dump_manager::get_dump_file_info): ...method.
2119         (get_dump_file_name): Convert to...
2120         (gcc::dump_manager::get_dump_file_name): ...method.
2121         (dump_start): Convert to...
2122         (gcc::dump_manager::dump_start): ...method.
2123         (dump_finish): Convert to...
2124         (gcc::dump_manager::dump_finish): ...method.
2125         (dump_begin): Replace body with...
2126         (gcc::dump_manager::dump_begin): ...new method.
2127         (dump_phase_enabled_p): Convert to...
2128         (gcc::dump_manager::dump_phase_enabled_p): ...method.
2129         (dump_phase_enabled_p): Convert to...
2130         (gcc::dump_manager::dump_phase_enabled_p): ...method.
2131         (dump_initialized_p):  Convert to...
2132         (gcc::dump_manager::dump_initialized_p): ...method.
2133         (dump_flag_name): Replace body with...
2134         (gcc::dump_manager::dump_flag_name): ...new method.
2135         (dump_enable_all): Convert to...
2136         (gcc::dump_manager::dump_enable_all): ...new method.
2137         (opt_info_enable_passes): Convert to...
2138         (gcc::dump_manager::opt_info_enable_passes): ...new method.
2139         (dump_switch_p_1): Convert to...
2140         (gcc::dump_manager::dump_switch_p_1): ...new method.
2141         (dump_switch_p):  Convert to...
2142         (gcc::dump_manager::dump_switch_p): ...new method.
2143         (opt_info_switch_p): Port to dump_manager API.
2144         (enable_rtl_dump_file): Likewise.
2145         * opts-global.c (handle_common_deferred_options): Port to new
2146         dump_manager API.
2147         * passes.c (pass_manager::finish_optimization_passes): Likewise.
2148         (pass_manager::register_one_dump_file): Likewise.
2149         (pass_manager::register_pass): Likewise.
2150         (pass_init_dump_file): Likewise.
2151         (pass_fini_dump_file): Likewise.
2152         * statistics.c (statistics_early_init): Likewise.
2154 2013-10-14  Richard Biener  <rguenther@suse.de>
2156         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
2157         iterative_hash_canonical_type, gimple_canonical_type_hash,
2158         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
2159         gimple_register_canonical_type, print_gimple_types_stats,
2160         free_gimple_type_tables): Move to lto/lto.c
2161         (gt-gimple.h): Do not include.
2162         * gimple.h (gimple_register_canonical_type,
2163         print_gimple_types_stats, free_gimple_type_tables): Remove.
2164         * Makefile.in (GTFILES): Remove gimple.c.
2166 2013-10-14  Travis Snoozy  <quandary@remstate.com>
2168         PR target/58716
2169         * config/msp430/msp430.c (msp430_option_override): Correct thinko
2170         scanning for msp430x targets.
2172 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
2174         PR bootstrap/58509
2175         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
2176         (registers_ok_for_ldd_peep): Move around.
2177         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
2178         * config/sparc/sparc.md (widening peepholes): Use it.
2180 2013-10-14  Richard Biener  <rguenther@suse.de>
2182         PR middle-end/58712
2183         PR middle-end/55358
2184         * gimple.c (iterative_hash_canonical_type): Make sure to
2185         record the hash into the correct hashtable slot.
2187 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
2189         PR rtl-optimization/58662
2190         * combine.c (try_combine): Take into account death nodes on I2 when
2191         splitting a PARALLEL of two independent SETs.  Fix dump message.
2193 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
2195         PR target/51244
2196         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
2197         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
2198         * config/sh/t-sh (sh_treg_combine.o): New entry.
2199         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
2200         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
2201         (register_sh_passes): New function.  Register sh_treg_combine pass.
2202         (sh_option_override): Invoke it.
2203         (sh_canonicalize_comparison): Handle op0_preserve_value.
2204         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
2205         opportunities.  Canonicalize branch condition.
2206         (nott): Allow only if pseudos can be created for non-SH2A.
2208 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
2210         PR target/58690
2211         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
2212         (ix86_expand_movmem): Replace copy_addr_to_reg with
2213         ix86_copy_addr_to_reg.
2214         (ix86_expand_setmem): Likewise.
2216 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
2218         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
2219         provided by ix86_fp_compare_mode instead of CCFPUmode.
2221 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
2223         * config/aarch64/arm_neon.h
2224         (vtbx<1,3>_<psu>8): Fix register constriants.
2226 2013-10-11  Jeff Law  <law@redhat.com>
2228         PR tree-optimization/58640
2229         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
2230         threading paths that cross over two loop entry points.
2232 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2234         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
2235         handle vector float as well.
2236         (*vsx_le_perm_load_v4si): Likewise.
2237         (*vsx_le_perm_store_v2di): Likewise.
2238         (*vsx_le_perm_store_v4si): Likewise.
2240 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2242         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
2243         directly to circumvent subtract from splat{31} workaround.
2244         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
2245         prototype.
2246         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
2247         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
2248         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
2249         (altivec_vperm_<mode>): Convert to define_insn_and_split to
2250         separate big and little endian logic.
2251         (*altivec_vperm_<mode>_internal): New define_insn.
2252         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
2253         separate big and little endian logic.
2254         (*altivec_vperm_<mode>_uns_internal): New define_insn.
2255         (vec_permv16qi): Add little endian logic.
2257 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
2259         * doc/extend.texi (returns_nonnull): Remove arguments.
2261 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2262             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2263             Sergey Lega  <sergey.s.lega@intel.com>
2264             Anna Tikhonova  <anna.tikhonova@intel.com>
2265             Ilya Tocar  <ilya.tocar@intel.com>
2266             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2267             Ilya Verbin  <ilya.verbin@intel.com>
2268             Kirill Yukhin  <kirill.yukhin@intel.com>
2269             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2271         * config/i386/sse.md (VI48F_256_512): New.
2272         (avx2_permvar<mode>): Change to ...
2273         (<avx2_avx512f>_permvar<mode>): This.
2275 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2276             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2277             Sergey Lega  <sergey.s.lega@intel.com>
2278             Anna Tikhonova  <anna.tikhonova@intel.com>
2279             Ilya Tocar  <ilya.tocar@intel.com>
2280             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2281             Ilya Verbin  <ilya.verbin@intel.com>
2282             Kirill Yukhin  <kirill.yukhin@intel.com>
2283             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2285         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
2286         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
2287         * config/i386/sse.md (VI4_AVX): New.
2288         (sf2simodelower): Ditto.
2289         (sse2_cvtps2dq): Change to ...
2290         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
2292 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2293             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2294             Sergey Lega  <sergey.s.lega@intel.com>
2295             Anna Tikhonova  <anna.tikhonova@intel.com>
2296             Ilya Tocar  <ilya.tocar@intel.com>
2297             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2298             Ilya Verbin  <ilya.verbin@intel.com>
2299             Kirill Yukhin  <kirill.yukhin@intel.com>
2300             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2302         * config/i386/sse.md (V_512): New.
2303         (VI_512): Ditto.
2304         (vcond<V_512:mode><VF_512:mode>): Ditto.
2305         (vcond<V_512:mode><VI_512:mode>): Ditto.
2306         (vcondu<V_512:mode><VI_512:mode>): Ditto.
2308 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2309             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2310             Sergey Lega  <sergey.s.lega@intel.com>
2311             Anna Tikhonova  <anna.tikhonova@intel.com>
2312             Ilya Tocar  <ilya.tocar@intel.com>
2313             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2314             Ilya Verbin  <ilya.verbin@intel.com>
2315             Kirill Yukhin  <kirill.yukhin@intel.com>
2316             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2318         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
2319         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
2320         (FMAMODE): Ditto.
2321         (fma<mode>4): Removed condition.
2322         (fms<mode>4): Ditto.
2323         (fnma<mode>4): Ditto.
2324         (fnms<mode>4): Ditto.
2325         (fma4i_fmadd_<mode>): Ditto.
2326         (*fma_fmadd_<mode>): Ditto.
2327         (*fma_fmsub_<mode>): Ditto.
2328         (*fma_fnmadd_<mode>): Ditto.
2329         (*fma_fnmsub_<mode>): Ditto.
2330         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
2331         (*fma_fmaddsub_<mode>): Ditto.
2332         (*fma_fmsubadd_<mode>): Ditto.
2333         (*fmai_fmadd_<mode>): Ditto.
2334         (*fmai_fmsub_<mode>): Ditto.
2335         (*fmai_fnmadd_<mode>): Ditto.
2336         (*fmai_fnmsub_<mode>): Ditto.
2338 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2339             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2340             Sergey Lega  <sergey.s.lega@intel.com>
2341             Anna Tikhonova  <anna.tikhonova@intel.com>
2342             Ilya Tocar  <ilya.tocar@intel.com>
2343             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2344             Ilya Verbin  <ilya.verbin@intel.com>
2345             Kirill Yukhin  <kirill.yukhin@intel.com>
2346             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2348         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
2349         (VI124_256): Changed to ...
2350         (VI124_256_48_512): This.
2351         (ssepackmode): Extended with wider modes.
2352         (<code><mode>3): Changed iterator.
2353         (*avx2_<code><mode>3): Ditto.
2354         (vec_pack_trunc_<mode>): Ditto.
2356 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2357             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2358             Sergey Lega  <sergey.s.lega@intel.com>
2359             Anna Tikhonova  <anna.tikhonova@intel.com>
2360             Ilya Tocar  <ilya.tocar@intel.com>
2361             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2362             Ilya Verbin  <ilya.verbin@intel.com>
2363             Kirill Yukhin  <kirill.yukhin@intel.com>
2364             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2366         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
2367         (VI8F_256_512): Ditto.
2368         (abs<mode>2): Changed iterator.
2369         (avx2_perm<mode>): Changed to ...
2370         (<avx2_avx512f>_perm<mode>): This.
2371         (avx2_perm<mode>_1): Changed to ...
2372         (<avx2_avx512f>_perm<mode>_1): This.
2374 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2375             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2376             Sergey Lega  <sergey.s.lega@intel.com>
2377             Anna Tikhonova  <anna.tikhonova@intel.com>
2378             Ilya Tocar  <ilya.tocar@intel.com>
2379             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2380             Ilya Verbin  <ilya.verbin@intel.com>
2381             Kirill Yukhin  <kirill.yukhin@intel.com>
2382             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2384         * config/i386/sse.md (VI48_AVX512F): New.
2385         (VI48_AVX2): Changed to ...
2386         (VI48_AVX2_48_AVX512F): This.
2387         (avx2_ashrv<mode>): Changed to ...
2388         (<avx2_avx512f>_ashrv<mode>): This.
2389         (avx2_<shift_insn>v<mode>): Changed to ...
2390         (<avx2_avx512f>_<shift_insn>v<mode>): This.
2392 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2393             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2394             Sergey Lega  <sergey.s.lega@intel.com>
2395             Anna Tikhonova  <anna.tikhonova@intel.com>
2396             Ilya Tocar  <ilya.tocar@intel.com>
2397             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2398             Ilya Verbin  <ilya.verbin@intel.com>
2399             Kirill Yukhin  <kirill.yukhin@intel.com>
2400             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2402         * config/i386/sse.md (VI4_AVX512F): New.
2403         (VI8_AVX2_AVX512F): Ditto.
2404         (mul<mode>3): Extended with wider modes.
2405         (*<sse4_1_avx2>_mul<mode>3): Ditto.
2406         (mul<mode>3): Ditto.
2407         (vec_widen_<s>mult_odd_<mode>): Ditto.
2409 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2410             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2411             Sergey Lega  <sergey.s.lega@intel.com>
2412             Anna Tikhonova  <anna.tikhonova@intel.com>
2413             Ilya Tocar  <ilya.tocar@intel.com>
2414             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2415             Ilya Verbin  <ilya.verbin@intel.com>
2416             Kirill Yukhin  <kirill.yukhin@intel.com>
2417             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2419         * config/i386/sse.md (VI2_AVX512F): New.
2420         (VI124_AVX512F): Ditto.
2421         (sseunpackmode): Extended with wider modes.
2422         (sseunpackfltmode): Ditto.
2423         (vec_unpacks_float_hi_<mode>): Ditto.
2424         (vec_unpacks_float_lo_<mode>): Ditto.
2425         (vec_unpacku_float_hi_<mode>): Ditto.
2426         (vec_unpacku_float_lo_<mode>): Ditto.
2427         (vec_unpacks_lo_<mode>): Ditto.
2428         (vec_unpacks_hi_<mode>): Ditto.
2429         (vec_unpacku_lo_<mode>): Ditto.
2430         (vec_unpacku_hi_<mode>): Ditto.
2432 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2433             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2434             Sergey Lega  <sergey.s.lega@intel.com>
2435             Anna Tikhonova  <anna.tikhonova@intel.com>
2436             Ilya Tocar  <ilya.tocar@intel.com>
2437             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2438             Ilya Verbin  <ilya.verbin@intel.com>
2439             Kirill Yukhin  <kirill.yukhin@intel.com>
2440             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2442         * config/i386/i386.md (multdiv): New.
2443         (multdiv_mnemonic): Ditto.
2444         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
2445         (<sse>_vm<multdiv_mnemonic><mode>3): This.
2446         (<sse>_vmdiv<mode>3): Removed.
2448 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2449             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2450             Sergey Lega  <sergey.s.lega@intel.com>
2451             Anna Tikhonova  <anna.tikhonova@intel.com>
2452             Ilya Tocar  <ilya.tocar@intel.com>
2453             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2454             Ilya Verbin  <ilya.verbin@intel.com>
2455             Kirill Yukhin  <kirill.yukhin@intel.com>
2456             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2458         * config/i386/sse.md (V): Extended with wider modes.
2459         (VF2): Ditto.
2460         (ssehalfvecmode): Ditto.
2461         (i128): Ditto.
2462         (ssepackfltmode): Ditto.
2463         (avx_vec_concat<mode>): Ditto.
2464         (V_256_512): New iterator.
2465         (VF2_512_256): Ditto.
2466         (si2dfmode): New attribute.
2467         (si2dfmodelower): Ditto.
2468         (sf2dfmode): Ditto.
2469         (concat_tg_mode): Ditto.
2470         (floatv4siv4df2): Changed to ...
2471         (float<si2dfmodelower><mode>2): This.
2472         (avx_cvtps2pd256): Changed to ...
2473         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
2474         (vec_pack_trunc_v4df): Changed to ...
2475         (vec_pack_trunc_<mode>): This.
2476         (avx_vpermil<mode>): Changed to ...
2477         (<sse2_avx_avx512f>_vpermil<mode>): This.
2478         (<fixsuffix>fix_truncv8dfv8si2): New.
2479         (vec_pack_sfix_trunc_v8df): Ditto.
2480         (avx512f_rndscale<mode>): Ditto.
2481         (avx512f_roundpd512): Ditto.
2482         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2484 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2485             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2486             Sergey Lega  <sergey.s.lega@intel.com>
2487             Anna Tikhonova  <anna.tikhonova@intel.com>
2488             Ilya Tocar  <ilya.tocar@intel.com>
2489             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2490             Ilya Verbin  <ilya.verbin@intel.com>
2491             Kirill Yukhin  <kirill.yukhin@intel.com>
2492             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2494         * config/i386/i386.md (any_fix): New iterator.
2495         (fixsuffix): New attribute.
2496         * config/i386/sse.md (VF1): Extened with wider modes.
2497         (VI): Ditto.
2498         (VI_AVX2): Ditto.
2499         (VI8): Ditto.
2500         (sseintvecmodelower): Ditto.
2501         (ssescalarmode): Ditto.
2502         (ssescalarnum): Ditto.
2503         (VF1_128_256): New.
2504         (ssexmmmode): Ditto.
2505         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2506         (<sse>_rcp<mode>2): Change iterator.
2507         (rsqrt<mode>2): Ditto.
2508         (<sse>_rsqrt<mode>2): Ditto.
2509         (avx2_vec_dup<mode>): Ditto.
2510         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2511         (round<mode>2_sfix): Ditto.
2512         (avx2_pbroadcast<mode>): Ditto.
2513         (*andnot<mode>3): Handle XI mode.
2514         (*<code><mode>3): Ditto.
2515         (AVXTOSSEMODE): Removed.
2516         (avx_vpermil<mode>): Changed to ...
2517         (<sse2_avx_avx512f>_vpermil<mode>): This.
2519 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2520             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2521             Sergey Lega  <sergey.s.lega@intel.com>
2522             Anna Tikhonova  <anna.tikhonova@intel.com>
2523             Ilya Tocar  <ilya.tocar@intel.com>
2524             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2525             Ilya Verbin  <ilya.verbin@intel.com>
2526             Kirill Yukhin  <kirill.yukhin@intel.com>
2527             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2529         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2530         (<sse>_comi): Ditto.
2531         (<sse>_ucomi): Ditto.
2532         (sse_cvtss2siq_2): Ditto.
2533         (sse2_cvtsd2si): Ditto.
2534         (sse2_cvtsd2siq): Ditto.
2535         (sse2_cvttsd2si): Ditto.
2536         (sse2_cvttsd2siq): Ditto.
2537         (<shift_insn><mode>3): Ditto.
2538         (sse2_cvtsi2sdq): Update constraint and prefix.
2539         (sse_cvtsi2ss): Update prefix.
2540         (sse_cvtsi2ssq): Ditto.
2542 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2544         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2545         ignore internal calls.
2547 2013-10-11  Richard Biener  <rguenther@suse.de>
2549         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2550         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2551         TREE_OVERFLOW is set.
2553 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2555         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2557         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2558         * gimple.h: Likewise.
2560         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2562         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2563         macros.
2565         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2567         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2568         * configure: Regenerate.
2570 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2572         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2573         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2574         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2575         similarly pp_right_paren.
2576         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2577         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2578         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2579         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2580         clauses.
2581         (omp_declare_simd_clauses_equal,
2582         omp_remove_redundant_declare_simd_attrs): New functions.
2583         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2584         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2585         * tree.h (OMP_LOOP_CHECK): Define.
2586         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2587         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2588         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2589         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2590         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2591         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2592         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2593         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2594         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2595         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2596         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2597         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2598         (omp_remove_redundant_declare_simd_attrs): New prototype.
2599         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2600         GIMPLE_OMP_TEAMS): New codes.
2601         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2602         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2603         fields.
2604         (target_nesting_level): New variable.
2605         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2606         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2607         for collapse > 1 static schedule unless ordered.
2608         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2609         (determine_parallel_type): Adjust get_ws_args_for caller.
2610         (install_var_field): Handle mask & 4 for double indirection.
2611         (scan_sharing_clauses): Ignore shared clause on teams construct.
2612         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2613         (create_omp_child_function): If inside target or declare target
2614         constructs, set "omp declare target" attribute on the child function.
2615         (find_combined_for): New function.
2616         (scan_omp_parallel): Handle combined loops.
2617         (scan_omp_target, scan_omp_teams): New functions.
2618         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2619         restrictions and set ctx->cancellable for cancellable constructs.
2620         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2621         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2622         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2623         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2624         (omp_clause_aligned_alignment): New function.
2625         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2626         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2627         on teams constructs.  Handle user defined reductions and new
2628         OpenMP 4.0 clauses.
2629         (lower_reduction_clauses): Don't set placeholder to address of ref
2630         if it has already the right type.
2631         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2632         (expand_parallel_call): Use the new non-_start suffixed builtins,
2633         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2634         and GOMP_parallel_end after the call.
2635         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2636         (expand_omp_for_init_counts): Handle combined loops.
2637         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2638         loops.
2639         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2640         end of cancellable loops.
2641         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2642         Likewise.  Handle collapse > 1 loops.
2643         (expand_omp_simd): Handle combined loops.
2644         (expand_omp_for): Add inner_stmt argument, adjust callers of
2645         expand_omp_for* functions, use expand_omp_for_static*chunk even
2646         for collapse > 1 unless ordered.
2647         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2648         of cancellable sections.
2649         (expand_omp_single): Remove need_barrier variable, just rely on
2650         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2651         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2652         (expand_omp_atomic_load, expand_omp_atomic_store,
2653         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2654         (expand_omp_target): New function.
2655         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2656         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2657         (build_omp_regions_1): Immediately close region for
2658         GF_OMP_TARGET_KIND_UPDATE.
2659         (maybe_add_implicit_barrier_cancel): New function.
2660         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2661         cancellation.
2662         (lower_omp_single): Likewise.  Add clobber after the barrier.
2663         (lower_omp_taskgroup): New function.
2664         (lower_omp_for): Handle combined loops.  Adjust
2665         lower_rec_input_clauses caller.  Handle cancellation.
2666         (lower_depend_clauses): New function.
2667         (lower_omp_taskreg): Lower depend clauses.  Adjust
2668         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2669         cancellation.
2670         (lower_omp_target, lower_omp_teams): New functions.
2671         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2672         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2673         and GOMP_cancellation_point calls.
2674         (lower_omp): Fold stmts inside of target region.
2675         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2676         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2677         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2678         (BT_FN_VOID_OMPFN_PTR_UINT,
2679         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2680         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2681         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2682         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2683         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2684         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2685         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2686         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2687         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2688         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2689         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2690         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2691         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2692         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2693         BUILT_IN_GOMP_PARALLEL_END.
2694         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2695         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2696         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2697         GF_OMP_FOR_KIND_DISTRIBUTE.
2698         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2699         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2700         (dump_gimple_omp_return): Print lhs if it has any.
2701         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2702         gimple_omp_atomic_seq_cst_p.
2703         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2704         and GIMPLE_OMP_TEAMS.
2705         * langhooks.c (lhd_omp_mappable_type): New function.
2706         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2707         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2708         hook.
2709         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2710         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2711         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2712         (struct gimplify_omp_ctx): Add combined_loop field.
2713         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2714         on stmts inside of target region.
2715         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2716         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2717         ORT_TARGET and ORT_TARGET_DATA.
2718         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2719         Handle GOVD_MAP.
2720         (omp_notice_threadprivate_variable): Complain about threadprivate
2721         variables in target region.
2722         (omp_notice_variable): Complain about vars with non-mappable type
2723         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2724         (omp_check_private): Ignore ORT_TARGET* regions.
2725         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2726         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2727         (find_combined_omp_for): New function.
2728         (gimplify_omp_for): Handle gimplification of combined loops.
2729         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2730         OMP_TEAMS.
2731         (gimplify_omp_target_update): New function.
2732         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2733         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2734         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2735         (gimplify_body): If fndecl has "omp declare target" attribute, add
2736         implicit ORT_TARGET context around it.
2737         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2738         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2739         * tree-nested.c (convert_nonlocal_reference_stmt,
2740         convert_local_reference_stmt, convert_gimple_call): Handle
2741         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2742         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2743         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2744         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2745         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2746         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2747         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2748         BUILT_IN_GOMP_LOOP_END_CANCEL,
2749         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2750         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2751         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2752         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2753         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2754         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2755         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2756         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2757         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2758         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2759         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2760         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2761         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2762         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2763         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2764         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2765         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2766         gimple_build_omp_teams): New functions.
2767         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2768         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2769         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2770         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2771         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2772         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2773         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2774         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2775         GF_OMP_ATOMIC_SEQ_CST): New.
2776         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2777         gimple_build_omp_teams): New prototypes.
2778         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2779         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2780         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2781         GIMPLE_OMP_SINGLE as end of range.
2782         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2783         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2784         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2785         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2786         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2787         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2788         gimple_omp_target_kind, gimple_omp_target_set_kind,
2789         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2790         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2791         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2792         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2793         gimple_omp_teams_set_clauses): New inlines.
2794         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2795         and GIMPLE_OMP_TASKGROUP.
2796         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2797         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2798         enum omp_clause_proc_bind_kind): New.
2799         (union omp_clause_subcode): Add depend_kind, map_kind and
2800         proc_bind_kind fields.
2801         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2802         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2803         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2804         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2805         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2807 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2809         * predict.c (tree_estimate_probability): Add new parameter
2810         for estimate_bb_frequencies.
2811         (estimate_bb_frequencies): Add new parameter to force estimation.
2812         (rebuild_frequencies): When max frequency in function is small,
2813         recompute counts from frequencies.
2814         * predict.h (estimate_bb_frequencies): New parameter.
2816 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2818         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2819         optimizations are disabled.
2821 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2823         * coverage.c (coverage_finish): Fix leak of da_file_name.
2825 2013-10-10  Jan Hubicka  <jh@suse.cz>
2827         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2828         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2829         Bobcat and generic.
2831 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2833         PR middle-end/58670
2834         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2835         if any labels are in FALLTHRU_BB, use a special label emitted
2836         immediately after the asm goto insn rather than label_rtx
2837         of the LABEL_DECL.
2838         (expand_asm_stmt): Adjust caller.
2839         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2840         edge if the last insn in predecessor is a jump with single successor,
2841         but it isn't simplejump_p.
2843 2013-10-10  Richard Biener  <rguenther@suse.de>
2845         PR tree-optimization/58656
2846         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2848 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2850         * gimplify.c: Include expr.h and tm_p.h for targets with special
2851         va-arg padding requirements.
2853 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2855         * tree-flow.h: Move some prototypes to gimple.h.
2856         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2857         * gimple.h: Relocate some prototypes from tree-flow.h
2858         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2859         Move to gimplify.c.
2860         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2861         (build_va_arg_indirect_ref): Relocate and make static.
2862         (std_gimplify_va_arg_expr): Relocate here.
2863         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2864         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2866 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2868         * doc/md.texi: Document the mnemonic attribute.
2870 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2872         PR target/57377
2873         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2874         (set_attr_alternative x ...) when searching for user defined
2875         mnemonic attribute.
2877 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2879         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2881 2013-10-09  Easwaran Raman <eraman@google.com>
2883         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2884         * cfgexpand.c (defer_stack_allocation): ...use here
2885         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2887 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2889         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2890         (operand_equal_for_value_replacement): New function, extracted from
2891         value_replacement and enhanced to catch more cases.
2892         (value_replacement): Use operand_equal_for_value_replacement.
2894 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2896         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2897         get_max_loop_iterations.
2899 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2901         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2902         Place comment above function.
2904 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2906         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2907         are not related to tree-cfg.c.
2908         * tree-ssa-address.h: New file.  Relocate prototypes.
2909         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2910         (addr_for_mem_ref): New.  Combine call to get_address_description and
2911         return addr_for_mem_ref.
2912         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2913         * tree-ssa-live.h: Adjust prototypes.
2914         * passes.c: Include tree-ssa-live.h.
2915         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2916         * graphite.c (graphite_transform_loops): Make static.
2917         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2918         make_pass_graphite, pass_data_graphite_transforms,
2919         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2920         * ipa-pure-const.c (warn_function_noreturn): Make static.
2921         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2922         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2923         Relocate from tree-cfg.c
2924         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2925         static.
2926         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2927         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2928         Move to ipa-pure-const.c.
2929         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2930         Relocate from tree-optimize.c.
2931         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2932         make_pass_fixup_cfg): Move to tree-cfg.c.
2933         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2934         Relocate some prototypes.
2935         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2936         * tree-dump.c (dump_function_to_file): Remove prototype.
2937         Add tree-flow.h to the include file.
2938         * tree-dump.h: Remove prototype.
2939         * tree-parloops.h: New File.  Add prototypes.
2940         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2941         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2942         from tree-ssa-loop.c.
2943         * tree-predcom.c (run_tree_predictive_commoning,
2944         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2945         Relocate here from tree-ssa-loop.c.
2946         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2947         ssa_name_values.release ().
2948         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2949         (ssa_name_values): Relocate from tree-flow.h.
2950         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2951         * tree-ssa-loop.c (run_tree_predictive_commoning,
2952         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2953         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2954         make_pass_graphite, pass_data_graphite_transforms,
2955         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2956         tree_parallelize_loops, pass_data_parallelize_loops,
2957         make_pass_parallelize_loops): Move to other files.
2958         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2959         moved here.
2960         * tree.h: Remove prototypes from tree-address.c.
2962 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2964         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2965         (struct int_tree_map): Move to tree-hasher.h
2966         (SCALE, LABEL, PERCENT): Move to gimple.h
2967         * tree-flow-inline.h: Delete.  Move functions to other files.
2968         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2969         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2970         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2971         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2972         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2973         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2974         * tree-hasher.h: Don't include tree-flow.h.
2975         (struct int_tree_map): Relocate from tree-flow.h.
2976         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2977         tree-flow-inline.h and make static.
2978         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2979         tree-flow-inline.h.
2980         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2981         tree-flow-inline.h and make static.
2982         * tree.h (is_global_var, may_be_aliased): Relocate from
2983         tree-flow-inline.h.
2984         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2985         * value-prof.c: No longer include tree-flow-inline.h.
2986         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2988 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2990         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2991         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2992         (enum move_pos): Move to tree-ssa-loop-im.h
2993         * cfgloop.h: Move some prototypes.
2994         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2995         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2996         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2997         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2998         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2999         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
3000         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
3001         (enum move_pos): Relocate here.
3002         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
3003         tree-ssa-loop.c.
3004         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
3005         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
3006         make_pass_lim): Relocate here from tree-ssa-loop.c.
3007         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
3008         tree-ssa-loop.c.
3009         (loop_edge_to_cancel, unloop_loops): Make static.
3010         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
3011         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
3012         (tree_complete_unroll, gate_tree_complete_unroll,
3013         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
3014         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
3015         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
3016         * tree-ssa-loop-ivopts.c: Remove local prototypes.
3017         (stmt_invariant_in_loop_p): Remove unused function.
3018         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
3019         * tree-ssa-loop-manip.h: New file.  Add prototypes.
3020         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
3021         (gcov_type_to_double_int): Move to cfgloop.h.
3022         (double_int_cmp, bound_index,
3023         estimate_numbers_of_iterations_loop): Make static.
3024         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
3025         (max_loop_iterations): Factor out get_max_loop_iterations.
3026         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
3027         cfgloop.c.
3028         * tree-ssa-loop-niter.h: New file.  Add prototypes.
3029         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
3030         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
3031         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
3032         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
3033         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
3034         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
3035         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
3036         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
3037         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
3038         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
3039         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
3040         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
3041         pass_data_complete_unroll, make_pass_complete_unroll,
3042         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
3043         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
3044         tree-ssa-loop-ivcanon.c.
3045         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
3046         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
3047         tree-ssa-loop-prefetch.c.
3048         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
3049         tree-ssa-loop-im.c.
3050         (get_lsm_tmp_name): Relocate and add suffix parameter.
3051         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
3052         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
3053         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
3054         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
3055         (get_estimated_loop_iterations): Factor out accessor from
3056         estimated_loop_iterations in tree-ssa-loop-niter.c.
3057         (get_max_loop_iterations): Factor out accessor from
3058         _max_loop_iterations in tree-ssa-niter.c.
3059         * loop-unroll.c (decide_unroll_constant_iterations,
3060         decide_unroll_runtime_iterations, decide_peel_simple,
3061         decide_unroll_stupid): Use new get_* accessors.
3063 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
3065         PR tree-optimization/20318
3066         * doc/extend.texi (returns_nonnull): New function attribute.
3067         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
3068         attribute.
3069         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
3070         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
3072 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
3074         PR middle-end/58570
3075         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
3076         false if both components are bitfields.
3078 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3080         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
3081         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
3082         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
3083         * config/aarch64/aarch64.h
3084         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
3085         * config/aarch64/aarch64-simd-builtins.def
3086         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
3088 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3090         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
3092 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3094         * config/aarch64/arm_neon.h (vdiv_f64): Added.
3096 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3098         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
3099         (vneg_f64): New intrinsic.
3100         (vneg_s8): Asm replaced with C.
3101         (vneg_s16): Likewise.
3102         (vneg_s32): Likewise.
3103         (vneg_s64): New intrinsic.
3104         (vnegq_f32): Asm replaced with C.
3105         (vnegq_f64): Likewise.
3106         (vnegq_s8): Likewise.
3107         (vnegq_s16): Likewise.
3108         (vnegq_s32): Likewise.
3109         (vnegq_s64): Likewise.
3111 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
3113         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
3115 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3117         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
3118         packed stack special handling.
3119         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
3120         back to fixed stack slots for FPRs saved due to stdarg.
3122 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3124         * config/s390/s390.c (s390_frame_info): Restructure function.
3126 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3128         * config/s390/s390.c (struct s390_frame_layout): New field
3129         gpr_save_slots.
3130         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
3131         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
3132         regs_ever_clobbered to char*.
3133         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
3134         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
3135         (s390_register_info): Enable FPR save slots.  Move/Copy some
3136         functionality into ...
3137         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
3138         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
3139         function.
3140         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
3141         not imply a stack frame.
3142         (s390_init_frame_layout): Remove variable clobbered_regs.
3143         (s390_update_register_info): Remove function.
3144         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
3145         cannot be used for register renaming.
3146         (s390_hard_regno_scratch_ok): New function.
3147         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
3148         (s390_initial_elimination_offset): Change offset calculation of
3149         the return address pointer.
3150         (save_gprs): Deal with only r6 being saved from the call-saved regs.
3151         (restore_gprs): Set frame related flag.
3152         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
3153         (s390_emit_prologue): Call s390_register_info instead of
3154         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
3155         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
3156         (s390_optimize_prologue): Call s390_optimize_register_info.
3157         Try to remove also FPR slot save/restore INSNs.  Remove frame
3158         related flags from restore INSNs.
3160 2013-10-08  DJ Delorie  <dj@redhat.com>
3162         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
3163         (movhi): Likewise.
3165         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
3166         avoid conflict with the MI use of %c.
3167         * config/rl78/rl78-real.md: change %c to %C throughout.
3168         * config/rl78/rl78-virt.md: Likewise.
3170 2013-10-08  Jan Hubicka  <jh@suse.cz>
3172         * config/i386/i386.c (ix86_option_override_internal): Switch
3173         to SSE math for -ffast-math when target ISA supports SSE2.
3175 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
3177         * tree-flow.h: Remove some prototypes.
3178         * tree.h: Remove some protypes, add a couple.
3179         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
3180         using_eh_for_cleanups_p): Add interface routines for front ends.
3181         * tree-eh.h: New file.  Add protoptyes.
3182         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
3183         (add_stmt_to_eh_lp_fn): Make static.
3184         (lower_try_finally): Use new using_eh_for_cleanups_p.
3185         * emit-rtl.c: Include tree-eh.h.
3186         * gimple.h: Include tree-eh.h.
3188 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
3190         PR tree-optimization/58480
3191         * tree-vrp.c (infer_nonnull_range): New function.
3192         (infer_value_range): Call infer_nonnull_range.
3194 2013-10-08  Dehao Chen  <dehao@google.com>
3196         PR tree-optimization/58619
3197         * tree-inline.c (copy_phis_for_bb): Combine location data
3198         only if non-null.
3200 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3202         PR target/58423
3203         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
3204         RTX_FRAME_RELATED_P on INSN.
3206 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3208         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
3209         (altivec_expand_vec_perm_const): Call it.
3211 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3213         * config/rs6000/vector.md (mov<mode>): Emit permuted move
3214         sequences for LE VSX loads and stores at expand time.
3215         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
3216         prototype.
3217         * config/rs6000/rs6000.c (rs6000_const_vec): New.
3218         (rs6000_gen_le_vsx_permute): New.
3219         (rs6000_gen_le_vsx_load): New.
3220         (rs6000_gen_le_vsx_store): New.
3221         (rs6000_gen_le_vsx_move): New.
3222         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
3223         (*vsx_le_perm_load_v4si): New.
3224         (*vsx_le_perm_load_v8hi): New.
3225         (*vsx_le_perm_load_v16qi): New.
3226         (*vsx_le_perm_store_v2di): New.
3227         (*vsx_le_perm_store_v4si): New.
3228         (*vsx_le_perm_store_v8hi): New.
3229         (*vsx_le_perm_store_v16qi): New.
3230         (*vsx_xxpermdi2_le_<mode>): New.
3231         (*vsx_xxpermdi4_le_<mode>): New.
3232         (*vsx_xxpermdi8_le_V8HI): New.
3233         (*vsx_xxpermdi16_le_V16QI): New.
3234         (*vsx_lxvd2x2_le_<mode>): New.
3235         (*vsx_lxvd2x4_le_<mode>): New.
3236         (*vsx_lxvd2x8_le_V8HI): New.
3237         (*vsx_lxvd2x16_le_V16QI): New.
3238         (*vsx_stxvd2x2_le_<mode>): New.
3239         (*vsx_stxvd2x4_le_<mode>): New.
3240         (*vsx_stxvd2x8_le_V8HI): New.
3241         (*vsx_stxvd2x16_le_V16QI): New.
3243 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
3245         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
3247 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3249         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
3250         loop to work also for 31bit ABI.
3251         Save the stack pointer for frame_size > 0.
3253 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3255         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
3256         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
3257         constraint letters from expanders.
3258         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
3259         retry count to general_operand.
3260         ("tabort"): Give operand 0 a mode.
3261         ("tabort_1"): Add mode and constraint letter for operand 0.
3262         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
3264 2013-10-04  Jeff Law  <law@redhat.com>
3266         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
3268         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
3269         out of ...
3270         (thread_across_edge): Here.  Call it.
3272 2013-10-04  Cary Coutant  <ccoutant@google.com>
3274         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
3275         discarding a location list expression (or a piece of one).
3277 2013-10-03  Jan Hubicka  <jh@suse.cz>
3279         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
3280         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
3281         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
3283 2013-10-03  Jan Hubicka  <jh@suse.cz>
3285         * config/i386/i386.c (ix86_option_override_internal): Do not enable
3286         accumulate-outgoing-args when producing unwind info.
3288 2013-10-03  Wei Mi  <wmi@google.com>
3290         * lra-constraints.c (insert_move_for_subreg): New function
3291         extracted from simplify_operand_subreg.
3292         (simplify_operand_subreg): Add reload for paradoxical subreg.
3294 2013-10-03  Rong Xu  <xur@google.com>
3296         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
3297         the candidate of builtin_expect such that we should fix the
3298         size/time estimation.
3299         (estimate_function_body_sizes): Do the acutally size/time fix-up
3300         for builtin_expect.
3302 2013-10-03  Rong Xu  <xur@google.com>
3304         * predict.c (tree_predict_by_opcode): Get the probability
3305         for builtin_expect from param builtin_expect_probability.
3306         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
3307         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
3308         * doc/invoke.texi: Add documentation for builtin-expect-probability.
3310 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
3312         PR c++/19476
3313         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
3314         * calls.c (alloca_call_p): Use get_callee_fndecl.
3315         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
3316         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
3317         Likewise.
3318         (vrp_visit_stmt): Remove duplicated code.
3320 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
3322         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
3323         ceildf2, btruncdf2, instead of vsx_* name.
3325         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
3326         iterators to only do V2DF and V4SF here.  Move the DF code to
3327         rs6000.md where it is combined with SF mode.  Replace <VSv> with
3328         just 'v' since only vector operations are handled with these insns
3329         after moving the DF support to rs6000.md.
3330         (vsx_sub<mode>3): Likewise.
3331         (vsx_mul<mode>3): Likewise.
3332         (vsx_div<mode>3): Likewise.
3333         (vsx_fre<mode>2): Likewise.
3334         (vsx_neg<mode>2): Likewise.
3335         (vsx_abs<mode>2): Likewise.
3336         (vsx_nabs<mode>2): Likewise.
3337         (vsx_smax<mode>3): Likewise.
3338         (vsx_smin<mode>3): Likewise.
3339         (vsx_sqrt<mode>2): Likewise.
3340         (vsx_rsqrte<mode>2): Likewise.
3341         (vsx_fms<mode>4): Likewise.
3342         (vsx_nfma<mode>4): Likewise.
3343         (vsx_copysign<mode>3): Likewise.
3344         (vsx_btrunc<mode>2): Likewise.
3345         (vsx_floor<mode>2): Likewise.
3346         (vsx_ceil<mode>2): Likewise.
3347         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
3348         (vsx_sminsf3): Likewise.
3349         (vsx_fmadf4): Likewise.
3350         (vsx_fmsdf4): Likewise.
3351         (vsx_nfmadf4): Likewise.
3352         (vsx_nfmsdf4): Likewise.
3353         (vsx_cmpdf_internal1): Likewise.
3355         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
3356         simpler to select whether a target has SPE or traditional floating
3357         point support in iterators.
3358         (TARGET_DF_SPE): Likewise.
3359         (TARGET_SF_FPR): Likewise.
3360         (TARGET_DF_FPR): Likewise.
3361         (TARGET_SF_INSN): Macros to say whether floating point support
3362         exists for a given operation for expanders.
3363         (TARGET_DF_INSN): Likewise.
3365         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
3366         combining of SF/DF mode operations, using both traditional and VSX
3367         registers.
3368         (Fvsx): Likewise.
3369         (Ff): Likewise.
3370         (Fv): Likewise.
3371         (Fs): Likewise.
3372         (Ffre): Likewise.
3373         (FFRE): Likewise.
3374         (abs<mode>2): Combine SF/DF modes using traditional floating point
3375         instructions.  Add support for using the upper DF registers with
3376         VSX support, and SF registers with power8-vector support.  Update
3377         expanders for operations supported by both the SPE and traditional
3378         floating point units.
3379         (abs<mode>2_fpr): Likewise.
3380         (nabs<mode>2): Likewise.
3381         (nabs<mode>2_fpr): Likewise.
3382         (neg<mode>2): Likewise.
3383         (neg<mode>2_fpr): Likewise.
3384         (add<mode>3): Likewise.
3385         (add<mode>3_fpr): Likewise.
3386         (sub<mode>3): Likewise.
3387         (sub<mode>3_fpr): Likewise.
3388         (mul<mode>3): Likewise.
3389         (mul<mode>3_fpr): Likewise.
3390         (div<mode>3): Likewise.
3391         (div<mode>3_fpr): Likewise.
3392         (sqrt<mode>3): Likewise.
3393         (sqrt<mode>3_fpr): Likewise.
3394         (fre<Fs>): Likewise.
3395         (rsqrt<mode>2): Likewise.
3396         (cmp<mode>_fpr): Likewise.
3397         (smax<mode>3): Likewise.
3398         (smin<mode>3): Likewise.
3399         (smax<mode>3_vsx): Likewise.
3400         (smin<mode>3_vsx): Likewise.
3401         (negsf2): Delete SF operations that are merged with DF.
3402         (abssf2): Likewise.
3403         (addsf3): Likewise.
3404         (subsf3): Likewise.
3405         (mulsf3): Likewise.
3406         (divsf3): Likewise.
3407         (fres): Likewise.
3408         (fmasf4_fpr): Likewise.
3409         (fmssf4_fpr): Likewise.
3410         (nfmasf4_fpr): Likewise.
3411         (nfmssf4_fpr): Likewise.
3412         (sqrtsf2): Likewise.
3413         (rsqrtsf_internal1): Likewise.
3414         (smaxsf3): Likewise.
3415         (sminsf3): Likewise.
3416         (cmpsf_internal1): Likewise.
3417         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
3418         (negdf2): Delete DF operations that are merged with SF.
3419         (absdf2): Likewise.
3420         (nabsdf2): Likewise.
3421         (adddf3): Likewise.
3422         (subdf3): Likewise.
3423         (muldf3): Likewise.
3424         (divdf3): Likewise.
3425         (fred): Likewise.
3426         (rsqrtdf_internal1): Likewise.
3427         (fmadf4_fpr): Likewise.
3428         (fmsdf4_fpr): Likewise.
3429         (nfmadf4_fpr): Likewise.
3430         (nfmsdf4_fpr): Likewise.
3431         (sqrtdf2): Likewise.
3432         (smaxdf3): Likewise.
3433         (smindf3): Likewise.
3434         (cmpdf_internal1): Likewise.
3435         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
3436         (btrunc<mode>2): Delete separate expander, and combine with the
3437         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
3438         (btrunc<mode>2_fpr): Likewise.
3439         (ceil<mode>2): Likewise.
3440         (ceil<mode>2_fpr): Likewise.
3441         (floor<mode>2): Likewise.
3442         (floor<mode>2_fpr): Likewise.
3443         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
3444         Add support for using the upper registers with VSX and
3445         power8-vector.  Move insns to be closer to the define_expands. On
3446         VSX systems, prefer the traditional form of FMA over the VSX
3447         version, since the traditional form allows the target not to
3448         overlap with the inputs.
3449         (fms<mode>4_fpr): Likewise.
3450         (nfma<mode>4_fpr): Likewise.
3451         (nfms<mode>4_fpr): Likewise.
3453 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3454             Richard Earnshaw  <richard.earnshaw@arm.com>
3456         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
3457         (struct mult_cost_table): Likewise.
3458         (struct mem_cost_table): Likewise.
3459         (struct fp_cost_table): Likewise.
3460         (struct vector_cost_table): Likewise.
3461         (cpu_cost_table): Likewise.
3462         * config/arm/arm.opt (mold-rts-costs): New option.
3463         (mnew-generic-costs): Likewise.
3464         * config/arm/arm.c (generic_extra_costs): New table.
3465         (cortexa15_extra_costs): Likewise.
3466         (arm_slowmul_tune): Use NULL as new costs.
3467         (arm_fastmul_tune): Likewise.
3468         (arm_strongarm_tune): Likewise.
3469         (arm_xscale_tune): Likewise.
3470         (arm_9e_tune): Likewise.
3471         (arm_v6t2_tune): Likewise.
3472         (arm_cortex_a5_tune): Likewise.
3473         (arm_cortex_a9_tune): Likewise.
3474         (arm_v6m_tune): Likewise.
3475         (arm_fa726te_tune): Likewise.
3476         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3477         (arm_cortex_tune): Use generict_extra_costs.
3478         (shifter_op_p): New function.
3479         (arm_unspec_cost): Likewise.
3480         (LIBCALL_COST): Define.
3481         (arm_new_rtx_costs): New function.
3482         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3483         table is available. Use old costs otherwise unless mnew-generic-costs
3484         is specified.
3485         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3486         (cpu_cost_table): Declare.
3488 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3490         PR target/58460
3491         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3492         (*subs_mul_imm_<mode>)
3493         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3494         (*sub_<shift>_<mode>)
3495         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3496         Remove k constraint.
3498 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3500         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3501         code.
3502         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3504 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3506         * predict.c (probably_never_executed): New function.
3507         (probably_never_executed_bb_p): Invoke probably_never_executed.
3508         (probably_never_executed_edge_p): Ditto.
3509         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3510         Treat profile insanities conservatively.
3512 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3514         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3516 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3518         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3519         use it.  Use smaller increase for scratch.  Don't increase reject
3520         for early clobber scratch.
3521         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3522         setting eliminated regs except setting fp from hfp.
3523         (lra_eliminate): Check lra_insn_recog_data on NULL.
3525 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3527         PR target/58587
3528         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3529         setting -mvsx-timode by default until the underlying problem is fixed.
3530         (RS6000_CPU, power7 defaults): Likewise.
3532 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3534         * config/x-linux (host-linux.o): Remove header dependencies.
3535         Use $(COMPILE) and $(POSTCOMPILE).
3536         * config/t-linux-android (linux-android.o): Ditto.
3538 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3540         * Makefile.in (expmed.o-warn): Remove.
3542 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3544         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3545         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3547 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3549         * dojump.c (do_jump_1): Divide probability between
3550         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3552 2013-10-02  Tom Tromey  <tromey@redhat.com>
3554         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3556 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3558         * tree-flow.h: Remove some prototypes.
3559         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3560         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3561         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3562         mark_virtual_phi_result_for_renaming): Relocate here.
3563         * tree-into-ssa.h: Add prototypes.
3564         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3565         single_pred_before_succ_order.
3566         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3567         (nonfreeing_call_p) Move to gimple.c.
3568         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3569         tree-ssa-phiopt.c.
3570         * basic-block.h (single_pred_before_succ_order): Add prototype.
3571         * gimple.c (nonfreeing_call_p): Relocate here.
3572         * gimple.h: Add prototype.
3573         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3574         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3575         * tree-ssa.h: Include tree-ssa-dom.h.
3577 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3579         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3580         Use $(COMPILE) and $(POSTCOMPILE).
3582         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3584 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3586         * tree-flow.h: Remove some prototypes.
3587         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3588         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3589         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3590         propagate_tree_value*): Move from here to...
3591         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3592         propagate_tree_value*): Relocate here.
3593         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3594         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3595         * gimple-fold.c: Remove gimple-fold.h from include list.
3596         * tree-vrp.c: Remove gimple-fold.h from include list.
3597         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3598         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3599         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3600         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3601         * sese.c: Add tree-ssa-propagate.h to include list.
3603 2013-10-02  Richard Biener  <rguenther@suse.de>
3605         * tree-loop-distribution.c: Include tree-vectorizer.h for
3606         find_loop_location.
3607         (enum partition_kind): Remove PKIND_REDUCTION.
3608         (struct partition_s): Remove has_writes member, add reduction_p member.
3609         (partition_alloc): Adjust.
3610         (partition_builtin_p): Likewise.
3611         (partition_has_writes): Remove.
3612         (partition_reduction_p): New function.
3613         (partition_merge_into): Likewise.
3614         (generate_code_for_partition): Commonize builtin partition
3615         handling tail.
3616         (rdg_cannot_recompute_vertex_p): Remove.
3617         (already_processed_vertex_p): Likewise.
3618         (rdg_flag_vertex): Do not set has_writes.
3619         (classify_partition): Adjust.
3620         (rdg_build_partitions): Do not set has_writes, treat all
3621         partitions as useful.
3622         (distribute_loop): Record number of library calls generated.  Adjust.
3623         (tree_loop_distribution): Report number of loops and library
3624         calls generated as opt-info.
3626 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3628         * tree-flow.h: Include new .h files.  Move prototypes.
3629         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3630         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3631         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3632         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3633         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3634         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3635         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3636         * tree.h (get_ref_base_and_extent): Move prototype out.
3637         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3638         tree-dfa.h.
3639         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3640         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3641         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3642         * tree-scalar-evolution.c: Include tree.h.
3643         * sese.c: Include tree.h.
3644         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3645         * dwarf2out.c: Include tree-dfa.h.
3646         * tree-chrec.c: Include tree.h.
3647         * tree-data-ref.c: Include tree.h.
3649 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3651         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3652         Fix whitespace.
3654 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3656         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3657         Use $(COMPILE) and $(POSTCOMPILE).
3658         (sol2-cxx.o): Likewise.
3659         (sol2-stubs.o): Likewise.
3660         (sol2.o): Likewise.
3661         * config/x-solaris (host-solaris.o): Likewise.
3663         * config/sparc/t-sparc (sparc.o): Remove.
3664         (sparc-c.o): Remove header dependencies.
3665         Use $(COMPILE) and $(POSTCOMPILE).
3666         * config/sparc/x-sparc: Likewise.
3668 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3670         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3671         * config/arc/arc700.md: Likewise.
3672         * config/arc/arc-modes.def: Likewise.
3673         * config/arc/arc-simd.h: Likewise.
3674         * config/arc/t-arc-uClibc: Likewise.
3675         * config/arc/t-arc-newlib: Likewise.
3677 2013-10-02  Renlin Li  <renlin.li@arm.com>
3679         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3680         plus_constant.
3681         (aarch64_expand_epilogue): Likewise.
3683 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3684             Yufeng Zhang  <yufeng.zhang@arm.com>
3686         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3687         declaration.
3688         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3689         'base_in' represent a conversion and legal_cast_p_1 holds; set
3690         'base_in' with the returned value from get_unwidened.
3692 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3694         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3695         plus_constant is not used.
3697 2013-10-01  Wei Mi  <wmi@google.com>
3699         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3700         * config/i386/i386.md: Add define_peephole2 to
3701         break partial reg stall for cvtss2sd/cvtsd2ss.
3703 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3705         * config/arc/arc.c (pass_arc_ifcvt::clone):
3706         Update for ctxt_ -> m_ctxt change.
3708 2013-10-01  Jeff Law  <law@redhat.com>
3710         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3711         outgoing_edge and intermediate_edge fields.  Instead store the path.
3712         (redirection_data::hash): Hash on the last edge's destination index.
3713         (redirection_data::equal): Check the entire thread path.
3714         (lookup_redirectio_data): Corresponding changes.
3715         (create_edge_and_update_destination_phis): Likewise.
3716         (thread_single_edge): Likewise.
3718 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3719             Diego Novillo <dnovillo@google.com>
3721         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3722         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3723         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3724         versions of these patterns.
3726         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3727         (__builtin_arc_aligned): Likewise.
3729         * config/arc/arc.md: Expand adc_0 comment stating the intended
3730         purpose and why it isn't ready.
3731         Replace commented out call_value_via_label_mixed with a
3732         plain comment about bl_s.
3734         * config/arc/arc.c (stdio.h): Don't include directly.
3735         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3736         Remove if (1) condition.
3737         (arc_encode_section_info): Fix comment.
3739 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3741         * config/arc/arc.c (arc_conditional_register_usage):
3742         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3743         Also set reg_alloc_order for DMA config regs.
3745 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3746             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3748         * doc/install.texi (--with-cpu): Mention ARC.
3749         (arc-*-elf32): New paragraph.
3750         (arc-linux-uclibc): Likewise.
3751         * doc/md.texi (Machine Constraints): Add ARC part.
3752         * doc/invoke.texi: (menu): Add ARC Options.
3753         (Machine Dependent Options) <ARC Options>: Add synopsis.
3754         (node ARC Options): Add.
3755         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3756         (ARC Built-in Functions): New section defining
3757         generic ARC built-in functions.
3758         (ARC SIMD Built-in Functions): New section defining SIMD specific
3759         built-in functions.
3760         (Declaring Attributes of Functions): Extended
3761         description of short_call and long_call attributes for ARC and
3762         added index entries.
3764 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3765             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3766             Joern Rennecke  <joern.rennecke@embecosm.com>
3767             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3768             Brendan Kehoe  <brendan@zen.org>
3769             Michael Eager  <eager@eagercon.com>
3770             Simon Cook  <simon.cook@embecosm.com>
3771             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3773         * config/arc, common/config/arc: New directories.
3775 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3776             Brendan Kehoe  <brendan@zen.org>
3777             Simon Cook  <simon.cook@embecosm.com>
3779         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3781 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3783         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3784         * tree-ssa-coalesce.h: New. Move prototype to here.
3785         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3786         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3787         (gimple_can_coalesce_p): Move to...
3788         * gimple.c (gimple_can_coalesce_p): Here.
3790 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3792         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3793         (dump_decl_set): Move to gimple.c.
3794         * gimple.h: Don't include tree-ssa-operands.h.
3795         (dump_decl_set): Add prototype.
3796         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3797         Move to gimple-ssa.h.
3798         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3799         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3800         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3801         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3802         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3803         than PHI_ARG_DEF.
3804         (dump_decl_set): Relocate here.
3805         * gimple-ssa.h: New file.
3806         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3807         Relocate from gimple.h.
3808         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3809         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3810         swap_tree_operands and remove non-ssa path.
3811         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3812         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3813         swap_ssa_operands.
3814         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3815         vect_is_simple_reduction_1): Use swap_ssa_operands.
3816         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3817         (enum need_phi_state): Move to tree-into-ssa.c.
3818         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3819         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3820         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3821         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3822         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3823         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3824         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3825         num_imm_uses): Move to ssa-iterators.h.
3826         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3827         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3828         tree-phinodes.h.
3829         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3830         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3831         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3832         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3833         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3834         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3835         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3836         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3837         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3838         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3839         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3840         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3841         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3842         (set_phi_nodes): Move to tree-phinodes.h.
3843         * tree-ssa-operands.h (enum ssa_op_iter_type,
3844         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3845         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3846         (dump_decl_set): Remove prototype.
3847         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3848         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3849         (set_phi_nodes): Relocate from tree-flow-inline.h.
3850         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3851         tree-flow-inline.h
3852         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3853         include list.  Temporarily add gimple.h to include list.
3854         * ssa-iterators.h: New file.
3855         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3856         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3857         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3858         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3859         Relocate from tree-ssa-operands.h.
3860         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3861         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3862         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3863         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3864         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3865         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3866         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3867         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3868         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3869         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3870         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3871         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3872         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3873         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3874         Relocate from tree-flow-inline.h.
3875         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3877 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3879         * aarch64-simd.md
3880         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3881         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3882         Insert '\t' to output template.
3883         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3884         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3885         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3886         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3888 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3890         * doc/install.texi (Host/target specific installation notes for GCC):
3891         Put @anchor before @heading.
3892         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3893         Use @email for email addresses.
3895 2013-10-01  Jeff Law  <law@redhat.com>
3897         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3898         a vec.  Only delete the path if we create one without successfully
3899         registering a jump thread.
3900         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3901         as a pointer.
3902         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3903         (paths): New vector of jump threading paths.
3904         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3905         (THREAD_PATH): New accessor macro for the entire thread path.
3906         (lookup_redirection_data): Get intermediate and final outgoing edge
3907         from the thread path.
3908         (create_edge_and_update_destination_phis): Copy the threading path.
3909         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3910         jump threading path.
3911         (ssa_redirect_edges): Get edges and block types from the jump threading
3912         path.  Free the path vector.
3913         (thread_block): Get edges from the jump threading path.  Look at the
3914         entire path to see if we thread to a loop exit.  If we cancel a jump
3915         thread request, then free the path vector.
3916         (thread_single_edge): Get edges and block types from the jump threading
3917         path.  Free the path vector.
3918         (thread_through_loop_header): Get edges and block types from the jump
3919         threading path.  Free the path vector.
3920         (mark_threaded_blocks): Iterate over the vector of paths and store
3921         the path on the appropriate edge.  Get edges and block types from the
3922         jump threading path.
3923         (mark_threaded_blocks): Get edges and block types from the jump
3924         threading path.  Free the path vector.
3925         (thread_through_all_blocks): Use the vector of paths rather than
3926         a vector of 3-edge sets.
3927         (register_jump_thread): Accept pointer to a path vector rather
3928         than the path vector itself.  Store the path vector for later use.
3929         Simplify.
3931 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3932             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3934         PR target/58574
3935         * config/s390/s390.c (s390_split_branches): Modify check for table
3936         jump insns.
3937         (s390_chunkify_start): Rearrange table jump insn check in order to
3938         deal with compare and branch insns correctly.
3940 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3942         PR target/58578
3943         Revert
3944         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3945         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3946         define_insn_and_split.
3947         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3948         (shiftsi3_compare): New pattern.
3949         (rrx): New pattern.
3950         * config/arm/unspecs.md (UNSPEC_RRX): New.
3952 2013-10-01  Alan Modra  <amodra@gmail.com>
3954         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3955         casing inout operands.
3957 2013-10-01  Richard Biener  <rguenther@suse.de>
3959         PR tree-optimization/58553
3960         * tree-loop-distribution.c (struct partition_s): Add niter member.
3961         (classify_partition): Populate niter member for the partition
3962         and properly identify whether the relevant store happens before
3963         or after the loop exit.
3964         (generate_memset_builtin): Use niter member from the partition.
3965         (generate_memcpy_builtin): Likewise.
3967 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3969         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3970         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3972 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3974         * basic-block.h (edge_list): Prefix member names with "m_".
3975         * context.h (context): Likewise.
3976         * domwalk.h (dom_walker): Likewise.
3977         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3978         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3979         * hash-table.h (hash_table): Likewise.
3980         * machmode.h (bit_field_mode_iterator): Likewise.
3981         * pass_manager.h (pass_list): Likewise.
3982         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3983         * tree-pass.h (pass_data): Likewise.
3984         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3985         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3986         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3987         * asan.c (pass_data_asan): Update accordingly.
3988         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3989         (control_dependences::control_dependences): Likewise.
3990         (control_dependences::~control_dependences): Likewise.
3991         (control_dependences::~control_dependences): Likewise.
3992         (control_dependences::get_edges_dependent_on): Likewise.
3993         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3994         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3995         * context.c (gcc::context::context): Likewise.
3996         * cprop.c (pass_rtl_cprop::clone): Likewise.
3997         * domwalk.c (dom_walker::walk): Likewise.
3998         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3999         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
4000         * mode-switching.c (pass_mode_switching::clone): Likewise.
4001         * passes.c (opt_pass::opt_pass): Likewise.
4002         (pass_manager::pass_manager): Likewise.
4003         * predict.c (pass_strip_predict_hints::clone): Likewise.
4004         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
4005         (pass_split_all_insns::clone): Likewise.
4006         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4007         Likewise.
4008         (bit_field_mode_iterator::next_mode): Likewise.
4009         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
4010         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
4011         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
4012         * tree-complex.c (pass_lower_complex::clone): Likewise.
4013         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
4014         * tree-object-size.c (pass_object_sizes::clone): Likewise.
4015         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
4016         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
4017         (pass_fold_builtins::clone): Likewise.
4018         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
4019         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
4020         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
4021         (pass_cd_dce::clone): Likewise.
4022         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
4023         (pass_phi_only_cprop::clone): Likewise.
4024         * tree-ssa-dse.c (pass_dse::clone): Likewise.
4025         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
4026         * tree-ssa-loop.c (pass_lim::clone): Likewise.
4027         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
4028         * tree-ssa-pre.c (pass_fre::clone): Likewise.
4029         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
4030         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
4031         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
4032         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
4033         * tree-vrp.c (pass_vrp::clone): Likewise.
4034         * tsan.c (pass_tsan::clone): Likewise.
4036 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
4038         PR middle-end/58564
4039         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
4040         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
4042         PR middle-end/58564
4043         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
4044         optimization, punt if sign_bit_p looked through any zero extension.
4046 2013-09-30  Teresa Johnson  <tejohnson@google.com>
4048         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
4049         Update redirected out edge count in joiner case.
4050         (ssa_redirect_edges): Common the joiner and non-joiner cases
4051         so that joiner case gets profile updates.
4053 2013-09-30  Richard Biener  <rguenther@suse.de>
4055         PR tree-optimization/58554
4056         * tree-loop-distribution.c (classify_partition): Require
4057         unconditionally executed stores for memcpy and memset recognition.
4058         (tree_loop_distribution): Calculate dominance info.
4060 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
4062         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
4063         (NO_PROFILE_COUNTERS): Likewise.
4064         (PROFILE_HOOK): Likewise.
4065         (FUNCTION_PROFILER): Likewise.
4066         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
4068 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
4070         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
4071         calls and defines in TARGET_MACHO conditional.
4072         (load_macho_picbase_di): Likewise.
4073         (reload_macho_picbase): Likewise.
4074         (reload_macho_picbase_si): Likewise.
4075         (reload_macho_picbase_di): Likewise.
4076         (nonlocal_goto_receiver): Likewise.
4078 2013-09-30  Nick Clifton  <nickc@redhat.com>
4080         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
4081         that use the MSP430X ISA.
4082         (msp430_option_override): Scan -mmcu command line option for any
4083         MCU name that supports the MSP430X ISA.
4084         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
4085         -mmcu options which enable the MSP430X ISA.
4087 2013-09-30  Richard Biener  <rguenther@suse.de>
4089         PR middle-end/58532
4090         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
4091         before looking for setjmp-like calls.
4093 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
4095         PR target/10901
4096         * config/darwin-protos.h (machopic_get_function_picbase): New.
4097         * config/darwin.c (machopic_get_function_picbase): New.
4098         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
4099         label for a new func.  (load_macho_picbase_di): Likewise.
4100         (reload_macho_picbase): New expand.
4101         (reload_macho_picbase_si): New insn.
4102         (reload_macho_picbase_di): New insn.
4103         (nonlocal_goto_receiver): New define and split.
4104         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
4105         (unspecv enum): Add UNSPECV_NLGR.
4107 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
4109         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
4110         that altivec registers are correctly sized on Darwin.
4112 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
4114         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
4115         darwin-driver.o): Use COMPILE and POSTCOMPILE.
4116         * config/x-darwin (host-darwin.o): Likewise.
4117         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
4118         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
4119         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
4121 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
4123         * doc/invoke.texi: Fix usage of @tie{} command.
4125 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
4127         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
4129 2013-09-28  Jan Hubicka  <jh@suse.cz>
4131         * config/i386/x86-tune.def: Add documentation for each of the options;
4132         add whitespace.
4134 2013-09-28  Jan Hubicka  <jh@suse.cz>
4136         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
4137         generic.
4138         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
4139         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
4140         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
4141         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
4142         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
4144 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
4146         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
4147         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
4148         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
4149         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
4150         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
4151         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
4152         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
4153         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
4154         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
4155         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
4156         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
4157         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
4158         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
4159         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
4160         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
4161         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
4162         gimple.h, godump.c, graphite-clast-to-gimple.c,
4163         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
4164         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
4165         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
4166         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
4167         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
4168         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
4169         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
4170         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
4171         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
4172         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
4173         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
4174         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
4175         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
4176         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
4177         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
4178         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
4179         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
4180         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
4181         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
4182         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
4183         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
4184         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
4185         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
4186         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
4187         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
4188         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4189         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
4190         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
4191         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
4192         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4193         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4194         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
4195         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
4196         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
4197         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
4198         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
4199         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
4200         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
4201         whitespace before "(".
4203 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
4205         * expr.h (extract_bit_field): Remove packedp parameter.
4206         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
4207         from forward declaration.
4208         (store_split_bit_field): Remove packedp arg from calls to
4209         extract_fixed_bit_field.
4210         (extract_bit_field_1): Remove packedp parameter and packedp
4211         argument from recursive calls and calls to extract_fixed_bit_field.
4212         (extract_bit_field): Remove packedp parameter and corresponding
4213         arg to extract_bit_field_1.
4214         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
4215         to issue warnings.
4216         (extract_split_bit_field): Remove packedp arg from call to
4217         extract_fixed_bit_field.
4218         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
4219         (copy_blkmode_from_reg): Likewise.
4220         (copy_blkmode_to_reg): Likewise.
4221         (read_complex_part): Likewise.
4222         (store_field): Likewise.
4223         (expand_expr_real_1): Likewise.
4224         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
4225         to extract_bit_field.
4226         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
4227         call to extract_bit_field.
4228         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
4229         call to extract_bit_field.
4230         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
4231         and special packedp behavior from -fstrict-volatile-bitfields
4232         documentation.
4234 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4236         * lra-eliminations.c (init_elim_table): Guard value_p.
4238 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
4240         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
4241         DFmode, DImode, and SFmode in the upper VSX registers based on the
4242         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
4243         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
4244         of the VSX register handling.
4246         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
4247         (f32_sv): Likewise.
4248         (zero_extendsidi2_lfiwzx): Add support for loading into the
4249         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
4250         only do VSX memory options on Altivec registers.
4251         (extendsidi2_lfiwax): Likewise.
4252         (extendsfdf2_fpr): Likewise.
4253         (mov<mode>_hardfloat, SF/SD modes): Likewise.
4254         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
4255         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
4256         (movdi_internal64): Likewise.
4258 2013-09-27  Xinliang David Li  <davidxl@google.com>
4260         * opts.c (finish_options): Adjust parameters
4261         according to vect cost model.
4262         (common_handle_option): Set dynamic vect cost
4263         model for FDO.
4264         targhooks.c (default_add_stmt_cost): Compute stmt cost
4265         unconditionally.
4266         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
4267         Use helper function.
4268         * tree-vectorizer.h (unlimited_cost_model): New function.
4269         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
4270         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
4271         function.
4272         (vect_enhance_data_refs_alignment): Ditto.
4273         * flag-types.h: New enum.
4274         * common/config/i386/i386-common.c (ix86_option_init_struct):
4275         No need to initialize vect_cost_model flag.
4276         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
4277         unconditionally.
4279 2013-09-27  Diego Novillo  <dnovillo@google.com>
4281         * gimple.h (enum ssa_mode): Remove.
4283 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
4285         * cfgloop.h (number_of_loops): Fix typo in check for null.
4287 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
4289         PR middle-end/58551
4290         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
4291         are children of outermost saved_cfun's loop, and set it up to
4292         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
4293         if loop != loop0 and SESE region contains bbs that belong to loop0.
4295 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
4297         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
4298         (binary_scale_code_p, get_base_term, get_index_term): New functions.
4299         (set_address_segment, set_address_base, set_address_index)
4300         (set_address_disp): Accept the argument unconditionally.
4301         (baseness): Remove must_be_base_p and must_be_index_p checks.
4302         (decompose_normal_address): Classify as much as possible in the
4303         main loop.
4305 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
4307         * cse.c (count_reg_usage): Handle INT_LIST.
4308         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
4309         * reginfo.c (reg_scan_mark_refs): Likewise.
4310         * reload1.c (eliminate_regs_1): Likewise.
4312 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
4314         PR middle-end/58547
4315         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
4316         signed.
4318 2013-09-27  Richard Biener  <rguenther@suse.de>
4320         PR tree-optimization/58459
4321         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
4322         restriction not propagating into loops.
4324 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
4326         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
4327         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
4328         * doc/tree-ssa.texi (Walking use-def chains): Delete.
4330 2013-09-26  Richard Biener  <rguenther@suse.de>
4332         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
4334 2013-09-26  Richard Biener  <rguenther@suse.de>
4336         * alias.h (component_uses_parent_alias_set): Rename to ...
4337         (component_uses_parent_alias_set_from): ... this.
4338         * alias.c (component_uses_parent_alias_set): Rename to ...
4339         (component_uses_parent_alias_set_from): ... this and return
4340         the desired parent.
4341         (reference_alias_ptr_type_1): Use the result from
4342         component_uses_parent_alias_set_from instead of stripping
4343         components one at a time.
4344         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
4346 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
4348         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
4349         Move prototypes to...
4350         * tree-ssa-ter.h: New File.  Move prototypes here.
4351         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
4352         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
4353         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
4354         from is_replaceable_p.
4355         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
4356         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
4357         (process_replaceable): Use ter_is_replaceable_p.
4358         (find_replaceable_in_bb): Use ter_is_replaceable_p.
4359         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
4360         newly refactored ssa_is_replaceable_p.
4361         * cfgexpand.c: Include tree-outof-ssa.h.
4362         * ssaexpand.h: Delete.
4364 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
4366         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
4367         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
4368         tree-ssa.c
4369         (create_gimple_tmp): Delete.
4370         (get_expr_type, build_assign, build_type_cast): Move to...
4371         * gimple-builder.c: New File.
4372         (get_expr_type): Relocate from gimple.c.
4373         (build_assign, build_type_cast): Change to only create ssanames.
4374         * gimple.h: Move prototypes to...
4375         * gimple-builder.h: New File. Here.
4376         * tree-ssa.h: And here.
4377         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
4378         count_uses_and_derefs): Relocate from gimple.c.
4379         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
4380         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
4381         * tree-ssa-math-opts (execute_cse_reciprocals): Use
4382         gimple_replace_ssa_lhs.
4383         * asan.c: Include gimple-builder.h.
4384         * Makefile.in: Add gimple-builder.o.
4386 2013-09-26  Richard Biener  <rguenther@suse.de>
4388         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
4389         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
4390         (loe_visit_block): Use gcc_checking_assert.
4391         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
4392         gimple_assign_ssa_name_copy_p.
4393         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
4394         change.
4396 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
4398         * config/rs6000/t-rs6000 (rs6000.o): Remove.
4399         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
4401 2013-09-26  Richard Biener  <rguenther@suse.de>
4403         PR tree-optimization/58539
4404         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
4405         the fact that debug statements are not taking part in loop-closed
4406         SSA construction.
4408 2013-09-26  Nick Clifton  <nickc@redhat.com>
4410         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
4411         time warning message.
4412         (msp430_print_operand_raw): Delete unused letter parameter.
4413         (TARGET_PRINT_OPERAND_ADDRESS): Define.
4414         (msp430_print_operand_address): New function.
4415         (msp430_print_operand): Move address printing code from here to
4416         new function.
4417         * config/msp430/msp430.md (movsipsi2): Add comment in generated
4418         assembler.
4419         (zero_extendpsisi2): Likewise.
4420         (extendpsisi2): New pattern.
4421         (andneghi3): New pattern.
4423 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
4425         * config/aarch64/aarch64.opt (mlra): New option.
4426         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
4427         (TARGET_LRA_P): Define.
4429 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
4431         * expr.c (expand_assignment): Remove obsolete comment.
4433 2013-09-25  Jeff Law  <law@redhat.com>
4435         * tree-flow.h (thread_through_all_blocks): Prototype moved into
4436         tree-ssa-threadupdate.h.
4437         (register_jump_thread): Similarly.
4438         * tree-ssa-threadupdate.h: New header file.
4439         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
4440         * tree-vrp.c: Likewise.
4441         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
4442         (thread_around_empty_blocks): Change type of path vector argument to
4443         an edge,type pair from just an edge.  Initialize both elements when
4444         appending to a jump threading path.  Tweak references to elements
4445         appropriately.
4446         (thread_across_edge): Similarly.  Release memory for the elements
4447         as needed.
4448         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
4449         (dump_jump_thread_path): New function broken out from
4450         register_jump_thread.
4451         (register_jump_thread): Use dump_jump_thread_path.  Change type of
4452         path vector entries.  Search the path for NULL edges and dump
4453         the path if one is found.  Tweak the conversion of path to 3-edge
4454         form to use the block copy type information embedded in the path.
4456 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4458         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
4460 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4461             Vladimir Makarov  <vmakarov@redhat.com>
4463         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
4464         from the least significant bit.
4465         (strip_address_mutations): Add bitfield operations handling.
4466         (must_be_index_p): Add shifting and rotate operations handling.
4467         (set_address_base): Use must_be_base_p predicate.
4468         (set_address_index): Use must_be_index_p predicate.
4470 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4471             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4472             Sergey Lega  <sergey.s.lega@intel.com>
4473             Anna Tikhonova  <anna.tikhonova@intel.com>
4474             Ilya Tocar  <ilya.tocar@intel.com>
4475             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4476             Ilya Verbin  <ilya.verbin@intel.com>
4477             Kirill Yukhin  <kirill.yukhin@intel.com>
4478             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4480         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4481         Use new names.
4482         (ix86_expand_vector_move_misalign): Support new unaligned load and
4483         stores and use new names.
4484         (CODE_FOR_sse2_storedqu): Rename to ...
4485         (CODE_FOR_sse2_storedquv16qi): ... this.
4486         (CODE_FOR_sse2_loaddqu): Rename to ...
4487         (CODE_FOR_sse2_loaddquv16qi): ... this.
4488         (CODE_FOR_avx_loaddqu256): Rename to ...
4489         (CODE_FOR_avx_loaddquv32qi): ... this.
4490         (CODE_FOR_avx_storedqu256): Rename to ...
4491         (CODE_FOR_avx_storedquv32qi): ... this.
4492         * config/i386/i386.md (fpint_logic): New.
4493         * config/i386/sse.md (VMOVE): Extend for AVX512.
4494         (VF): Ditto.
4495         (VF_128_256): New.
4496         (VF_512): Ditto.
4497         (VI_UNALIGNED_LOADSTORE): Ditto.
4498         (sse2_avx_avx512f): Ditto.
4499         (sse2_avx2): Extend for AVX512.
4500         (sse4_1_avx2): Ditto.
4501         (avx2_avx512f): New.
4502         (sse): Extend for AVX512.
4503         (sse2): Ditto.
4504         (sse4_1): Ditto.
4505         (avxsizesuffix): Ditto.
4506         (sseintvecmode): Ditto.
4507         (ssePSmode): Ditto.
4508         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4509         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4510         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4511         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4512         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4513         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4514         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4515         (STORENT_MODE): Extend for AVX512.
4516         (*absneg<mode>2): Replace constraint "x" with "v".
4517         (*mul<mode>3): Ditto.
4518         (*ieee_smin<mode>3): Ditto.
4519         (*ieee_smax<mode>3): Ditto.
4520         (avx_cmp<mode>3): Replace VF with VF_128_256.
4521         (*<sse>_maskcmp<mode>3_comm): Ditto.
4522         (<sse>_maskcmp<mode>3): Ditto.
4523         (<sse>_andnot<mode>3): Extend for AVX512.
4524         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4525         (<code><mode>3, fpint_logic): New.
4526         (*<code><mode>3): Extend for AVX512.
4527         (avx512flogicsuff): New.
4528         (avx512f_<logic><mode>): Ditto.
4529         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4530         VF_128_256.
4531         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4532         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4533         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4534         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4535         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4536         (xop_vpermil2<mode>3): Ditto.
4537         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4538         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4539         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4540         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4542 2013-09-25  Tom Tromey  <tromey@redhat.com>
4544         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4545         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4546         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4547         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4548         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4549         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4550         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4551         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4552         (GRAPHITE_HTAB_H): Remove.
4554 2013-09-25  Tom Tromey  <tromey@redhat.com>
4556         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4558 2013-09-25  Tom Tromey  <tromey@redhat.com>
4560         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4562 2013-09-25  Tom Tromey  <tromey@redhat.com>
4564         * config/i386/t-i386 (i386.o): Remove.
4565         (i386-c.o): Use COMPILE and POSTCOMPILE.
4567 2013-09-25  Tom Tromey  <tromey@redhat.com>
4569         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4571 2013-09-25  Tom Tromey  <tromey@redhat.com>
4573         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4574         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4575         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4576         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4577         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4578         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4579         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4580         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4581         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4582         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4583         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4584         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4585         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4586         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4587         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4588         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4589         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4590         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4591         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4592         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4593         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4594         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4595         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4596         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4597         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4598         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4599         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4600         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4601         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4602         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4603         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4604         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4605         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4606         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4607         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4608         (graphite-clast-to-gimple.o, graphite-dependences.o)
4609         (graphite-interchange.o, graphite-poly.o)
4610         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4611         (graphite-optimize-isl.o, tree-vect-loop.o)
4612         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4613         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4614         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4615         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4616         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4617         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4618         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4619         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4620         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4621         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4622         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4623         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4624         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4625         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4626         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4627         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4628         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4629         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4630         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4631         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4632         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4633         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4634         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4635         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4636         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4637         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4638         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4639         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4640         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4641         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4642         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4643         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4644         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4645         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4646         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4647         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4648         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4649         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4650         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4651         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4652         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4653         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4654         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4655         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4656         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4657         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4658         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4659         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4660         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4661         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4662         (gcov-dump.o): Remove.
4663         (default-c.o): Use COMPILE and POSTCOMPILE.
4664         (CFLAGS-gcc.o): New variable.
4665         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4667 2013-09-25  Tom Tromey  <tromey@redhat.com>
4669         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4670         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4671         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4672         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4673         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4674         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4675         (c-family/array-notation-common.o, c-family/stub-objc.o)
4676         (c-family/c-ubsan.o): Remove.
4678 2013-09-25  Tom Tromey  <tromey@redhat.com>
4680         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4682 2013-09-25  Tom Tromey  <tromey@redhat.com>
4684         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4685         to add -DENABLE_SHARED_LIBGCC.
4686         (gcc.o): Don't use subshell.
4688 2013-09-25  Tom Tromey  <tromey@redhat.com>
4690         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4691         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4692         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4693         * configure, config.in: Rebuild.
4695 2013-09-25  Tom Tromey  <tromey@redhat.com>
4697         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4698         (COMPILE, POSTCOMPILE): New variables.
4699         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4700         (DEPFILES): New variable.
4701         Include ".Po" files.
4702         * configure.ac: Add checks for dependency checking.
4703         * configure, aclocal.m4: Regenerate.
4705 2013-09-25  Tom Tromey  <tromey@redhat.com>
4707         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4708         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4710 2013-09-25  Tom Tromey  <tromey@redhat.com>
4712         * Makefile.in (generated_files): Add options.h,
4713         target-hooks-def.h, insn-opinit.h,
4714         common/common-target-hooks-def.h, pass-instances.def,
4715         c-family/c-target-hooks-def.h.
4717 2013-09-25  Jeff Law  <law@redhat.com>
4719         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4720         than foo[foo.length () - 1] to access last member in a vec.
4721         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4723 2013-09-25  Richard Biener  <rguenther@suse.de>
4725         PR middle-end/58521
4726         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4728 2013-09-25  Jan Hubicka  <jh@suse.cz>
4730         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4731         test.
4733 2013-09-25  Marek Polacek  <polacek@redhat.com>
4735         PR sanitizer/58420
4736         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4737         when determining the type name.
4739 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4741         * config/sh/sh.md: Fix formatting.
4743 2013-09-24  Xinliang David Li  <davidxl@google.com>
4745         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4746         max peel iterations parameter.
4747         * param.def: New parameter.
4748         * doc/invoke.texi: Document New parameter.
4750 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4752         * gimple-pretty-print.c: Various whitespace tweaks.
4753         * tree-core.h: Likewise.
4754         * tree-pretty-print.c: Likewise.
4755         * tree-ssa-alias.c: Likewise.
4756         * tree-ssa-copy.c: Likewise.
4757         * tree-ssanames.c: Likewise.
4758         * tree-ssanames.h: Likewise.
4759         * tree-vrp.c: Likewise.
4761 2013-09-24  Alan Modra  <amodra@gmail.com>
4763         PR middle-end/57134
4764         PR middle-end/57586
4765         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4766         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4767         inout operands.
4769 2013-09-24  Richard Biener  <rguenther@suse.de>
4771         PR middle-end/58513
4772         * tree.c (reference_alias_ptr_type): Move ...
4773         * alias.c (reference_alias_ptr_type): ... here and implement
4774         in terms of the new reference_alias_ptr_type_1.
4775         (ref_all_alias_ptr_type_p): New helper.
4776         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4777         use ref_all_alias_ptr_type_p.
4778         (get_deref_alias_set): Add flag_strict_aliasing check here.
4779         (reference_alias_ptr_type_1): New function, split out from ...
4780         (get_alias_set): ... here.
4781         (alias_ptr_types_compatible_p): New function.
4782         * alias.h (reference_alias_ptr_type): Declare.
4783         (alias_ptr_types_compatible_p): Likewise.
4784         * tree.h (reference_alias_ptr_type): Remove.
4785         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4786         to compare MEM_REF alias types.
4788 2013-09-24  Richard Biener  <rguenther@suse.de>
4790         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4792 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4794         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4795         reload helper function arrays into a single array reg_addr.
4796         (reload_fpr_gpr): Likewise.
4797         (reload_gpr_vsx): Likewise.
4798         (reload_vsx_gpr): Likewise.
4799         (struct rs6000_reg_addr): Likewise.
4800         (reg_addr): Likewise.
4801         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4802         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4803         (rs6000_init_hard_regno_mode_ok): Likewise.
4804         (rs6000_secondary_reload_direct_move): Likewise.
4805         (rs6000_secondary_reload): Likewise.
4807         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4808         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4809         power8 changes.  Put wg constraint in alphabetical order.
4811         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4812         for future work to add ISA 2.07 VSX single precision support.
4813         (-mvsx-scalar-double): Change default from -1 to 1, update
4814         documentation comment.
4815         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4816         (-mupper-regs-df): New debug switch to control whether DF values
4817         can go in the traditional Altivec registers.
4818         (-mupper-regs-sf): New debug switch to control whether SF values
4819         can go in the traditional Altivec registers.
4821         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4822         and wy constraints.
4823         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4824         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4825         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4826         (rs6000_debug_legitimate_address_p): Print if we are running
4827         before, during, or after reload.
4828         (rs6000_secondary_reload): Add a comment.
4829         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4831         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4832         constraints.  Update documentation string.
4833         (wd constraint): Likewise.
4834         (wf constraint): Likewise.
4835         (wg constraint): Likewise.
4836         (wn constraint): Likewise.
4837         (ws constraint): Likewise.
4838         (wt constraint): Likewise.
4839         (wx constraint): Likewise.
4840         (wz constraint): Likewise.
4841         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4842         instructions.
4843         (ww constraint): Likewise.
4844         (wy constraint): Likewise.
4845         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4846         in the previous submissions.
4847         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4849 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4851         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4852         (REG_BR_PROB): Say that the probability is stored in an int_list.
4853         * reg-notes.def: Update commentary to mention INT_LIST.
4854         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4855         (INT_LIST): New rtx.
4856         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4857         * rtlanal.c (int_reg_note_p): New function.
4858         (alloc_reg_note): Assert that the note does not have an int argument.
4859         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4860         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4861         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4862         rather than an INSN_LIST.  Handle INT_LIST.
4863         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4864         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4865         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4866         Manipulate them as ints rather than rtxes.
4867         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4868         * regmove.c (copy_src_to_dest): Likewise.
4869         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4871         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4872         into the cases that need it.
4873         * config/arm/arm.c (arm_unwind_emit): Likewise.
4875         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4876         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4877         * loop-doloop.c (add_test, doloop_modify): Likewise.
4878         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4879         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4880         * predict.c (combine_predictions_for_insn): Likewise.
4881         * print-rtl.c (print_rtx): Handle INT_LIST.
4882         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4883         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4884         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4885         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4886         (ix86_split_fp_branch, predict_jump): Likewise.
4887         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4888         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4889         * config/spu/spu.c (ea_load_store_inline): Likewise.
4891         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4892         value of a REG_BR_PROB note.
4893         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4894         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4895         * emit-rtl.c (try_split): Likewise.
4896         * predict.c (br_prob_note_reliable_p): Likewise.
4897         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4898         * reorg.c (mostly_true_jump): Likewise.
4899         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4900         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4901         * config/i386/i386.c (ix86_print_operand): Likewise.
4902         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4903         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4904         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4905         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4906         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4907         * config/sparc/sparc.c (output_cbranch): Likewise.
4908         * config/spu/spu.c (get_branch_target): Likewise.
4909         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4910         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4912 2013-09-23  Jan Hubicka  <jh@suse.cz>
4914         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4915         for ipa-devirt.
4916         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4917         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4918         of external calls
4919         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4920         (gimple_fold_call): Dump inconsistent devirtualizations; add
4921         sanity check for type based devirtualizations.
4922         * ipa-prop.c: Include ipa-utils.h
4923         (ipa_intraprocedural_devirtualization): Add sanity check.
4924         (try_make_edge_direct_virtual_call): Likewise.
4926 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4928         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4929         assignment statements.
4931 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4933         * gimple-pretty-print.c (dump_ssaname_info): New function.
4934         (dump_gimple_phi): Call it.
4935         (pp_gimple_stmt_1): Likewise.
4936         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4937         (range_info_def): Declare.
4938         * tree-pretty-print.c (pp_double_int): New function.
4939         (dump_generic_node): Call it.
4940         * tree-pretty-print.h (pp_double_int): Declare.
4941         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4942         * tree-ssanames.h (range_info_def): New structure.
4943         (value_range_type): Move definition here.
4944         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4945         Declare.
4946         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4947         initialization.
4948         (set_range_info): New function.
4949         (get_range_info): Likewise.
4950         (duplicate_ssa_name_range_info): Likewise.
4951         (duplicate_ssa_name_fn): Check pointer type and call
4952         duplicate_ssa_name_range_info.
4953         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4954         * tree-vrp.c (value_range_type): Remove definition, now in
4955         tree-ssanames.h.
4956         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4957         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4958         (SSA_NAME_RANGE_INFO): New macro.
4960 2013-09-23  Richard Biener  <rguenther@suse.de>
4962         PR tree-optimization/58464
4963         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4964         (phi_trans_add): Change to add conditionally on being not present.
4965         (phi_translate_1): Remove recursion detection here.
4966         (phi_translate): Pre-seed the cache with NULL to catch
4967         recursion here in a more generic way.
4968         (bitmap_find_leader): Adjust comment.
4969         (get_representative_for): Dump value-numbers.
4970         (create_expression_by_pieces): Likewise.
4971         (insert_into_preds_of_block): Likewise.
4973 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4975         PR target/58475
4976         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4977         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4979 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4981         Revert r202780:
4982         2013-09-20  Renlin Li  <renlin.li@arm.com>
4984         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4985         plus_constant.
4986         (aarch64_expand_epilogue): Likewise.
4987         (aarch64_legitimize_reload_address): Likewise.
4989 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4991         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4992         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4994 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4996         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4997         comparison operators when -fno-trapping-math is in effect.
4998         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4999         comparison operators in TFmode and assert that unsupported operators
5000         cannot reach here.
5001         (ia64_print_operand): Likewise.
5003 2013-09-21  Jan Hubicka  <jh@suse.cz>
5005         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
5006         (sse_typeless_stores): Enable for core
5007         (sse_load0_by_pxor): Likewise.
5008         (four_jump_limit): Disable for core.
5009         (pad_returns): Likewise.
5010         (avoid_vector_decode): Likewise.
5011         (fuse_cmp_and_branch): Enable for cores.
5012         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
5014 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
5016         PR middle-end/56791
5017         * config/pa/pa.c (pa_option_override): Disable auto increment and
5018         decrement instructions until reload is completed.
5020         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
5021         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
5022         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
5024 2013-09-20  DJ Delorie  <dj@redhat.com>
5025             Nick Clifton  <nickc@redhat.com>
5027         * config/rl78/rl78.c: Various whitespace and comment tweaks.
5028         (need_to_save): Save bank 0 on interrupts.
5029         (characterize_address): Strip far address wrappers.
5030         (rl78_as_legitimate_address): Likewise.
5031         (transcode_memory_rtx): Likewise.
5032         (rl78_peep_movhi_p): Disable this peephole after devirt.
5033         (rl78_propogate_register_origins): Forget all origins when a
5034         CLOBBER is seen.
5035         * config/rl78/rl78-virt.md: Various whitespace tweaks.
5036         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
5037         peephole2's.
5038         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
5039         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
5040         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
5041         relocating.
5042         * config/rl78/constraints.md: Various whitespace and paren tweaks.
5044 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
5046         * config/pa/pa.md: In "scc" insn patterns, change output template to
5047         handle const0_rtx in reg_or_0_operand operands.
5049 2013-09-20  Martin Husemann  <martin@NetBSD.org>
5051         PR target/56875
5052         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
5053         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
5055 2013-09-20  Richard Biener  <rguenther@suse.de>
5057         PR middle-end/58484
5058         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
5059         remembering SSA name version and block index.
5060         (new_scev_info_str): Adjust.
5061         (hash_scev_info): Likewise.  Also hash the block index.
5062         (eq_scev_info): Adjust.
5063         (find_var_scev_info): Likewise.
5064         (struct instantiate_cache_entry): Remove.
5065         (struct instantiate_cache_type): Use a htab to map name, block
5066         to chrec.
5067         (instantiate_cache_type::~instantiate_cache_type): Adjust.
5068         (get_instantiated_value_entry): Likewise.
5069         (hash_idx_scev_info, eq_idx_scev_info): New functions.
5070         (instantiate_scev_name): Adjust.
5072 2013-09-20  Jeff Law  <law@redhat.com>
5074         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
5076 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
5078         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5079         Call aarch64_simd_expand_args to update op[argc].
5081 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
5083         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
5084         plugin argument.
5086 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
5088         * gengtype.c (file_rules): Added rule for *.cc files.
5089         (get_output_file_with_visibility): Give fatal message when no
5090         rules found.
5092 2013-09-20  Renlin Li  <renlin.li@arm.com>
5094         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
5095         (aarch64_expand_epilogue): Likewise.
5096         (aarch64_legitimize_reload_address): Likewise.
5098 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5100         PR middle-end/57748
5101         * expr.c (expand_assignment): Remove misalignp code path.
5103 2013-09-20  Marek Polacek  <polacek@redhat.com>
5105         PR sanitizer/58413
5106         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
5107         TYPE_PRECISION.  Add asserts.
5109 2013-09-20  Richard Biener  <rguenther@suse.de>
5111         PR tree-optimization/58453
5112         * tree-loop-distribution.c (distribute_loop): Apply the cost
5113         model for -ftree-loop-distribute-patterns, too.
5115 2013-09-20  Richard Biener  <rguenther@suse.de>
5117         PR middle-end/58473
5118         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
5119         make type comparison less strict.
5121 2013-09-20  Alan Modra  <amodra@gmail.com>
5123         * configure: Regenerate.
5124         * aclocal.m4: Regenerate.
5126 2013-09-20  Marek Polacek  <polacek@redhat.com>
5128         PR other/58467
5129         * doc/extend.texi: Document that attribute used is meant to be used
5130         on variables with static storage duration.
5132 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
5134         PR tree-optimization/58472
5135         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
5136         simd_lane_access set inv_p = false.
5137         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
5138         the simduid magic VAR_DECL.
5140 2013-09-19  Jan Hubicka  <jh@suse.cz>
5142         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
5144 2013-09-17  Jeff Law  <law@redhat.com>
5146         * tree-ssa-dom.c (record_temporary_equivalences): New function
5147         split out of dom_opt_dom_walker::after_dom_children.
5148         (dom_opt_dom_walker::thread_across_edge): Move common code
5149         in here from dom_opt_dom_walker::after_dom_children.
5150         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
5152 2013-09-19  Jan Hubicka  <jh@suse.cz>
5154         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
5155         (TARGET_GENERIC): Use PROCESOR_GENERIC
5156         (enum processor_type): Unify generic32 and 64.
5157         * i386.md (cpu): Likewise.
5158         * x86-tune.def (use_leave): Enable for generic32.
5159         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
5160         * athlon.md: Change generic64 to generic in all occurences.
5161         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
5162         (ix86_target_macros_internal): Likewise.
5163         * driver-i386.c (host_detect_local_cpu): Likewise.
5164         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
5165         to ..
5166         (generic_memcpy, generic_memset, generic_cost): This one.
5167         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
5168         (m_GENERIC32, m_GENERIC64): Remove.
5169         (m_GENERIC): Turn into one flag.
5170         (processor_target): Unify generic tunnings.
5171         (ix86_option_override_internal): Replace generic32/64 by generic.
5172         (ix86_issue_rate): Likewise.
5173         (ix86_adjust_cost): Likewise.
5175 2013-09-19  Jan Hubicka  <jh@suse.cz>
5177         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
5178         of speculative flag.
5180 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
5182         * omp-low.c (expand_omp_sections): Always pass len - 1 to
5183         GOMP_sections_start, even if !exit_reachable.
5185 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
5187         * lra-constraints.c (need_for_all_save_p): Use macro
5188         HARD_REGNO_CALL_PART_CLOBBERED.
5189         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
5190         set up pseudo conflict hard regs.
5192 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
5194         PR target/58452
5195         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
5196         operands.
5198 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
5200         PR rtl-optimization/58438
5201         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
5202         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
5203         from previous subpasses.
5205 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
5207         * arm.c (arm_get_frame_offsets): Validate architecture supports
5208         LDRD/STRD before accepting the tuning preference.
5209         (arm_expand_prologue): Likewise.
5210         (arm_expand_epilogue): Likewise.
5212 2013-09-18  Richard Biener  <rguenther@suse.de>
5214         PR tree-optimization/58417
5215         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
5216         have chrecs with symbols defined in the loop as operands.
5217         (chrec_fold_multiply): Likewise.
5218         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
5219         parameters before folding binary operations.
5220         (struct instantiate_cache_entry_hasher): Remove.
5221         (struct instantiate_cache_type): Use a pointer-map.
5222         (instantiate_cache_type::instantiate_cache_type): New function.
5223         (instantiate_cache_type::get): Likewise.
5224         (instantiate_cache_type::set): Likewise.
5225         (instantiate_cache_type::~instantiate_cache_type): Adjust.
5226         (get_instantiated_value_entry): Likewise.
5227         (global_cache): New global.
5228         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
5229         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
5230         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
5231         (instantiate_scev_name): Adjust.
5232         (instantiate_scev): Construct global instead of local cache.
5233         (resolve_mixers): Likewise.
5235 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
5236             Paolo Carlini  <paolo.carlini@oracle.com>
5238         PR c++/58458
5239         * doc/implement-cxx.texi: Fix references to the C++ standards.
5241 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
5243         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
5244         * tree-vect-data-refs.c (vect_analyze_data_refs): For
5245         simd_lane_access drs, update also DR_ALIGNED_TO.
5247 2013-09-18  Marek Polacek  <polacek@redhat.com>
5249         PR sanitizer/58411
5250         * doc/extend.texi: Document no_sanitize_undefined attribute.
5251         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
5252         no_sanitize_undefined attribute.
5254 2013-09-18  Nick Clifton  <nickc@redhat.com>
5256         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
5257         (ASM_DECLARE_FUNCTION_NAME): Define.
5259 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
5261         * compare-elim.c (find_comparison_dom_walker): New class
5262         (find_comparisons_in_bb): Rename to
5263         find_comparison_dom_walker::before_dom_children
5264         (find_comparisons): Adjust
5265         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
5266         adjust.
5267         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
5268         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
5269         (init_walk_dominator_tree): Remove declaration.
5270         (fini_walk_dominator_tree): Remove declaration.
5271         * fwprop.c (single_def_use_dom_walker): New class
5272         (single_def_use_enter_block): Convert to
5273         single_def_use_dom_walker::before_dom_children.
5274         (single_def_use_leave_block): Convert to
5275         single_def_use_dom_walker::after_dom_children.
5276         (build_single_def_use_links): Adjust.
5277         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
5278         class.
5279         (find_candidates_in_block): Convert to
5280         find_candidates_dom_walker::before_dom_children.
5281         (execute_strength_reduction): Adjust.
5282         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
5283         (sese_dom_walker): New class.
5284         (sese_dom_walker::sese_dom_walker): New constructor.
5285         (sese_dom_walker::~sese_dom_walker): New destructor.
5286         (build_sese_conditions_before): Convert to
5287         sese_dom_walker::before_dom_children.
5288         (build_sese_conditions_after): Convert to
5289         sese_dom_walker::after_dom_children.
5290         (build_poly_scop): Adjust
5291         * tree-into-ssa.c (rewrite_dom_walker): New class
5292         (rewrite_enter_block): Convert to
5293         rewrite_dom_walker::before_dom_children.
5294         (rewrite_leave_block): Convert to
5295         rewrite_dom_walker::after_dom_children.
5296         (rewrite_update_dom_walker): New class.
5297         (rewrite_update_enter_block): Convert to
5298         rewrite_update_dom_walker::before_dom_children.
5299         (rewrite_update_leave_block): Convert to
5300         rewrite_update_dom_walker::after_dom_children.
5301         (rewrite_blocks, rewrite_into_ssa): Adjust.
5302         (mark_def_dom_walker): New class.
5303         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
5304         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
5305         (mark_def_sites_blocks): Convert to
5306         mark_def_dom_walker::before_dom_children.
5307         (mark_def_site_blocks): Remove.
5308         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
5309         (tree_ssa_dominator_optimize): Adjust.
5310         (dom_thread_across_edge): Convert to method
5311         dom_opt_dom_walker::thread_across_edge.
5312         (dom_opt_enter_block): Convert to member function
5313         dom_opt_dom_walker::before_dom_children.
5314         (dom_opt_leave_block): Convert to member function
5315         dom_opt_dom_walker::after_dom_children.
5316         * tree-ssa-dse.c (dse_dom_walker): New class.
5317         (dse_enter_block): Convert to member function
5318         dse_dom_walker::before_dom_children.
5319         (tree_ssa_dse): Adjust.
5320         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
5321         (determine_invariantness_stmt): Convert to method
5322         invariantness_dom_walker::before_dom_children.
5323         (determine_invariantness): Remove
5324         (move_computations_dom_walker): New class.
5325         (move_computations_stmt): Convert to method
5326         move_computations_dom_walker::before_dom_children.
5327         (move_computations, tree_ssa_lim): Adjust.
5328         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
5329         (nt_init_block): Convert to method
5330         notrappping_dom_walker::before_dom_children.
5331         (nt_fini_block): Convert to method
5332         method nontrapping_dom_walker::after_dom_children.
5333         (get_non_trapping): Adjust.
5334         * tree-ssa-pre.c (eliminate_dom_walker): New class.
5335         (eliminate_bb): Convert to method
5336         eliminate_dom_walker::before_dom_children.
5337         (eliminate_leave_block): Convert to method
5338         eliminate_dom_walker::after_dom_children.
5339         (eliminate): Adjust.
5340         * tree-ssa-strlen.c (strlen_dom_walker): New class.
5341         (strlen_enter_block): Convert to method
5342         strlen_dom_walker::before_dom_children.
5343         (strlen_leave_block): Convert to method
5344         method strlen_dom_walker::after_dom_children.
5345         (tree_ssa_strlen): Adjust.
5346         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
5347         (tree_ssa_uncprop): Adjust.
5348         (uncprop_leave_block): Convert to method
5349         uncprop_dom_walker::after_dom_children.
5350         (uncprop_leave_block): Convert to method
5351         uncprop_dom_walker::before_dom_children.
5353 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
5355         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
5356         branch in same basic block.  Check both src and dest of the move insn.
5358 2013-09-17  Nick Clifton  <nickc@redhat.com>
5360         * config/rl78/rl78-real.md (bf): New pattern.
5361         (bt): New pattern.
5362         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
5363         (rl78_print_operand): Do not put a # before a %B.
5364         * config/rl78/rl78.opt: Tweak doc strings.
5366 2013-09-17  DJ Delorie  <dj@redhat.com>
5368         * config/rl78/constraints.md (Wcv): Allow up to $r31.
5369         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
5370         (rl78_option_override): Likewise, if -mallregs.
5371         (is_virtual_register): Likewise.
5372         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
5373         (REGNO_OK_FOR_BASE_P): Likewise.
5374         * config/rl78/rl78.opt (-mallregs): New.
5376 2013-09-17  Nick Clifton  <nickc@redhat.com>
5378         * config/rl78/rl78.c (need_to_save): Change return type to bool.
5379         For interrupt functions: save all call clobbered registers if the
5380         interrupt handler is not a leaf function.
5381         (rl78_expand_prologue): Always recompute the frame information.
5382         For interrupt functions: only select bank 0 if one of the bank 0
5383         registers is going to be psuhed.
5385 2013-09-17  DJ Delorie  <dj@redhat.com>
5387         * config/rl78/constraints.md: For each W* constraint, rename to C*
5388         and create a W* constraint that checks for an optional ES: prefix
5389         pattern also.
5390         * config/rl78/rl78.md (UNS_ES_ADDR): New.
5391         (es_addr): New.  Used to wrap far addresses.
5392         * config/rl78/rl78-protos.h (rl78_es_addr): New.
5393         (rl78_es_base): New.
5394         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
5395         wrapped far addresses.
5396         (rl78_print_operand_1): Unwrap far addresses before processing.
5397         (rl78_lo16): Wrap far addresses in unspecs.
5398         (rl78_es_addr): New.
5399         (rl78_es_base): New.
5400         (insn_ok_now): Check for not-yet-wrapped far addresses.
5401         (transcode_memory_rtx): Properly re-wrap far addresses.
5403 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5405         * config/sparc/t-rtems: Add leon3 multilibs.
5407 2013-09-17  Cong Hou  <congh@google.com>
5409         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
5410         when checking the dot production pattern. The type of rhs operand
5411         of multiply is now checked correctly.
5413 2013-09-17  Jeff Law  <law@redhat.com>
5415         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5416         edge implied equivalences into successor phis.
5417         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
5418         here from tree-ssa-threadedge.c.
5419         (mark_threaded_blocks): When threading through a joiner, if both
5420         successors of the joiner's clone reach the same block, verify the
5421         PHI arguments are equal.  If not, cancel the jump threading request.
5422         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
5423         tree-ssa-threadupdate.c
5424         (thread_across_edge): Don't check PHI argument equality when
5425         threading through joiner block here.
5427 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
5429         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
5430         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
5431         (warn_uninit, warn_uninitialized_vars,
5432         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5433         Move to tree-ssa-uninit.c.
5434         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
5435         (has_undefined_value_p): New.  Pass dependant parts of
5436         ssa_undefined_value_p.
5437         (uninit_undefined_value_p): Use has_undefined_value_p.
5438         (warn_uninit, warn_uninitialized_vars,
5439         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5440         Move from tree-ssa.c.
5441         * tree-ssa.h: Adjust prototypes.
5443 2013-09-17  Jan Hubicka  <jh@suse.cz>
5445         PR middle-end/58332
5446         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
5447         * ipa-inline.c (can_inline_edge_p): Do not downgrade
5448         FUNCTION_NOT_OPTIMIZED.
5449         * ipa-inline-analysis.c (compute_inline_parameters): Function
5450         not optimized is not inlinable unless it is alwaysinline.
5451         (inline_analyze_function): Force calls in not optimized
5452         function not inlinable.
5454 2013-09-17  Jan Hubicka  <jh@suse.cz>
5456         PR middle-end/58329
5457         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
5458         to return NULL.
5459         * ipa.c (function_and_variable_visibility): Likewise.
5460         * ipa-profile.c (ipa_profile): Likewise.
5462 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5464         PR ipa/58398
5465         * cgraph.c (cgraph_function_body_availability): Check for ifunc
5466         attribute, and don't inline the resolver in this case.
5468 2013-09-17  Teresa Johnson  <tejohnson@google.com>
5470         * coverage.c (get_coverage_counts): Add missing newline.
5472 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5474         PR tree-optimization/58088
5475         * fold-const.c (mask_with_trailing_zeros): New function.
5476         (fold_binary_loc): Make sure we don't recurse infinitely
5477         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5478         Use mask_with_trailing_zeros where appropriate.
5480 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5482         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5483         of 'prev' var to get better distance estimation.
5485 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5487         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5488         * tree-inline.c (is_parameter_of): New predicate.
5489         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5490         a parameter has been remapped.
5491         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5492         (optimize_inline_calls): Initialize transform_parameter.
5493         (copy_gimple_seq_and_replace_locals): Likewise.
5494         (tree_function_versioning): Likewise.
5495         (maybe_inline_call_in_expr): Likewise.
5497 2013-09-17  Nick Clifton  <nickc@redhat.com>
5499         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5500         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5501         interrupt handlers.
5502         (is_attr_func): New function.
5503         (msp430_is_interrupt_func): New function.
5504         (is_naked_func): New function.
5505         (is_reentrant_func): New function.
5506         (is_critical_func): New function.
5507         (msp430_start_function): Add annotations for function attributes.
5508         (msp430_attr): New function.
5509         (msp430_attribute_table): New.
5510         (msp430_function_section): New function.
5511         (TARGET_ASM_FUNCTION_SECTION): Define.
5512         (msp430_builtin): New enum.
5513         (msp430_init_builtins): New function.
5514         (msp430_builtin_devl): New function.
5515         (msp430_expand_builtin): New function.
5516         (TARGET_INIT_BUILTINS): Define.
5517         (TARGET_EXPAND_BUILTINS): Define.
5518         (TARGET_BUILTIN_DECL): Define.
5519         (msp430_expand_prologue): Add support for naked, interrupt,
5520         critical and reentrant functions.
5521         (msp430_expand_epilogue): Likewise.
5522         (msp430_print_operand): Handle 'O' character.
5523         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5524         NO_TRAMPOLINES.
5525         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5526         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5527         (pushm): Use a 'n' rather than an 'i' constraint.
5528         (msp_return): Add generation of the interrupt return instruction.
5529         (disable_interrupts): New pattern.
5530         (enable_interrupts): New pattern.
5531         (push_intr_state): New pattern.
5532         (pop_intr_state): New pattern.
5533         (bic_SR): New pattern.
5534         (bis_SR): New pattern.
5535         * doc/extend.texi: Document MSP430 function attributes and builtin
5536         functions.
5538 2013-09-17  Richard Biener  <rguenther@suse.de>
5540         PR tree-optimization/58432
5541         * tree-loop-distribution.c (tree_loop_distribution): Also
5542         scan PHIs for outside loop uses and seed a partition from them.
5544 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5546         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5547         (restructure_reference): Call backtrace_base_for_ref.
5549 2013-09-17  Alan Modra  <amodra@gmail.com>
5551         PR target/57589
5552         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5553         patch.
5555 2013-09-16  DJ Delorie  <dj@redhat.com>
5557         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5558         vregs location for RL78/G10.
5559         (rl78_expand_prologue): Avoid SEL on G10.
5560         (rl78_expand_epilogue): Likewise.
5561         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5562         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5563         __RL78_G10__ when appropriate.
5564         (ASM_SPEC): Pass -mg10 along to the assembler.
5565         * config/rl78/rl78.md (sel_rb): Disable for G10.
5566         * config/rl78/rl78.opt: Add -mg10 option.
5567         * config/rl78/t-rl78: Add -mg10 multilib.
5569 2013-09-16  Xinliang David Li  <davidxl@google.com>
5571         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5572         * omp-low.c (omp_max_vf): Ditto.
5573         (expand_omp_simd): Ditto.
5574         * tree-vectorizer.c (vectorize_loops): Ditto.
5575         (gate_vect_slp): Ditto.
5576         (gate_increase_alignment): Ditto.
5577         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5578         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5579         (gate_tree_vectorize): Name change.
5580         (tree_vectorize): Ditto.
5581         (pass_vectorize::gate): Call new function.
5582         (pass_vectorize::execute): Ditto.
5583         * opts.c: O3 default setting change.
5584         (finish_options): Check new flag.
5585         * doc/invoke.texi: Document new flags.
5586         * common.opt: New flags.
5588 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5590         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5591         (Condition Code): Don't reference it.
5592         * doc/tm.texi: Regenerate.
5594 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5596         PR middle-end/58418
5597         * lra-constraints.c (undo_optional_reloads): Consider all optional
5598         reload even if it did not get a hard reg.
5600 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5602         * dumpfile.c (dump_loc): Remove newline emission.
5603         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5604         emission to dump_printf_loc calls where missing.
5605         (vect_mark_for_runtime_alias_test): Ditto.
5606         (vect_analyze_data_ref_dependence): Ditto.
5607         (vect_analyze_data_ref_dependences): Ditto.
5608         (vect_slp_analyze_data_ref_dependence): Ditto.
5609         (vect_slp_analyze_data_ref_dependences): Ditto.
5610         (vect_compute_data_ref_alignment): Ditto.
5611         (vect_update_misalignment_for_peel): Ditto.
5612         (vect_verify_datarefs_alignment): Ditto.
5613         (vector_alignment_reachable_p): Ditto.
5614         (vect_get_data_access_cost): Ditto.
5615         (vect_enhance_data_refs_alignment): Ditto.
5616         (vect_find_same_alignment_drs): Ditto.
5617         (vect_analyze_data_refs_alignment): Ditto.
5618         (vect_analyze_group_access): Ditto.
5619         (vect_analyze_data_ref_access): Ditto.
5620         (vect_analyze_data_ref_accesses): Ditto.
5621         (vect_prune_runtime_alias_test_list): Ditto.
5622         (vect_analyze_data_refs): Ditto.
5623         (vect_create_addr_base_for_vector_ref): Ditto.
5624         (vect_create_data_ref_ptr): Ditto.
5625         (vect_grouped_store_supported): Ditto.
5626         (vect_grouped_load_supported): Ditto.
5627         * value-prof.c (check_counter): Ditto.
5628         (check_ic_target): Ditto.
5629         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5630         (vect_recog_widen_mult_pattern): Ditto.
5631         (vect_recog_widen_sum_pattern): Ditto.
5632         (vect_recog_over_widening_pattern): Ditto.
5633         (vect_recog_widen_shift_pattern): Ditto.
5634         (vect_recog_rotate_pattern): Ditto.
5635         (vect_recog_vector_vector_shift_pattern): Ditto.
5636         (vect_recog_divmod_pattern): Ditto.
5637         (vect_recog_mixed_size_cond_pattern): Ditto.
5638         (vect_recog_bool_pattern): Ditto.
5639         (vect_pattern_recog_1): Ditto.
5640         (vect_pattern_recog): Ditto.
5641         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5642         (vect_is_simple_iv_evolution): Ditto.
5643         (vect_analyze_scalar_cycles_1): Ditto.
5644         (vect_get_loop_niters): Ditto.
5645         (vect_analyze_loop_1): Ditto.
5646         (vect_analyze_loop_form): Ditto.
5647         (vect_analyze_loop_operations): Ditto.
5648         (vect_analyze_loop_2): Ditto.
5649         (vect_analyze_loop): Ditto.
5650         (report_vect_op): Ditto.
5651         (vect_is_slp_reduction): Ditto.
5652         (vect_is_simple_reduction_1): Ditto.
5653         (vect_get_known_peeling_cost): Ditto.
5654         (vect_estimate_min_profitable_iters): Ditto.
5655         (vect_model_reduction_cost): Ditto.
5656         (vect_model_induction_cost): Ditto.
5657         (get_initial_def_for_induction): Ditto.
5658         (vect_create_epilog_for_reduction): Ditto.
5659         (vectorizable_reduction): Ditto.
5660         (vectorizable_induction): Ditto.
5661         (vectorizable_live_operation): Ditto.
5662         (vect_loop_kill_debug_uses): Ditto.
5663         (vect_transform_loop): Ditto.
5664         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5665         (vect_stmt_relevant_p): Ditto.
5666         (process_use): Ditto.
5667         (vect_mark_stmts_to_be_vectorized): Ditto.
5668         (vect_model_simple_cost): Ditto.
5669         (vect_model_promotion_demotion_cost): Ditto.
5670         (vect_model_store_cost): Ditto.
5671         (vect_get_store_cost): Ditto.
5672         (vect_model_load_cost): Ditto.
5673         (vect_get_load_cost): Ditto.
5674         (vect_init_vector_1): Ditto.
5675         (vect_get_vec_def_for_operand): Ditto.
5676         (vect_finish_stmt_generation): Ditto.
5677         (vectorizable_call): Ditto.
5678         (vectorizable_conversion): Ditto.
5679         (vectorizable_assignment): Ditto.
5680         (vectorizable_shift): Ditto.
5681         (vectorizable_operation): Ditto.
5682         (vectorizable_store): Ditto.
5683         (vectorizable_load): Ditto.
5684         (vectorizable_condition): Ditto.
5685         (vect_analyze_stmt): Ditto.
5686         (vect_transform_stmt): Ditto.
5687         (vect_is_simple_use): Ditto.
5688         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5689         (vect_can_advance_ivs_p): Ditto.
5690         (vect_update_ivs_after_vectorizer): Ditto.
5691         (vect_do_peeling_for_loop_bound): Ditto.
5692         (vect_gen_niters_for_prolog_loop): Ditto.
5693         (vect_update_inits_of_drs): Ditto.
5694         (vect_create_cond_for_alias_checks): Ditto.
5695         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5696         (vect_build_slp_tree_1): Ditto.
5697         (vect_supported_load_permutation_p): Ditto.
5698         (vect_analyze_slp_instance): Ditto.
5699         (vect_analyze_slp): Ditto.
5700         (vect_make_slp_decision): Ditto.
5701         (vect_detect_hybrid_slp): Ditto.
5702         (vect_bb_vectorization_profitable_p): Ditto.
5703         (vect_slp_analyze_bb_1): Ditto.
5704         (vect_update_slp_costs_according_to_vf): Ditto.
5705         (vect_get_mask_element): Ditto.
5706         (vect_transform_slp_perm_load): Ditto.
5707         (vect_schedule_slp_instance): Ditto.
5708         (vect_schedule_slp): Ditto.
5709         (vect_slp_transform_bb): Ditto.
5710         * profile.c (read_profile_edge_counts): Ditto.
5711         (compute_branch_probabilities): Ditto.
5712         * coverage.c (get_coverage_counts): Ditto.
5714 2013-09-16  Diego Novillo  <dnovillo@google.com>
5716         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5718 2013-09-16  Jan Hubicka  <jh@suse.cz>
5720         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5721         abstract functions; for static functions check the presence of body.
5723 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5725         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5726         * config/aarch64/aarch64-simd.md
5727         (aarch64_mla_elt<mode>): New.
5728         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5729         (aarch64_mls_elt<mode>): Likewise.
5730         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5731         (aarch64_fma4_elt<mode>): Likewise.
5732         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5733         (aarch64_fma4_elt_to_128v2df): Likewise.
5734         (aarch64_fma4_elt_to_64df): Likewise.
5735         (fnma<mode>4): Likewise.
5736         (aarch64_fnma4_elt<mode>): Likewise.
5737         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5738         (aarch64_fnma4_elt_to_128v2df): Likewise.
5739         (aarch64_fnma4_elt_to_64df): Likewise.
5740         * config/aarch64/iterators.md (VDQSF): New.
5741         * config/aarch64/arm_neon.h
5742         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5743         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5745 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5747         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5748         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5749         (aarch64_mul3_elt_to_128df): Likewise.
5750         (aarch64_mul3_elt_to_64v2df): Likewise.
5751         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5752         (VMUL): New.
5753         (VMUL_CHANGE_NLANES) Likewise.
5754         (h_con): Likewise.
5755         (f): Likewise.
5756         * config/aarch64/arm_neon.h
5757         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5759 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5761         * config/aarch64/arm_neon.h
5762         (vcvtx_high_f32_f64): Fix parameters.
5764 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5765             Uros Bizjak  <ubizjak@gmail.com>
5767         * config/alpha.c: Include tree-ssanames.h.
5769 2013-09-16  Richard Biener  <rguenther@suse.de>
5771         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5772         (dot_rdg_1): Handle control_dd.
5773         (create_edge_for_control_dependence): New function.
5774         (create_rdg_edges): Add control dependences if asked for.
5775         (build_rdg): Likewise.
5776         (generate_loops_for_partition): If there are not necessary
5777         control stmts remove all their dependencies.
5778         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5779         (distribute_loop): Pass on control dependences.
5780         (tree_loop_distribution): Compute control dependences and remove
5781         restriction on number of loop nodes.
5783 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5785         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5786         for internal calls.
5788 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5790         * cse.c (try_const_anchors): Punt on CC modes.
5792 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5794         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5796 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5798         PR target/58382
5799         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5800         calls to word_mode.
5802 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5804         PR target/48094
5805         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5806         seen.
5807         (darwin_objc1_section): Likewise.
5808         (darwin_file_end): Emit Image Info section when required.
5810 2013-09-14  Jan Hubicka  <jh@suse.cz>
5812         * tree-into-ssa.c (gate_into_ssa): New.
5813         (pass_data_build_ssa): Use it.
5814         * cgraph.h (expand_thunk): Update prototype.
5815         * cgraphunit.c (analyze_function): Expand thunks early.
5816         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5817         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5818         set lowered flag; do not add new function.
5819         (assemble_thunks_and_aliases): Update.
5820         * tree-ssa.c (gate_init_datastructures): New gate.
5821         (pass_data_init_datastructures): Use it.
5823 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5825         PR target/58269
5826         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5827         xmm register set described in the psABI.
5829 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5831         * dwarf2out.c (should_emit_struct_debug): Add check
5832         for type_decl variable is not NULL.
5834 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5836         * config.gcc: Use new winnt-c.c target hooks
5837         * config/t-winnt: New file
5838         * config/winnt-c.c: New file
5839         * doc/tm.texi.in: Document new hook
5840         * doc/tm.texi: Regenerated
5842 2013-09-13  Jan Hubicka  <jh@suse.cz>
5844         PR middle-end/58094
5845         * ipa-inline.c (check_callers): New function.
5846         (check_caller_edge): Remove.
5847         (want_inline_function_to_all_callers_p): Also permit alises that are
5848         called dirrectly.
5849         (inline_to_all_callers): Terminate the walk when devirtualization
5850         introduce new calls.
5852 2013-09-13  Jan Hubicka  <jh@suse.cz>
5854         * ipa-inline-analysis.c (struct growth_data): Add node.
5855         (do_estimate_growth_1): Fix detection of recursion.
5857 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5859         PR tree-optimization/58392
5860         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5861         to avoid shadowing of outer loop variable.  If
5862         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5863         replace_by_duplicate_decl simduid of loops that have it set and
5864         set dest_cfun->has_simduid_loops and/or
5865         dest_cfun->has_force_vect_loops.
5866         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5867         instead of maybe_lookup_decl.
5868         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5869         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5870         call.  Copy over force_vect and copy and remap simduid.  Set
5871         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5872         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5873         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5874         caller.  Don't set cfun->has_simduid_loops and/or
5875         cfun->has_force_vect_loops here.
5876         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5877         caller.
5878         (expand_call_inline, tree_function_versioning): Adjust copy_body
5879         callers.
5881 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5883         PR bootstrap/58388
5884         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5885         the assert if the edge was a speculative one.
5887 2013-09-13  Richard Biener  <rguenther@suse.de>
5889         * tree-data-ref.h (known_dependences_p): Move here ...
5890         * tree-loop-distribution.c (known_dependences_p): ... from here.
5891         (dump_rdg_component, debug_rdg_component): Remove.
5892         (dump_rdg): Adjust.
5893         (generate_loops_for_partition): Use gimple_uid instead of
5894         relying on matching stmt visit order.
5895         (rdg_build_partitions): Take starting stmt vector.
5896         (ldist_gen): Merge into ...
5897         (distribute_loop): ... this function.  Do not compute starting
5898         vertices vector.
5899         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5901 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5903         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5904         Set type attribute correctly. Set predicable_short_it attribute.
5905         (cmpsi_shiftsi): Remove %? from output template.
5907 2013-09-13  Richard Biener  <rguenther@suse.de>
5909         * tree-loop-distribution.c (struct rdg_component,
5910         rdg_defs_used_in_other_loops_p, free_rdg_components,
5911         rdg_build_components): Remove.
5912         (stmts_from_loop): Do not record virtual PHIs.
5913         (generate_loops_for_partition): Skip virtual PHIs.
5914         (build_rdg_partition_for_component): Rename to ...
5915         (build_rdg_partition_for_vertex): ... this and adjust.
5916         (rdg_build_partitions): Take a vector of starting vertices
5917         instead of components.  Remove unnecessary leftover handling.
5918         (ldist_gen): Do not build components or record other stores.
5919         (distribute_loop): Do not distribute loops containing stmts
5920         with side-effects.
5922 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5924         PR target/58314
5925         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5927 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5929         * config.gcc: Separate cases for mingw and cygwin targets,
5930         and add 64-bit cygwin target case.
5932         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5933         dll-export inline-functions.
5934         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5937 2013-09-13  Jeff Law  <law@redhat.com>
5939         PR middle-end/58387
5940         Revert:
5941         2013-09-06  Jeff Law  <law@redhat.com>
5943         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5944         edge implied equivalences into successor phis.
5946 2013-09-12  DJ Delorie  <dj@redhat.com>
5948         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5949         separators.
5951 2013-09-12  Brooks Moses  <bmoses@google.com>
5953         PR driver/42955
5954         * Makefile.in: Do not install driver binaries in $(target)/bin.
5956 2013-09-12  DJ Delorie  <dj@redhat.com>
5958         * config/rl78/rl78.opt (mrelax): New.
5959         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5960         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5962         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5963         between SP and FP.
5964         (rl78_expand_epilogue): Likewise.
5966 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5968         PR middle-end/58335
5969         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5970         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5971         pointer elimination with using remove_reg_equal_offset_note.
5973 2013-09-12  DJ Delorie  <dj@redhat.com>
5975         * config/msp430/: New port.
5976         * config.gcc (msp430): Added.
5977         * doc/invoke.texi: Document MSP430 options.
5978         * doc/install.texi: Document msp430-elf
5979         * doc/md.texi: Document msp430-elf
5980         * doc/contrib.texi: Document msp430-elf
5982         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5983         PSImode.
5985 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5987         PR ipa/58389
5988         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5989         reference descriptor is NULL.
5990         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5991         edge to NULL.
5993 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5995         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5996         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5997         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5998         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5999         make_temp_ssa_name): move to tree-ssanames.h
6000         * tree-ssa-alias.h: Move prototype.
6001         * tree-ssa.h: Include tree-ssanames.h.
6002         * tree-ssanames.c (FREE_SSANAMES): Move to here.
6003         * tree-ssanames.h: New.  Move items from tree-flow*.h
6004         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
6006 2013-09-12  Richard Biener  <rguenther@suse.de>
6008         PR tree-optimization/58404
6009         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
6010         propagate non-invariant addresses into dereferences wrapped
6011         in component references.
6013 2013-09-12  Richard Biener  <rguenther@suse.de>
6015         PR tree-optimization/58402
6016         * passes.def: Move pass_late_warn_uninitialized later.
6018 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
6020         * tree-ssa.h: New.  Move content from tree-flow.h and
6021         tree-flow-inline.h.
6022         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
6023         Move prototypes belonging to tree-ssa.c.
6024         * tree-flow-inline.h (redirect_edge_var_map_def,
6025         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
6026         tree-ssa.h.
6027         * gimple.h: Adjust prototypes.
6028         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
6029         to...
6030         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
6031         * tree.h: Move prototype to tree-ssa.h.
6032         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
6033         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
6034         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
6035         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
6036         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
6037         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
6038         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
6039         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
6040         graphite-blocking.c, graphite-clast-to-gimple.c,
6041         graphite-dependences.c, graphite-interchange.c,
6042         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
6043         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
6044         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
6045         ipa-reference.c, ipa-split.c, ipa-utils.c,
6046         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
6047         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
6048         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
6049         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
6050         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
6051         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
6052         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
6053         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
6054         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
6055         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
6056         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
6057         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
6058         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
6059         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
6060         value-prof.c, var-tracking.c,
6061         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
6063 2013-09-12  Richard Biener  <rguenther@suse.de>
6065         PR tree-optimization/58396
6066         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
6067         (build_rdg): Take a loop-nest parameter, fix memleaks.
6068         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
6070 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
6072         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
6073         for SLM.
6075 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
6077         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
6078         The FMA instruction names should have a 'v' prefix.
6080 2013-09-12  Richard Biener  <rguenther@suse.de>
6082         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
6083         (dot_rdg): Use popen instead of system in optional code.
6084         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
6085         (already_processed_vertex_p): Adjust.
6086         (has_anti_or_output_dependence, predecessor_has_mem_write,
6087         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
6088         rdg_flag_uses): Remove.
6089         (rdg_flag_vertex): Simplify.
6090         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
6091         remove recursion.
6092         (build_rdg_partition_for_component): Process the first vertex
6093         of a component only.
6094         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
6096 2013-09-12  Alan Modra  <amodra@gmail.com>
6098         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
6100 2013-09-11  DJ Delorie  <dj@redhat.com>
6101             Nick Clifton  <nickc@redhat.com>
6103         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
6104         (rl78_stack_based_mem): New.
6105         * config/rl78/constraints.md (Iv08): New.
6106         (Iv16): New.
6107         (Iv24): New.
6108         (Is09): New.
6109         (Is17): New.
6110         (Is25): New.
6111         (ISsi): New.
6112         (IShi): New.
6113         (ISqi): New.
6114         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
6115         (movhi): Likewise.
6116         (movsi): Change from expand to insn-and-split.
6117         (ashrsi3): Clobber AX.
6118         (lshrsi3): New.
6119         (ashlsi3): New.
6120         (cbranchsi4): New.
6121         * config/rl78/rl78.md (CC_REG): Fix.
6122         (addsi3): Allow memory and immediate operands.
6123         (addsi3_internal): Split into...
6124         (addsi3_internal_virt): ...new, and ...
6125         (addsi3_internal_real): ...new.
6126         (subsi): New.
6127         (subsi3_internal_virt): New.
6128         (subsi3_internal_real): New.
6129         (mulsi3): Add memory operand.
6130         (mulsi3_rl78): Likewise.
6131         (mulsi3_g13): Likewise.
6132         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
6133         (cbranchqi4_real): Add more constraint options.
6134         (cbranchhi4_real): Expand pattern.
6135         (cbranchhi4_real_signed): New.
6136         (cbranchhi4_real_inverted): New.
6137         (cbranchsi4_real_lt): New.
6138         (cbranchsi4_real_ge): New.
6139         (cbranchsi4_real_signed): New.
6140         (cbranchsi4_real): New.
6141         (peephole2): New.
6142         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
6143         constant shifts.
6144         (lshrsi3_virt): Likewise.
6145         (ashlsi3_virt): Likewise.
6146         (cbranchqi4_virt_signed): New.
6147         (cbranchhi4_virt_signed): New.
6148         (cbranchsi4_virt): New.
6149         * config/rl78/rl78.c: Whitespace fixes throughout.
6150         (move_elim_pass): New.
6151         (pass_data_rl78_move_elim): New.
6152         (pass_rl78_move_elim): New.
6153         (make_pass_rl78_move_elim): New.
6154         (rl78_devirt_info): Run devirt earlier.
6155         (rl78_move_elim_info): New.
6156         (rl78_asm_file_start): Register it.
6157         (rl78_split_movsi): New.
6158         (rl78_as_legitimate_address): Allow virtual base registers when
6159         appropriate.
6160         (rl78_addr_space_convert): Remove spurious debug stuff.
6161         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
6162         (rl78_print_operand): More cases for not printing '#'.
6163         (rl78_expand_compare): Remove most of the logic.
6164         (content_memory): New.
6165         (clear_content_memory): New.
6166         (get_content_index): New.
6167         (get_content_name): New.
6168         (display_content_memory): New.
6169         (update_content): New.
6170         (record_content): New.
6171         (already_contains): New.
6172         (insn_ok_now): Re-recog insns with virtual registers.
6173         (add_postponed_content_update): New.
6174         (process_postponed_content_update): New.
6175         (gen_and_emit_move): New.
6176         (transcode_memory_rtx): Record new location content.
6177         Use gen_and_emit_move.
6178         (force_into_acc): New.
6179         (move_to_acc): Use gen_and_emit_move.
6180         (move_from_acc): Likewise.
6181         (move_acc_to_reg): Likewise.
6182         (move_to_x): Likewise.
6183         (move_to_hl): Likewise.
6184         (move_to_de): Likewise.
6185         (rl78_alloc_physical_registers_op1): Record location content.
6186         (has_constraint): New.
6187         (rl78_alloc_physical_registers_op2): Record location content.
6188         Optimize use of HL.
6189         (rl78_alloc_physical_registers_ro1): Likewise.
6190         (rl78_alloc_physical_registers_cmp): Likewise.
6191         (rl78_alloc_physical_registers_umul): Likewise.
6192         (rl78_alloc_address_registers_macax): New.
6193         (rl78_alloc_physical_registers): Initialize and set location
6194         content memory as needed.
6195         (rl78_reorg): Make sure split2 is called.
6196         (rl78_rtx_costs): New.
6198 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
6200         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
6201         for (not (neg ...)) and (neg (not ...)) cases.
6203 2013-09-11  Richard Biener  <rguenther@suse.de>
6205         PR middle-end/58377
6206         * passes.def: Split critical edges before late uninit warning passes.
6207         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
6209 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
6211         PR tree-optimization/58385
6212         * fold-const.c (build_range_check): If both low and high are NULL,
6213         use omit_one_operand_loc to preserve exp side-effects.
6215 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6217         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
6219 2013-09-11  Richard Biener  <rguenther@suse.de>
6221         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
6222         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
6223         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
6224         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
6225         stmts_from_loop, known_dependences_p, build_empty_rdg,
6226         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
6227         * tree-loop-distribution.c: ... here.
6228         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
6229         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
6230         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
6231         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
6232         * tree-loop-distribution.c: ... here.
6233         * tree-loop-distribution.c: Include gimple-pretty-print.h.
6234         (struct partition_s): Add loops member.
6235         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
6236         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
6237         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
6239 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6240             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
6241             Sergey Lega  <sergey.s.lega@intel.com>
6242             Anna Tikhonova  <anna.tikhonova@intel.com>
6243             Ilya Tocar  <ilya.tocar@intel.com>
6244             Andrey Turetskiy  <andrey.turetskiy@intel.com>
6245             Ilya Verbin  <ilya.verbin@intel.com>
6246             Kirill Yukhin  <kirill.yukhin@intel.com>
6247             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
6249         * config/i386/constraints.md (k): New.
6250         (Yk): Ditto.
6251         * config/i386/i386.c (const regclass_map): Add new mask registers.
6252         (dbx_register_map): Ditto.
6253         (dbx64_register_map): Ditto.
6254         (svr4_dbx_register_map): Ditto.
6255         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
6256         disabled.
6257         (ix86_preferred_reload_class): Disable constants for mask registers.
6258         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
6259         (ix86_hard_regno_mode_ok): Support new mask registers.
6260         (x86_order_regs_for_local_alloc): Ditto.
6261         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
6262         (FIXED_REGISTERS): Add new mask registers.
6263         (CALL_USED_REGISTERS): Ditto.
6264         (REG_ALLOC_ORDER): Ditto.
6265         (VALID_MASK_REG_MODE): New.
6266         (FIRST_MASK_REG): Ditto.
6267         (LAST_MASK_REG): Ditto.
6268         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
6269         (MAYBE_MASK_CLASS_P): New.
6270         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
6271         (REG_CLASS_CONTENTS): Ditto.
6272         (MASK_REGNO_P): New.
6273         (ANY_MASK_REG_P): Ditto.
6274         (HI_REGISTER_NAMES): Add new mask registers.
6275         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
6276         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
6277         mask registers.
6278         (attribute "type"): Add mskmov, msklog.
6279         (attribute "length_immediate"): Support them.
6280         (attribute "memory"): Ditto.
6281         (attribute "prefix_0f"): Ditto.
6282         (*movhi_internal): Support new mask registers.
6283         (*movqi_internal): Ditto.
6284         (define_split): Split out clobber pattern is a logic
6285         insn on mask registers.
6286         (*k<logic><mode>): New.
6287         (*andhi_1): Extend to support mask regs.
6288         (*andqi_1): Extend to support mask regs.
6289         (kandn<mode>): New.
6290         (define_split): Split and-not to and and not if operands
6291         are not mask regs.
6292         (*<code><mode>_1): Separate HI mode to new pattern...
6293         (*<code>hi_1): This.
6294         (*<code>qi_1): Extend to support mask regs.
6295         (kxnor<mode>): New.
6296         (kortestzhi): Ditto.
6297         (kortestchi): Ditto.
6298         (kunpckhi): Ditto.
6299         (*one_cmpl<mode>2_1): Remove HImode and handle it...
6300         (*one_cmplhi2_1): ...Here, now with mask registers support.
6301         (*one_cmplqi2_1): Support new mask registers.
6302         (HI/QImode arithmetics splitter): Don't split if mask registers
6303         are used.
6304         (HI/QImode not splitter): Ditto.
6305         * config/i386/predicated.md (mask_reg_operand): New.
6306         (general_reg_operand): Ditto.
6308 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6310         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
6311         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
6313 2013-09-10  Jeff Law  <law@redhat.com>
6315         PR tree-optimization/58380
6316         * tree-ssa-threadupdate.c (thread_block): Recognize another case
6317         of threading through a buried loop header.
6319         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
6320         return value for single successor case.
6322 2013-09-10  Jan Hubicka  <jh@suse.cz>
6324         * ipa-devirt.c (ipa_devirt): Enable with LTO.
6326 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
6328         PR target/58361
6329         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
6330         support conditional execution.
6331         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
6333 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
6335         * lra.c (lra): Clear lra_optional_reload_pseudos before every
6336         constraint pass.
6337         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
6338         Check destination too to check move insn.
6339         (undo_optional_reloads): Add check that the original peudo did not
6340         changed its allocation and the optional reload was inherited on last
6341         inheritance pass.  Break loop after deciding to keep optional reload.
6342         (lra_undo_inheritance): Add check that inherited pseudo still in
6343         memory.
6345 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
6347         * config/aarch64/aarch64.md (generic_sched): New.
6348         * config/aarch64/aarch64-generic.md (load): Make conditional
6349         on generic_sched attribute.
6350         (nonload): Likewise.
6352 2013-09-10  Jan Hubicka  <jh@suse.cz>
6354         * lto-cgraph.c: Include ipa-utils.h.
6355         (compute_ltrans_boundary): Also add possible targets into the boundary.
6357 2013-09-10  Jan Hubicka  <jh@suse.cz>
6359         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
6360         VAR_DECL of vtable rather than full expression.
6362 2013-09-10  Jan Hubicka  <jh@suse.cz>
6363             Paolo Carlini  <paolo.carlini@oracle.com>
6365         * cgraphunit.c (analyze_functions): Save input_location, set it
6366         to UNKNOWN_LOCATION and restore it at the end.
6368 2013-09-10  Martin Jambor  <mjambor@suse.cz>
6370         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
6371         represented by a thunk.
6373 2013-09-10  Jeff Law  <law@redhat.com>
6375         PR tree-optimization/58343
6376         * tree-ssa-threadupdate.c (thread_block): Identify and disable
6377         jump threading requests through loop headers buried in the middle
6378         of a jump threading path.
6380         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
6381         in return value/type.
6383 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
6385         PR rtl-optimization/58365
6386         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
6387         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
6388         it differs.
6390 2013-09-10  Richard Biener  <rguenther@suse.de>
6392         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
6393         * tree-data-ref.c (create_rdg_vertices): Collect all data
6394         references, signal failure to the caller, use data-ref API.
6395         (build_rdg): Compute data references only once.  Maintain lifetime
6396         of data references and data dependences from within RDG.
6397         (free_rdg): Free dependence relations.
6398         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
6399         inventing extra dependences.
6400         (distribute_loop): Update for RDG API changes.
6402 2013-09-10  Kai Tietz  <ktietz@redhat.com>
6404         * doc/invoke.texi (fms-extensions): Document changed
6405         behavior for ms-abi targets.
6406         * config/i386/i386.c (ix86_option_override_internal):
6407         Set default value of option -fms-extension for ms-abi targets.
6409 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
6411         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
6413 2013-09-10  Alan Modra  <amodra@gmail.com>
6415         PR target/58330
6416         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
6418 2013-09-10  Alan Modra  <amodra@gmail.com>
6420         * config/rs6000/predicates.md (add_cint_operand): New.
6421         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
6422         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
6423         using add_cint_operand.
6424         (largetoc_high_plus_aix): Likewise.
6426 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
6428         PR tree-optimization/58364
6429         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
6430         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
6431         the current range can't be an unconditional true or false.
6433 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
6435         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
6437 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
6439         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
6441 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
6443         PR c++/43452
6444         * doc/invoke.texi (-Wdelete-incomplete): Document it.
6446 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
6448         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
6449         NO_REGS for immediate that can't be moved directly into FP_REGS.
6451 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6453         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6454         comparison with negated operand.
6455         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
6456         RTL form.
6458 2013-09-09  Richard Biener  <rguenther@suse.de>
6460         PR middle-end/58326
6461         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
6462         of a subloop record all its block as affecting loop-closed SSA form.
6464 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6466         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
6467         of an rtx/bitpos pair.
6468         (store_fixed_bit_field): Update accordingly.
6470 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6472         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
6473         GEN_INT.
6474         * builtins.c (expand_errno_check): Likewise.
6475         * dwarf2cfi.c (init_return_column_size): Likewise.
6476         * except.c (sjlj_mark_call_sites): Likewise.
6477         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6478         * lra-constraints.c (emit_inc): Likewise.
6479         * ree.c (combine_set_extension): Likewise.
6480         * regmove.c (fixup_match_2): Likewise.
6481         * reload1.c (inc_for_reload): Likewise.
6483 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6485         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6486         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6487         Use gen_int_mode with the mode of the associated simplify_* call.
6488         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6489         * expmed.c (expand_shift_1): Likewise.
6490         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6491         * loop-iv.c (iv_number_of_iterations): Likewise.
6492         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6493         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6495 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6497         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6498         of the associated expand_* call.
6499         (asan_emit_stack_protection): Likewise.
6500         * builtins.c (round_trampoline_addr): Likewise.
6501         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6502         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6503         (emit_store_flag): Likewise.
6504         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6505         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6506         Likewise.
6507         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6508         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6509         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6510         * modulo-sched.c (generate_prolog_epilog): Likewise.
6511         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6512         (expand_ctz, expand_ffs, expand_unop): Likewise.
6514 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6516         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6517         of the associated gen_rtx_* call.
6518         * caller-save.c (init_caller_save): Likewise.
6519         * combine.c (find_split_point, make_extraction): Likewise.
6520         (make_compound_operation): Likewise.
6521         * dwarf2out.c (mem_loc_descriptor): Likewise.
6522         * explow.c (plus_constant, probe_stack_range): Likewise.
6523         * expmed.c (expand_mult_const): Likewise.
6524         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6525         * reload1.c (init_reload): Likewise.
6526         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6527         * var-tracking.c (adjust_mems): Likewise.
6528         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6529         rather than gen_rtx_fmt_ee.
6531 2013-09-09  Jan Hubicka  <jh@suse.cz>
6533         PR middle-end/58294
6534         * value-prof.c (gimple_ic): Copy also abnormal edges.
6536 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6538         * asan.c (asan_shadow_cst): Use gen_int_mode.
6540 2013-09-08  Jan Hubicka  <jh@suse.cz>
6542         * ipa-profile.c: Add toplevel comment.
6543         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6544         (contains_hot_call_p): New function.
6545         (ipa_propagate_frequency): Set frequencies based on counts when
6546         profile is read.
6547         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6548         profile; do not tamper with profile after inlining if it is read.
6550 2013-09-08  Jan Hubicka  <jh@suse.cz>
6552         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6553         speculative edges.
6555 2013-09-08  Jan Hubicka  <jh@suse.cz>
6557         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6558         summary generation.
6560 2013-09-08  Jeff Law  <law@redhat.com>
6562         PR bootstrap/58340
6563         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6564         of 'found'.
6566 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6568         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6570 2013-09-08  Jan Hubicka  <jh@suse.cz>
6572         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6574 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6576         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6577         for non-debug insns.
6578         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6579         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6580         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6581         accordingly.
6583 2013-09-08  Jan Hubicka  <jh@suse.cz>
6585         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6586         targets and devirtualize to BUILT_IN_UNREACHABLE.
6587         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6588         * ipa.c (walk_polymorphic_call_targets): New function.
6589         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6590         functions; use the new timevar.
6591         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6592         was removed from the program.
6593         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6594         not consider it in the walk when its vtable is dead.
6595         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6596         record_binfo.
6597         (devirt_variable_node_removal_hook): New function.
6598         (possible_polymorphic_call_targets): Also register
6599         devirt_variable_node_removal_hook.
6600         (ipa_devirt): Do not do non-speculative devirtualization.
6601         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6603 2013-09-08  Jan Hubicka  <jh@suse.cz>
6605         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6606         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6607         varpool_remove_variable_insertion_hook): Declare.
6608         * varpool.c (varpool_node_hook_list): New structure.
6609         (first_varpool_node_removal_hook,
6610         first_varpool_variable_insertion_hook): New variables.
6611         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6612         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6613         varpool_remove_variable_insertion_hook,
6614         varpool_call_variable_insertion_hooks): New functions.
6615         (varpool_remove_node): Use it.
6617 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6619         PR c++/54941
6620         * diagnostic.c (diagnostic_build_prefix): When s.file is
6621         "<built-in>" don't output line and column numbers.
6623 2013-09-06  Jan Hubicka  <jh@suse.cz>
6625         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6626         * lto-streamer-out.c: Stream thunks, too.
6627         * lto-streamer-in.c (input_function): Pop cfun here
6628         (lto_read_body): Instead of here.
6630 2013-09-06  Caroline Tice  <cmtice@google.com>
6632         * doc/install.texi: Add documentation for the --enable-vtable-verify
6633         and the --disable-libvtv configure options.
6635 2013-09-06  Jeff Law  <law@redhat.com>
6637         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6638         edge implied equivalences into successor phis.
6640 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6642         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6644 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6646         * resource.c (mark_target_live_regs): Compute resources taking
6647         into account if a call is predicated or not.
6649 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6651         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6652         the compiler in the function names.
6654 2013-09-06  Jan Hubicka  <jh@suse.cz>
6656         PR middle-end/58094
6657         * ipa-inline.c (has_caller_p): New function.
6658         (want_inline_function_to_all_callers_p): Use it.
6659         (sum_callers, inline_to_all_callers): Break out from ...
6660         (ipa_inline): ... here.
6662 2013-09-06  Jan Hubicka  <jh@suse.cz>
6664         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6665         only when AVX is enabled.
6667 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6669         * config/aarch64/aarch64.md
6670         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6671         is fpsimd_<load/store>2.
6672         (load_pair<mode>): Likewise.
6673         (store_pair<mode>): Likewise.
6675 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6677         * config/arm/types.md (type): Add "mrs" type.
6678         * config/aarch64/aarch64.md
6679         (aarch64_load_tp_hard): Make type "mrs".
6680         * config/arm/arm.md
6681         (load_tp_hard): Make type "mrs".
6682         * config/arm/cortex-a15.md: Update with new attributes.
6683         * config/arm/cortex-a5.md: Update with new attributes.
6684         * config/arm/cortex-a53.md: Update with new attributes.
6685         * config/arm/cortex-a7.md: Update with new attributes.
6686         * config/arm/cortex-a8.md: Update with new attributes.
6687         * config/arm/cortex-a9.md: Update with new attributes.
6688         * config/arm/cortex-m4.md: Update with new attributes.
6689         * config/arm/cortex-r4.md: Update with new attributes.
6690         * config/arm/fa526.md: Update with new attributes.
6691         * config/arm/fa606te.md: Update with new attributes.
6692         * config/arm/fa626te.md: Update with new attributes.
6693         * config/arm/fa726te.md: Update with new attributes.
6695 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6697         * config/aarch64/aarch64.md
6698         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6699         (*movtf_aarch64): Likewise.
6700         * config/arm/arm.md
6701         (thumb1_movdi_insn): Use "multiple" for type where more than one
6702         instruction is used for a move.
6703         (*arm32_movhf): Likewise.
6704         (*thumb_movdf_insn): Likewise.
6706 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6708         * config/arm/types.md (type): Rename fcpys to fmov.
6709         * config/arm/vfp.md
6710         (*arm_movsi_vfp): Rename type fcpys as fmov.
6711         (*thumb2_movsi_vfp): Likewise
6712         (*movhf_vfp_neon): Likewise
6713         (*movhf_vfp): Likewise
6714         (*movsf_vfp): Likewise
6715         (*thumb2_movsf_vfp): Likewise
6716         (*movsfcc_vfp): Likewise
6717         (*thumb2_movsfcc_vfp): Likewise
6718         * config/aarch64/aarch64-simd.md
6719         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6720         * config/aarch64/aarch64.md
6721         (*movsi_aarch64): Replace type mov_reg with fmovs.
6722         (*movdi_aarch64): Likewise
6723         (*movsf_aarch64): Likewise
6724         (*movdf_aarch64): Likewise
6725         * config/arm/arm.c
6726         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6727         * config/arm/iwmmxt.md
6728         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6729         * config/arm/arm1020e.md: Update with new attributes.
6730         * config/arm/cortex-a15-neon.md: Update with new attributes.
6731         * config/arm/cortex-a5.md: Update with new attributes.
6732         * config/arm/cortex-a53.md: Update with new attributes.
6733         * config/arm/cortex-a7.md: Update with new attributes.
6734         * config/arm/cortex-a8-neon.md: Update with new attributes.
6735         * config/arm/cortex-a9.md: Update with new attributes.
6736         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6737         * config/arm/cortex-r4f.md: Update with new attributes.
6738         * config/arm/marvell-pj4.md: Update with new attributes.
6739         * config/arm/vfp11.md: Update with new attributes.
6741 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6743         * config/aarch64/aarch64.md
6744         (*madd<mode>): Fix type attribute.
6745         (*maddsi_uxtw): Likewise.
6746         (*msub<mode>): Likewise.
6747         (*msubsi_uxtw): Likewise.
6748         (<su_optab>maddsidi4): Likewise.
6749         (<su_optab>msubsidi4): Likewise.
6751 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6753         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6754         * config/arm/arm.md (core_cycles): Remove fdiv.
6755         * config/arm/vfp.md:
6756         (*sqrtsf2_vfp): Update for attribute changes.
6757         (*sqrtdf2_vfp): Likewise.
6758         * config/aarch64/aarch64.md:
6759         (sqrt<mode>2): Update for attribute changes.
6760         * config/arm/arm1020e.md: Update with new attributes.
6761         * config/arm/cortex-a15-neon.md: Update with new attributes.
6762         * config/arm/cortex-a5.md: Update with new attributes.
6763         * config/arm/cortex-a53.md: Update with new attributes.
6764         * config/arm/cortex-a7.md: Update with new attributes.
6765         * config/arm/cortex-a8-neon.md: Update with new attributes.
6766         * config/arm/cortex-a9.md: Update with new attributes.
6767         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6768         * config/arm/cortex-r4f.md: Update with new attributes.
6769         * config/arm/marvell-pj4.md: Update with new attributes.
6770         * config/arm/vfp11.md: Update with new attributes.
6772 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6774         * config/arm/types.md
6775         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6776         * config/aarch64/aarch64.md
6777         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6778         new attributes.
6779         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6780         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6781         (float<GPI:mode><GPF:mode>2): Likewise.
6782         * config/arm/vfp.md
6783         (*truncsisf2_vfp): Update with new attributes.
6784         (*truncsidf2_vfp): Likewise.
6785         (fixuns_truncsfsi2): Likewise.
6786         (fixuns_truncdfsi2): Likewise.
6787         (*floatsisf2_vfp): Likewise.
6788         (*floatsidf2_vfp): Likewise.
6789         (floatunssisf2): Likewise.
6790         (floatunssidf2): Likewise.
6791         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6792         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6793         * config/arm/arm1020e.md: Update with new attributes.
6794         * config/arm/cortex-a15-neon.md: Update with new attributes.
6795         * config/arm/cortex-a5.md: Update with new attributes.
6796         * config/arm/cortex-a53.md: Update with new attributes.
6797         * config/arm/cortex-a7.md: Update with new attributes.
6798         * config/arm/cortex-a8-neon.md: Update with new attributes.
6799         * config/arm/cortex-a9.md: Update with new attributes.
6800         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6801         * config/arm/cortex-r4f.md: Update with new attributes.
6802         * config/arm/marvell-pj4.md: Update with new attributes.
6803         * config/arm/vfp11.md: Update with new attributes.
6805 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6807         * config/aarch64/arm_neon.h
6808         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6809         (vqtbx<1,2,3,4><q>_s8): Likewise.
6811 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6813         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6814         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6815         (add<mode>3): Add type attribute.
6816         (add<mode>3): Likewise.
6817         (usadd<mode>3): Likewise.
6818         (ssadd<mode>3): Likewise.
6819         (sub<mode>3): Likewise.
6820         (sub<mode>3): Likewise.
6821         (ussub<mode>3): Likewise.
6822         (sssub<mode>3): Likewise.
6823         (ssmulsa3): Likewise.
6824         (usmulusa3): Likewise.
6825         (arm_usatsihi): Likewise.
6826         * config/arm/vfp.md
6827         (*movdi_vfp): Add types for all instructions.
6828         (*movdi_vfp_cortexa8): Likewise.
6829         (*movhf_vfp_neon): Likewise.
6830         (*movhf_vfp): Likewise.
6831         (*movdf_vfp): Likewise.
6832         (*thumb2_movdf_vfp): Likewise.
6833         (*thumb2_movdfcc_vfp): Likewise.
6834         * config/arm/arm.md: Add type attribute to all insn patterns.
6835         (*thumb1_adddi3): Add type attribute.
6836         (*arm_adddi3): Likewise.
6837         (*adddi_sesidi_di): Likewise.
6838         (*adddi_zesidi_di): Likewise.
6839         (*thumb1_addsi3): Likewise.
6840         (addsi3_compare0): Likewise.
6841         (*addsi3_compare0_scratch): Likewise.
6842         (*compare_negsi_si): Likewise.
6843         (cmpsi2_addneg): Likewise.
6844         (*addsi3_carryin_<optab>): Likewise.
6845         (*addsi3_carryin_alt2_<optab>): Likewise.
6846         (*addsi3_carryin_clobercc_<optab>): Likewise.
6847         (*subsi3_carryin): Likewise.
6848         (*subsi3_carryin_const): Likewise.
6849         (*subsi3_carryin_compare): Likewise.
6850         (*subsi3_carryin_compare_const): Likewise.
6851         (*arm_subdi3): Likewise.
6852         (*thumb_subdi3): Likewise.
6853         (*subdi_di_zesidi): Likewise.
6854         (*subdi_di_sesidi): Likewise.
6855         (*subdi_zesidi_di): Likewise.
6856         (*subdi_sesidi_di): Likewise.
6857         (*subdi_zesidi_ze): Likewise.
6858         (thumb1_subsi3_insn): Likewise.
6859         (*arm_subsi3_insn): Likewise.
6860         (*anddi3_insn): Likewise.
6861         (*anddi_zesidi_di): Likewise.
6862         (*anddi_sesdi_di): Likewise.
6863         (*ne_zeroextracts): Likewise.
6864         (*ne_zeroextracts): Likewise.
6865         (*ite_ne_zeroextr): Likewise.
6866         (*ite_ne_zeroextr): Likewise.
6867         (*anddi_notdi_di): Likewise.
6868         (*anddi_notzesidi): Likewise.
6869         (*anddi_notsesidi): Likewise.
6870         (andsi_notsi_si): Likewise.
6871         (thumb1_bicsi3): Likewise.
6872         (*iordi3_insn): Likewise.
6873         (*iordi_zesidi_di): Likewise.
6874         (*iordi_sesidi_di): Likewise.
6875         (*thumb1_iorsi3_insn): Likewise.
6876         (*xordi3_insn): Likewise.
6877         (*xordi_zesidi_di): Likewise.
6878         (*xordi_sesidi_di): Likewise.
6879         (*arm_xorsi3): Likewise.
6880         (*andsi_iorsi3_no): Likewise.
6881         (*smax_0): Likewise.
6882         (*smax_m1): Likewise.
6883         (*arm_smax_insn): Likewise.
6884         (*smin_0): Likewise.
6885         (*arm_smin_insn): Likewise.
6886         (*arm_umaxsi3): Likewise.
6887         (*arm_uminsi3): Likewise.
6888         (*minmax_arithsi): Likewise.
6889         (*minmax_arithsi_): Likewise.
6890         (*satsi_<SAT:code>): Likewise.
6891         (arm_ashldi3_1bit): Likewise.
6892         (arm_ashrdi3_1bit): Likewise.
6893         (arm_lshrdi3_1bit): Likewise.
6894         (*arm_negdi2): Likewise.
6895         (*thumb1_negdi2): Likewise.
6896         (*arm_negsi2): Likewise.
6897         (*thumb1_negsi2): Likewise.
6898         (*negdi_extendsid): Likewise.
6899         (*negdi_zero_extend): Likewise.
6900         (*arm_abssi2): Likewise.
6901         (*thumb1_abssi2): Likewise.
6902         (*arm_neg_abssi2): Likewise.
6903         (*thumb1_neg_abss): Likewise.
6904         (one_cmpldi2): Likewise.
6905         (extend<mode>di2): Likewise.
6906         (*compareqi_eq0): Likewise.
6907         (*arm_extendhisi2addsi): Likewise.
6908         (*arm_movdi): Likewise.
6909         (*thumb1_movdi_insn): Likewise.
6910         (*arm_movt): Likewise.
6911         (*thumb1_movsi_insn): Likewise.
6912         (pic_add_dot_plus_four): Likewise.
6913         (pic_add_dot_plus_eight): Likewise.
6914         (tls_load_dot_plus_eight): Likewise.
6915         (*thumb1_movhi_insn): Likewise.
6916         (*thumb1_movsf_insn): Likewise.
6917         (*movdf_soft_insn): Likewise.
6918         (*thumb_movdf_insn): Likewise.
6919         (cbranchsi4_insn): Likewise.
6920         (cbranchsi4_scratch): Likewise.
6921         (*negated_cbranchsi4): Likewise.
6922         (*tbit_cbranch): Likewise.
6923         (*tlobits_cbranch): Likewise.
6924         (*tstsi3_cbranch): Likewise.
6925         (*cbranchne_decr1): Likewise.
6926         (*addsi3_cbranch): Likewise.
6927         (*addsi3_cbranch_scratch): Likewise.
6928         (*arm_cmpdi_insn): Likewise.
6929         (*arm_cmpdi_unsig): Likewise.
6930         (*arm_cmpdi_zero): Likewise.
6931         (*thumb_cmpdi_zero): Likewise.
6932         (*deleted_compare): Likewise.
6933         (*mov_scc): Likewise.
6934         (*mov_negscc): Likewise.
6935         (*mov_notscc): Likewise.
6936         (*cstoresi_eq0_thumb1_insn): Likewise.
6937         (cstoresi_nltu_thumb1): Likewise.
6938         (cstoresi_ltu_thu): Likewise.
6939         (thumb1_addsi3_addgeu): Likewise.
6940         (*arm_jump): Likewise.
6941         (*thumb_jump): Likewise.
6942         (*check_arch2): Likewise.
6943         (arm_casesi_internal): Likewise.
6944         (thumb1_casesi_dispatch): Likewise.
6945         (*arm_indirect_jump): Likewise.
6946         (*thumb1_indirect_jump): Likewise.
6947         (nop): Likewise.
6948         (*and_scc): Likewise.
6949         (*ior_scc): Likewise.
6950         (*compare_scc): Likewise.
6951         (*cond_move): Likewise.
6952         (*cond_arith): Likewise.
6953         (*cond_sub): Likewise.
6954         (*cmp_ite0): Likewise.
6955         (*cmp_ite1): Likewise.
6956         (*cmp_and): Likewise.
6957         (*cmp_ior): Likewise.
6958         (*ior_scc_scc): Likewise.
6959         (*ior_scc_scc_cmp): Likewise.
6960         (*and_scc_scc): Likewise.
6961         (*and_scc_scc_cmp): Likewise.
6962         (*and_scc_scc_nod): Likewise.
6963         (*negscc): Likewise.
6964         (movcond_addsi): Likewise.
6965         (movcond): Likewise.
6966         (*ifcompare_plus_move): Likewise.
6967         (*if_plus_move): Likewise.
6968         (*ifcompare_move_plus): Likewise.
6969         (*if_move_plus): Likewise.
6970         (*ifcompare_arith_arith): Likewise.
6971         (*if_arith_arith): Likewise.
6972         (*ifcompare_arith_move): Likewise.
6973         (*if_arith_move): Likewise.
6974         (*ifcompare_move_arith): Likewise.
6975         (*if_move_arith): Likewise.
6976         (*ifcompare_move_not): Likewise.
6977         (*if_move_not): Likewise.
6978         (*ifcompare_not_move): Likewise.
6979         (*if_not_move): Likewise.
6980         (*ifcompare_shift_move): Likewise.
6981         (*if_shift_move): Likewise.
6982         (*ifcompare_move_shift): Likewise.
6983         (*if_move_shift): Likewise.
6984         (*ifcompare_shift_shift): Likewise.
6985         (*ifcompare_not_arith): Likewise.
6986         (*ifcompare_arith_not): Likewise.
6987         (*if_arith_not): Likewise.
6988         (*ifcompare_neg_move): Likewise.
6989         (*if_neg_move): Likewise.
6990         (*ifcompare_move_neg): Likewise.
6991         (*if_move_neg): Likewise.
6992         (prologue_thumb1_interwork): Likewise.
6993         (*cond_move_not): Likewise.
6994         (*sign_extract_onebit): Likewise.
6995         (*not_signextract_onebit): Likewise.
6996         (stack_tie): Likewise.
6997         (align_4): Likewise.
6998         (align_8): Likewise.
6999         (consttable_end): Likewise.
7000         (consttable_1): Likewise.
7001         (consttable_2): Likewise.
7002         (consttable_4): Likewise.
7003         (consttable_8): Likewise.
7004         (consttable_16): Likewise.
7005         (*thumb1_tablejump): Likewise.
7006         (prefetch): Likewise.
7007         (force_register_use): Likewise.
7008         (thumb_eh_return): Likewise.
7009         (load_tp_hard): Likewise.
7010         (load_tp_soft): Likewise.
7011         (tlscall): Likewise.
7012         (*arm_movtas_ze): Likewise.
7013         (*arm_rev): Likewise.
7014         (*arm_revsh): Likewise.
7015         (*arm_rev16): Likewise.
7016         * config/arm/thumb2.md
7017         (*thumb2_smaxsi3): Likewise.
7018         (*thumb2_sminsi3): Likewise.
7019         (*thumb32_umaxsi3): Likewise.
7020         (*thumb2_uminsi3): Likewise.
7021         (*thumb2_negdi2): Likewise.
7022         (*thumb2_abssi2): Likewise.
7023         (*thumb2_neg_abss): Likewise.
7024         (*thumb2_movsi_insn): Likewise.
7025         (tls_load_dot_plus_four): Likewise.
7026         (*thumb2_movhi_insn): Likewise.
7027         (*thumb2_mov_scc): Likewise.
7028         (*thumb2_mov_negs): Likewise.
7029         (*thumb2_mov_negs): Likewise.
7030         (*thumb2_mov_nots): Likewise.
7031         (*thumb2_mov_nots): Likewise.
7032         (*thumb2_movsicc_): Likewise.
7033         (*thumb2_movsfcc_soft_insn): Likewise.
7034         (*thumb2_indirect_jump): Likewise.
7035         (*thumb2_and_scc): Likewise.
7036         (*thumb2_ior_scc): Likewise.
7037         (*thumb2_ior_scc_strict_it): Likewise.
7038         (*thumb2_cond_move): Likewise.
7039         (*thumb2_cond_arith): Likewise.
7040         (*thumb2_cond_ari): Likewise.
7041         (*thumb2_cond_sub): Likewise.
7042         (*thumb2_negscc): Likewise.
7043         (*thumb2_movcond): Likewise.
7044         (thumb2_casesi_internal): Likewise.
7045         (thumb2_casesi_internal_pic): Likewise.
7046         (*thumb2_alusi3_short): Likewise.
7047         (*thumb2_mov<mode>_shortim): Likewise.
7048         (*thumb2_addsi_short): Likewise.
7049         (*thumb2_subsi_short): Likewise.
7050         (thumb2_addsi3_compare0): Likewise.
7051         (*thumb2_cbz): Likewise.
7052         (*thumb2_cbnz): Likewise.
7053         (*thumb2_one_cmplsi2_short): Likewise.
7054         (*thumb2_negsi2_short): Likewise.
7055         (*orsi_notsi_si): Likewise.
7056         * config/arm/arm1020e.md: Update with new attributes.
7057         * config/arm/arm1026ejs.md: Update with new attributes.
7058         * config/arm/arm1136jfs.md: Update with new attributes.
7059         * config/arm/arm926ejs.md: Update with new attributes.
7060         * config/arm/cortex-a15.md: Update with new attributes.
7061         * config/arm/cortex-a5.md: Update with new attributes.
7062         * config/arm/cortex-a53.md: Update with new attributes.
7063         * config/arm/cortex-a7.md: Update with new attributes.
7064         * config/arm/cortex-a8.md: Update with new attributes.
7065         * config/arm/cortex-a9.md: Update with new attributes.
7066         * config/arm/cortex-m4.md: Update with new attributes.
7067         * config/arm/cortex-r4.md: Update with new attributes.
7068         * config/arm/fa526.md: Update with new attributes.
7069         * config/arm/fa606te.md: Update with new attributes.
7070         * config/arm/fa626te.md: Update with new attributes.
7071         * config/arm/fa726te.md: Update with new attributes.
7073 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
7075         * config/aarch64/aarch64-simd.md
7076         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
7077         <vwx> iterator to ensure correct register choice.
7078         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
7079         (aarch64_sqdmull_n<mode>): Likewise.
7080         (aarch64_sqdmull2_n<mode>_internal): Likewise.
7081         * config/aarch64/arm_neon.h
7082         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
7083         (vml<as><q>_n_<su>16): Likewise.
7084         (vml<as>l_high_lane<q>_<su>16): Likewise.
7085         (vml<as>l_high_n_<su>16): Likewise.
7086         (vml<as>l_lane<q>_<su>16): Likewise.
7087         (vml<as>l_n_<su>16): Likewise.
7088         (vmul<q>_lane<q>_<su>16): Likewise.
7089         (vmul<q>_n_<su>16): Likewise.
7090         (vmull_lane<q>_<su>16): Likewise.
7091         (vmull_n_<su>16): Likewise.
7092         (vmull_high_lane<q>_<su>16): Likewise.
7093         (vmull_high_n_<su>16): Likewise.
7094         (vqrdmulh<q>_n_s16): Likewise.
7096 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
7098         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
7099         have the correct lane parameter.
7101 2013-09-06  Richard Biener <rguenther@suse.de>
7103         * cfganal.c (control_dependences::~control_dependences):
7104         Properly free all of the vector.
7106 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
7108         PR target/58269
7109         * config/i386/i386.c (ix86_conditional_register_usage):
7110         Proper initialize extended SSE registers.
7112 2013-09-06  Jan Hubicka  <jh@suse.cz>
7114         PR tree-optimization/58311
7115         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
7117 2013-09-06  Jan Hubicka  <jh@suse.cz>
7119         * Makefile.in (tree-sra.o): Update dependencies.
7120         * tree-sra.c: Include ipa-utils.h
7121         (scan_function): Use recursive_call_p.
7122         (has_caller_p): New function.
7123         (cgraph_for_node_and_aliases): Count also callers of aliases.
7125 2013-09-06  Jan Hubicka  <jh@suse.cz>
7127         PR middle-end/58094
7128         * cgraph.h (symtab_semantically_equivalent_p): Declare.
7129         * tree-tailcall.c: Include ipa-utils.h.
7130         (find_tail_calls): Use it.
7131         * ipa-pure-const.c (check_call): Likewise.
7132         * ipa-utils.c (recursive_call_p): New function.
7133         * ipa-utils.h (recursive_call_p): Dclare.
7134         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
7135         (symtab_semantically_equivalent_p): New function.
7136         * Makefile.in (tree-tailcall.o): Update dependencies.
7138 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
7140         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
7141         non-inlinable part.
7143 2013-09-06  Richard Biener  <rguenther@suse.de>
7145         * lto-streamer.h (lto_global_var_decls): Remove.
7146         * Makefile.in (OBJS): Remove lto-symtab.o.
7147         (lto-symtab.o): Remove.
7148         (GTFILES): Remove lto-symtab.c
7149         * lto-symtab.c: Move to lto/
7151 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7153         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
7154         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
7155         (UNSPEC_FPINT_RINT): New constant definitions.
7156         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
7157         definition with 2 attributes.
7158         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
7159         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
7160         definitions.
7162 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7164         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
7165         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
7166         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
7167         attribute to "z196_alone".
7168         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
7169         "zEC12_simple".
7171 2013-09-06  Richard Biener  <rguenther@suse.de>
7173         * basic-block.h (class control_dependences): New.
7174         * tree-ssa-dce.c (control_dependence_map): Remove.
7175         (cd): New global.
7176         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
7177         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
7178         find_pdom, find_control_dependence, find_all_control_dependences):
7179         Move to cfganal.c.
7180         (mark_control_dependent_edges_necessary,
7181         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
7182         tree_dce_done, perform_tree_ssa_dce): Adjust.
7183         * cfganal.c (set_control_dependence_map_bit,
7184         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
7185         find_all_control_dependences): Move from tree-ssa-dce.c and
7186         implement as methods of control_dependences class.
7187         (control_dependences::control_dependences): New.
7188         (control_dependences::~control_dependences): Likewise.
7189         (control_dependences::get_edges_dependent_on): Likewise.
7190         (control_dependences::get_edge): Likewise.
7192 2013-09-04  Jan Hubicka  <jh@suse.cz>
7194         * tree.c (types_same_for_odr): Drop overactive check.
7195         * ipa-devirt.c (hash_type_name): Likewise.
7197 2013-09-04  Jan Hubicka  <jh@suse.cz>
7199         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
7200         (analyze_functions): ... here.
7202 2013-09-04  Jan Hubicka  <jh@suse.cz>
7204         PR middle-end/58201
7205         * cgraphunit.c (analyze_functions): Clear AUX fields
7206         after processing; initialize assembler name has.
7208 2013-09-05  Jeff Law  <law@redhat.com>
7210         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
7211         from thread_around_empty_block.  Record threading path into PATH.
7212         Recurse if threading through the initial block is successful.
7213         (thread_across_edge): Corresponding changes to slightly simplify.
7215 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7217         * config/aarch64/aarch64.md
7218         (type): Remove frecpe, frecps, frecpx.
7219         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
7220         fix to be a TARGET_SIMD instruction.
7221         (aarch64_frecps): Remove.
7222         * config/aarch64/aarch64-simd.md
7223         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
7224         (aarch64_frecps<mode>): Handle all float/vector of float modes.
7226 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7227             Sofiane Naci  <sofiane.naci@arm.com>
7229         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
7230         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
7231         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
7232         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
7233         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
7234         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
7235         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
7236         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
7237         into "clz, "rbit".  Rename "shift" to "shift_imm".
7238         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
7239         changes.  Update for attribute changes all occurrences of arlo_* and
7240         shift* types.
7241         * config/arm/arm-fixed.md: Update for attribute changes
7242         all occurrences of arlo_* types.
7243         * config/arm/thumb2.md: Update for attribute changes all occurrences
7244         of arlo_* types.
7245         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
7246         (cortexa7_older_only): Likewise.
7247         (cortexa7_younger):  Likewise.
7248         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7249         (1020alu_shift_op): Likewise.
7250         (1020alu_shift_reg_op): Likewise.
7251         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7252         (alu_shift_op): Likewise.
7253         (alu_shift_reg_op): Likewise.
7254         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7255         (11_alu_shift_op): Likewise.
7256         (11_alu_shift_reg_op): Likewise.
7257         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7258         (9_alu_shift_reg_op): Likewise.
7259         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
7260         attribute changes.
7261         (cortex_a15_alu_shift): Likewise.
7262         (cortex_a15_alu_shift_reg): Likewise.
7263         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
7264         attribute changes.
7265         (cortex_a5_alu_shift): Likewise.
7266         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
7267         attribute changes.
7268         (cortex_a53_alu_shift): Likewise.
7269         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
7270         attribute changes.
7271         (cortex_a7_alu_reg): Likewise.
7272         (cortex_a7_alu_shift): Likewise.
7273         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
7274         attribute changes.
7275         (cortex_a8_alu_shift): Likewise.
7276         (cortex_a8_alu_shift_reg): Likewise.
7277         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7278         (cortex_a9_dp_shift): Likewise.
7279         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
7280         attribute changes.
7281         * config/arm/cortex-r4.md
7282         (cortex_r4_alu): Update for attribute changes.
7283         (cortex_r4_mov): Likewise.
7284         (cortex_r4_alu_shift_reg): Likewise.
7285         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7286         (526_alu_shift_op): Likewise.
7287         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7288         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7289         (626te_alu_shift_op): Likewise.
7290         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
7291         (726te_alu_shift_op): Likewise.
7292         (726te_alu_shift_reg_op): Likewise.
7293         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7294         (mp626_alu_shift_op): Likewise.
7295         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
7296         (pj4_alu_conds): Likewise.
7297         (pj4_shift): Likewise.
7298         (pj4_shift_conds): Likewise.
7299         (pj4_alu_shift): Likewise.
7300         (pj4_alu_shift_conds): Likewise.
7301         * config/aarch64/aarch64.md: Update for attribute change
7302         all occurrences of arlo_* and shift* types.
7304 2013-09-05  Mike Stump  <mikestump@comcast.net>
7306         * tree.h: Move documentation for tree_function_decl to tree-core.h
7307         with the declaration.
7309 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
7311         PR target/58139
7312         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
7313         looking for widest mode.
7315 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
7317         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
7319 2013-09-05  Richard Biener  <rguenther@suse.de>
7321         PR tree-optimization/58137
7322         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
7323         Do not create vectors of pointers.
7324         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
7325         types for the components of the vector initializer.
7326         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
7327         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
7329 2013-09-05  Martin Jambor  <mjambor@suse.cz>
7331         * ipa-prop.c (remove_described_reference): Accept missing references,
7332         return false if that hppens, otherwise return true.
7333         (cgraph_node_for_jfunc): New function.
7334         (try_decrement_rdesc_refcount): Likewise.
7335         (try_make_edge_direct_simple_call): Use them.
7336         (ipa_edge_removal_hook): Remove references from rdescs.
7337         (ipa_edge_duplication_hook): Clone rdescs and their references
7338         when the new edge has the same caller as the old one.
7339         * cgraph.c (cgraph_resolve_speculation): Remove speculative
7340         reference before removing any edges.
7342 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
7344         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
7345         initial store.
7346         * thumb2.md (thumb2_storewb_parisi): New pattern.
7348 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
7350         * config/aarch64/aarch64-option-extensions.def: Add
7351         AARCH64_OPT_EXTENSION of 'crc'.
7352         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
7353         (AARCH64_ISA_CRC): Ditto.
7354         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
7355         description of the CRC extension.
7357 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7359         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
7360         * config/rs6000/linux.h: Ditto.
7361         * alpha/linux.h: Ditto.
7362         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
7363         no_c99_libc_has_function.
7364         * config/c6x/uclinux-elf.h: Ditto.
7365         * config/lm32/uclinux-elf.h: Ditto.
7366         * config/m68k/uclinux.h: Ditto.
7367         * config/moxie/uclinux.h: Ditto.
7368         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
7369         (crisv32-*-linux*, cris-*-linux*): Ditto.
7370         * config/bfin/bfin.c: Include "tm_p.h".
7372 2013-09-05  Richard Biener  <rguenther@suse.de>
7374         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
7375         check for a definition without a basic-block.
7377 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7378             Sofiane Naci <sofiane.naci@arm.com>
7380         * config/aarch64/aarch64.md
7381         (*movti_aarch64): Rename r_2_f and f_2_r.
7382         (*movsf_aarch64): Likewise.
7383         (*movdf_aarch64): Likewise.
7384         (*movtf_aarch64): Likewise.
7385         (aarch64_movdi_<mode>low): Likewise.
7386         (aarch64_movdi_<mode>high): Likewise.
7387         (aarch64_mov<mode>high_di): Likewise.
7388         (aarch64_mov<mode>low_di): Likewise.
7389         (aarch64_movtilow_tilow): Likewise.
7390         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
7391         values to config/arm/types.md
7392         (attribute "conds"): Update for attribute change.
7393         (anddi3_insn): Likewise.
7394         (iordi3_insn): Likewise.
7395         (xordi3_insn): Likewise.
7396         (one_cmpldi2): Likewise.
7397         * config/arm/types.md (type): Add Neon types.
7398         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
7399         use "type" attribute.
7400         (movmisalign<mode>_neon_store): Likewise.
7401         (movmisalign<mode>_neon_load): Likewise.
7402         (vec_set<mode>_internal): Likewise.
7403         (vec_setv2di_internal): Likewise.
7404         (vec_extract<mode>): Likewise.
7405         (vec_extractv2di): Likewise.
7406         (add<mode>3_neon): Likewise.
7407         (adddi3_neon): Likewise.
7408         (sub<mode>3_neon): Likewise.
7409         (subdi3_neon): Likewise.
7410         (mul<mode>3_neon): Likewise.
7411         (mul<mode>3add<mode>_neon): Likewise.
7412         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
7413         (fma<VCVTF:mode>4)): Likewise.
7414         (fma<VCVTF:mode>4_intrinsic): Likewise.
7415         (fmsub<VCVTF:mode>4)): Likewise.
7416         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
7417         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
7418         (ior<mode>3): Likewise.
7419         (and<mode>3): Likewise.
7420         (anddi3_neon): Likewise.
7421         (orn<mode>3_neon): Likewise.
7422         (orndi3_neon): Likewise.
7423         (bic<mode>3_neon): Likewise.
7424         (bicdi3_neon): Likewise.
7425         (xor<mode>3): Likewise.
7426         (one_cmpl<mode>2): Likewise.
7427         (abs<mode>2): Likewise.
7428         (neg<mode>2): Likewise.
7429         (umin<mode>3_neon): Likewise.
7430         (umax<mode>3_neon): Likewise.
7431         (smin<mode>3_neon): Likewise.
7432         (smax<mode>3_neon): Likewise.
7433         (vashl<mode>3): Likewise.
7434         (vashr<mode>3_imm): Likewise.
7435         (vlshr<mode>3_imm): Likewise.
7436         (ashl<mode>3_signed): Likewise.
7437         (ashl<mode>3_unsigned): Likewise.
7438         (neon_load_count): Likewise.
7439         (ashldi3_neon_noclobber): Likewise.
7440         (signed_shift_di3_neon): Likewise.
7441         (unsigned_shift_di3_neon): Likewise.
7442         (ashrdi3_neon_imm_noclobber): Likewise.
7443         (lshrdi3_neon_imm_noclobber): Likewise.
7444         (widen_ssum<mode>3): Likewise.
7445         (widen_usum<mode>3): Likewise.
7446         (quad_halves_<code>v4si): Likewise.
7447         (quad_halves_<code>v4sf): Likewise.
7448         (quad_halves_<code>v8hi): Likewise.
7449         (quad_halves_<code>v16qi): Likewise.
7450         (reduc_splus_v2di): Likewise.
7451         (neon_vpadd_internal<mode>): Likewise.
7452         (neon_vpsmin<mode>): Likewise.
7453         (neon_vpsmax<mode>): Likewise.
7454         (neon_vpumin<mode>): Likewise.
7455         (neon_vpumax<mode>): Likewise.
7456         (ss_add<mode>_neon): Likewise.
7457         (us_add<mode>_neon): Likewise.
7458         (ss_sub<mode>_neon): Likewise.
7459         (us_sub<mode>_neon): Likewise.
7460         (neon_vadd<mode>_unspec): Likewise.
7461         (neon_vaddl<mode>): Likewise.
7462         (neon_vaddw<mode>): Likewise.
7463         (neon_vhadd<mode>): Likewise.
7464         (neon_vqadd<mode>): Likewise.
7465         (neon_vaddhn<mode>): Likewise.
7466         (neon_vmul<mode>): Likewise.
7467         (neon_vmla<mode>): Likewise.
7468         (neon_vmlal<mode>): Likewise.
7469         (neon_vmls<mode>): Likewise.
7470         (neon_vmlsl<mode>): Likewise.
7471         (neon_vqdmulh<mode>): Likewise.
7472         (neon_vqdmlal<mode>): Likewise.
7473         (neon_vqdmlsl<mode>): Likewise.
7474         (neon_vmull<mode>): Likewise.
7475         (neon_vqdmull<mode>): Likewise.
7476         (neon_vsub<mode>_unspec): Likewise.
7477         (neon_vsubl<mode>): Likewise.
7478         (neon_vsubw<mode>): Likewise.
7479         (neon_vqsub<mode>): Likewise.
7480         (neon_vhsub<mode>): Likewise.
7481         (neon_vsubhn<mode>): Likewise.
7482         (neon_vceq<mode>): Likewise.
7483         (neon_vcge<mode>): Likewise.
7484         (neon_vcgeu<mode>): Likewise.
7485         (neon_vcgt<mode>): Likewise.
7486         (neon_vcgtu<mode>): Likewise.
7487         (neon_vcle<mode>): Likewise.
7488         (neon_vclt<mode>): Likewise.
7489         (neon_vcage<mode>): Likewise.
7490         (neon_vcagt<mode>): Likewise.
7491         (neon_vtst<mode>): Likewise.
7492         (neon_vabd<mode>): Likewise.
7493         (neon_vabdl<mode>): Likewise.
7494         (neon_vaba<mode>): Likewise.
7495         (neon_vabal<mode>): Likewise.
7496         (neon_vmax<mode>): Likewise.
7497         (neon_vmin<mode>): Likewise.
7498         (neon_vpaddl<mode>): Likewise.
7499         (neon_vpadal<mode>): Likewise.
7500         (neon_vpmax<mode>): Likewise.
7501         (neon_vpmin<mode>): Likewise.
7502         (neon_vrecps<mode>): Likewise.
7503         (neon_vrsqrts<mode>): Likewise.
7504         (neon_vqabs<mode>): Likewise.
7505         (neon_vqneg<mode>): Likewise.
7506         (neon_vcls<mode>): Likewise.
7507         (clz<mode>2): Likewise.
7508         (popcount<mode>2): Likewise.
7509         (neon_vrecpe): Likewise.
7510         (neon_vrsqrte): Likewise.
7511         (neon_vget_lane<mode>_sext_internal): Likewise.
7512         (neon_vget_lane<mode>_zext_internal): Likewise.
7513         (neon_vdup_n<mode>): Likewise.
7514         (neon_vdup_nv2di): Likewise.
7515         (neon_vdpu_lane<mode>_internal): Likewise.
7516         (neon_vswp<mode>): Likewise.
7517         (float<mode><V_cvtto>2): Likewise.
7518         (floatuns<mode><V_cvtto>2): Likewise.
7519         (fix_trunc<mode><V_cvtto>)2): Likewise
7520         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7521         (neon_vcvt<mode>): Likewise.
7522         (neon_vcvtv4sfv4hf): Likewise.
7523         (neon_vcvtv4hfv4sf): Likewise.
7524         (neon_vcvt_n<mode>): Likewise.
7525         (neon_vmovn<mode>): Likewise.
7526         (neon_vqmovn<mode>): Likewise.
7527         (neon_vqmovun<mode>): Likewise.
7528         (neon_vmovl<mode>): Likewise.
7529         (neon_vmul_lane<mode>): Likewise.
7530         (neon_vmull_lane<mode>): Likewise.
7531         (neon_vqdmull_lane<mode>): Likewise.
7532         (neon_vqdmulh_lane<mode>): Likewise.
7533         (neon_vmla_lane<mode>): Likewise.
7534         (neon_vmlal_lane<mode>): Likewise.
7535         (neon_vqdmlal_lane<mode>): Likewise.
7536         (neon_vmls_lane<mode>): Likewise.
7537         (neon_vmlsl_lane<mode>): Likewise.
7538         (neon_vqdmlsl_lane<mode>): Likewise.
7539         (neon_vext<mode>): Likewise.
7540         (neon_vrev64<mode>): Likewise.
7541         (neon_vrev32<mode>): Likewise.
7542         (neon_vrev16<mode>): Likewise.
7543         (neon_vbsl<mode>_internal): Likewise.
7544         (neon_vshl<mode>): Likewise.
7545         (neon_vqshl<mode>): Likewise.
7546         (neon_vshr_n<mode>): Likewise.
7547         (neon_vshrn_n<mode>): Likewise.
7548         (neon_vqshrn_n<mode>): Likewise.
7549         (neon_vqshrun_n<mode>): Likewise.
7550         (neon_vshl_n<mode>): Likewise.
7551         (neon_vqshl_n<mode>): Likewise.
7552         (neon_vqshlu_n<mode>): Likewise.
7553         (neon_vshll_n<mode>): Likewise.
7554         (neon_vsra_n<mode>): Likewise.
7555         (neon_vsri_n<mode>): Likewise.
7556         (neon_vsli_n<mode>): Likewise.
7557         (neon_vtbl1v8qi): Likewise.
7558         (neon_vtbl2v8qi): Likewise.
7559         (neon_vtbl3v8qi): Likewise.
7560         (neon_vtbl4v8qi): Likewise.
7561         (neon_vtbx1v8qi): Likewise.
7562         (neon_vtbx2v8qi): Likewise.
7563         (neon_vtbx3v8qi): Likewise.
7564         (neon_vtbx4v8qi): Likewise.
7565         (neon_vtrn<mode>_internal): Likewise.
7566         (neon_vzip<mode>_internal): Likewise.
7567         (neon_vuzp<mode>_internal): Likewise.
7568         (neon_vld1<mode>): Likewise.
7569         (neon_vld1_lane<mode>): Likewise.
7570         (neon_vld1_dup<mode>): Likewise.
7571         (neon_vld1_dupv2di): Likewise.
7572         (neon_vst1<mode>): Likewise.
7573         (neon_vst1_lane<mode>): Likewise.
7574         (neon_vld2<mode>): Likewise.
7575         (neon_vld2_lane<mode>): Likewise.
7576         (neon_vld2_dup<mode>): Likewise.
7577         (neon_vst2<mode>): Likewise.
7578         (neon_vst2_lane<mode>): Likewise.
7579         (neon_vld3<mode>): Likewise.
7580         (neon_vld3qa<mode>): Likewise.
7581         (neon_vld3qb<mode>): Likewise.
7582         (neon_vld3_lane<mode>): Likewise.
7583         (neon_vld3_dup<mode>): Likewise.
7584         (neon_vst3<mode>): Likewise.
7585         (neon_vst3qa<mode>): Likewise.
7586         (neon_vst3qb<mode>): Likewise.
7587         (neon_vst3_lane<mode>): Likewise.
7588         (neon_vld4<mode>): Likewise.
7589         (neon_vld4qa<mode>): Likewise.
7590         (neon_vld4qb<mode>): Likewise.
7591         (neon_vld4_lane<mode>): Likewise.
7592         (neon_vld4_dup<mode>): Likewise.
7593         (neon_vst4<mode>): Likewise.
7594         (neon_vst4qa<mode>): Likewise.
7595         (neon_vst4qb<mode>): Likewise.
7596         (neon_vst4_lane<mode>): Likewise.
7597         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7598         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7599         (neon_vec_<US>mult_lo_<mode>): Likewise.
7600         (neon_vec_<US>mult_hi_<mode>): Likewise.
7601         (neon_vec_<US>shiftl_<mode>): Likewise.
7602         (neon_unpack<US>_<mode>): Likewise.
7603         (neon_vec_<US>mult_<mode>): Likewise.
7604         (vec_pack_trunc_<mode>): Likewise.
7605         (neon_vec_pack_trunk_<mode>): Likewise.
7606         (neon_vabd<mode>_2): Likewise.
7607         (neon_vabd<mode>_3): Likewise.
7608         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7609         (thumb2_movsi_vfp): Likewise.
7610         (movdi_vfp): Likewise.
7611         (movdi_vfp_cortexa8): Likewise.
7612         (movhf_vfp_neon): Likewise.
7613         (movhf_vfp): Likewiwse.
7614         (movsf_vfp): Likewiwse.
7615         (thumb2_movsf_vfp): Likewiwse.
7616         (movdf_vfp): Likewise.
7617         (thumb2_movdf_vfp): Likewise.
7618         (movsfcc_vfp): Likewise.
7619         (thumb2_movsfcc_vfp): Likewise.
7620         (movdfcc_vfp): Likewise.
7621         (thumb2_movdfcc_vfp): Likewise.
7622         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7623         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7624         (v10_v2c): Likewise.
7625         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7626         attribute change.
7627         (cortex_a15_neon_int_2): Likewise.
7628         (cortex_a15_neon_int_3): Likewise.
7629         (cortex_a15_neon_int_4): Likewise.
7630         (cortex_a15_neon_int_5): Likewise.
7631         (cortex_a15_neon_vqneg_vqabs): Likewise.
7632         (cortex_a15_neon_vmov): Likewise.
7633         (cortex_a15_neon_vaba): Likewise.
7634         (cortex_a15_neon_vaba_qqq): Likewise.
7635         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7636         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7637         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7638         scalar_64_32_long_scalar): Likewise.
7639         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7640         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7641         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7642         lotype_qdd_64_32_long): Likewise.
7643         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7644         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7645         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7646         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7647         (cortex_a15_neon_shift_1): Likewise.
7648         (cortex_a15_neon_shift_2): Likewise.
7649         (cortex_a15_neon_shift_3): Likewise.
7650         (cortex_a15_neon_vshl_ddd): Likewise.
7651         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7652         (cortex_a15_neon_vsra_vrsra): Likewise.
7653         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7654         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7655         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7656         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7657         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7658         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7659         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7660         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7661         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7662         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7663         (cortex_a15_neon_bp_simple): Likewise.
7664         (cortex_a15_neon_bp_2cycle): Likewise.
7665         (cortex_a15_neon_bp_3cycle): Likewise.
7666         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7667         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7668         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7669         (cortex_a15_neon_vld2_4_regs): Likewise.
7670         (cortex_a15_neon_vld3_vld4): Likewise.
7671         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7672         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7673         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7674         (cortex_a15_neon_vst3_vst4): Likewise.
7675         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7676         (cortex_a15_neon_vld3_vld4_lane" 10
7677         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7678         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7679         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7680         (cortex_a15_neon_ldm_2): Likewise.0
7681         (cortex_a15_neon_stm_2): Likewise.
7682         (cortex_a15_neon_mcr): Likewise.
7683         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7684         (cortex_a15_neon_mrc): Likewise.
7685         (cortex_a15_neon_mrrc): Likewise.
7686         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7687         change.
7688         (cortex_a15_alu_shift): Likewise.
7689         (cortex_a15_alu_shift_reg): Likewise.
7690         (cortex_a15_mult32): Likewise.
7691         (cortex_a15_mult64): Likewise.
7692         (cortex_a15_block): Likewise.
7693         (cortex_a15_branch): Likewise.
7694         (cortex_a15_load1): Likewise.
7695         (cortex_a15_load3): Likewise.
7696         (cortex_a15_store1): Likewise.
7697         (cortex_a15_store3): Likewise.
7698         (cortex_a15_call): Likewise.
7699         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7700         (cortex_a5_f2r): Likewise.
7701         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7702         change.
7703         (cortex_a53_f2r): Likewise.
7704         * config/arm/cortex-a7.md
7705         (cortex_a7_branch): Update for attribute change.
7706         (cortex_a7_call): Likewise.
7707         (cortex_a7_alu_imm): Likewise.
7708         (cortex_a7_alu_reg): Likewise.
7709         (cortex_a7_alu_shift): Likewise.
7710         (cortex_a7_mul): Likewise.
7711         (cortex_a7_load1): Likewise.
7712         (cortex_a7_store1): Likewise.
7713         (cortex_a7_load2): Likewise.
7714         (cortex_a7_store2): Likewise.
7715         (cortex_a7_load3): Likewise.
7716         (cortex_a7_store3): Likewise.
7717         (cortex_a7_load4): Likewise.
7718         (cortex_a7_store4): Likewise.
7719         (cortex_a7_fpalu): Likewise.
7720         (cortex_a7_fconst): Likewise.
7721         (cortex_a7_fpmuls): Likewise.
7722         (cortex_a7_neon_mul): Likewise.
7723         (cortex_a7_fpmacs): Likewise.
7724         (cortex_a7_neon_mla: Likewise.
7725         (cortex_a7_fpmuld: Likewise.
7726         (cortex_a7_fpmacd: Likewise.
7727         (cortex_a7_fpfmad: Likewise.
7728         (cortex_a7_fdivs: Likewise.
7729         (cortex_a7_fdivd: Likewise.
7730         (cortex_a7_r2f: Likewise.
7731         (cortex_a7_f2r: Likewise.
7732         (cortex_a7_f_flags: Likewise.
7733         (cortex_a7_f_loads: Likewise.
7734         (cortex_a7_f_loadd: Likewise.
7735         (cortex_a7_f_stores: Likewise.
7736         (cortex_a7_f_stored: Likewise.
7737         (cortex_a7_neon): Likewise.
7738         * config/arm/cortex-a8-neon.md
7739         (cortex_a8_neon_mrc): Update for attribute change.
7740         (cortex_a8_neon_mrrc): Likewise.
7741         (cortex_a8_neon_int_1): Likewise.
7742         (cortex_a8_neon_int_2): Likewise.
7743         (cortex_a8_neon_int_3): Likewise.
7744         (cortex_a8_neon_int_4): Likewise.
7745         (cortex_a8_neon_int_5): Likewise.
7746         (cortex_a8_neon_vqneg_vqabs): Likewise.
7747         (cortex_a8_neon_vmov): Likewise.
7748         (cortex_a8_neon_vaba): Likewise.
7749         (cortex_a8_neon_vaba_qqq): Likewise.
7750         (cortex_a8_neon_vsma): Likewise.
7751         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7752         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7753         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7754         long_scalar): Likewise.
7755         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7756         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7757         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7758         long_scalar_qdd_64_32_long): Likewise.
7759         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7760         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7761         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7762         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7763         (cortex_a8_neon_shift_1): Likewise.
7764         (cortex_a8_neon_shift_2): Likewise.
7765         (cortex_a8_neon_shift_3): Likewise.
7766         (cortex_a8_neon_vshl_ddd): Likewise.
7767         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7768         (cortex_a8_neon_vsra_vrsra): Likewise.
7769         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7770         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7771         (cortex_a8_neon_fp_vsum): Likewise.
7772         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7773         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7774         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7775         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7776         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7777         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7778         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7779         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7780         (cortex_a8_neon_bp_simple): Likewise.
7781         (cortex_a8_neon_bp_2cycle): Likewise.
7782         (cortex_a8_neon_bp_3cycle): Likewise.
7783         (cortex_a8_neon_ldr): Likewise.
7784         (cortex_a8_neon_str): Likewise.
7785         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7786         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7787         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7788         (cortex_a8_neon_vld2_4_regs): Likewise.
7789         (cortex_a8_neon_vld3_vld4): Likewise.
7790         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7791         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7792         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7793         (cortex_a8_neon_vst3_vst4): Likewise.
7794         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7795         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7796         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7797         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7798         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7799         (cortex_a8_neon_mcr): Likewise.
7800         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7801         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7802         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7803         change.
7804         (ca9_neon_mrrc): Likewise.
7805         (cortex_a9_neon_int_1): Likewise.
7806         (cortex_a9_neon_int_2): Likewise.
7807         (cortex_a9_neon_int_3): Likewise.
7808         (cortex_a9_neon_int_4): Likewise.
7809         (cortex_a9_neon_int_5): Likewise.
7810         (cortex_a9_neon_vqneg_vqabs): Likewise.
7811         (cortex_a9_neon_vmov): Likewise.
7812         (cortex_a9_neon_vaba): Likewise.
7813         (cortex_a9_neon_vaba_qqq): Likewise.
7814         (cortex_a9_neon_vsma): Likewise.
7815         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7816         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7817         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7818         long_scalar): Likewise.
7819         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7820         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7821         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7822         long_scalar_qdd_64_32_long): Likewise.
7823         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7824         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7825         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7826         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7827         (cortex_a9_neon_shift_1): Likewise.
7828         (cortex_a9_neon_shift_2): Likewise.
7829         (cortex_a9_neon_shift_3): Likewise.
7830         (cortex_a9_neon_vshl_ddd): Likewise.
7831         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7832         (cortex_a9_neon_vsra_vrsra): Likewise.
7833         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7834         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7835         (cortex_a9_neon_fp_vsum): Likewise.
7836         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7837         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7838         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7839         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7840         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7841         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7842         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7843         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7844         (cortex_a9_neon_bp_simple): Likewise.
7845         (cortex_a9_neon_bp_2cycle): Likewise.
7846         (cortex_a9_neon_bp_3cycle): Likewise.
7847         (cortex_a9_neon_ldr): Likewise.
7848         (cortex_a9_neon_str): Likewise.
7849         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7850         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7851         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7852         (cortex_a9_neon_vld2_4_regs): Likewise.
7853         (cortex_a9_neon_vld3_vld4): Likewise.
7854         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7855         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7856         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7857         (cortex_a9_neon_vst3_vst4): Likewise.
7858         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7859         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7860         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7861         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7862         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7863         (cortex_a9_neon_mcr): Likewise.
7864         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7865         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7866         (cortex_a9_fps): Likewise.
7867         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7868         change.
7869         (cortex_m4_fmuls): Likewise.
7870         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7871         change.
7872         (cortex_r4_mrc): Likewise.
7873         * config/arm/iterators.md: Update comment referring to neon_type.
7874         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7875         (iwmmxt_movsi_insn): Likewise.
7876         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7877         attribute change.
7878         (pj4_core_to_vfp): Likewise.
7879         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7880         attribute change.
7881         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7882         (vfp_fstore): Likewise.
7883         * doc/md.texi: Change references to neon_type to refer to type.
7885 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7887         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7889 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7891         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7892         lglobl if not weak.
7894 2013-09-04  Easwaran Raman  <eraman@google.com>
7896         PR middle-end/57370
7897         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7898         (build_and_add_sum): Use it.
7899         (appears_later_in_bb): Simplify code.
7901 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7903         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7905 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7907         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7909 2013-09-04  Jan Hubicka  <jh@suse.cz>
7911         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7912         * ipa-devirt.c: Include diganostic.h
7913         (odr_type_d): Add types and types_set.
7914         (hash_type_name): Work for types with vtables during LTO.
7915         (odr_hasher::remove): Fix comment; destroy types_set.
7916         (add_type_duplicate): New function,
7917         (get_odr_type): Use it.
7918         (dump_type_inheritance_graph): Dump type duplicates.
7919         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7920         graph.
7921         * tree.c (types_same_for_odr): Give exact answers on types with
7922         virtual tables.
7924 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7926         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7927         explaining their differences.
7929 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7931         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7933 2013-09-03  Jeff Law  <law@redhat.com>
7935         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7936         when not threading through a joiner block.  Pass joiner/no joiner
7937         state to register_jump_thread.
7938         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7939         state from argument rather than implying on path length.
7940         Dump the entire jump thread path into debugging dump.
7941         * tree-flow.h (register_jump_thread): Update prototype.
7943 2013-08-29  Xinliang David Li  <davidxl@google.com>
7945         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7946         Remove a trivial gcc_assert.
7948 2013-08-29  Xinliang David Li  <davidxl@google.com>
7950         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7951         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7952         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7953         Delay base decl alignment adjustment.
7954         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7955         * tree-vectorizer.h: New data structure.
7956         (set_dr_misalignment): New function.
7957         (dr_misalignment): Ditto.
7958         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7959         (vectorizable_load): Ditto.
7960         (ensure_base_align): New function.
7961         (vectorize_loops): Add dbg_cnt support.
7962         (execute_vect_slp): Ditto.
7963         * dbgcnt.def: New debug counter.
7964         * Makefile: New dependency.
7966 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7968         Revert:
7970         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7972         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7974 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7976         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7977         function descriptor.
7979 2013-09-03  Richard Biener  <rguenther@suse.de>
7981         * tree-affine.c (add_elt_to_tree): Fix association issue,
7982         avoid useless converts and make sure to always return a
7983         properly typed result.
7985 2013-09-03  Richard Biener  <rguenther@suse.de>
7987         PR middle-end/57656
7988         * fold-const.c (negate_expr_p): Fix division case.
7989         (negate_expr): Likewise.
7991 2013-09-03  Richard Biener  <rguenther@suse.de>
7993         PR lto/58285
7994         * tree-streamer-out.c: Include tm.h.
7995         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7997 2013-09-03  Jan Hubicka  <jh@suse.cz>
7999         * tree-profile.c (tree_profiling): Cleanup CFG when done.
8001 2013-09-03  Alan Modra  <amodra@gmail.com>
8003         * config.gcc (powerpc*-*-linux*): Add support for little-endian
8004         multilibs to big-endian target and vice versa.
8005         * config/rs6000/t-linux64: Use := assignment on all vars.
8006         (MULTILIB_EXTRA_OPTS): Remove fPIC.
8007         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
8008         * config/rs6000/t-linux64le: New file.
8009         * config/rs6000/t-linux64bele: New file.
8010         * config/rs6000/t-linux64lebe: New file.
8012 2013-09-02  Jan Hubicka  <jh@suse.cz>
8014         * ipa-inline-transform.c (inline_transform): Do not
8015         optimize_inline_calls when not optimizing.
8017 2013-09-02  Jan Hubicka  <jh@suse.cz>
8019         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
8020         duplicated nodes for assembler names.
8021         * symtab.c (symtab_unregister_node): Do not attempt to unlink
8022         hard registers from assembler name hash.
8024 2013-09-02  Jan Hubicka  <jh@suse.cz>
8026         * ipa-split.c (execute_split_functions): Split externally visible
8027         functions called once.
8029 2013-09-02  Martin Jambor  <mjambor@suse.cz>
8031         PR ipa/58106
8032         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
8033         linked list.  When finding the correct duplicate, also consider also
8034         the caller in additon to its inlined_to node.
8036 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
8038         * config/aarch64/aarch64-simd-builtins.def
8039         (dup_lane_scalar): Remove.
8040         * config/aarch64/aarch64-simd.md
8041         (aarch64_simd_dup): Add 'w->w' alternative.
8042         (aarch64_dup_lane<mode>): Allow for VALL.
8043         (aarch64_dup_lane_scalar<mode>): Remove.
8044         (aarch64_dup_lane_<vswap_width_name><mode>): New.
8045         (aarch64_get_lane_signed<mode>): Add w->w altenative.
8046         (aarch64_get_lane_unsigned<mode>): Likewise.
8047         (aarch64_get_lane<mode>): Likewise.
8048         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
8049         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
8050         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
8051         (VCON): Change container of V2SF.
8052         (vswap_width_name): Likewise.
8053         * config/aarch64/arm_neon.h
8054         (__aarch64_vdup_lane_any): New.
8055         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
8056         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
8057         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
8059 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
8061         PR middle-end/56382
8062         * expr.c (emit_move_complex): Do not move complex FP values as parts if
8063         the source or the destination is a single hard register.
8065 2013-09-02  Richard Biener  <rguenther@suse.de>
8067         PR middle-end/57511
8068         * tree-scalar-evolution.c (instantiate_scev_name): Allow
8069         non-linear SCEVs.
8071 2013-09-02  Richard Biener  <rguenther@suse.de>
8073         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
8074         arithmetic to sizetype.
8076 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
8078         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
8079         Find auto-increment use both before and after candidate.
8081 2013-09-02  Marek Polacek  <polacek@redhat.com>
8083         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
8085 2013-09-01  Jan Hubicka  <jh@suse.cz>
8087         * Makefile.in: Add ipa-profile.o
8088         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
8089         * cgraph.c (struct cgraph_propagate_frequency_data,
8090         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
8091         ipa-profile.c; replace cgraph_ by ipa_ prefix.
8092         * cgraph.h (cgraph_propagate_frequency): Remove.
8093         * ipa-inline-analysis.c: Include ipa-utils.h;
8094         drop duplicated cfgloop.h.
8095         (inline_update_callee_summaries): Update.
8096         * ipa-profile.c: New file.
8097         * ipa-utils.h (ipa_propagate_frequency): Declare.
8098         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
8099         data-streamer.h, value-prof.h.
8100         (symtab_remove_unreachable_nodes): Update profile.
8101         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
8102         account_time_size, cmp_counts, dump_histogram,
8103         ipa_profile_generate_summary, ipa_profile_write_summary,
8104         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
8105         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
8106         Move to ipa-profile.c.
8108 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
8110         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
8112 2013-09-01  Jan Hubicka  <jh@suse.cz>
8114         * common.opt (fdevirtualize-speculatively): New function.
8115         * invoke.texi (fdevirtualize-speculatively): Document.
8116         * ipa-devirt.c: Include ipa-inline.h
8117         (likely_target_p): New function.
8118         (ipa_devirt): New function.
8119         (gate_ipa_devirt): New function.
8120         (pass_data_ipa_devirt): New static var.
8121         (pass_ipa_devirt): Likewise.
8122         (make_pass_ipa_devirt): New function.
8123         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
8124         (common_handle_option): Disable devirtualization when
8125         value range profiling is available.
8126         * passes.def (pass_ipa_devirt): Add.
8127         * timever.def (TV_IPA_DEVIRT): New timevar.
8128         * tree-pass.h (make_pass_ipa_devirt):
8130 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
8132         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
8133         include sanitize(undefined).
8135 2013-08-31  Diego Novillo  <dnovillo@google.com>
8137         * Makefile.in (TREE_CORE_H): Define.
8138         (TREE_H): Use.
8139         (GTFILES): Add tree-core.h.
8140         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
8141         size the array.
8142         * tree-core.h: New file.
8143         Move all data structures, enum, typedefs, global
8144         declarations and constants from ...
8145         * tree.h: ... here.
8147 2013-08-31  Jan Hubicka  <jh@suse.cz>
8149         * bulitins.c (expand_builtin): Do not early exit for gcov
8150         instrumented functions.
8152 2013-08-31  Marek Polacek  <polacek@redhat.com>
8154         * ubsan.c: Include tm_p.h.
8156 2013-08-31  Jan Hubicka  <jh@suse.cz>
8158         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
8159         warning.
8161         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
8162         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
8163         * tree-cfg.c (verify_gimple_label): ... here.
8164         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
8165         (ipa_merge_profiles): New function.
8166         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
8167         (lto_input_function_body): Likewise.
8168         * ipa-utils.h (ipa_merge_profiles): Declare.
8169         * lto-streamer.h (lto_input_function_body): Update prototype.
8170         (emit_label_in_global_context_p): Remove.
8171         * lto-symtab.c: Include ipa-utils.h
8172         (lto_cgraph_replace_node): Use ipa_merge_profiles.
8174 2013-08-31  Jan Hubicka  <jh@suse.cz>
8176         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
8178 2013-08-31  Jan Hubicka  <jh@suse.cz>
8180         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
8182 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
8184         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
8185         "cmp" RTX before signed_comparison_operator check to account
8186         for "code" changes.
8188 2013-08-30  Jan Hubicka  <jh@suse.cz>
8190         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
8191         (detect_type_change_1): Rename to ...
8192         (detect_type_change): ... this one; early return on non-polymorphic
8193         types.
8194         (detect_type_change_ssa): Add comp_type parameter; update
8195         use of detect_type_change.
8196         (compute_complex_assign_jump_func): Add param_type parameter;
8197         update use of detect_type_change_ssa.
8198         (compute_complex_ancestor_jump_func): Likewise.
8199         (ipa_get_callee_param_type): New function.
8200         (ipa_compute_jump_functions_for_edge): Compute parameter type;
8201         update calls to the jump function computation functions.
8203 2013-08-30  Teresa Johnson  <tejohnson@google.com>
8204             Steven Bosscher  <steven@gcc.gnu.org>
8206         * cfgrtl.c (fixup_new_cold_bb): New routine.
8207         (commit_edge_insertions): Invoke fixup_partitions.
8208         (find_partition_fixes): New routine.
8209         (fixup_partitions): Ditto.
8210         (verify_hot_cold_block_grouping): Update comments.
8211         (rtl_verify_edges): Invoke find_partition_fixes.
8212         (rtl_verify_bb_pointers): Update comments.
8213         (rtl_verify_bb_layout): Ditto.
8214         * basic-block.h (probably_never_executed_edge_p): Declare.
8215         (fixup_partitions): Ditto.
8216         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
8217         * bb-reorder.c (sanitize_hot_paths): New function.
8218         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
8219         sanitize_hot_paths.
8220         * predict.c (probably_never_executed_edge_p): New routine.
8221         * cfg.c (check_bb_profile): Add partition insanity warnings.
8223 2013-08-30  Meador Inge  <meadori@codesourcery.com>
8225         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
8227 2013-08-30  Marek Polacek  <polacek@redhat.com>
8229         * Makefile.in (ubsan.o): Add.
8230         (c-family/c-ubsan.o): Add.
8231         (builtins.o): Add ubsan.h dependency.
8232         * ubsan.h: New file.
8233         * ubsan.c: New file.
8234         * common.opt: Add -fsanitize=undefined option.
8235         (flag_sanitize): Add variable.
8236         (fsanitize=): Add option.  Add Driver.
8237         (fsanitize=thread): Remove option.
8238         (fsanitize=address): Likewise.
8239         (static-libubsan): New option.
8240         * doc/invoke.texi: Document the new flag and -static-libubsan.
8241         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
8242         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
8243         * builtin-attrs.def (ATTR_COLD): Define.
8244         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
8245         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
8246         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
8247         * flag-types.h (sanitize_code): New enum.
8248         * opts.c (common_handle_option): Parse command line arguments
8249         of -fsanitize=.  Add -fsanitize=unreachable option.
8250         * varasm.c (get_variable_section): Adjust.
8251         (assemble_noswitch_variable): Likewise.
8252         (assemble_variable): Likewise.
8253         (output_constant_def_contents): Likewise.
8254         (categorize_decl_for_section): Likewise.
8255         (place_block_symbol): Likewise.
8256         (output_object_block): Likewise.
8257         * builtins.def: Likewise.
8258         * toplev.c (compile_file): Likewise.
8259         (process_options): Likewise.
8260         * cppbuiltin.c: Likewise.
8261         * tsan.c (tsan_pass): Likewise.
8262         (tsan_gate): Likewise.
8263         (tsan_gate_O0): Likewise.
8264         * cfgexpand.c (partition_stack_vars): Likewise.
8265         (expand_stack_vars): Likewise.
8266         (defer_stack_allocation): Likewise.
8267         (expand_used_vars): Likewise.
8268         * cfgcleanup.c (old_insns_match_p): Likewise.
8269         * asan.c (asan_finish_file): Likewise.
8270         (asan_instrument): Likewise.
8271         (gate_asan): Likewise.
8272         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
8273         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
8274         (asan_global_struct): Use pointer_sized_int_node instead
8275         calling build_nonstandard_integer_type.
8276         (initialize_sanitizer_builtins): Likewise.
8277         (asan_finish_file): Likewise.
8278         * gcc.c: Document %{%:function(args):X}.
8279         (static_spec_functions): Add sanitize.
8280         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
8281         store funcval != NULL there.
8282         (do_spec_1): Adjust handle_spec_function caller.
8283         (handle_braces): Allow %:function(args) as condition.
8284         (sanitize_spec_function): New function.
8285         (ADD_STATIC_LIBUBSAN_LIBS): Define.
8286         (LIBUBSAN_SPEC): Likewise.
8287         (LIBUBSAN_EARLY_SPEC): Likewise.
8288         (SANITIZER_SPEC): Handle libubsan.
8289         (SANITIZER_EARLY_SPEC): Likewise.
8290         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
8291         instead of fsanitize=address.
8292         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
8293         instead of fsanitize=address*.
8294         * builtins.c: Include ubsan.h.
8295         (fold_builtin_0): Instrument __builtin_unreachable.
8296         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
8297         instead of flag_asan.
8298         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
8299         (pointer_sized_int_node): Define.
8300         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
8302 2013-08-30  Mike Stump  <mikestump@comcast.net>
8304         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
8305         with RE patterns.
8307 2013-08-29  Jan Hubicka  <jh@suse.cz>
8309         * cgraph.c (cgraph_function_body_availability): Handle weakref
8310         correctly.
8311         * passes.def: Remove pass_fixup_cfg.
8312         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
8313         track when we need to remove functions.
8314         (gate_ipa_inline): Execute inlining always; add comment why.
8315         (pass_data_ipa_inline): Remove TODO_remove_functions.
8316         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
8317         do not produce summaries.
8318         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
8319         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
8320         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
8321         constant pool and vtable.
8323 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
8325         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
8326         New arm_neon.h's internal macros to specify 64-bit constants.
8327         Avoid using stdint.h's macros.
8329 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
8331         * recog.c (verify_changes): Verify that changes[i].old is non-zero
8332         before applying REG_P.
8334 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
8336         PR tree-optimization/58277
8337         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
8338         after seeing too many stmts with vdef in between dombb and current
8339         bb, invalidate everything.
8341 2013-08-30  Richard Biener  <rguenther@suse.de>
8343         * fold-const.c (fold_single_bit_test): Fix overflow test.
8345 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
8347         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
8348         and which can live in a register, always retrieve the value on entry.
8349         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
8350         passed by invisible reference specially.
8351         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
8352         (vt_add_function_parameter): Correctly deal with a parameter passed by
8353         invisible reference.
8355 2013-08-30  Jan Hubicka  <jh@suse.cz>
8357         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
8359 2013-08-30  Richard Biener  <rguenther@suse.de>
8361         PR tree-optimization/58228
8362         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
8363         allow invariant loads in nested loop vectorization.
8365 2013-08-30  Richard Biener  <rguenther@suse.de>
8367         PR tree-optimization/58223
8368         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
8369         (has_anti_or_output_dependence): ... this and adjust to also
8370         look for output dependences.
8371         (mark_nodes_having_upstream_mem_writes): Adjust.
8372         (rdg_flag_uses): Likewise.
8374 2013-08-30  Richard Biener  <rguenther@suse.de>
8376         PR tree-optimization/58010
8377         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
8378         assert that we have a loop-closed PHI.
8380 2013-08-29  Jan Hubicka  <jh@suse.cz>
8382         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
8383         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
8384         * lto-section-in.c (lto_free_function_in_decl_state): New function.
8385         (lto_free_function_in_decl_state_for_node): New function.
8387 2013-08-29  Xinliang David Li  <davidxl@google.com>
8389         * loop-unroll.c (report_unroll_peel): Minor message change.
8390         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
8391         Emit alignment peeling message with default -fopt-info.
8392         (vect_loop_versioning): Emit loop version info message.
8393         * tree-vectorizer.c (vectorize_loops): Minor message change.
8394         (execute_vect_slp): Ditto.
8396 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
8398         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
8399         of the clone from the DECL_NAME of the original function.
8401 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
8403         * passes.c (register_pass): Add overload.
8404         * tree-pass.h (register_pass): Forward declare it.  Add comment.
8406 2013-08-29  Jan Hubicka  <jh@suse.cz>
8408         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
8409         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
8410         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
8411         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8412         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
8413         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
8414         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8415         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
8417 2013-08-29  Teresa Johnson  <tejohnson@google.com>
8419         * dumpfile.c (dump_loc): Output column number.
8420         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
8421         * doc/invoke.texi: Document optall -fopt-info flag.
8422         * profile.c (read_profile_edge_counts): Use new dump framework.
8423         (compute_branch_probabilities): Ditto.
8424         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
8425         when pass not in any opt group.
8426         * pass_manager.h (pass_manager::get_pass_profile): New method.
8427         * value-prof.c (check_counter): Use new dump framework.
8428         (check_ic_target): Ditto.
8429         * coverage.c (get_coverage_counts): Ditto.
8430         (coverage_init): Setup new dump framework.
8432 2013-08-29  Richard Biener  <rguenther@suse.de>
8434         PR tree-optimization/58246
8435         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
8436         handle the dominance check inside a basic-block.
8438 2013-08-29  Richard Biener  <rguenther@suse.de>
8440         PR middle-end/57287
8441         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
8442         of default defs that appear in abnormal PHI nodes.
8444 2013-08-29  Richard Biener  <rguenther@suse.de>
8446         PR tree-optimization/57685
8447         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
8448         single-use operands to avoid exponential complexity.
8450 2013-08-28  Dehao Chen  <dehao@google.com>
8452         * ipa-inline.c (edge_badness): Fix integer underflow.
8454 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
8456         * gtm-builtins.def (_ITM_free): Declare leaf.
8458 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8460         PR target/58067
8461         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
8462         (*tls_local_dynamic_base_64_largepic): Likewise.
8463         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
8464         Remove predicate from call operand.
8465         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
8466         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
8468 2013-08-28  Jeff Law  <law@redhat.com>
8470         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
8471         checks for the number of predecessors and successors allowed.
8472         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
8473         which require copying a joiner block if there is a request which
8474         is a subpath that requires no joiner block copying.
8476 2013-08-28  Jan Hubicka  <jh@suse.cz>
8478         * lto-streamer-out.c (DFS_write_tree_body): Drop
8479         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8480         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8481         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8482         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8483         Do not read DECL_ERROR_ISSUED.
8484         (unpack_ts_decl_with_vis_value_fields): Do not read
8485         DECL_DEFER_OUTPUT.
8486         (lto_input_ts_binfo_tree_pointers): Do not read
8487         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8488         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8489         write DECL_ERROR_ISSUED..
8490         (pack_ts_decl_with_vis_value_fields): Do not write
8491         DECL_DEFER_OUTPUT.
8492         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8493         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8494         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8495         * tree.h (tree_decl_common): Update comment.
8496         (DECL_ERROR_ISSUED): Remove.
8498 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8500         PR middle-end/58257
8501         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8503 2013-08-28  Jan Hubicka  <jh@suse.cz>
8505         * builtins.def (free): Declare leaf.
8507 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8509         * gdbhooks.py: New.
8510         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8511         * configure: Regenerate.
8513 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8515         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8516         (ipa_ancestor_jf_data): Likewise.
8517         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8518         (ipa_get_jf_pass_through_type_preserved): New function.
8519         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8520         (ipa_get_jf_ancestor_type_preserved): New function.
8521         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8522         (ipa_get_jf_ancestor_result): Likewise.
8523         (propagate_vals_accross_pass_through): Use
8524         ipa_get_jf_pass_through_result to do all the value mappings.
8525         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8526         type_preserved flag.
8527         (ipa_set_jf_cst_copy): New function.
8528         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8529         (ipa_set_jf_arith_pass_through): Likewise.
8530         (ipa_set_ancestor_jf): Likewise.
8531         (compute_complex_assign_jump_func): Set type_preserved instead of
8532         punting.
8533         (ipa_compute_jump_functions_for_edge): Likewise.
8534         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8535         (update_jump_functions_after_inlining): Update type_preserved.
8536         Explicitely create jump functions when combining one with pass_through.
8537         (ipa_write_jump_function): Stream the type_preserved flags.
8538         (ipa_read_jump_function): Likewise.
8540 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8541             Aldy Hernandez  <aldyh@redhat.com>
8543         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8544         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8545         * function.h (struct function): Add has_force_vect_loops and
8546         has_simduid_loops.
8547         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8548         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8549         handle it.
8550         * gimple.def: Update CLAUSES comments.
8551         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8552         (gimple_build_omp_for): Add argument to prototype.
8553         (gimple_omp_for_kind): New.
8554         (gimple_omp_for_set_kind): New.
8555         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8556         GOVD_DATA_SHARE_CLASS.
8557         (enum omp_region_type): Add ORT_SIMD.
8558         (gimple_add_tmp_var): Handle ORT_SIMD.
8559         (gimplify_var_or_parm_decl): Same.
8560         (is_gimple_stmt): Same.
8561         (omp_firstprivatize_variable): Same.
8562         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8563         (omp_notice_variable): Handle ORT_SIMD.
8564         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8565         (omp_check_private): Handle ORT_SIMD.
8566         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8567         OMP_CLAUSE_SAFELEN.
8568         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8569         Handle OMP_CLAUSE_LASTPRIVATE.
8570         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8571         OMP_CLAUSE_SAFELEN.
8572         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8573         (gimplify_expr): Handle OMP_SIMD.
8574         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8575         (expand_GOMP_SIMD_VF): New.
8576         (expand_GOMP_SIMD_LAST_LANE): New.
8577         * internal-fn.def (GOMP_SIMD_LANE): New.
8578         (GOMP_SIMD_VF): New.
8579         (GOMP_SIMD_LAST_LANE): New.
8580         * omp-low.c: Include target.h.
8581         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8582         OMP_CLAUSE_SAFELEN.
8583         (check_omp_nesting_restrictions): Same.
8584         (omp_max_vf): New.
8585         (lower_rec_simd_input_clauses): New.
8586         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8587         OMP_CLAUSE_LINEAR.
8588         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8589         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8590         (expand_omp_build_assign): New.
8591         (expand_omp_for_init_counts): New.
8592         (expand_omp_for_init_vars): New.
8593         (extract_omp_for_update_vars): New.
8594         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8595         and rewrite accordingly.
8596         (expand_omp_simd): New.
8597         (expand_omp_for): Use expand_omp_simd.
8598         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8599         (lower_omp_for): Do not lower the body.
8600         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8601         in their own loops.
8602         * tree-flow.h (find_omp_clause): Remove prototype.
8603         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8604         forcing vectorization of the loop, or if flag_tree_vectorize.
8605         (gate_tree_if_conversion): Similarly.
8606         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8607         gimple_build_omp_for.
8608         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8609         * tree-parloops (create_parallel_loop): Pass kind argument to
8610         gimple_build_omp_for.
8611         * tree-pretty-print.c (dump_omp_clause): Add cases for
8612         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8613         OMP_CLAUSE__SIMDUID_.
8614         (dump_generic_node): Handle OMP_SIMD.
8615         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8616         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8617         unroll OMP_SIMD loops here.
8618         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8619         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8620         loop->safelen.
8621         (vect_analyze_data_refs): Handle simd loops.
8622         * tree-vect-loop.c (vectorizable_live_operation): Handle
8623         IFN_GOMP_SIMD*.
8624         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8625         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8626         (vectorizable_load): Same.
8627         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8628         (struct simduid_to_vf): New.
8629         (simduid_to_vf::hash): New.
8630         (simduid_to-vf::equal): New.
8631         (struct simd_array_to_simduid): New.
8632         (simd_array_to_simduid::hash): New.
8633         (simd_array_to_simduid::equal): New.
8634         (adjust_simduid_builtins): New.
8635         (struct note_simd_array_uses_struct): New.
8636         (note_simd_array_uses_cb): New.
8637         (note_simd_array_uses): New.
8638         (vectorize_loops): Handle simd hints and adjust simd builtins
8639         accordingly.
8640         * tree-vectorizer.h (struct _stmt_vec_info): Add
8641         simd_lane_access_p field.
8642         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8643         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8644         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8645         (omp_clause_code_name): Same.
8646         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8647         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8648         * tree.def (OMP_SIMD): New entry.
8649         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8650         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8651         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8652         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8653         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8654         (OMP_CLAUSE_LINEAR_STEP): New.
8655         (OMP_CLAUSE_SAFELEN_EXPR): New.
8656         (OMP_CLAUSE__SIMDUID__DECL): New.
8657         (find_omp_clause): New prototype.
8659 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8661         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8662         Haswell processor detection.
8664 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8666         PR target/57927
8667         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8668         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8669         AVX2 capable processors.
8671 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8673         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8674         of vget_low_* with implementations in terms of other intrinsics.
8676 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8678         PR middle-end/57219
8679         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8680         values to -1, 0 and 1.
8682 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8684         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8685         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8686         (<optab><mode>3_insn): Remove.
8687         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8688         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8689         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8690         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8691         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8692         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8693         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8694         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8695         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8696         (ror<mode>3_insn): Likewise.
8697         * config/aarch64/predicates.md (aarch64_simd_register): New.
8699 2013-08-27  Richard Biener  <rguenther@suse.de>
8701         PR tree-optimization/57521
8702         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8703         one edge is non-critical.
8704         (find_phi_replacement_condition): Make sure to use a non-critical
8705         edge.  Cleanup and remove old bug workarounds.
8706         (bb_postdominates_preds): Remove.
8707         (if_convertible_loop_p_1): Do not compute post-dominators.
8708         (combine_blocks): Do not free post-dominators.
8709         (main_tree_if_conversion): Likewise.
8710         (pass_data_if_conversion): Add TODO_verify_ssa.
8712 2013-08-27  DJ Delorie  <dj@redhat.com>
8714         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8716 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8718         * function.c (assign_parm_find_data_types): Set passed_mode and
8719         nominal_mode to the TYPE_MODE of nominal_type for the built
8720         pointer type in case of the struct-pass-by-reference.
8722 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8724         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8725         (UINT16_TYPE): Change default to "unsigned int".
8727         * config/avr/avr.opt (mfract-convert-truncate): New option.
8728         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8729         is set, round negative fractional integers according to n1169
8730         when converting to integer types.
8732 2013-08-26  Jan Hubicka  <jh@suse.cz>
8734         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8735         methods can not be called indirectly when their address is not taken.
8737 2013-08-26  Jan Hubicka  <jh@suse.cz>
8739         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8740         ctor_for_folding.
8742 2013-08-26  Jan Hubicka  <jh@suse.cz>
8744         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8745         can be unshared.
8747 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8749         * reload.c (find_valid_class): Allow classes that do not include
8750         FIRST_PSEUDO_REGISTER - 1.
8752 2013-08-26  Jan Hubicka  <jh@suse.cz>
8754         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8755         fix edge count/frequency when speculation failed; fix type check
8756         for the direct call.
8758 2013-08-26  Jan Hubicka  <jh@suse.cz>
8760         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8762 2013-08-26  Jan Hubicka  <jh@suse.cz>
8764         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8765         to be changed by edge redirection.
8767 2013-08-26  Jan Hubicka  <jh@suse.cz>
8769         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8770         formating; add sanity check.
8771         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8772         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8773         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8774         (resolve_noninline_speculation): Update callee keys, too.
8776 2013-08-26  Jan Hubicka  <jh@suse.cz>
8778         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8779         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8781 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8783         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8784         into proper place.
8786 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8788         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8789         (x86_64_elf_select_section): Ditto.
8790         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8791         arguments.
8792         (ix86_pass_by_reference): Ditto.
8793         (output_set_got): Ditto.
8794         (ix86_unary_operator_ok): Ditto.
8795         (ix86_expand_builtin): Ditto.
8797 2013-08-23  Jan Hubicka  <jh@suse.cz>
8799         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8801 2013-08-23  Jan Hubicka  <jh@suse.cz>
8803         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8804         (tree_decl_with_vis): Add FINAL field.
8806 2013-08-23  Jeff Law  <law@redhat.com>
8808         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8809         the debugging dump when the expression is fully redundant.
8811 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8813         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8814         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8815         * pretty-print.c (pp_formatted_text_data): Likewise.
8816         (pp_write_text_to_stream): Likewise.
8817         (pp_write_text_as_dot_label_to_stream): Likewise.
8818         (pp_append_r): Likewise.
8819         (pp_format): Likewise.
8820         (pp_flush): Likewise.
8821         (pp_clear_output_area): Likewise.
8822         (pp_append_text): Likewise.
8823         (pp_formatted_text): Likewise.
8824         (pp_remaining_character_count_for_line): Likewise.
8825         (pp_newline): Likewise.
8826         (pp_character): Likewise.
8827         (output_buffer::~output_buffer): Define.
8828         (pretty_printer::~pretty_printer): Destruct output buffer.
8829         * pretty-print.h (output_buffer::~output_buffer): Declare.
8830         (pretty_printer::~pretty_printer): Declare virtual.
8832 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8834         PR other/57324
8835         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8836         HOST_WIDE_INT_M1U): New macros.
8837         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8838         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8839         unsigned -1 for lshift.
8840         * cse.c (cse_insn): Likewise.
8841         * double-int.c (rshift_double, lshift_double): Likewise.
8842         * builtins.c (fold_builtin_bitop): Likewise.
8843         * combine.c (force_to_mode): Likewise.
8844         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8845         * simplify-rtx.c (simplify_const_unary_operation,
8846         simplify_const_binary_operation): Likewise.
8847         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8848         check_va_list_escapes): Likewise.
8849         * rtlanal.c (nonzero_bits1): Likewise.
8850         * expmed.c (expand_smod_pow2): Likewise.
8851         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8853 2013-08-23  Jan Hubicka  <jh@suse.cz>
8855         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8856         as having address taken.
8858 2013-08-23  Jan Hubicka  <jh@suse.cz>
8860         * ipa-utils.h (method_class_type): Declare.
8861         * ipa-devirt.c (method_class_type): Export.
8863         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8864         do not walk base classes of anonymous types.
8866 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8868         PR rtl-optimization/58220
8869         PR regression/58221
8870         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8871         handle SEQUENCE insns properly.
8873 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8875         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8876         definition.
8877         (pp_newline_and_indent): Likewise.
8878         (pp_separate_with): Likewise.
8879         * pretty-print.c (pp_newline_and_flush): Define.
8880         (pp_newline_and_indent): Likewise.
8881         (pp_separate_with): Likewise.
8883 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8885         PR target/58218
8886         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8887         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8889 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8891         * config/i386/predicates.md (ext_sse_reg_operand): New.
8892         * config/i386/i386.md (*movti_internal): Use
8893         predicate to determine if EVEX is needed.
8894         (*movsi_internal): Ditto.
8895         (*movdf_internal): Ditto.
8896         (*movsf_internal): Ditto.
8897         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8899 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8901         PR tree-optimization/58209
8902         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8903         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8904         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8905         emit POINTER_PLUS_EXPR.
8906         (create_tailcall_accumulator): For pointer result type accumulate in
8907         sizetype type.
8909 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8911         * configure.ac: Add backslashes missing from the last change.
8912         * configure: Regenerate.
8914 2013-08-22  Jan Hubicka  <jh@susue.cz>
8916         * ipa.c (function_and_variable_visibility): First remember function
8917         was global and then make it local.
8919 2013-08-22  Julian Brown  <julian@codesourcery.com>
8921         * configure.ac: Add aarch64 to list of arches which use "nop" in
8922         debug_line test.
8923         * configure: Regenerate.
8925 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8927         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8928         gnu_libc_has_function.
8929         * config/s390/tpf.h: Likewise.
8931 2013-08-22  Jan Hubicka  <jh@susue.cz>
8933         * timevar.c (validate_phases): Add cast.
8935 2013-08-22  Jan Hubicka  <jh@susue.cz>
8937         * timevar.c (validate_phases): Use size_t for memory.
8938         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8940 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8942         * pretty-print.h (output_buffer::output_buffer): Declare.
8943         (pretty_printer::pretty_printer): Likewise.
8944         (pp_construct): Remove.
8945         * pretty-print.c (output_buffer::output_buffer): Define.
8946         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8947         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8948         (print_gimple_expr): Likewise.
8949         (print_gimple_seq): Likewise.
8950         (gimple_dump_bb): Likewise.
8951         * sched-vis.c (dump_value_slim): Likewise.
8952         (dump_insn_slim): Likewise.
8953         (dump_rtl_slim): Likewise.
8954         (str_pattern_slim): Likewise.
8955         * tree-mudflap.c (mf_varname_tree): Likewise.
8956         * graph.c (print_graph_cfg): Likewise.
8957         (start_graph_dump): Likewise.
8958         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8959         placement-new.
8960         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8961         pretty printer initialization.
8962         * coretypes.h (diagnostic_context): Remove superflous type alias
8963         declaration.
8964         (pretty_printer): Likewise.  Declare directly as a class.
8965         (pretty_print_info): Remove declaration as class.
8966         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8967         and pp_clear_output_area.
8968         (asan_add_global): Likewise.
8970 2013-08-22  Jan Hubicka  <jh@suse.cz>
8972         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8973         * ipa-utils.h (update_type_inheritance_graph): Declare.
8974         (possible_polymorphic_call_target_p): Declare.
8975         (possible_polymorphic_call_target_p): New.
8976         * ipa-devirt.c: Update toplevel comments.
8977         (cached_polymorphic_call_targets): Move up.
8978         (odr_type_d): Move ID down.
8979         (polymorphic_type_binfo_p): Update comment.
8980         (odr_hasher::remove): Likewise;
8981         (get_odr_type): Set anonymous_namespace.
8982         (dump_odr_type): Dump it.
8983         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8984         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8985         (record_binfo): Add comment.
8986         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8987         built.
8988         (devirt_node_removal_hook): Do not iCE when cache is freed.
8989         (possible_polymorphic_call_target_p): New predicate.
8990         (update_type_inheritance_graph): New function.
8992 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8993             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8994             Sergey Lega  <sergey.s.lega@intel.com>
8995             Anna Tikhonova  <anna.tikhonova@intel.com>
8996             Ilya Tocar  <ilya.tocar@intel.com>
8997             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8998             Ilya Verbin  <ilya.verbin@intel.com>
8999             Kirill Yukhin  <kirill.yukhin@intel.com>
9000             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
9002         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
9003         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
9004         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
9005         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
9006         (OPTION_MASK_ISA_AVX2_UNSET): Update.
9007         (OPTION_MASK_ISA_AVX512F_UNSET): New.
9008         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
9009         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
9010         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
9011         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
9012         OPT_mavx512pf, OPT_mavx512er cases.
9013         * config/i386/constraints.md (v): New constraint.
9014         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
9015         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
9016         (bit_AVX512CD): New.
9017         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9018         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
9019         * config/i386/i386-c.c (ix86_target_macros_internal):
9020         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
9021         __AVX512PF__.
9022         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
9023         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
9024         * config/i386/i386.c (regclass_map, dbx_register_map)
9025         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
9026         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
9027         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
9028         -mavx512pf options.
9029         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
9030         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
9031         -mavx512pf options.  Fix formatting.
9032         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
9033         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
9034         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
9035         -mavx512cd, -mavx512pf options.
9036         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
9037         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
9038         (ix86_preferred_output_reload_class): Replace SSE_REGS with
9039         ALL_SSE_REGS.
9040         (ix86_hard_regno_mode_ok): Support 512-bit registers.
9041         (ix86_set_reg_reg_cost): Ditto.
9042         (x86_order_regs_for_local_alloc): Ditto.
9043         (MAX_VECT_LEN): Extend to 64-byte.
9044         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
9045         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
9046         (TARGET_AVX512ER, TARGET_AVX512CD): New.
9047         (BIGGEST_ALIGNMENT): Extend to 512-bits.
9048         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
9049         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
9050         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
9051         (SSE_REG_MODE_P): Support new modes.
9052         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
9053         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
9054         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
9055         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
9056         (REG_CLASS_CONTENTS): Add new registers.
9057         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
9058         (EXT_REX_SSE_REGNO_P): New.
9059         (HI_REGISTER_NAMES): Add new registers.
9060         * config/i386/i386.md: Define constants for new registers.
9061         (mode): Add new 512-bit modes.
9062         (prefix): Support evex prefix.
9063         (isa): Support avx512f, noavx512f, fma_avx512f.
9064         (ssemodesuffix): Add new 512-bit modes.
9065         (movxi): New.
9066         (*movxi_internal_avx512f): Ditto.
9067         (*movdi_internal): Replace constraint "x" with the new constraint "v".
9068         Support MODE_XI.
9069         (*movsi_internal): Likewise.
9070         (*movdf_internal): Likewise.
9071         (*movsf_internal): Likewise.
9072         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
9073         (<code><mode>3): Likewise.
9074         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
9075         New.
9076         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
9077         with the new constraint "v".
9078         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
9079         modes.
9080         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
9081         with the new constraint "v".
9082         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
9083         (<sse2>_storedqu<avxsizesuffix>): Likewise.
9084         (*<plusminus_insn><mode>3): Likewise.
9085         (<sse>_vm<plusminus_insn><mode>3): Likewise.
9086         (*mul<mode>3): Likewise.
9087         (<sse>_vmmul<mode>3): Likewise.
9088         (<sse>_div<mode>3): Likewise.
9089         (<sse>_vmdiv<mode>3): Likewise.
9090         (<sse>_sqrt<mode>2): Likewise.
9091         (<sse>_vmsqrt<mode>2): Likewise.
9092         (*<code><mode>3_finite): Likewise.
9093         (*<code><mode>3) <smaxmin>: Likewise.
9094         (<sse>_vm<code><mode>3): Likewise.
9095         (*<code><mode>3) <any_logic>: Likewise.
9096         (*fma_fmadd_<mode>): Likewise.
9097         (*fma_fmsub_<mode>): Likewise.
9098         (*fma_fnmadd_<mode>): Likewise.
9099         (*fma_fnmsub_<mode>): Likewise.
9100         (*fma_fmaddsub_<mode>): Likewise.
9101         (*fma_fmsubadd_<mode>): Likewise.
9102         (*fmai_fmadd_<mode>): Likewise.
9103         (*fmai_fmsub_<mode>): Likewise.
9104         (*fmai_fnmadd_<mode>): Likewise.
9105         (*fmai_fnmsub_<mode>): Likewise.
9106         (sse_cvtsi2ss): Likewise.
9107         (sse_cvtsi2ssq): Likewise.
9108         (sse_cvtss2si): Likewise.
9109         (sse_cvtss2si_2): Likewise.
9110         (sse_cvtss2siq): Likewise.
9111         (sse_cvtss2siq_2): Likewise.
9112         (sse_cvttss2si): Likewise.
9113         (sse_cvtss2siq_2): Likewise.
9114         (float<sseintvecmodelower><mode>2): Likewise.
9115         (sse2_cvtsd2si_2): Likewise.
9116         (sse2_cvtsd2siq_2): Likewise.
9117         (*<plusminus_insn><mode>3): Likewise.
9118         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
9119         (*<sse4_1_avx2>_mul<mode>3): Likewise.
9120         (ashr<mode>3): Likewise.
9121         (<shift_insn><mode>3): Likewise.
9122         (avx2_<code><mode>3): Likewise.
9123         (*avx2_<code><mode>3): Likewise.
9124         (*andnot<mode>3): Likewise.
9125         (*<code><mode>3) <any_logic>: Likewise.
9126         (abs<mode>2): Likewise.
9127         (avx2_permvar<mode>): Likewise.
9128         (avx2_perm<mode>_1): Likewise.
9129         (*avx_vpermilp<mode>): Likewise.
9130         (avx_vpermilvar<mode>3): Likewise.
9131         (avx2_ashrv<mode>): Likewise.
9132         (avx2_<shift_insn>v<mode>): Likewise.
9133         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
9134         -mavx512cd.
9135         * doc/rtl.texi: Document XImode.
9137 2013-08-21  Jeff Law  <law@redhat.com>
9139         * tree-flow.h (register_jump_thread): Pass vector of edges
9140         instead of each important edge.
9141         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
9142         thread path into a vector and pass that to register_jump_thread.
9143         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
9144         passed in edge vector to the current 3-edge form.
9146         Revert:
9147         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
9149         * dce.c (fini_dce): Call df_analyze again just in case
9150         delete_unmarked_insns removed anything.
9152 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
9154         * reload.h (struct reg_equivs): Rename to ..
9155         (struct reg_equivs_s): .. this.
9157 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
9159         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
9161 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9163         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
9165 2013-08-21  Jeff Law  <law@redhat.com>
9167         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
9168         simplify assignments too.  If the RHS collapses to a singleton
9169         range, then return the value for the range.
9171 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
9173         * config/i386/sse.md (V16): Rename to...
9174         (VMOVE): this.
9175         (mov<mode>): Update iterator name.
9176         (*mov<mode>_internal): Ditto.
9177         (push<mode>1): Ditto.
9178         (movmisalign<mode>): Ditto.
9180 2013-08-20  Jan Hubicka  <jh@suse.cz>
9182         PR bootstrap/58186
9183         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
9184         entry for direct edges.
9185         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
9187 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9189         Revert my last two changes, r201865 and r201864:
9191         Revert r201865:
9192         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9194         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
9195         instances can own GC refs.
9197         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
9198         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
9199         (gcc::context::gt_pch_nx): Likewise.
9200         (gcc::context::gt_pch_nx):  Likewise.
9201         * ggc.h (gt_ggc_mx <T>): New.
9202         (gt_pch_nx_with_op <T>): New.
9203         (gt_pch_nx <T>): New.
9204         * passes.c (opt_pass::gt_ggc_mx): New.
9205         (opt_pass::gt_pch_nx): New.
9206         (opt_pass::gt_pch_nx_with_op): New.
9207         (pass_manager::gt_ggc_mx): New.
9208         (pass_manager::gt_pch_nx): New.
9209         (pass_manager::gt_pch_nx_with_op): New.
9210         (pass_manager::operator new): Use
9211         ggc_internal_cleared_alloc_stat rather than xcalloc.
9212         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
9213         (pass_manager::gt_ggc_mx): New.
9214         (pass_manager::gt_pch_nx): New.
9215         (pass_manager::gt_pch_nx_with_op): New.
9216         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
9217         (opt_pass::operator new): New.
9218         (opt_pass::gt_ggc_mx): New.
9219         (opt_pass::gt_pch_nx): New.
9220         (opt_pass::gt_pch_nx_with_op): New.
9222         Revert r201864:
9223         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9225         * Makefile.in (GTFILES): Add context.h.
9226         * context.c (gcc::context::operator new): New.
9227         (gcc::context::gt_ggc_mx): New.
9228         (gcc::context::gt_pch_nx): New.
9229         (gcc::context::gt_pch_nx): New.
9230         * context.h (gcc::context): Add GTY((user)) marking.
9231         (gcc::context::operator new): New.
9232         (gcc::context::gt_ggc_mx): New.
9233         (gcc::context::gt_pch_nx): New.
9234         (gcc::context::gt_pch_nx): New.
9235         (g): Add GTY marking.
9236         (gt_ggc_mx (gcc::context *)): New.
9237         (gt_pch_nx (gcc::context *)): New.
9238         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
9239         void *cookie)): New.
9240         * gengtype.c (open_base_files) <ifiles>: Add context.h.
9242 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
9244         * dce.c (fini_dce): Call df_analyze again just in case
9245         delete_unmarked_insns removed anything.
9247 2013-08-20  Teresa Johnson  <tejohnson@google.com>
9249         PR rtl-optimizations/57451
9250         * final.c (reemit_insn_block_notes): Prevent lexical blocks
9251         from crossing split section boundaries.
9253 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
9255         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
9256         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
9257         with MULTLIB_DEFAULTS.
9259 2013-08-20  Nick Clifton  <nickc@redhat.com>
9261         * target.def (narrow_volatile_bitfield): Note that the default
9262         value is false, not !TARGET_STRICT_ALIGN.
9263         * doc/tm.texi: Regenerate.
9265 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
9267         Fix LIB_SPEC for systems without libpthread.
9269         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
9270         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
9271         for Android.
9272         * config/i386/linux-common.h: Likewise.
9273         * config/mips/linux-common.h: Likewise.
9275 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
9277         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
9278         checks.
9280 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9282         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
9283         instances can own GC refs.
9285         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
9286         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
9287         (gcc::context::gt_pch_nx): Likewise.
9288         (gcc::context::gt_pch_nx):  Likewise.
9289         * ggc.h (gt_ggc_mx <T>): New.
9290         (gt_pch_nx_with_op <T>): New.
9291         (gt_pch_nx <T>): New.
9292         * passes.c (opt_pass::gt_ggc_mx): New.
9293         (opt_pass::gt_pch_nx): New.
9294         (opt_pass::gt_pch_nx_with_op): New.
9295         (pass_manager::gt_ggc_mx): New.
9296         (pass_manager::gt_pch_nx): New.
9297         (pass_manager::gt_pch_nx_with_op): New.
9298         (pass_manager::operator new): Use
9299         ggc_internal_cleared_alloc_stat rather than xcalloc.
9300         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
9301         (pass_manager::gt_ggc_mx): New.
9302         (pass_manager::gt_pch_nx): New.
9303         (pass_manager::gt_pch_nx_with_op): New.
9304         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
9305         (opt_pass::operator new): New.
9306         (opt_pass::gt_ggc_mx): New.
9307         (opt_pass::gt_pch_nx): New.
9308         (opt_pass::gt_pch_nx_with_op): New.
9310 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9312         * Makefile.in (GTFILES): Add context.h.
9313         * context.c (gcc::context::operator new): New.
9314         (gcc::context::gt_ggc_mx): New.
9315         (gcc::context::gt_pch_nx): New.
9316         (gcc::context::gt_pch_nx): New.
9317         * context.h (gcc::context): Add GTY((user)) marking.
9318         (gcc::context::operator new): New.
9319         (gcc::context::gt_ggc_mx): New.
9320         (gcc::context::gt_pch_nx): New.
9321         (gcc::context::gt_pch_nx): New.
9322         (g): Add GTY marking.
9323         (gt_ggc_mx (gcc::context *)): New.
9324         (gt_pch_nx (gcc::context *)): New.
9325         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
9326         void *cookie)): New.
9327         * gengtype.c (open_base_files) <ifiles>: Add context.h.
9329 2013-08-20  Alan Modra  <amodra@gmail.com>
9331         PR target/57865
9332         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
9333         (rs6000_emit_epilogue): Likewise.
9335 2013-08-19  Dehao Chen  <dehao@google.com>
9337         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
9339 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
9340             Jakub Jelinek  <jakub@redhat.com>
9342         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
9343         (BUILT_IN_FABSD64): Likewise.
9344         (BUILT_IN_FABSD128): Likewise.
9345         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
9346         (fold_builtin_1): Likewise.
9347         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
9348         destination and source operands.
9349         (*abstd2_fpr): Likewise.
9350         (*nabstd2_fpr): Likewise.
9352 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
9354         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
9355         JUMP_P and INSN_P.
9357 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
9359         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
9360         incomplete.
9362 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9364         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
9365         * builtins.c (default_libc_has_function): New.
9366         (gnu_libc_has_function): Ditto.
9367         (no_c99_libc_has_function): Ditto.
9368         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
9369         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
9370         (fold_builtin_sincos): Likewise.
9371         (fold_builtin_cexp): Likewise.
9372         * builtins.def (DEF_C94_BUILTIN): Likewise.
9373         (DEF_C99_BUILTIN): Likewise.
9374         (DEF_C99_C90RES_BUILTIN): Likewise.
9375         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
9376         definitions to using this define.
9377         * config/darwin-protos.h (darwin_libc_has_function): New.
9378         * config/darwin.c (darwin_libc_has_function): Ditto.
9379         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
9380         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
9381         * config/darwin.h: Ditto.
9382         * config/elfos.h: Ditto.
9383         * config/freebsd.h: Ditto.
9384         * config/i386/cygming.h: Ditto.
9385         * config/i386/djgpp.h: Ditto.
9386         * config/i386/i386-interix.h: Ditto.
9387         * config/microblaze/microblaze.h: Ditto.
9388         * config/mmix/mmix.h: Ditto.
9389         * config/gnu-user.h: Ditto.
9390         * config/ia64/hpux.h: Ditto.
9391         * config/pa/pa-hpux.h: Ditto.
9392         * config/pdp11/pdp11.h: Ditto.
9393         * config/picochip/picochip.h: Ditto.
9394         * config/linux.h: Ditto.
9395         * config/netbsd.h: Ditto.
9396         * config/openbsd.h: Ditto.
9397         * config/rs6000/aix43.h: Ditto.
9398         * config/rs6000/aix51.h: Ditto.
9399         * config/rs6000/aix52.h: Ditto.
9400         * config/rs6000/aix53.h: Ditto.
9401         * config/rs6000/aix61.h: Ditto.
9402         * config/rs6000/darwin.h: Ditto.
9403         * config/rs6000/linux.h: Ditto.
9404         * config/rs6000/linux64.h: Ditto.
9405         * config/s390/tpf.h: Ditto.
9406         * config/sol2-10.h: Ditto.
9407         * config/sol2.h: Ditto.
9408         * config/vms/vms.h: Ditto.
9409         * config/vxworks.h: Ditto.
9410         * config/linux-android.c (linux_android_libc_has_function):
9411         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
9412         * config/linux-protos.h (linux_android_libc_has_function):
9413         New declaration.
9414         * config/i386/i386.c (ix86_libc_has_function): New.
9415         * config/i386/i386-protos.h
9416         (ix86_libc_has_function): New declaration.
9417         * config/i386/i386.md
9418         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
9419         ("isinf<mode>2): Likewise.
9420         * convert.c (convert_to_integer): Using new target hook
9421         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
9422         TARGET_C99_FUNCTIONS.
9423         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
9424         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
9425         * coretypes.h (function_class): New enum for different
9426         classes of functions.
9427         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
9428         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
9429         (TARGET_HAS_SINCOS): Likewise.
9430         (TARGET_LIBC_HAS_FUNCTION): New.
9431         * doc/tm.texi: Regenerated.
9432         * targhooks.h (default_libc_has_function): New declaration.
9433         (no_c99_libc_has_function): Ditto.
9434         (gnu_libc_has_function): Ditto.
9435         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
9436         and TARGET_HAS_SINCOS.
9438 2013-08-18  Jan Hubicka  <jh@suse.cz>
9440         * Makeifle-in (ipa-devirt.o): New.
9441         (GTFILES): Add ipa-utils.h and ipa-devirt.c
9442         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
9443         (analyze_functions): Look into possible targets of polymorphic call.
9444         * dumpfile.c (dump_files): Add type-inheritance dump.
9445         * dumpfile.h (TDI_inheritance): New.
9446         * ipa-devirt.c: New file.
9447         * ipa-utils.h (odr_type_d): Forward declare.
9448         (odr_type): New type.
9449         (build_type_inheritance_graph): Declare.
9450         (possible_polymorphic_call_targets): Declare and introduce inline
9451         variant when only edge is pased.
9452         (dump_possible_polymorphic_call_targets): Likewise.
9453         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
9454         * tree.c (type_in_anonymous_namespace_p): Break out from ...
9455         (types_same_for_odr): ... here.
9456         * tree.h (type_in_anonymous_namespace_p): Declare.
9458 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
9460         PR tree-optimization/58006
9461         * tree-parloops.c (take_address_of): Don't ICE if get_name
9462         returns NULL.
9463         (eliminate_local_variables_stmt): Remove clobber stmts.
9465 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
9467         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
9468         error message is issued for an alias to undefined symbol.
9470 2013-08-18  Jan Hubicka  <jh@suse.cz>
9472         * cgraph.c (cgraph_create_indirect_edge): Discover
9473         polymorphic calls and record basic info into indirect_info.
9474         * gimple-fold.c (gimple_fold_call): When doing BINFO based
9475         devirtualization, ignore objc function calls.
9476         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9477         call with no parm index info.
9478         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9479         * tree.c (virtual_method_call_p): New function.
9480         * tree.h (virtual_method_call_p): Declare.
9482 2013-08-16  Jan Hubicka  <jh@suse.cz>
9484         PR middle-end/58179
9485         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9487 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9489         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9490         attribute.
9492 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9494         * gengtype.c (type_for_name): Add special-case support for
9495         locating types within the "gcc::" namespace.
9496         (open_base_files): Emit a "using namespace gcc" directive.
9498 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9500         PR target/58160
9501         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9502         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9504         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9505         array instead of each individual operand as a separate argument.
9506         (emit_fusion_gpr_load): Likewise.
9507         (expand_fusion_gpr_load): Add new function declaration.
9509         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9510         signature to have the operands passed as an array, instead of as
9511         separate arguments.  Allow ZERO_EXTEND to be in the memory
9512         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9513         depend on the register live/dead flags when peepholes are run.
9514         (expand_fusion_gpr_load): New function to be called from the
9515         peephole2 pass, to change the register that addis sets to be the
9516         target register.
9517         (emit_fusion_gpr_load): Change the calling signature to have the
9518         operands passed as an array, instead of as separate arguments.
9519         Allow ZERO_EXTEND to be in the memory address, and also
9520         SIGN_EXTEND if -mpower8-fusion-sign.
9522         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9523         unspec enumeration.
9524         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9525         adjust the register addis loads up in the peephole2 pass.  Do not
9526         depend on the register live/dead state when the peephole pass is done.
9528 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9530         * gengtype.c (create_user_defined_type): Ensure that the kind
9531         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9532         declaration is seen before the GTY((user)) marking.
9534 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9536         PR target/58105
9537         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9539 2013-08-16  Jan Hubicka  <jh@suse.cz>
9541         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9542         arugment expected_type.
9543         (gimple_fold_call): Use it.
9544         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9545         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9546         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9547         (try_make_edge_direct_virtual_call): Likewise.
9548         * tree.c (obj_type_ref_class): New.
9549         * tree.h (obj_type_ref_class): Use it.
9551 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9553         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9554         (rtl_slim_pp): Likewise.
9555         (init_rtl_slim_pretty_print): Likewise.
9556         (dump_value_slim):  Don't call it.  Use local pretty printer.
9557         (dump_insn_slim): Likewise.
9558         (dump_rtl_slim): Likewise.
9559         (str_pattern_slim): Likewise.
9560         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9561         Simplify.
9563 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9565         PR tree-optimization/58164
9566         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9567         walk gimple_goto_dest of GIMPLE_GOTO.
9569         PR tree-optimization/58165
9570         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9571         bi_call must be the last stmt in a bb, don't split_block, instead
9572         use fallthru edge from it and give up if there is none.
9573         Release conds vector when returning early.
9575 2013-08-14  Xinliang David Li  <davidxl@google.com>
9577         * config/i386/i386.c (ix86_option_override_internal):
9578         Remove unused variable and field.
9580 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9582         PR target/57949
9583         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9584         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9585         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9586         and Linux, correct BLKmode alignment when 128-bit alignment is
9587         required and compatibility flag is not set.
9588         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9589         for zero-size arguments when compatibility flag is not set.
9591 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9593         PR tree-optimization/58145
9594         * tree-sra.c (build_ref_for_offset): If prev_base has
9595         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9597 2013-08-14  Xinliang David Li  <davidxl@google.com>
9599         * config/i386/i386.c (ix86_option_override_internal):
9600         Fix uninitialized variable error.
9602 2013-08-14  Xinliang David Li  <davidxl@google.com>
9604         * config/i386/i386.opt: Define two new options.
9605         * config/i386/x86-tune.def: Add arch selector field in macros.
9606         * config/i386/i386.h: Adjust macro definition.
9607         * config/i386/i386.c (ix86_option_override_internal):
9608         Refactor the code.
9609         (parse_mtune_ctrl_str): New function.
9610         (set_ix86_tune_features): New function.
9611         (ix86_function_specific_restore): Call the new helper function.
9613 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9615         PR rtl-optimization/57662
9616         * sel-sched.c (code_motion_process_successors): When the current insn
9617         is removed after the recursive traversal, break from the loop.
9618         Add comments and debug printouts.
9620 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9621             Alexandre Oliva  <aoliva@redhat.com>
9623         PR target/58067
9624         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9625         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9626         there also UNSPEC_PLTOFF.
9628 2013-08-14  Marek Polacek  <polacek@redhat.com>
9630         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9631         NUM_CONDITIONS bit positions.
9633 2013-08-13  Cary Coutant  <ccoutant@google.com>
9635         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9636         (attr_checksum): Hash vector contents instead of pointer.
9637         (attr_checksum_ordered): Likewise.
9639 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9641         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9642         when Pmode != word_mode.  Add length_address attribute.
9643         (sse3_monitor_<mode>): Merge from sse3_monitor and
9644         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9645         Pmode != word_mode.  Update insn length attribute.
9646         * config/i386/i386.c (ix86_option_override_internal): Update
9647         ix86_gen_monitor selection for merged sse3_monitor insn.
9649 2013-08-13  Julian Brown  <julian@codesourcery.com>
9651         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9652         perform invalid legitimization on greater-than-word-size modes for
9653         TARGET_E500_DOUBLE.
9655 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9657         * ira.c (setup_class_translate_array): Use aclass instead of cl
9658         for classes not fully covered by allocno classes.
9660 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9662         PR tree-optimization/57661
9663         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9664         * tree-inline.c (tree_function_versioning): Initialize it.
9665         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9666         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9667         that is not being copied.
9669         PR sanitizer/56417
9670         * asan.c (instrument_strlen_call): Fix typo in comment.
9671         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9673 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9675         * config/mips/mips.md (prefetch): Use lw instead of ld on
9676         loongson in 32bit mode.
9678 2013-08-13  Nick Clifton  <nickc@redhat.com>
9680         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9682 2013-08-13  Jan Hubicka  <jh@suse.cz>
9684         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9685         introduced edge; fix typo in sanity check.
9686         (cgraph_resolve_speculation): Export; improve diagnostic.
9687         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9688         speculation at type mismatch.
9689         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9690         (cgraph_resolve_speculation): Declare.
9691         (symtab_can_be_discarded): New function.
9692         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9693         * ipa-inline-transform.c (speculation_removed): New global var.
9694         (clone_inlined_nodes): See if speculation can be removed.
9695         (inline_call): If speculations was removed, we growths may not match.
9696         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9697         (speculation_useful_p): New function.
9698         (resolve_noninline_speculation): New function.
9699         (inline_small_functions): Resolve useless speculations.
9700         * ipa-inline.h (speculation_useful_p): Declare
9701         * ipa.c (can_replace_by_local_alias): Simplify.
9702         (ipa_profile): Produce speculative calls in non-lto, too;
9703         add simple cost model; produce local aliases.
9705 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9707         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9708         PASS_MANAGER_H.
9710 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9712         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9714 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9716         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9717         iterator instead of X87MODEF.
9719 2013-08-12  Perez Read  <netfirewall@gmail.com>
9721         PR target/58132
9722         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9723         operand 0 for intel asm alternative.
9724         (*movabs<mode>_2): Ditto for operand 1.
9726 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9728         * config/aarch64/arm_none.h
9729         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9731 2013-08-12  Nick Clifton  <nickc@redhat.com>
9733         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9735 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9737         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9738         expand for QI/HImode operand to produce more effictive code for
9739         unsigned char(short) --> float(double) conversion.
9741 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9743         * doc/invoke.texi: Mention that -ftls-model does not force the final
9744         model.
9746 2013-08-12  Marek Polacek  <polacek@redhat.com>
9747             Marc Glisse  <marc.glisse@inria.fr>
9749         PR tree-optimization/57980
9750         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9751         when creating -1 constant.
9753 2013-08-10  Jan Hubicka  <jh@suse.cz>
9755         Workaround binutils PR14342.
9756         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9757         (gimple_init_edge_profiler): Likewise.
9758         (gimple_gen_ic_func_profiler): Likewise.
9760 2013-08-09  Jan Hubicka  <jh@suse.cz>
9762         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9764 2013-08-09  Xinliang David Li  <davidxl@google.com>
9766         * config/i386/stringop.def: New file.
9767         * config/i386/stringop.opt: New file.
9768         * config/i386/i386-opts.h: Include stringopt.def.
9769         * config/i386/i386.opt: Include stringopt.opt.
9770         * config/i386/i386.c (ix86_option_override_internal):
9771         Override default size based stringop inline strategies with options.
9772         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9773         New function.
9775 2013-08-09  Jan Hubicka  <jh@suse.cz>
9777         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9779 2013-08-09  Jan Hubicka  <jh@suse.cz>
9781         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9782         CGRAPH_FREQ_MAX.
9783         (dump_cgraph_node): Dump profile-id.
9784         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9785         and common_target_probability.
9786         * lto-cgraph.c (lto_output_edge): Stream common targets.
9787         (lto_output_node): Stream profile ids.
9788         (input_node): Stream profile ids.
9789         (input_edge): Stream common targets.
9790         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9791         * ipa.c: Include value-prof.h
9792         (ipa_profile_generate_summary): Turn indirect call statement histograms
9793         into common targets.
9794         (ipa_profile): Turn common targets into speculative edges.
9796 2013-08-09  Jan Hubicka  <jh@suse.cz>
9798         * cgraph.h (cgraph_node): Add profile_id.
9799         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9800         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9801         (del_node_map): Update.
9802         (find_func_by_funcdef_no): Replace by ...
9803         (find_func_by_profile_id): ... this one.
9804         (gimple_ic_transform): Do not remove useful histograms when
9805         speculation is not done; dump info when indirect call removal
9806         can happen at LTO.
9807         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9808         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9809         (__gcov_indirect_call_profiler_v2): .. this one.
9810         * profile.h (init_node_map): Update.
9811         * coverage.c (coverage_compute_profile_id): New function.
9812         * coverage.h (coverage_compute_profile_id): Declare.
9813         * tree-profile.c (init_ic_make_global_vars): Make
9814         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9815         (gimple_init_edge_profiler): Update prototype of
9816         __gcov_indirect_call_profiler.
9817         (gimple_gen_ic_func_profiler): Simplify.
9818         (tree_profiling): Use init_node_map
9820 2013-08-09  Jan Hubicka  <jh@suse.cz>
9822         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9823         non-speculative refs.
9824         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9825         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9826         (cgraph_set_call_stmt): Likewise.
9827         (cgraph_create_edge_1): Fix release checking compilatoin;
9828         clear lto_stmt_uid.
9829         (cgraph_free_edge): Free indirect info.
9830         (cgraph_turn_edge_to_speculative): New function.
9831         (cgraph_speculative_call_info): New function.
9832         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9833         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9834         (dump_cgraph_node): Dump speculation.
9835         (verify_edge_count_and_frequency): Accept speculative edges.
9836         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9837         (verify_cgraph_node): Handle speculation.
9838         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9839         (cgraph_set_call_stmt): Update prototype.
9840         (cgraph_make_edge_direct): Update prototype.
9841         (cgraph_speculative_call_info): Declare.
9842         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9843         to change; update call of ipa_find_references.
9844         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9845         lto_stmt_uid and speculative flags.
9846         (ipa_dump_references): Dump speculation.
9847         (ipa_clone_references): Clone speculative flag.
9848         (ipa_clone_referring): Likewise.
9849         (ipa_clone_ref): New function.
9850         (ipa_find_reference): Look into lto_stmt_uids
9851         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9852         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9853         (ipa_find_reference): Update declaration.
9854         (ipa_clone_ref): Declare.
9855         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9856         stream speculative flag.
9857         (lto_output_ref): Stream statements uids and speculation.
9858         (input_ref): Likewise.
9859         (input_edge): Stream speuclation.
9860         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9861         (cgraph_set_call_stmt_including_clones): Handle speculation.
9862         * ipa-inline.c (heap_edge_removal_hook): New function.
9863         (inline_small_functions): Register it.
9864         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9865         also initialize refs.
9866         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9867         edge to change.
9868         (try_make_edge_direct_simple_call): Likewise.
9869         (try_make_edge_direct_simple_call): Likewise.
9870         (update_indirect_edges_after_inlining): Likewise.
9871         (remove_described_reference): Look proper lto_stmt_uid.
9872         (propagate_controlled_uses): Likewise.
9873         (propagate_controlled_uses): Liekwise.
9874         * tree-inline.c (copy_bb): Copy speculative edges.
9875         (redirect_all_calls): New function.
9876         (copy_cfg_body): Do redirection after loop info is updated.
9877         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9879 2013-08-09  Jan Hubicka  <jh@suse.cz>
9881         * lto-streamer-out.c (output_function): Renumber PHIs.
9882         * lto-streamer-in.c (input_function): Likewise.
9884 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9886         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9887         (get_lane_unsigned): Likewise.
9888         (dup_lane_scalar): Likewise.
9889         (get_lane): enable for VALL.
9890         * config/aarch64/aarch64-simd.md
9891         (aarch64_dup_lane_scalar<mode>): Remove.
9892         (aarch64_get_lane_signed<mode>): Likewise.
9893         (aarch64_get_lane_unsigned<mode>): Likewise.
9894         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9895         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9896         (aarch64_get_lane<mode>): Enable for all vector modes.
9897         (aarch64_get_lanedi): Remove misleading constraints.
9898         * config/aarch64/arm_neon.h
9899         (__aarch64_vget_lane_any): Define.
9900         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9901         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9902         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9903         * config/aarch64/iterators.md (VDQQH): New.
9904         (VDQQHS): Likewise.
9905         (vwcore): Likewise.
9907 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9909         * configure.ac: Add GAS check for LEON instructions on SPARC.
9910         * configure: Regenerate.
9911         * config.in: Likewise.
9912         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9913         sparc*-*-* block.
9914         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9915         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9916         for LEON or LEON3.
9917         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9918         (AS_LEON_FLAG): New macro.
9919         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9920         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9921         Deal with LEON and LEON3 for the memory model.
9922         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9923         (atomic_compare_and_swap<mode>_1): Likewise.
9924         (*atomic_compare_and_swap<mode>_1): Likewise.
9926 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9928         * config/arm/neon.md (vcond): Fix floating-point vector
9929         comparisons against 0.
9931 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9933         * lra-constraints.c (emit_spill_move): Remove assert.
9934         (process_alt_operands): Add more debugging
9935         output.  Increase reject for spilling into memory.  Decrease
9936         reject for reloading scratch.
9937         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9939 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9941         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9942         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9943         micromips incompatible.  Add nan2008.
9944         (MULTILIB_DIRNAMES): Add nan2008.
9945         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9946         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9947         and micromips incompatible.  Add nan2008.
9948         (MULTILIB_DIRNAMES): Add nan2008.
9949         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9951 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9953         PR rtl-optimization/58079
9954         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9955         simplify_comparison has widened a comparison with an integer.
9957 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9959         * config/arm/neon.md (movmisalign<mode>): Disable when we
9960         don't allow unaligned accesses.
9961         (*movmisalign<mode>_neon_store): Likewise.
9962         (*movmisalign<mode>_neon_load): Likewise.
9963         (*movmisalign<mode>_neon_store): Likewise.
9964         (*movmisalign<mode>_neon_load): Likewise.
9966 2013-08-08  Jan Hubicka  <jh@suse.cz>
9968         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9969         (make_pass_rebuild_cgraph_edges): Also clear references.
9970         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9971         * ipa-inline-transform.c (inline_transform): Remove all references
9972         after inlining.
9973         * cgraphunit.c (expand_function): Remove all references after
9974         expansion.
9975         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9976         (ipa_find_reference): Rewrite to iterator.
9977         (remove_stmt_references): Likewise.
9978         (ipa_clear_stmts_in_references): New function.
9979         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9980         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9981         clear references.
9982         * ipa-split.c (split_function): Remove references in split function.
9984 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9986         PR target/57431
9987         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9988         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9990 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9992         PR target/56979
9993         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9994         suggested mode for the assignment isn't compatible with the
9995         registers required.
9997 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9999         PR target/58065
10000         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
10002 2013-08-07  Xinliang David Li  <davidxl@google.com>
10004         * config/i386/i386.opt: New option -mtune-ctrl=.
10005         * config/i386/x86-tune.def: New file.
10006         * config/i386/i386.h: include x86-tune.def.
10007         * config/i386/i386.c (ix86_option_override_internal):
10008         Parsing -mtune-ctrl= option and set tune features.
10010 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
10012         PR other/12081
10013         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
10014         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
10015         to gen_2arg_fn_t.
10017 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
10019         * rtl.h (update_alignments): Declare.
10020         * final.c (grow_label_align): New function extracted from...
10021         (shorten_branches): ...here.  Call it.
10022         (update_alignments): New function.
10023         * reorg.c (sibling_labels): New variable.
10024         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
10025         the new label along with it onto the sibling_labels vector.
10026         (fill_simple_delay_slots): Adjust call to get_label_before.
10027         (fill_slots_from_thread): Likewise.
10028         (relax_delay_slots): Likewise.
10029         (make_return_insns): Likewise.
10030         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
10032 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
10034         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
10035         document its semantics.
10036         (diagnostic_report_diagnostic): Adjust accordingly.
10038 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
10040         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
10041         (sparc_option_override): ...and port to new C++ pass API.
10042         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
10044 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
10046         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
10048 2013-08-06  Caroline Tice  <cmtice@google.com>
10050         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
10051         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
10052         * tree-pass.h: Add pass_vtable_verify.
10053         * varasm.c (assemble_variable): Add code to properly set the comdat
10054         section and name for the .vtable_map_vars section.
10055         (assemble_vtyv_preinit_initializer): New function.
10056         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
10057         LINK_ONCE flag.
10058         * output.h: Add function decl for assemble_vtv_preinit_initializer.
10059         * vtable-verify.c: New file.
10060         * vtable-verify.h: New file.
10061         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
10062         initialiation levels.
10063         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
10064         * passes.def: Insert pass_vtable_verify.
10065         * aclocal.m4: Reorder includes.
10066         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
10067         -fvtv-counts options.
10068         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
10069         as appropriate, if -fvtable-verify=... is used.
10070         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
10071         -fvtable-verify=... is used.
10072         * Makefile.in (OBJS):  Add vtable-verify.o to list.
10073         (vtable-verify.o): Add new build rule.
10074         (GTFILES): Add vtable-verify.c to list.
10075         * common.opt (fvtable-verify=): New flag.
10076         (vtv_priority): Values for fvtable-verify= flag.
10077         (fvtv-counts): New flag.
10078         (fvtv-debug): New flag.
10079         * tree.h (save_vtable_map_decl): New extern function decl.
10081 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
10083         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
10084         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
10085         (pass_data_rl78_devirt): ...new pass_data instance and...
10086         (make_pass_rl78_devirt): ...new function.
10087         (rl78_asm_file_start): Port pass registration to new C++ API.
10089 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
10091         * coretypes.h (rtl_opt_pass): Add.
10092         (gcc::context): Add.
10093         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
10094         (epiphany_init): Port to new C++ pass API.
10095         (epiphany_optimize_mode_switching): Likewise.
10096         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
10097         (pass_manager::get_pass_mode_switching): New.
10098         (pass_manager::get_pass_peephole2): New.
10099         * mode-switching.c (pass_mode_switching): Add clone method.
10100         * recog.c (pass_peephole2): Add clone method.
10101         (pass_split_all_insns): Add clone method.
10103 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
10105         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
10106         (mips_option_override): ...here, porting to new C++ API for passes.
10108 2013-08-06  Jan Hubicka  <jh@suse.cz>
10110         * cgraph.c (cgraph_get_body): New function based on lto.c
10111         implementation.
10112         * cgraph.h (cgraph_get_body): Declare.
10113         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
10114         LTO paths.
10115         * cgraphunit.c (expand_function): Get body prior expanding.
10116         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
10117         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
10118         really need.
10119         * passes.c (do_per_function_toporder): Get body.
10120         * tree-inline.c (expand_call_inline): Get body prior inlining it.
10121         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
10123 2013-08-06  Martin Jambor  <mjambor@suse.cz>
10125         PR fortran/57987
10126         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
10127         is not re-finalized.  Rename second parameter to no_collect.
10129 2013-08-06  Martin Jambor  <mjambor@suse.cz>
10131         PR middle-end/58041
10132         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
10133         MEM_REF has proper alignment information.
10135 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
10137         PR other/12081
10138         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
10139         class insn_gen_fn.
10140         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
10141         rtx (*) (rtx, ...) with insn_gen_fn.
10142         * genoutput.c (output_insn_data): Cast gen_? function pointers to
10143         insn_gen_fn::stored_funcptr.  Add initializer braces.
10145 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10147         Rewrite how instances of passes are cloned to remove assumptions
10148         about their sizes (thus allowing pass subclasses to have
10149         additional data fields, albeit non-GC-managed ones at this point).
10151         * passes.c (make_pass_instance): Now that passes have clone
10152         methods, rewrite this function to eliminate XNEW and memcpy
10153         calls that used hardcoded sizes.  Since this function no longer
10154         creates pass instances, rename it to...
10155         (add_pass_instance): ...this.  Document the old way that passes were
10156         numbered and flagged, and rework this function to continue using it.
10157         (next_pass_1): Add an initial_pass argument for use by
10158         add_pass_instance.
10159         (position_pass): When adding multiple instances of a pass, use
10160         the pass's clone method, rather than relying on the XNEW/memcpy
10161         within the former make_pass_instance (now add_pass_instance).
10162         (pass_manager::pass_manager): When invoking next_pass_1, also supply
10163         the initial instance of the current pass within the pass manager.
10165 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10167         This is the automated part of the conversion of passes from C
10168         structs to C++ classes.
10170         Patch autogenerated by refactor_passes.py from
10171         https://github.com/davidmalcolm/gcc-refactoring-scripts
10172         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
10174         * asan.c (pass_asan): Convert from a global struct to a subclass of
10175         gimple_opt_pass along with...
10176         (pass_data_asan): ...new pass_data instance and...
10177         (make_pass_asan): ...new function.
10178         (pass_asan_O0): Convert from a global struct to a subclass of
10179         gimple_opt_pass along with...
10180         (pass_data_asan_O0): ...new pass_data instance and...
10181         (make_pass_asan_O0): ...new function.
10182         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
10183         subclass of rtl_opt_pass along with...
10184         (pass_data_inc_dec): ...new pass_data instance and...
10185         (make_pass_inc_dec): ...new function.
10186         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
10187         a subclass of rtl_opt_pass along with...
10188         (pass_data_reorder_blocks): ...new pass_data instance and...
10189         (make_pass_reorder_blocks): ...new function.
10190         (pass_duplicate_computed_gotos): Convert from a global struct to a
10191         subclass of rtl_opt_pass along with...
10192         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
10193         (make_pass_duplicate_computed_gotos): ...new function.
10194         (pass_partition_blocks): Convert from a global struct to a subclass of
10195         rtl_opt_pass along with...
10196         (pass_data_partition_blocks): ...new pass_data instance and...
10197         (make_pass_partition_blocks): ...new function.
10198         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
10199         struct to a subclass of rtl_opt_pass along with...
10200         (pass_data_branch_target_load_optimize1): ...new pass_data instance
10201         and...
10202         (make_pass_branch_target_load_optimize1): ...new function.
10203         (pass_branch_target_load_optimize2): Convert from a global struct to a
10204         subclass of rtl_opt_pass along with...
10205         (pass_data_branch_target_load_optimize2): ...new pass_data instance
10206         and...
10207         (make_pass_branch_target_load_optimize2): ...new function.
10208         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
10209         of rtl_opt_pass along with...
10210         (pass_data_jump): ...new pass_data instance and...
10211         (make_pass_jump): ...new function.
10212         (pass_jump2): Convert from a global struct to a subclass of
10213         rtl_opt_pass along with...
10214         (pass_data_jump2): ...new pass_data instance and...
10215         (make_pass_jump2): ...new function.
10216         * cfgexpand.c (pass_expand): Convert from a global struct to a
10217         subclass of rtl_opt_pass along with...
10218         (pass_data_expand): ...new pass_data instance and...
10219         (make_pass_expand): ...new function.
10220         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
10221         of rtl_opt_pass along with...
10222         (pass_data_free_cfg): ...new pass_data instance and...
10223         (make_pass_free_cfg): ...new function.
10224         (pass_into_cfg_layout_mode): Convert from a global struct to a
10225         subclass of rtl_opt_pass along with...
10226         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
10227         (make_pass_into_cfg_layout_mode): ...new function.
10228         (pass_outof_cfg_layout_mode): Convert from a global struct to a
10229         subclass of rtl_opt_pass along with...
10230         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
10231         (make_pass_outof_cfg_layout_mode): ...new function.
10232         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
10233         struct to a subclass of gimple_opt_pass along with...
10234         (pass_data_build_cgraph_edges): ...new pass_data instance and...
10235         (make_pass_build_cgraph_edges): ...new function.
10236         (pass_rebuild_cgraph_edges): Convert from a global struct to a
10237         subclass of gimple_opt_pass along with...
10238         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
10239         (make_pass_rebuild_cgraph_edges): ...new function.
10240         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
10241         subclass of gimple_opt_pass along with...
10242         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
10243         and...
10244         (make_pass_remove_cgraph_callee_edges): ...new function.
10245         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
10246         struct to a subclass of rtl_opt_pass along with...
10247         (pass_data_stack_adjustments): ...new pass_data instance and...
10248         (make_pass_stack_adjustments): ...new function.
10249         * combine.c (pass_combine): Convert from a global struct to a subclass
10250         of rtl_opt_pass along with...
10251         (pass_data_combine): ...new pass_data instance and...
10252         (make_pass_combine): ...new function.
10253         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
10254         global struct to a subclass of rtl_opt_pass along with...
10255         (pass_data_compare_elim_after_reload): ...new pass_data instance
10256         and...
10257         (make_pass_compare_elim_after_reload): ...new function.
10258         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
10259         of rtl_opt_pass along with...
10260         (pass_data_rtl_cprop): ...new pass_data instance and...
10261         (make_pass_rtl_cprop): ...new function.
10262         * cse.c (pass_cse): Convert from a global struct to a subclass of
10263         rtl_opt_pass along with...
10264         (pass_data_cse): ...new pass_data instance and...
10265         (make_pass_cse): ...new function.
10266         (pass_cse2): Convert from a global struct to a subclass of
10267         rtl_opt_pass along with...
10268         (pass_data_cse2): ...new pass_data instance and...
10269         (make_pass_cse2): ...new function.
10270         (pass_cse_after_global_opts): Convert from a global struct to a
10271         subclass of rtl_opt_pass along with...
10272         (pass_data_cse_after_global_opts): ...new pass_data instance and...
10273         (make_pass_cse_after_global_opts): ...new function.
10274         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
10275         of rtl_opt_pass along with...
10276         (pass_data_ud_rtl_dce): ...new pass_data instance and...
10277         (make_pass_ud_rtl_dce): ...new function.
10278         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
10279         rtl_opt_pass along with...
10280         (pass_data_fast_rtl_dce): ...new pass_data instance and...
10281         (make_pass_fast_rtl_dce): ...new function.
10282         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
10283         a subclass of rtl_opt_pass along with...
10284         (pass_data_df_initialize_opt): ...new pass_data instance and...
10285         (make_pass_df_initialize_opt): ...new function.
10286         (pass_df_initialize_no_opt): Convert from a global struct to a
10287         subclass of rtl_opt_pass along with...
10288         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
10289         (make_pass_df_initialize_no_opt): ...new function.
10290         (pass_df_finish): Convert from a global struct to a subclass of
10291         rtl_opt_pass along with...
10292         (pass_data_df_finish): ...new pass_data instance and...
10293         (make_pass_df_finish): ...new function.
10294         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
10295         rtl_opt_pass along with...
10296         (pass_data_rtl_dse1): ...new pass_data instance and...
10297         (make_pass_rtl_dse1): ...new function.
10298         (pass_rtl_dse2): Convert from a global struct to a subclass of
10299         rtl_opt_pass along with...
10300         (pass_data_rtl_dse2): ...new pass_data instance and...
10301         (make_pass_rtl_dse2): ...new function.
10302         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
10303         subclass of rtl_opt_pass along with...
10304         (pass_data_dwarf2_frame): ...new pass_data instance and...
10305         (make_pass_dwarf2_frame): ...new function.
10306         * except.c (pass_set_nothrow_function_flags): Convert from a global
10307         struct to a subclass of rtl_opt_pass along with...
10308         (pass_data_set_nothrow_function_flags): ...new pass_data instance
10309         and...
10310         (make_pass_set_nothrow_function_flags): ...new function.
10311         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
10312         subclass of rtl_opt_pass along with...
10313         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
10314         and...
10315         (make_pass_convert_to_eh_region_ranges): ...new function.
10316         * final.c (pass_compute_alignments): Convert from a global struct to a
10317         subclass of rtl_opt_pass along with...
10318         (pass_data_compute_alignments): ...new pass_data instance and...
10319         (make_pass_compute_alignments): ...new function.
10320         (pass_final): Convert from a global struct to a subclass of
10321         rtl_opt_pass along with...
10322         (pass_data_final): ...new pass_data instance and...
10323         (make_pass_final): ...new function.
10324         (pass_shorten_branches): Convert from a global struct to a subclass of
10325         rtl_opt_pass along with...
10326         (pass_data_shorten_branches): ...new pass_data instance and...
10327         (make_pass_shorten_branches): ...new function.
10328         (pass_clean_state): Convert from a global struct to a subclass of
10329         rtl_opt_pass along with...
10330         (pass_data_clean_state): ...new pass_data instance and...
10331         (make_pass_clean_state): ...new function.
10332         * function.c (pass_instantiate_virtual_regs): Convert from a global
10333         struct to a subclass of rtl_opt_pass along with...
10334         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
10335         (make_pass_instantiate_virtual_regs): ...new function.
10336         (pass_leaf_regs): Convert from a global struct to a subclass of
10337         rtl_opt_pass along with...
10338         (pass_data_leaf_regs): ...new pass_data instance and...
10339         (make_pass_leaf_regs): ...new function.
10340         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
10341         subclass of rtl_opt_pass along with...
10342         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
10343         and...
10344         (make_pass_thread_prologue_and_epilogue): ...new function.
10345         (pass_match_asm_constraints): Convert from a global struct to a
10346         subclass of rtl_opt_pass along with...
10347         (pass_data_match_asm_constraints): ...new pass_data instance and...
10348         (make_pass_match_asm_constraints): ...new function.
10349         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
10350         subclass of rtl_opt_pass along with...
10351         (pass_data_rtl_fwprop): ...new pass_data instance and...
10352         (make_pass_rtl_fwprop): ...new function.
10353         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
10354         rtl_opt_pass along with...
10355         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
10356         (make_pass_rtl_fwprop_addr): ...new function.
10357         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
10358         rtl_opt_pass along with...
10359         (pass_data_rtl_pre): ...new pass_data instance and...
10360         (make_pass_rtl_pre): ...new function.
10361         (pass_rtl_hoist): Convert from a global struct to a subclass of
10362         rtl_opt_pass along with...
10363         (pass_data_rtl_hoist): ...new pass_data instance and...
10364         (make_pass_rtl_hoist): ...new function.
10365         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
10366         subclass of gimple_opt_pass along with...
10367         (pass_data_lower_cf): ...new pass_data instance and...
10368         (make_pass_lower_cf): ...new function.
10369         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
10370         from a global struct to a subclass of gimple_opt_pass along with...
10371         (pass_data_strength_reduction): ...new pass_data instance and...
10372         (make_pass_strength_reduction): ...new function.
10373         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
10374         of rtl_opt_pass along with...
10375         (pass_data_rtl_ifcvt): ...new pass_data instance and...
10376         (make_pass_rtl_ifcvt): ...new function.
10377         (pass_if_after_combine): Convert from a global struct to a subclass of
10378         rtl_opt_pass along with...
10379         (pass_data_if_after_combine): ...new pass_data instance and...
10380         (make_pass_if_after_combine): ...new function.
10381         (pass_if_after_reload): Convert from a global struct to a subclass of
10382         rtl_opt_pass along with...
10383         (pass_data_if_after_reload): ...new pass_data instance and...
10384         (make_pass_if_after_reload): ...new function.
10385         * init-regs.c (pass_initialize_regs): Convert from a global struct to
10386         a subclass of rtl_opt_pass along with...
10387         (pass_data_initialize_regs): ...new pass_data instance and...
10388         (make_pass_initialize_regs): ...new function.
10389         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
10390         of ipa_opt_pass_d along with...
10391         (pass_data_ipa_cp): ...new pass_data instance and...
10392         (make_pass_ipa_cp): ...new function.
10393         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
10394         global struct to a subclass of gimple_opt_pass along with...
10395         (pass_data_inline_parameters): ...new pass_data instance and...
10396         (make_pass_inline_parameters): ...new function.
10397         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
10398         subclass of gimple_opt_pass along with...
10399         (pass_data_early_inline): ...new pass_data instance and...
10400         (make_pass_early_inline): ...new function.
10401         (pass_ipa_inline): Convert from a global struct to a subclass of
10402         ipa_opt_pass_d along with...
10403         (pass_data_ipa_inline): ...new pass_data instance and...
10404         (make_pass_ipa_inline): ...new function.
10405         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
10406         struct to a subclass of gimple_opt_pass along with...
10407         (pass_data_local_pure_const): ...new pass_data instance and...
10408         (make_pass_local_pure_const): ...new function.
10409         (pass_ipa_pure_const): Convert from a global struct to a subclass of
10410         ipa_opt_pass_d along with...
10411         (pass_data_ipa_pure_const): ...new pass_data instance and...
10412         (make_pass_ipa_pure_const): ...new function.
10413         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
10414         to a subclass of ipa_opt_pass_d along with...
10415         (pass_data_ipa_reference): ...new pass_data instance and...
10416         (make_pass_ipa_reference): ...new function.
10417         * ipa-split.c (pass_split_functions): Convert from a global struct to
10418         a subclass of gimple_opt_pass along with...
10419         (pass_data_split_functions): ...new pass_data instance and...
10420         (make_pass_split_functions): ...new function.
10421         (pass_feedback_split_functions): Convert from a global struct to a
10422         subclass of gimple_opt_pass along with...
10423         (pass_data_feedback_split_functions): ...new pass_data instance and...
10424         (make_pass_feedback_split_functions): ...new function.
10425         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
10426         global struct to a subclass of simple_ipa_opt_pass along with...
10427         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
10428         instance and...
10429         (make_pass_ipa_function_and_variable_visibility): ...new function.
10430         (pass_ipa_free_inline_summary): Convert from a global struct to a
10431         subclass of simple_ipa_opt_pass along with...
10432         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
10433         (make_pass_ipa_free_inline_summary): ...new function.
10434         (pass_ipa_whole_program_visibility): Convert from a global struct to a
10435         subclass of ipa_opt_pass_d along with...
10436         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
10437         and...
10438         (make_pass_ipa_whole_program_visibility): ...new function.
10439         (pass_ipa_profile): Convert from a global struct to a subclass of
10440         ipa_opt_pass_d along with...
10441         (pass_data_ipa_profile): ...new pass_data instance and...
10442         (make_pass_ipa_profile): ...new function.
10443         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
10444         ipa_opt_pass_d along with...
10445         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
10446         (make_pass_ipa_cdtor_merge): ...new function.
10447         * ira.c (pass_ira): Convert from a global struct to a subclass of
10448         rtl_opt_pass along with...
10449         (pass_data_ira): ...new pass_data instance and...
10450         (make_pass_ira): ...new function.
10451         (pass_reload): Convert from a global struct to a subclass of
10452         rtl_opt_pass along with...
10453         (pass_data_reload): ...new pass_data instance and...
10454         (make_pass_reload): ...new function.
10455         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
10456         subclass of rtl_opt_pass along with...
10457         (pass_data_cleanup_barriers): ...new pass_data instance and...
10458         (make_pass_cleanup_barriers): ...new function.
10459         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
10460         of rtl_opt_pass along with...
10461         (pass_data_loop2): ...new pass_data instance and...
10462         (make_pass_loop2): ...new function.
10463         (pass_rtl_loop_init): Convert from a global struct to a subclass of
10464         rtl_opt_pass along with...
10465         (pass_data_rtl_loop_init): ...new pass_data instance and...
10466         (make_pass_rtl_loop_init): ...new function.
10467         (pass_rtl_loop_done): Convert from a global struct to a subclass of
10468         rtl_opt_pass along with...
10469         (pass_data_rtl_loop_done): ...new pass_data instance and...
10470         (make_pass_rtl_loop_done): ...new function.
10471         (pass_rtl_move_loop_invariants): Convert from a global struct to a
10472         subclass of rtl_opt_pass along with...
10473         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
10474         (make_pass_rtl_move_loop_invariants): ...new function.
10475         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10476         rtl_opt_pass along with...
10477         (pass_data_rtl_unswitch): ...new pass_data instance and...
10478         (make_pass_rtl_unswitch): ...new function.
10479         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10480         subclass of rtl_opt_pass along with...
10481         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10482         and...
10483         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10484         (pass_rtl_doloop): Convert from a global struct to a subclass of
10485         rtl_opt_pass along with...
10486         (pass_data_rtl_doloop): ...new pass_data instance and...
10487         (make_pass_rtl_doloop): ...new function.
10488         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10489         a subclass of rtl_opt_pass along with...
10490         (pass_data_lower_subreg): ...new pass_data instance and...
10491         (make_pass_lower_subreg): ...new function.
10492         (pass_lower_subreg2): Convert from a global struct to a subclass of
10493         rtl_opt_pass along with...
10494         (pass_data_lower_subreg2): ...new pass_data instance and...
10495         (make_pass_lower_subreg2): ...new function.
10496         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10497         struct to a subclass of ipa_opt_pass_d along with...
10498         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10499         (make_pass_ipa_lto_gimple_out): ...new function.
10500         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10501         of ipa_opt_pass_d along with...
10502         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10503         (make_pass_ipa_lto_finish_out): ...new function.
10504         * mode-switching.c (pass_mode_switching): Convert from a global struct
10505         to a subclass of rtl_opt_pass along with...
10506         (pass_data_mode_switching): ...new pass_data instance and...
10507         (make_pass_mode_switching): ...new function.
10508         * modulo-sched.c (pass_sms): Convert from a global struct to a
10509         subclass of rtl_opt_pass along with...
10510         (pass_data_sms): ...new pass_data instance and...
10511         (make_pass_sms): ...new function.
10512         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10513         subclass of gimple_opt_pass along with...
10514         (pass_data_expand_omp): ...new pass_data instance and...
10515         (make_pass_expand_omp): ...new function.
10516         (pass_lower_omp): Convert from a global struct to a subclass of
10517         gimple_opt_pass along with...
10518         (pass_data_lower_omp): ...new pass_data instance and...
10519         (make_pass_lower_omp): ...new function.
10520         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10521         of gimple_opt_pass along with...
10522         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10523         (make_pass_diagnose_omp_blocks): ...new function.
10524         * passes.c (pass_early_local_passes): Convert from a global struct to
10525         a subclass of simple_ipa_opt_pass along with...
10526         (pass_data_early_local_passes): ...new pass_data instance and...
10527         (make_pass_early_local_passes): ...new function.
10528         (pass_all_early_optimizations): Convert from a global struct to a
10529         subclass of gimple_opt_pass along with...
10530         (pass_data_all_early_optimizations): ...new pass_data instance and...
10531         (make_pass_all_early_optimizations): ...new function.
10532         (pass_all_optimizations): Convert from a global struct to a subclass
10533         of gimple_opt_pass along with...
10534         (pass_data_all_optimizations): ...new pass_data instance and...
10535         (make_pass_all_optimizations): ...new function.
10536         (pass_all_optimizations_g): Convert from a global struct to a subclass
10537         of gimple_opt_pass along with...
10538         (pass_data_all_optimizations_g): ...new pass_data instance and...
10539         (make_pass_all_optimizations_g): ...new function.
10540         (pass_rest_of_compilation): Convert from a global struct to a subclass
10541         of rtl_opt_pass along with...
10542         (pass_data_rest_of_compilation): ...new pass_data instance and...
10543         (make_pass_rest_of_compilation): ...new function.
10544         (pass_postreload): Convert from a global struct to a subclass of
10545         rtl_opt_pass along with...
10546         (pass_data_postreload): ...new pass_data instance and...
10547         (make_pass_postreload): ...new function.
10548         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10549         subclass of rtl_opt_pass along with...
10550         (pass_data_gcse2): ...new pass_data instance and...
10551         (make_pass_gcse2): ...new function.
10552         * postreload.c (pass_postreload_cse): Convert from a global struct to
10553         a subclass of rtl_opt_pass along with...
10554         (pass_data_postreload_cse): ...new pass_data instance and...
10555         (make_pass_postreload_cse): ...new function.
10556         * predict.c (pass_profile): Convert from a global struct to a subclass
10557         of gimple_opt_pass along with...
10558         (pass_data_profile): ...new pass_data instance and...
10559         (make_pass_profile): ...new function.
10560         (pass_strip_predict_hints): Convert from a global struct to a subclass
10561         of gimple_opt_pass along with...
10562         (pass_data_strip_predict_hints): ...new pass_data instance and...
10563         (make_pass_strip_predict_hints): ...new function.
10564         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10565         of rtl_opt_pass along with...
10566         (pass_data_peephole2): ...new pass_data instance and...
10567         (make_pass_peephole2): ...new function.
10568         (pass_split_all_insns): Convert from a global struct to a subclass of
10569         rtl_opt_pass along with...
10570         (pass_data_split_all_insns): ...new pass_data instance and...
10571         (make_pass_split_all_insns): ...new function.
10572         (pass_split_after_reload): Convert from a global struct to a subclass
10573         of rtl_opt_pass along with...
10574         (pass_data_split_after_reload): ...new pass_data instance and...
10575         (make_pass_split_after_reload): ...new function.
10576         (pass_split_before_regstack): Convert from a global struct to a
10577         subclass of rtl_opt_pass along with...
10578         (pass_data_split_before_regstack): ...new pass_data instance and...
10579         (make_pass_split_before_regstack): ...new function.
10580         (pass_split_before_sched2): Convert from a global struct to a subclass
10581         of rtl_opt_pass along with...
10582         (pass_data_split_before_sched2): ...new pass_data instance and...
10583         (make_pass_split_before_sched2): ...new function.
10584         (pass_split_for_shorten_branches): Convert from a global struct to a
10585         subclass of rtl_opt_pass along with...
10586         (pass_data_split_for_shorten_branches): ...new pass_data instance
10587         and...
10588         (make_pass_split_for_shorten_branches): ...new function.
10589         * ree.c (pass_ree): Convert from a global struct to a subclass of
10590         rtl_opt_pass along with...
10591         (pass_data_ree): ...new pass_data instance and...
10592         (make_pass_ree): ...new function.
10593         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10594         subclass of rtl_opt_pass along with...
10595         (pass_data_stack_regs): ...new pass_data instance and...
10596         (make_pass_stack_regs): ...new function.
10597         (pass_stack_regs_run): Convert from a global struct to a subclass of
10598         rtl_opt_pass along with...
10599         (pass_data_stack_regs_run): ...new pass_data instance and...
10600         (make_pass_stack_regs_run): ...new function.
10601         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10602         subclass of rtl_opt_pass along with...
10603         (pass_data_cprop_hardreg): ...new pass_data instance and...
10604         (make_pass_cprop_hardreg): ...new function.
10605         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10606         subclass of rtl_opt_pass along with...
10607         (pass_data_reginfo_init): ...new pass_data instance and...
10608         (make_pass_reginfo_init): ...new function.
10609         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10610         of rtl_opt_pass along with...
10611         (pass_data_regmove): ...new pass_data instance and...
10612         (make_pass_regmove): ...new function.
10613         * regrename.c (pass_regrename): Convert from a global struct to a
10614         subclass of rtl_opt_pass along with...
10615         (pass_data_regrename): ...new pass_data instance and...
10616         (make_pass_regrename): ...new function.
10617         * reorg.c (pass_delay_slots): Convert from a global struct to a
10618         subclass of rtl_opt_pass along with...
10619         (pass_data_delay_slots): ...new pass_data instance and...
10620         (make_pass_delay_slots): ...new function.
10621         (pass_machine_reorg): Convert from a global struct to a subclass of
10622         rtl_opt_pass along with...
10623         (pass_data_machine_reorg): ...new pass_data instance and...
10624         (make_pass_machine_reorg): ...new function.
10625         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10626         of rtl_opt_pass along with...
10627         (pass_data_sched): ...new pass_data instance and...
10628         (make_pass_sched): ...new function.
10629         (pass_sched2): Convert from a global struct to a subclass of
10630         rtl_opt_pass along with...
10631         (pass_data_sched2): ...new pass_data instance and...
10632         (make_pass_sched2): ...new function.
10633         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10634         to a subclass of rtl_opt_pass along with...
10635         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10636         (make_pass_stack_ptr_mod): ...new function.
10637         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10638         to a subclass of rtl_opt_pass along with...
10639         (pass_data_rtl_store_motion): ...new pass_data instance and...
10640         (make_pass_rtl_store_motion): ...new function.
10641         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10642         of gimple_opt_pass along with...
10643         (pass_data_tracer): ...new pass_data instance and...
10644         (make_pass_tracer): ...new function.
10645         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10646         to a subclass of gimple_opt_pass along with...
10647         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10648         (make_pass_diagnose_tm_blocks): ...new function.
10649         (pass_lower_tm): Convert from a global struct to a subclass of
10650         gimple_opt_pass along with...
10651         (pass_data_lower_tm): ...new pass_data instance and...
10652         (make_pass_lower_tm): ...new function.
10653         (pass_tm_init): Convert from a global struct to a subclass of
10654         gimple_opt_pass along with...
10655         (pass_data_tm_init): ...new pass_data instance and...
10656         (make_pass_tm_init): ...new function.
10657         (pass_tm_mark): Convert from a global struct to a subclass of
10658         gimple_opt_pass along with...
10659         (pass_data_tm_mark): ...new pass_data instance and...
10660         (make_pass_tm_mark): ...new function.
10661         (pass_tm_edges): Convert from a global struct to a subclass of
10662         gimple_opt_pass along with...
10663         (pass_data_tm_edges): ...new pass_data instance and...
10664         (make_pass_tm_edges): ...new function.
10665         (pass_tm_memopt): Convert from a global struct to a subclass of
10666         gimple_opt_pass along with...
10667         (pass_data_tm_memopt): ...new pass_data instance and...
10668         (make_pass_tm_memopt): ...new function.
10669         (pass_ipa_tm): Convert from a global struct to a subclass of
10670         simple_ipa_opt_pass along with...
10671         (pass_data_ipa_tm): ...new pass_data instance and...
10672         (make_pass_ipa_tm): ...new function.
10673         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10674         subclass of gimple_opt_pass along with...
10675         (pass_data_call_cdce): ...new pass_data instance and...
10676         (make_pass_call_cdce): ...new function.
10677         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10678         subclass of gimple_opt_pass along with...
10679         (pass_data_build_cfg): ...new pass_data instance and...
10680         (make_pass_build_cfg): ...new function.
10681         (pass_split_crit_edges): Convert from a global struct to a subclass of
10682         gimple_opt_pass along with...
10683         (pass_data_split_crit_edges): ...new pass_data instance and...
10684         (make_pass_split_crit_edges): ...new function.
10685         (pass_warn_function_return): Convert from a global struct to a
10686         subclass of gimple_opt_pass along with...
10687         (pass_data_warn_function_return): ...new pass_data instance and...
10688         (make_pass_warn_function_return): ...new function.
10689         (pass_warn_function_noreturn): Convert from a global struct to a
10690         subclass of gimple_opt_pass along with...
10691         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10692         (make_pass_warn_function_noreturn): ...new function.
10693         (pass_warn_unused_result): Convert from a global struct to a subclass
10694         of gimple_opt_pass along with...
10695         (pass_data_warn_unused_result): ...new pass_data instance and...
10696         (make_pass_warn_unused_result): ...new function.
10697         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10698         a subclass of gimple_opt_pass along with...
10699         (pass_data_merge_phi): ...new pass_data instance and...
10700         (make_pass_merge_phi): ...new function.
10701         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10702         a subclass of gimple_opt_pass along with...
10703         (pass_data_lower_complex): ...new pass_data instance and...
10704         (make_pass_lower_complex): ...new function.
10705         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10706         gimple_opt_pass along with...
10707         (pass_data_lower_complex_O0): ...new pass_data instance and...
10708         (make_pass_lower_complex_O0): ...new function.
10709         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10710         subclass of gimple_opt_pass along with...
10711         (pass_data_lower_eh): ...new pass_data instance and...
10712         (make_pass_lower_eh): ...new function.
10713         (pass_refactor_eh): Convert from a global struct to a subclass of
10714         gimple_opt_pass along with...
10715         (pass_data_refactor_eh): ...new pass_data instance and...
10716         (make_pass_refactor_eh): ...new function.
10717         (pass_lower_resx): Convert from a global struct to a subclass of
10718         gimple_opt_pass along with...
10719         (pass_data_lower_resx): ...new pass_data instance and...
10720         (make_pass_lower_resx): ...new function.
10721         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10722         of gimple_opt_pass along with...
10723         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10724         (make_pass_lower_eh_dispatch): ...new function.
10725         (pass_cleanup_eh): Convert from a global struct to a subclass of
10726         gimple_opt_pass along with...
10727         (pass_data_cleanup_eh): ...new pass_data instance and...
10728         (make_pass_cleanup_eh): ...new function.
10729         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10730         to a subclass of simple_ipa_opt_pass along with...
10731         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10732         (make_pass_ipa_lower_emutls): ...new function.
10733         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10734         a subclass of gimple_opt_pass along with...
10735         (pass_data_if_conversion): ...new pass_data instance and...
10736         (make_pass_if_conversion): ...new function.
10737         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10738         subclass of gimple_opt_pass along with...
10739         (pass_data_build_ssa): ...new pass_data instance and...
10740         (make_pass_build_ssa): ...new function.
10741         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10742         global struct to a subclass of gimple_opt_pass along with...
10743         (pass_data_loop_distribution): ...new pass_data instance and...
10744         (make_pass_loop_distribution): ...new function.
10745         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10746         subclass of gimple_opt_pass along with...
10747         (pass_data_mudflap_1): ...new pass_data instance and...
10748         (make_pass_mudflap_1): ...new function.
10749         (pass_mudflap_2): Convert from a global struct to a subclass of
10750         gimple_opt_pass along with...
10751         (pass_data_mudflap_2): ...new pass_data instance and...
10752         (make_pass_mudflap_2): ...new function.
10753         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10754         subclass of gimple_opt_pass along with...
10755         (pass_data_mudflap_1): ...new pass_data instance and...
10756         (make_pass_mudflap_1): ...new function.
10757         (pass_mudflap_2): Convert from a global struct to a subclass of
10758         gimple_opt_pass along with...
10759         (pass_data_mudflap_2): ...new pass_data instance and...
10760         (make_pass_mudflap_2): ...new function.
10761         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10762         gimple_opt_pass along with...
10763         (pass_data_nrv): ...new pass_data instance and...
10764         (make_pass_nrv): ...new function.
10765         (pass_return_slot): Convert from a global struct to a subclass of
10766         gimple_opt_pass along with...
10767         (pass_data_return_slot): ...new pass_data instance and...
10768         (make_pass_return_slot): ...new function.
10769         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10770         to a subclass of gimple_opt_pass along with...
10771         (pass_data_object_sizes): ...new pass_data instance and...
10772         (make_pass_object_sizes): ...new function.
10773         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10774         global struct to a subclass of gimple_opt_pass along with...
10775         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10776         and...
10777         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10778         (pass_fixup_cfg): Convert from a global struct to a subclass of
10779         gimple_opt_pass along with...
10780         (pass_data_fixup_cfg): ...new pass_data instance and...
10781         (make_pass_fixup_cfg): ...new function.
10782         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10783         (make_pass_mudflap_1): ...new function.
10784         (pass_mudflap_2): Replace declaration with that of...
10785         (make_pass_mudflap_2): ...new function.
10786         (pass_asan): Replace declaration with that of...
10787         (make_pass_asan): ...new function.
10788         (pass_asan_O0): Replace declaration with that of...
10789         (make_pass_asan_O0): ...new function.
10790         (pass_tsan): Replace declaration with that of...
10791         (make_pass_tsan): ...new function.
10792         (pass_tsan_O0): Replace declaration with that of...
10793         (make_pass_tsan_O0): ...new function.
10794         (pass_lower_cf): Replace declaration with that of...
10795         (make_pass_lower_cf): ...new function.
10796         (pass_refactor_eh): Replace declaration with that of...
10797         (make_pass_refactor_eh): ...new function.
10798         (pass_lower_eh): Replace declaration with that of...
10799         (make_pass_lower_eh): ...new function.
10800         (pass_lower_eh_dispatch): Replace declaration with that of...
10801         (make_pass_lower_eh_dispatch): ...new function.
10802         (pass_lower_resx): Replace declaration with that of...
10803         (make_pass_lower_resx): ...new function.
10804         (pass_build_cfg): Replace declaration with that of...
10805         (make_pass_build_cfg): ...new function.
10806         (pass_early_tree_profile): Replace declaration with that of...
10807         (make_pass_early_tree_profile): ...new function.
10808         (pass_cleanup_eh): Replace declaration with that of...
10809         (make_pass_cleanup_eh): ...new function.
10810         (pass_sra): Replace declaration with that of...
10811         (make_pass_sra): ...new function.
10812         (pass_sra_early): Replace declaration with that of...
10813         (make_pass_sra_early): ...new function.
10814         (pass_early_ipa_sra): Replace declaration with that of...
10815         (make_pass_early_ipa_sra): ...new function.
10816         (pass_tail_recursion): Replace declaration with that of...
10817         (make_pass_tail_recursion): ...new function.
10818         (pass_tail_calls): Replace declaration with that of...
10819         (make_pass_tail_calls): ...new function.
10820         (pass_tree_loop): Replace declaration with that of...
10821         (make_pass_tree_loop): ...new function.
10822         (pass_tree_loop_init): Replace declaration with that of...
10823         (make_pass_tree_loop_init): ...new function.
10824         (pass_lim): Replace declaration with that of...
10825         (make_pass_lim): ...new function.
10826         (pass_tree_unswitch): Replace declaration with that of...
10827         (make_pass_tree_unswitch): ...new function.
10828         (pass_predcom): Replace declaration with that of...
10829         (make_pass_predcom): ...new function.
10830         (pass_iv_canon): Replace declaration with that of...
10831         (make_pass_iv_canon): ...new function.
10832         (pass_scev_cprop): Replace declaration with that of...
10833         (make_pass_scev_cprop): ...new function.
10834         (pass_empty_loop): Replace declaration with that of...
10835         (make_pass_empty_loop): ...new function.
10836         (pass_record_bounds): Replace declaration with that of...
10837         (make_pass_record_bounds): ...new function.
10838         (pass_graphite): Replace declaration with that of...
10839         (make_pass_graphite): ...new function.
10840         (pass_graphite_transforms): Replace declaration with that of...
10841         (make_pass_graphite_transforms): ...new function.
10842         (pass_if_conversion): Replace declaration with that of...
10843         (make_pass_if_conversion): ...new function.
10844         (pass_loop_distribution): Replace declaration with that of...
10845         (make_pass_loop_distribution): ...new function.
10846         (pass_vectorize): Replace declaration with that of...
10847         (make_pass_vectorize): ...new function.
10848         (pass_slp_vectorize): Replace declaration with that of...
10849         (make_pass_slp_vectorize): ...new function.
10850         (pass_complete_unroll): Replace declaration with that of...
10851         (make_pass_complete_unroll): ...new function.
10852         (pass_complete_unrolli): Replace declaration with that of...
10853         (make_pass_complete_unrolli): ...new function.
10854         (pass_parallelize_loops): Replace declaration with that of...
10855         (make_pass_parallelize_loops): ...new function.
10856         (pass_loop_prefetch): Replace declaration with that of...
10857         (make_pass_loop_prefetch): ...new function.
10858         (pass_iv_optimize): Replace declaration with that of...
10859         (make_pass_iv_optimize): ...new function.
10860         (pass_tree_loop_done): Replace declaration with that of...
10861         (make_pass_tree_loop_done): ...new function.
10862         (pass_ch): Replace declaration with that of...
10863         (make_pass_ch): ...new function.
10864         (pass_ccp): Replace declaration with that of...
10865         (make_pass_ccp): ...new function.
10866         (pass_phi_only_cprop): Replace declaration with that of...
10867         (make_pass_phi_only_cprop): ...new function.
10868         (pass_build_ssa): Replace declaration with that of...
10869         (make_pass_build_ssa): ...new function.
10870         (pass_build_alias): Replace declaration with that of...
10871         (make_pass_build_alias): ...new function.
10872         (pass_build_ealias): Replace declaration with that of...
10873         (make_pass_build_ealias): ...new function.
10874         (pass_dominator): Replace declaration with that of...
10875         (make_pass_dominator): ...new function.
10876         (pass_dce): Replace declaration with that of...
10877         (make_pass_dce): ...new function.
10878         (pass_dce_loop): Replace declaration with that of...
10879         (make_pass_dce_loop): ...new function.
10880         (pass_cd_dce): Replace declaration with that of...
10881         (make_pass_cd_dce): ...new function.
10882         (pass_call_cdce): Replace declaration with that of...
10883         (make_pass_call_cdce): ...new function.
10884         (pass_merge_phi): Replace declaration with that of...
10885         (make_pass_merge_phi): ...new function.
10886         (pass_split_crit_edges): Replace declaration with that of...
10887         (make_pass_split_crit_edges): ...new function.
10888         (pass_pre): Replace declaration with that of...
10889         (make_pass_pre): ...new function.
10890         (pass_profile): Replace declaration with that of...
10891         (make_pass_profile): ...new function.
10892         (pass_strip_predict_hints): Replace declaration with that of...
10893         (make_pass_strip_predict_hints): ...new function.
10894         (pass_lower_complex_O0): Replace declaration with that of...
10895         (make_pass_lower_complex_O0): ...new function.
10896         (pass_lower_complex): Replace declaration with that of...
10897         (make_pass_lower_complex): ...new function.
10898         (pass_lower_vector): Replace declaration with that of...
10899         (make_pass_lower_vector): ...new function.
10900         (pass_lower_vector_ssa): Replace declaration with that of...
10901         (make_pass_lower_vector_ssa): ...new function.
10902         (pass_lower_omp): Replace declaration with that of...
10903         (make_pass_lower_omp): ...new function.
10904         (pass_diagnose_omp_blocks): Replace declaration with that of...
10905         (make_pass_diagnose_omp_blocks): ...new function.
10906         (pass_expand_omp): Replace declaration with that of...
10907         (make_pass_expand_omp): ...new function.
10908         (pass_expand_omp_ssa): Replace declaration with that of...
10909         (make_pass_expand_omp_ssa): ...new function.
10910         (pass_object_sizes): Replace declaration with that of...
10911         (make_pass_object_sizes): ...new function.
10912         (pass_strlen): Replace declaration with that of...
10913         (make_pass_strlen): ...new function.
10914         (pass_fold_builtins): Replace declaration with that of...
10915         (make_pass_fold_builtins): ...new function.
10916         (pass_stdarg): Replace declaration with that of...
10917         (make_pass_stdarg): ...new function.
10918         (pass_early_warn_uninitialized): Replace declaration with that of...
10919         (make_pass_early_warn_uninitialized): ...new function.
10920         (pass_late_warn_uninitialized): Replace declaration with that of...
10921         (make_pass_late_warn_uninitialized): ...new function.
10922         (pass_cse_reciprocals): Replace declaration with that of...
10923         (make_pass_cse_reciprocals): ...new function.
10924         (pass_cse_sincos): Replace declaration with that of...
10925         (make_pass_cse_sincos): ...new function.
10926         (pass_optimize_bswap): Replace declaration with that of...
10927         (make_pass_optimize_bswap): ...new function.
10928         (pass_optimize_widening_mul): Replace declaration with that of...
10929         (make_pass_optimize_widening_mul): ...new function.
10930         (pass_warn_function_return): Replace declaration with that of...
10931         (make_pass_warn_function_return): ...new function.
10932         (pass_warn_function_noreturn): Replace declaration with that of...
10933         (make_pass_warn_function_noreturn): ...new function.
10934         (pass_cselim): Replace declaration with that of...
10935         (make_pass_cselim): ...new function.
10936         (pass_phiopt): Replace declaration with that of...
10937         (make_pass_phiopt): ...new function.
10938         (pass_forwprop): Replace declaration with that of...
10939         (make_pass_forwprop): ...new function.
10940         (pass_phiprop): Replace declaration with that of...
10941         (make_pass_phiprop): ...new function.
10942         (pass_tree_ifcombine): Replace declaration with that of...
10943         (make_pass_tree_ifcombine): ...new function.
10944         (pass_dse): Replace declaration with that of...
10945         (make_pass_dse): ...new function.
10946         (pass_nrv): Replace declaration with that of...
10947         (make_pass_nrv): ...new function.
10948         (pass_rename_ssa_copies): Replace declaration with that of...
10949         (make_pass_rename_ssa_copies): ...new function.
10950         (pass_sink_code): Replace declaration with that of...
10951         (make_pass_sink_code): ...new function.
10952         (pass_fre): Replace declaration with that of...
10953         (make_pass_fre): ...new function.
10954         (pass_check_data_deps): Replace declaration with that of...
10955         (make_pass_check_data_deps): ...new function.
10956         (pass_copy_prop): Replace declaration with that of...
10957         (make_pass_copy_prop): ...new function.
10958         (pass_vrp): Replace declaration with that of...
10959         (make_pass_vrp): ...new function.
10960         (pass_uncprop): Replace declaration with that of...
10961         (make_pass_uncprop): ...new function.
10962         (pass_return_slot): Replace declaration with that of...
10963         (make_pass_return_slot): ...new function.
10964         (pass_reassoc): Replace declaration with that of...
10965         (make_pass_reassoc): ...new function.
10966         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10967         (make_pass_rebuild_cgraph_edges): ...new function.
10968         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10969         (make_pass_remove_cgraph_callee_edges): ...new function.
10970         (pass_build_cgraph_edges): Replace declaration with that of...
10971         (make_pass_build_cgraph_edges): ...new function.
10972         (pass_local_pure_const): Replace declaration with that of...
10973         (make_pass_local_pure_const): ...new function.
10974         (pass_tracer): Replace declaration with that of...
10975         (make_pass_tracer): ...new function.
10976         (pass_warn_unused_result): Replace declaration with that of...
10977         (make_pass_warn_unused_result): ...new function.
10978         (pass_diagnose_tm_blocks): Replace declaration with that of...
10979         (make_pass_diagnose_tm_blocks): ...new function.
10980         (pass_lower_tm): Replace declaration with that of...
10981         (make_pass_lower_tm): ...new function.
10982         (pass_tm_init): Replace declaration with that of...
10983         (make_pass_tm_init): ...new function.
10984         (pass_tm_mark): Replace declaration with that of...
10985         (make_pass_tm_mark): ...new function.
10986         (pass_tm_memopt): Replace declaration with that of...
10987         (make_pass_tm_memopt): ...new function.
10988         (pass_tm_edges): Replace declaration with that of...
10989         (make_pass_tm_edges): ...new function.
10990         (pass_split_functions): Replace declaration with that of...
10991         (make_pass_split_functions): ...new function.
10992         (pass_feedback_split_functions): Replace declaration with that of...
10993         (make_pass_feedback_split_functions): ...new function.
10994         (pass_strength_reduction): Replace declaration with that of...
10995         (make_pass_strength_reduction): ...new function.
10996         (pass_ipa_lower_emutls): Replace declaration with that of...
10997         (make_pass_ipa_lower_emutls): ...new function.
10998         (pass_ipa_function_and_variable_visibility): Replace declaration with
10999         that of...
11000         (make_pass_ipa_function_and_variable_visibility): ...new function.
11001         (pass_ipa_tree_profile): Replace declaration with that of...
11002         (make_pass_ipa_tree_profile): ...new function.
11003         (pass_early_local_passes): Replace declaration with that of...
11004         (make_pass_early_local_passes): ...new function.
11005         (pass_ipa_whole_program_visibility): Replace declaration with that
11006         of...
11007         (make_pass_ipa_whole_program_visibility): ...new function.
11008         (pass_ipa_lto_gimple_out): Replace declaration with that of...
11009         (make_pass_ipa_lto_gimple_out): ...new function.
11010         (pass_ipa_increase_alignment): Replace declaration with that of...
11011         (make_pass_ipa_increase_alignment): ...new function.
11012         (pass_ipa_inline): Replace declaration with that of...
11013         (make_pass_ipa_inline): ...new function.
11014         (pass_ipa_free_lang_data): Replace declaration with that of...
11015         (make_pass_ipa_free_lang_data): ...new function.
11016         (pass_ipa_free_inline_summary): Replace declaration with that of...
11017         (make_pass_ipa_free_inline_summary): ...new function.
11018         (pass_ipa_cp): Replace declaration with that of...
11019         (make_pass_ipa_cp): ...new function.
11020         (pass_ipa_reference): Replace declaration with that of...
11021         (make_pass_ipa_reference): ...new function.
11022         (pass_ipa_pure_const): Replace declaration with that of...
11023         (make_pass_ipa_pure_const): ...new function.
11024         (pass_ipa_pta): Replace declaration with that of...
11025         (make_pass_ipa_pta): ...new function.
11026         (pass_ipa_lto_finish_out): Replace declaration with that of...
11027         (make_pass_ipa_lto_finish_out): ...new function.
11028         (pass_ipa_tm): Replace declaration with that of...
11029         (make_pass_ipa_tm): ...new function.
11030         (pass_ipa_profile): Replace declaration with that of...
11031         (make_pass_ipa_profile): ...new function.
11032         (pass_ipa_cdtor_merge): Replace declaration with that of...
11033         (make_pass_ipa_cdtor_merge): ...new function.
11034         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
11035         of...
11036         (make_pass_cleanup_cfg_post_optimizing): ...new function.
11037         (pass_init_datastructures): Replace declaration with that of...
11038         (make_pass_init_datastructures): ...new function.
11039         (pass_fixup_cfg): Replace declaration with that of...
11040         (make_pass_fixup_cfg): ...new function.
11041         (pass_expand): Replace declaration with that of...
11042         (make_pass_expand): ...new function.
11043         (pass_instantiate_virtual_regs): Replace declaration with that of...
11044         (make_pass_instantiate_virtual_regs): ...new function.
11045         (pass_rtl_fwprop): Replace declaration with that of...
11046         (make_pass_rtl_fwprop): ...new function.
11047         (pass_rtl_fwprop_addr): Replace declaration with that of...
11048         (make_pass_rtl_fwprop_addr): ...new function.
11049         (pass_jump): Replace declaration with that of...
11050         (make_pass_jump): ...new function.
11051         (pass_jump2): Replace declaration with that of...
11052         (make_pass_jump2): ...new function.
11053         (pass_lower_subreg): Replace declaration with that of...
11054         (make_pass_lower_subreg): ...new function.
11055         (pass_cse): Replace declaration with that of...
11056         (make_pass_cse): ...new function.
11057         (pass_fast_rtl_dce): Replace declaration with that of...
11058         (make_pass_fast_rtl_dce): ...new function.
11059         (pass_ud_rtl_dce): Replace declaration with that of...
11060         (make_pass_ud_rtl_dce): ...new function.
11061         (pass_rtl_dce): Replace declaration with that of...
11062         (make_pass_rtl_dce): ...new function.
11063         (pass_rtl_dse1): Replace declaration with that of...
11064         (make_pass_rtl_dse1): ...new function.
11065         (pass_rtl_dse2): Replace declaration with that of...
11066         (make_pass_rtl_dse2): ...new function.
11067         (pass_rtl_dse3): Replace declaration with that of...
11068         (make_pass_rtl_dse3): ...new function.
11069         (pass_rtl_cprop): Replace declaration with that of...
11070         (make_pass_rtl_cprop): ...new function.
11071         (pass_rtl_pre): Replace declaration with that of...
11072         (make_pass_rtl_pre): ...new function.
11073         (pass_rtl_hoist): Replace declaration with that of...
11074         (make_pass_rtl_hoist): ...new function.
11075         (pass_rtl_store_motion): Replace declaration with that of...
11076         (make_pass_rtl_store_motion): ...new function.
11077         (pass_cse_after_global_opts): Replace declaration with that of...
11078         (make_pass_cse_after_global_opts): ...new function.
11079         (pass_rtl_ifcvt): Replace declaration with that of...
11080         (make_pass_rtl_ifcvt): ...new function.
11081         (pass_into_cfg_layout_mode): Replace declaration with that of...
11082         (make_pass_into_cfg_layout_mode): ...new function.
11083         (pass_outof_cfg_layout_mode): Replace declaration with that of...
11084         (make_pass_outof_cfg_layout_mode): ...new function.
11085         (pass_loop2): Replace declaration with that of...
11086         (make_pass_loop2): ...new function.
11087         (pass_rtl_loop_init): Replace declaration with that of...
11088         (make_pass_rtl_loop_init): ...new function.
11089         (pass_rtl_move_loop_invariants): Replace declaration with that of...
11090         (make_pass_rtl_move_loop_invariants): ...new function.
11091         (pass_rtl_unswitch): Replace declaration with that of...
11092         (make_pass_rtl_unswitch): ...new function.
11093         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
11094         (make_pass_rtl_unroll_and_peel_loops): ...new function.
11095         (pass_rtl_doloop): Replace declaration with that of...
11096         (make_pass_rtl_doloop): ...new function.
11097         (pass_rtl_loop_done): Replace declaration with that of...
11098         (make_pass_rtl_loop_done): ...new function.
11099         (pass_web): Replace declaration with that of...
11100         (make_pass_web): ...new function.
11101         (pass_cse2): Replace declaration with that of...
11102         (make_pass_cse2): ...new function.
11103         (pass_df_initialize_opt): Replace declaration with that of...
11104         (make_pass_df_initialize_opt): ...new function.
11105         (pass_df_initialize_no_opt): Replace declaration with that of...
11106         (make_pass_df_initialize_no_opt): ...new function.
11107         (pass_reginfo_init): Replace declaration with that of...
11108         (make_pass_reginfo_init): ...new function.
11109         (pass_inc_dec): Replace declaration with that of...
11110         (make_pass_inc_dec): ...new function.
11111         (pass_stack_ptr_mod): Replace declaration with that of...
11112         (make_pass_stack_ptr_mod): ...new function.
11113         (pass_initialize_regs): Replace declaration with that of...
11114         (make_pass_initialize_regs): ...new function.
11115         (pass_combine): Replace declaration with that of...
11116         (make_pass_combine): ...new function.
11117         (pass_if_after_combine): Replace declaration with that of...
11118         (make_pass_if_after_combine): ...new function.
11119         (pass_ree): Replace declaration with that of...
11120         (make_pass_ree): ...new function.
11121         (pass_partition_blocks): Replace declaration with that of...
11122         (make_pass_partition_blocks): ...new function.
11123         (pass_match_asm_constraints): Replace declaration with that of...
11124         (make_pass_match_asm_constraints): ...new function.
11125         (pass_regmove): Replace declaration with that of...
11126         (make_pass_regmove): ...new function.
11127         (pass_split_all_insns): Replace declaration with that of...
11128         (make_pass_split_all_insns): ...new function.
11129         (pass_fast_rtl_byte_dce): Replace declaration with that of...
11130         (make_pass_fast_rtl_byte_dce): ...new function.
11131         (pass_lower_subreg2): Replace declaration with that of...
11132         (make_pass_lower_subreg2): ...new function.
11133         (pass_mode_switching): Replace declaration with that of...
11134         (make_pass_mode_switching): ...new function.
11135         (pass_sms): Replace declaration with that of...
11136         (make_pass_sms): ...new function.
11137         (pass_sched): Replace declaration with that of...
11138         (make_pass_sched): ...new function.
11139         (pass_ira): Replace declaration with that of...
11140         (make_pass_ira): ...new function.
11141         (pass_reload): Replace declaration with that of...
11142         (make_pass_reload): ...new function.
11143         (pass_clean_state): Replace declaration with that of...
11144         (make_pass_clean_state): ...new function.
11145         (pass_branch_prob): Replace declaration with that of...
11146         (make_pass_branch_prob): ...new function.
11147         (pass_value_profile_transformations): Replace declaration with that
11148         of...
11149         (make_pass_value_profile_transformations): ...new function.
11150         (pass_postreload_cse): Replace declaration with that of...
11151         (make_pass_postreload_cse): ...new function.
11152         (pass_gcse2): Replace declaration with that of...
11153         (make_pass_gcse2): ...new function.
11154         (pass_split_after_reload): Replace declaration with that of...
11155         (make_pass_split_after_reload): ...new function.
11156         (pass_branch_target_load_optimize1): Replace declaration with that
11157         of...
11158         (make_pass_branch_target_load_optimize1): ...new function.
11159         (pass_thread_prologue_and_epilogue): Replace declaration with that
11160         of...
11161         (make_pass_thread_prologue_and_epilogue): ...new function.
11162         (pass_stack_adjustments): Replace declaration with that of...
11163         (make_pass_stack_adjustments): ...new function.
11164         (pass_peephole2): Replace declaration with that of...
11165         (make_pass_peephole2): ...new function.
11166         (pass_if_after_reload): Replace declaration with that of...
11167         (make_pass_if_after_reload): ...new function.
11168         (pass_regrename): Replace declaration with that of...
11169         (make_pass_regrename): ...new function.
11170         (pass_cprop_hardreg): Replace declaration with that of...
11171         (make_pass_cprop_hardreg): ...new function.
11172         (pass_reorder_blocks): Replace declaration with that of...
11173         (make_pass_reorder_blocks): ...new function.
11174         (pass_branch_target_load_optimize2): Replace declaration with that
11175         of...
11176         (make_pass_branch_target_load_optimize2): ...new function.
11177         (pass_leaf_regs): Replace declaration with that of...
11178         (make_pass_leaf_regs): ...new function.
11179         (pass_split_before_sched2): Replace declaration with that of...
11180         (make_pass_split_before_sched2): ...new function.
11181         (pass_compare_elim_after_reload): Replace declaration with that of...
11182         (make_pass_compare_elim_after_reload): ...new function.
11183         (pass_sched2): Replace declaration with that of...
11184         (make_pass_sched2): ...new function.
11185         (pass_stack_regs): Replace declaration with that of...
11186         (make_pass_stack_regs): ...new function.
11187         (pass_stack_regs_run): Replace declaration with that of...
11188         (make_pass_stack_regs_run): ...new function.
11189         (pass_df_finish): Replace declaration with that of...
11190         (make_pass_df_finish): ...new function.
11191         (pass_compute_alignments): Replace declaration with that of...
11192         (make_pass_compute_alignments): ...new function.
11193         (pass_duplicate_computed_gotos): Replace declaration with that of...
11194         (make_pass_duplicate_computed_gotos): ...new function.
11195         (pass_variable_tracking): Replace declaration with that of...
11196         (make_pass_variable_tracking): ...new function.
11197         (pass_free_cfg): Replace declaration with that of...
11198         (make_pass_free_cfg): ...new function.
11199         (pass_machine_reorg): Replace declaration with that of...
11200         (make_pass_machine_reorg): ...new function.
11201         (pass_cleanup_barriers): Replace declaration with that of...
11202         (make_pass_cleanup_barriers): ...new function.
11203         (pass_delay_slots): Replace declaration with that of...
11204         (make_pass_delay_slots): ...new function.
11205         (pass_split_for_shorten_branches): Replace declaration with that of...
11206         (make_pass_split_for_shorten_branches): ...new function.
11207         (pass_split_before_regstack): Replace declaration with that of...
11208         (make_pass_split_before_regstack): ...new function.
11209         (pass_convert_to_eh_region_ranges): Replace declaration with that
11210         of...
11211         (make_pass_convert_to_eh_region_ranges): ...new function.
11212         (pass_shorten_branches): Replace declaration with that of...
11213         (make_pass_shorten_branches): ...new function.
11214         (pass_set_nothrow_function_flags): Replace declaration with that of...
11215         (make_pass_set_nothrow_function_flags): ...new function.
11216         (pass_dwarf2_frame): Replace declaration with that of...
11217         (make_pass_dwarf2_frame): ...new function.
11218         (pass_final): Replace declaration with that of...
11219         (make_pass_final): ...new function.
11220         (pass_rtl_seqabstr): Replace declaration with that of...
11221         (make_pass_rtl_seqabstr): ...new function.
11222         (pass_release_ssa_names): Replace declaration with that of...
11223         (make_pass_release_ssa_names): ...new function.
11224         (pass_early_inline): Replace declaration with that of...
11225         (make_pass_early_inline): ...new function.
11226         (pass_inline_parameters): Replace declaration with that of...
11227         (make_pass_inline_parameters): ...new function.
11228         (pass_update_address_taken): Replace declaration with that of...
11229         (make_pass_update_address_taken): ...new function.
11230         (pass_convert_switch): Replace declaration with that of...
11231         (make_pass_convert_switch): ...new function.
11232         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
11233         to a subclass of simple_ipa_opt_pass along with...
11234         (pass_data_ipa_tree_profile): ...new pass_data instance and...
11235         (make_pass_ipa_tree_profile): ...new function.
11236         * tree-sra.c (pass_sra_early): Convert from a global struct to a
11237         subclass of gimple_opt_pass along with...
11238         (pass_data_sra_early): ...new pass_data instance and...
11239         (make_pass_sra_early): ...new function.
11240         (pass_sra): Convert from a global struct to a subclass of
11241         gimple_opt_pass along with...
11242         (pass_data_sra): ...new pass_data instance and...
11243         (make_pass_sra): ...new function.
11244         (pass_early_ipa_sra): Convert from a global struct to a subclass of
11245         gimple_opt_pass along with...
11246         (pass_data_early_ipa_sra): ...new pass_data instance and...
11247         (make_pass_early_ipa_sra): ...new function.
11248         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
11249         subclass of gimple_opt_pass along with...
11250         (pass_data_ccp): ...new pass_data instance and...
11251         (make_pass_ccp): ...new function.
11252         (pass_fold_builtins): Convert from a global struct to a subclass of
11253         gimple_opt_pass along with...
11254         (pass_data_fold_builtins): ...new pass_data instance and...
11255         (make_pass_fold_builtins): ...new function.
11256         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
11257         subclass of gimple_opt_pass along with...
11258         (pass_data_copy_prop): ...new pass_data instance and...
11259         (make_pass_copy_prop): ...new function.
11260         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
11261         global struct to a subclass of gimple_opt_pass along with...
11262         (pass_data_rename_ssa_copies): ...new pass_data instance and...
11263         (make_pass_rename_ssa_copies): ...new function.
11264         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
11265         subclass of gimple_opt_pass along with...
11266         (pass_data_dce): ...new pass_data instance and...
11267         (make_pass_dce): ...new function.
11268         (pass_dce_loop): Convert from a global struct to a subclass of
11269         gimple_opt_pass along with...
11270         (pass_data_dce_loop): ...new pass_data instance and...
11271         (make_pass_dce_loop): ...new function.
11272         (pass_cd_dce): Convert from a global struct to a subclass of
11273         gimple_opt_pass along with...
11274         (pass_data_cd_dce): ...new pass_data instance and...
11275         (make_pass_cd_dce): ...new function.
11276         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
11277         subclass of gimple_opt_pass along with...
11278         (pass_data_dominator): ...new pass_data instance and...
11279         (make_pass_dominator): ...new function.
11280         (pass_phi_only_cprop): Convert from a global struct to a subclass of
11281         gimple_opt_pass along with...
11282         (pass_data_phi_only_cprop): ...new pass_data instance and...
11283         (make_pass_phi_only_cprop): ...new function.
11284         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
11285         subclass of gimple_opt_pass along with...
11286         (pass_data_dse): ...new pass_data instance and...
11287         (make_pass_dse): ...new function.
11288         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
11289         a subclass of gimple_opt_pass along with...
11290         (pass_data_forwprop): ...new pass_data instance and...
11291         (make_pass_forwprop): ...new function.
11292         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
11293         struct to a subclass of gimple_opt_pass along with...
11294         (pass_data_tree_ifcombine): ...new pass_data instance and...
11295         (make_pass_tree_ifcombine): ...new function.
11296         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
11297         subclass of gimple_opt_pass along with...
11298         (pass_data_ch): ...new pass_data instance and...
11299         (make_pass_ch): ...new function.
11300         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
11301         subclass of gimple_opt_pass along with...
11302         (pass_data_tree_loop): ...new pass_data instance and...
11303         (make_pass_tree_loop): ...new function.
11304         (pass_tree_loop_init): Convert from a global struct to a subclass of
11305         gimple_opt_pass along with...
11306         (pass_data_tree_loop_init): ...new pass_data instance and...
11307         (make_pass_tree_loop_init): ...new function.
11308         (pass_lim): Convert from a global struct to a subclass of
11309         gimple_opt_pass along with...
11310         (pass_data_lim): ...new pass_data instance and...
11311         (make_pass_lim): ...new function.
11312         (pass_tree_unswitch): Convert from a global struct to a subclass of
11313         gimple_opt_pass along with...
11314         (pass_data_tree_unswitch): ...new pass_data instance and...
11315         (make_pass_tree_unswitch): ...new function.
11316         (pass_predcom): Convert from a global struct to a subclass of
11317         gimple_opt_pass along with...
11318         (pass_data_predcom): ...new pass_data instance and...
11319         (make_pass_predcom): ...new function.
11320         (pass_vectorize): Convert from a global struct to a subclass of
11321         gimple_opt_pass along with...
11322         (pass_data_vectorize): ...new pass_data instance and...
11323         (make_pass_vectorize): ...new function.
11324         (pass_graphite): Convert from a global struct to a subclass of
11325         gimple_opt_pass along with...
11326         (pass_data_graphite): ...new pass_data instance and...
11327         (make_pass_graphite): ...new function.
11328         (pass_graphite_transforms): Convert from a global struct to a subclass
11329         of gimple_opt_pass along with...
11330         (pass_data_graphite_transforms): ...new pass_data instance and...
11331         (make_pass_graphite_transforms): ...new function.
11332         (pass_check_data_deps): Convert from a global struct to a subclass of
11333         gimple_opt_pass along with...
11334         (pass_data_check_data_deps): ...new pass_data instance and...
11335         (make_pass_check_data_deps): ...new function.
11336         (pass_iv_canon): Convert from a global struct to a subclass of
11337         gimple_opt_pass along with...
11338         (pass_data_iv_canon): ...new pass_data instance and...
11339         (make_pass_iv_canon): ...new function.
11340         (pass_scev_cprop): Convert from a global struct to a subclass of
11341         gimple_opt_pass along with...
11342         (pass_data_scev_cprop): ...new pass_data instance and...
11343         (make_pass_scev_cprop): ...new function.
11344         (pass_record_bounds): Convert from a global struct to a subclass of
11345         gimple_opt_pass along with...
11346         (pass_data_record_bounds): ...new pass_data instance and...
11347         (make_pass_record_bounds): ...new function.
11348         (pass_complete_unroll): Convert from a global struct to a subclass of
11349         gimple_opt_pass along with...
11350         (pass_data_complete_unroll): ...new pass_data instance and...
11351         (make_pass_complete_unroll): ...new function.
11352         (pass_complete_unrolli): Convert from a global struct to a subclass of
11353         gimple_opt_pass along with...
11354         (pass_data_complete_unrolli): ...new pass_data instance and...
11355         (make_pass_complete_unrolli): ...new function.
11356         (pass_parallelize_loops): Convert from a global struct to a subclass
11357         of gimple_opt_pass along with...
11358         (pass_data_parallelize_loops): ...new pass_data instance and...
11359         (make_pass_parallelize_loops): ...new function.
11360         (pass_loop_prefetch): Convert from a global struct to a subclass of
11361         gimple_opt_pass along with...
11362         (pass_data_loop_prefetch): ...new pass_data instance and...
11363         (make_pass_loop_prefetch): ...new function.
11364         (pass_iv_optimize): Convert from a global struct to a subclass of
11365         gimple_opt_pass along with...
11366         (pass_data_iv_optimize): ...new pass_data instance and...
11367         (make_pass_iv_optimize): ...new function.
11368         (pass_tree_loop_done): Convert from a global struct to a subclass of
11369         gimple_opt_pass along with...
11370         (pass_data_tree_loop_done): ...new pass_data instance and...
11371         (make_pass_tree_loop_done): ...new function.
11372         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
11373         struct to a subclass of gimple_opt_pass along with...
11374         (pass_data_cse_reciprocals): ...new pass_data instance and...
11375         (make_pass_cse_reciprocals): ...new function.
11376         (pass_cse_sincos): Convert from a global struct to a subclass of
11377         gimple_opt_pass along with...
11378         (pass_data_cse_sincos): ...new pass_data instance and...
11379         (make_pass_cse_sincos): ...new function.
11380         (pass_optimize_bswap): Convert from a global struct to a subclass of
11381         gimple_opt_pass along with...
11382         (pass_data_optimize_bswap): ...new pass_data instance and...
11383         (make_pass_optimize_bswap): ...new function.
11384         (pass_optimize_widening_mul): Convert from a global struct to a
11385         subclass of gimple_opt_pass along with...
11386         (pass_data_optimize_widening_mul): ...new pass_data instance and...
11387         (make_pass_optimize_widening_mul): ...new function.
11388         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
11389         subclass of gimple_opt_pass along with...
11390         (pass_data_phiopt): ...new pass_data instance and...
11391         (make_pass_phiopt): ...new function.
11392         (pass_cselim): Convert from a global struct to a subclass of
11393         gimple_opt_pass along with...
11394         (pass_data_cselim): ...new pass_data instance and...
11395         (make_pass_cselim): ...new function.
11396         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
11397         subclass of gimple_opt_pass along with...
11398         (pass_data_phiprop): ...new pass_data instance and...
11399         (make_pass_phiprop): ...new function.
11400         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
11401         subclass of gimple_opt_pass along with...
11402         (pass_data_pre): ...new pass_data instance and...
11403         (make_pass_pre): ...new function.
11404         (pass_fre): Convert from a global struct to a subclass of
11405         gimple_opt_pass along with...
11406         (pass_data_fre): ...new pass_data instance and...
11407         (make_pass_fre): ...new function.
11408         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
11409         subclass of gimple_opt_pass along with...
11410         (pass_data_reassoc): ...new pass_data instance and...
11411         (make_pass_reassoc): ...new function.
11412         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
11413         subclass of gimple_opt_pass along with...
11414         (pass_data_sink_code): ...new pass_data instance and...
11415         (make_pass_sink_code): ...new function.
11416         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
11417         subclass of gimple_opt_pass along with...
11418         (pass_data_strlen): ...new pass_data instance and...
11419         (make_pass_strlen): ...new function.
11420         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
11421         struct to a subclass of gimple_opt_pass along with...
11422         (pass_data_build_alias): ...new pass_data instance and...
11423         (make_pass_build_alias): ...new function.
11424         (pass_build_ealias): Convert from a global struct to a subclass of
11425         gimple_opt_pass along with...
11426         (pass_data_build_ealias): ...new pass_data instance and...
11427         (make_pass_build_ealias): ...new function.
11428         (pass_ipa_pta): Convert from a global struct to a subclass of
11429         simple_ipa_opt_pass along with...
11430         (pass_data_ipa_pta): ...new pass_data instance and...
11431         (make_pass_ipa_pta): ...new function.
11432         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
11433         subclass of gimple_opt_pass along with...
11434         (pass_data_uncprop): ...new pass_data instance and...
11435         (make_pass_uncprop): ...new function.
11436         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
11437         global struct to a subclass of gimple_opt_pass along with...
11438         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
11439         (make_pass_late_warn_uninitialized): ...new function.
11440         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
11441         to a subclass of gimple_opt_pass along with...
11442         (pass_data_init_datastructures): ...new pass_data instance and...
11443         (make_pass_init_datastructures): ...new function.
11444         (pass_early_warn_uninitialized): Convert from a global struct to a
11445         subclass of gimple_opt_pass along with...
11446         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
11447         (make_pass_early_warn_uninitialized): ...new function.
11448         (pass_update_address_taken): Convert from a global struct to a
11449         subclass of gimple_opt_pass along with...
11450         (pass_data_update_address_taken): ...new pass_data instance and...
11451         (make_pass_update_address_taken): ...new function.
11452         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
11453         struct to a subclass of gimple_opt_pass along with...
11454         (pass_data_release_ssa_names): ...new pass_data instance and...
11455         (make_pass_release_ssa_names): ...new function.
11456         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
11457         subclass of gimple_opt_pass along with...
11458         (pass_data_stdarg): ...new pass_data instance and...
11459         (make_pass_stdarg): ...new function.
11460         * tree-switch-conversion.c (pass_convert_switch): Convert from a
11461         global struct to a subclass of gimple_opt_pass along with...
11462         (pass_data_convert_switch): ...new pass_data instance and...
11463         (make_pass_convert_switch): ...new function.
11464         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
11465         to a subclass of gimple_opt_pass along with...
11466         (pass_data_tail_recursion): ...new pass_data instance and...
11467         (make_pass_tail_recursion): ...new function.
11468         (pass_tail_calls): Convert from a global struct to a subclass of
11469         gimple_opt_pass along with...
11470         (pass_data_tail_calls): ...new pass_data instance and...
11471         (make_pass_tail_calls): ...new function.
11472         * tree-vect-generic.c (pass_lower_vector): Convert from a global
11473         struct to a subclass of gimple_opt_pass along with...
11474         (pass_data_lower_vector): ...new pass_data instance and...
11475         (make_pass_lower_vector): ...new function.
11476         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11477         gimple_opt_pass along with...
11478         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11479         (make_pass_lower_vector_ssa): ...new function.
11480         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11481         to a subclass of gimple_opt_pass along with...
11482         (pass_data_slp_vectorize): ...new pass_data instance and...
11483         (make_pass_slp_vectorize): ...new function.
11484         (pass_ipa_increase_alignment): Convert from a global struct to a
11485         subclass of simple_ipa_opt_pass along with...
11486         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11487         (make_pass_ipa_increase_alignment): ...new function.
11488         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11489         gimple_opt_pass along with...
11490         (pass_data_vrp): ...new pass_data instance and...
11491         (make_pass_vrp): ...new function.
11492         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11493         subclass of simple_ipa_opt_pass along with...
11494         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11495         (make_pass_ipa_free_lang_data): ...new function.
11496         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11497         gimple_opt_pass along with...
11498         (pass_data_tsan): ...new pass_data instance and...
11499         (make_pass_tsan): ...new function.
11500         (pass_tsan_O0): Convert from a global struct to a subclass of
11501         gimple_opt_pass along with...
11502         (pass_data_tsan_O0): ...new pass_data instance and...
11503         (make_pass_tsan_O0): ...new function.
11504         * var-tracking.c (pass_variable_tracking): Convert from a global
11505         struct to a subclass of rtl_opt_pass along with...
11506         (pass_data_variable_tracking): ...new pass_data instance and...
11507         (make_pass_variable_tracking): ...new function.
11508         * web.c (pass_web): Convert from a global struct to a subclass of
11509         rtl_opt_pass along with...
11510         (pass_data_web): ...new pass_data instance and...
11511         (make_pass_web): ...new function.
11512         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11513         declaration with that of...
11514         (make_pass_mode_switch_use): ...new function.
11515         (pass_resolve_sw_modes): Replace declaration with that of...
11516         (make_pass_resolve_sw_modes): ...new function.
11517         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11518         from a global struct to a subclass of rtl_opt_pass along with...
11519         (pass_data_mode_switch_use): ...new pass_data instance and...
11520         (make_pass_mode_switch_use): ...new function.
11521         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11522         from a global struct to a subclass of rtl_opt_pass along with...
11523         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11524         (make_pass_resolve_sw_modes): ...new function.
11525         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11526         struct to a subclass of rtl_opt_pass along with...
11527         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11528         (make_pass_insert_vzeroupper): ...new function.
11529         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11530         global struct to a subclass of rtl_opt_pass along with...
11531         (pass_data_work_around_errata): ...new pass_data instance and...
11532         (make_pass_work_around_errata): ...new function.
11533         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11534         struct to a subclass of rtl_opt_pass along with...
11535         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11536         (make_pass_mips_machine_reorg2): ...new function.
11538 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11540         * passes.c (pass_manager::operator new): New.
11542 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11544         Handwritten part of conversion of passes to C++ classes.
11546         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11547         (toplev.o): Add dep on PASS_MANAGER_H.
11548         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11549         of early local pases to reflect this moving from a global to a
11550         member of gcc::pass_manager.
11551         (cgraph_add_new_function): Likewise.
11552         * lto-cgraph.c (lto_output_node): Update for conversion of
11553         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11554         * passes.c (opt_pass::clone): New.
11555         (opt_pass::gate): New.
11556         (opt_pass::execute): New.
11557         (opt_pass::opt_pass): New.
11558         (pass_manager::execute_early_local_passes): New.
11559         (pass_manager::execute_pass_mode_switching): new.
11560         (finish_optimization_passes): Convert to...
11561         (pass_manager::finish_optimization_passes): ...this.
11562         (finish_optimization_passes): Update for conversion of passes to
11563         C++ classes.
11564         (register_dump_files_1): Use has_gate since we cannot portably
11565         check a vtable entry against NULL.
11566         (dump_one_pass): Likewise.
11567         (ipa_write_summaries_2): Likewise.
11568         (ipa_write_optimization_summaries_1): Likewise.
11569         (ipa_read_summaries_1): Likewise.
11570         (ipa_read_optimization_summaries_1): Likewise.
11571         (execute_ipa_stmt_fixups): Likewise.
11572         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11573         pass-creation functions rather than wiring up globals, and
11574         storing the results in fields of pass_manager generated using
11575         pass-instances.def.
11576         (pass_manager::dump_profile_report): Update for conversion of
11577         passes to C++ classes.
11578         (pass_manager::execute_ipa_summary_passes): Likewise.
11579         (execute_one_ipa_transform_pass): Likewise.
11580         (execute_one_pass): Use has_gate and has_execute since we cannot
11581         portably check a vtable entry against NULL.
11582         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11583         (pass_manager): Use pass-instances.def to add fields for the
11584         various pass instances.
11585         * toplev.c (finalize): Update for move of
11586         finish_optimization_passes to a method of gcc::pass_manager.
11587         * toplev.h (finish_optimization_passes): Move to method of class
11588         pass_manager.
11589         * tree-pass.h (struct pass_data): New.
11590         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11591         (opt_pass::gate): Convert to virtual function.
11592         (opt_pass::~opt_pass): New.
11593         (opt_pass::clone): New.
11594         (opt_pass::execute): Convert to virtual function.
11595         (opt_pass::opt_pass): New.
11596         (opt_pass::ctxt_): new.
11597         (gimple_opt_pass): Convert to subclass of opt_pass.
11598         (gimple_opt_pass::gimple_opt_pass): New.
11599         (rtl_opt_pass): Convert to subclass of opt_pass.
11600         (rtl_opt_pass::rtl_opt_pass): New.
11601         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11602         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11603         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11604         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11605         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11606         invocation of pass_mode_switching to reflect this moving from a
11607         global to a member of gcc::pass_manager.
11608         (ix86_option_override): Rework how pass_insert_vzeroupper is
11609         added to the pass_manager to reflect autogenerated changes.
11610         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11612 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11614         PR rtl-optimization/57708
11615         * recog.c (peep2_find_free_register): Validate all regs in a
11616         multi-reg mode.
11618 2013-08-05  Jan Hubicka  <jh@suse.cz>
11620         PR lto/57602
11621         * cgraph.c (verify_cgraph_node): Accept local flags from other
11622         partitions.
11623         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11624         (function_and_variable_visibility): Likewise.
11625         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11627 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11629         * graph.c (init_graph_slim_pretty_print): Remove.
11630         (print_graph_cfg): Do not call it.  Use local pretty printer.
11631         (start_graph_dump): Likewise.
11633 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11635         * gimple-pretty-print.c (buffer): Remove.
11636         (initialized): Likewise.
11637         (maybe_init_pretty_print): Likewise.
11638         (print_gimple_stmt): Do not call it.  Use non-static local
11639         pretty_printer variable.
11640         (print_gimple_expr): Likewise.
11641         (print_gimple_seq): Likewise.
11642         (gimple_dump_bb): Likewise.
11644 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11646         * asan.c (asan_pp): Remove.
11647         (asan_pp_initialized): Likewise.
11648         (asan_pp_initialize): Likewise.
11649         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11650         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11651         (asan_add_global): Likewise.
11653 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11655         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11656         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11657         * pretty-print.c (pp_formatted_text_data): Likewise.
11658         (pp_indent): Rename from pp_base_indent.
11659         (pp_format): Rename from pp_base_format.
11660         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11661         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11662         (pp_flush): Rename from pp_base_flush.
11663         (pp_set_line_maximum_length): Rename from
11664         pp_base_set_line_maximum_length.
11665         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11666         (pp_set_prefix): Rename from pp_base_set_prefix.
11667         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11668         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11669         (pp_append_text): Rename from pp_base_append_text.
11670         (pp_formatted_text): Rename from pp_base_formatted_text.
11671         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11672         (pp_remaining_character_count_for_line): Rename from
11673         pp_base_remaining_character_count_for_line.
11674         (pp_newline): Rename from pp_base_newline.
11675         (pp_character): Rename from pp_base_character.
11676         (pp_string): Rename from pp_base_string.
11677         (pp_maybe_space): Rename from pp_base_maybe_space.
11678         * asan.c (asan_pp_string): Adjust.
11679         (asan_emit_stack_protection): Likewise.
11680         (asan_add_global): Likewise.
11681         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11682         * tree-mudflap.c (mf_varname_tree): Likewise.
11683         * tree-pretty-print.c (pp_tree_identifier): Rename from
11684         pp_base_tree_identifier.
11685         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11686         Declare as function.
11688 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11690         * pretty-print.h (pp_bar_bar): New.
11691         (pp_ampersand_ampersand): Likewise.
11692         (pp_less_equal): Likewise.
11693         (pp_greater_equal): Likewise.
11694         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11695         printer functions instead of pp_string or operators and punctuators.
11696         (dump_gimple_call): Likewise.
11697         (dump_gimple_omp_for): Likewise.
11698         (dump_gimple_transaction): Likewise.
11699         (dump_gimple_phi): Likewise.
11700         (pp_gimple_stmt_1): Likewise.
11701         * sched-vis.c (print_insn): Likewise.
11702         * tree-mudflap.c (mf_varname_tree): Likewise.
11703         * tree-pretty-print.c (dump_block_node): Likewise.
11704         (dump_generic_node): Likewise.
11706 2013-08-02  Jan Hubicka  <jh@suse.cz>
11708         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11709         boundaries.
11710         * lto-streamer-out.c (tree_is_indexable): Results decls and
11711         parm decls are not indexable.
11712         (DFS_write_tree_body): Do not follow args and results.
11713         (hash_tree): Likewise.
11714         (output_functions): Rearrange so struct function is needed
11715         only when real body is output; be able to also ouptut abstract
11716         functions; output DECL_ARGUMENTS and DECL_RESULT.
11717         (lto_output): When not in WPA, ale store abstract functions.
11718         (write_symbol): Do not care about RESULT_DECL.
11719         (output_symbol_p): Handle correctly sbtract decls.
11720         * lto-streamer-in.c (input_function): Rearrange so struct
11721         function can be NULL at entry; allow streaming of
11722         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11723         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11724         sanity check during LTO.
11725         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11726         RESULT_DECl and DECL_ARGUMENTS.
11727         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11728         Likewise.
11730 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11732         * pretty-print.h (pp_underscore): New.
11733         (pp_comma): Tidy.
11734         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11735         printer functions instead of pp_character.
11736         (dump_binary_rhs): Likewise.
11737         (dump_ternary_rhs): Likewise.
11738         (dump_gimple_call_args): Likewise.
11739         (pp_points_to_solution): Likewise.
11740         (dump_gimple_call): Likewise.
11741         (dump_gimple_switch): Likewise.
11742         (dump_gimple_cond): Likewise.
11743         (dump_gimple_bind): Likewise.
11744         (dump_gimple_try): Likewise.
11745         (dump_gimple_omp_for): Likewise.
11746         (dump_gimple_omp_continue): Likewise.
11747         (dump_gimple_omp_single): Likewise.
11748         (dump_gimple_omp_sections): Likewise.
11749         (dump_gimple_omp_block): Likewise.
11750         (dump_gimple_omp_critical): Likewise.
11751         (dump_gimple_transaction): Likewise.
11752         (dump_gimple_asm): Likewise.
11753         (dump_gimple_phi): Likewise.
11754         (dump_gimple_omp_parallel): Likewise.
11755         (dump_gimple_omp_task): Likewise.
11756         (dump_gimple_omp_atomic_load): Likewise.
11757         (dump_gimple_omp_atomic_store): Likewise.
11758         (dump_gimple_mem_ops): Likewise.
11759         (pp_gimple_stmt_1): Likewise.
11760         (pp_cfg_jump): Likewise.
11761         (dump_implicit_edges): Likewise.
11762         (gimple_dump_bb_for_graph): Likewise.
11763         * graph.c (draw_cfg_node): Likewise.
11764         * langhooks.c (lhd_print_error_function): Likewise.
11765         * sched-vis.c (print_exp): Likewise.
11766         (print_value): Likewise.
11767         (print_pattern): Likewise.
11768         (print_insn): Likewise.
11769         (rtl_dump_bb_for_graph): Likewise.
11770         * tree-pretty-print.c (dump_function_declaration): Likewise.
11771         (dump_array_domain): Likewise.
11772         (dump_omp_clause): Likewise.
11773         (dump_location): Likewise.
11774         (dump_generic_node): Likewise.
11775         (print_struct_decl): Likewise.
11776         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11778 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11780         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11781         candidate table when replacing a candidate statement.
11782         (replace_rhs_if_not_dup): Likewise.
11783         (replace_one_candidate): Likewise.
11785 2013-08-02  Jan Hubicka  <jh@suse.cz>
11786             Martin Liska  <marxin.liska@gmail.com>
11788         * cgraphunit.c (add_new_function): Fix logic when adding from
11789         late IPA pass.
11790         (assemble_thunk): Rename to ...
11791         (expand_thunk); .. this one; export; get it working with
11792         general functions; make produced gimple valid.
11793         * cgraph.h (expand_thunk): Declare.
11795 2013-08-02  Jan Hubicka  <jh@suse.cz>
11797         * ipa-cp.c (gather_context_independent_values): Use
11798         ipa_get_param_move_cost.
11799         (get_replacement_map): Remove PARAM; move parameter folding
11800         into tree-inline.c
11801         (create_specialized_node): Update.
11802         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11803         assert that we have gimple body; update move_cost.
11804         (count_formal_params): Assert that we have gimple body.
11805         (ipa_dump_param): New function.
11806         (ipa_alloc_node_params): Break out from ...
11807         (ipa_initialize_node_params): ... here.
11808         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11809         (ipa_write_node_info): Stream move costs.
11810         (ipa_read_node_info): Read move costs.
11811         (ipa_update_after_lto_read): Do not recompute node params.
11812         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11813         (ipa_get_param): Check we are not in WPA.
11814         (ipa_get_param_move_cost): New.
11815         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11816         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11817         parm numbers to be present.
11819 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11821         PR rtl-optimization/58048
11822         * lra-constraints.c (process_alt_operands): Don't check asm
11823         operand on register.
11825 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11827         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11828         the implied StoreLoad barrier for atomic operations if before.
11830 2013-08-02  Jan Hubicka  <jh@suse.cz>
11831             Martin Liska  <marxin.liska@gmail.com>
11833         * cgraph.c (cgraph_function_body_availability): Do not check
11834         cgraph flags.
11835         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11836         symtab_node_availability): Declare.
11837         * ipa.c (can_replace_by_local_alias): New.
11838         (function_and_variable_visibility): Use it.
11839         * symtab.c (symtab_for_node_and_aliases,
11840         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11842 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11844         PR rtl-optimization/57963
11845         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11846         (lra_constraints): Use them.
11848 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11850         * config/arm/types.md (define_attr "type"): Add "load_acq"
11851         and "store_rel".
11852         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11853         changes.
11854         (cortex_a53_store1): Likewise.
11856 2013-08-01  Jan Hubicka  <jh@suse.cz>
11858         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11859         partitions are not needed.
11861 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11863         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11864         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11865         MAYBE_NON_Q_CLASS_P where appropriate.
11867 2013-08-01  Jan Hubicka  <jh@suse.cz>
11869         * cgraph.h (release_function_body): Declare.
11870         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11871         of unused delcarations.
11873 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11875         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11876         RTL form when subtracting a constant.
11878 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11880         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11881         Generate canonical plus rtx with negated immediate instead of minus
11882         where appropriate.
11883         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11885 2013-08-01  Jan Hubicka  <jh@suse.cz>
11887         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11888         (cgraph_release_function_body): Likewise.
11889         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11890         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11891         to used_as_abstract_origin.
11892         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11893         symbols used as abstract origins.
11894         * cgraphunit.c (analyze_functions): Update.
11895         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11896         used_as_abstract_origin.
11897         * tree-inline.c (tree_function_versioning): Update
11898         used_as_abstract_origin; be ready for DECL_RESULT and
11899         DECL_ARGUMENTS to be NULL.
11901         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11902         for abstract functions.
11903         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11904         real symbols.
11906 2013-08-01  Jan Hubicka  <jh@suse.cz>
11908         * profile.c (compute_value_histograms): Fix thinko.
11910 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11912         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11913         aarch-common-protos.h to extra_headers.
11914         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11915         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11916         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11918 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11920         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11921         Include "../arm/types.md".  Define "type" attribute for all patterns.
11922         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11923         attribute changes.
11925 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11927         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11928         to support power8 load fusion.
11929         (fusion_gpr_mem_load): Likewise.
11931         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11933         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11934         declarations for power8 load fusion.
11935         (emit_fusion_gpr_load): Likewise.
11937         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11938         tuning for power8, turn on fusion mode by default.  Turn on sign
11939         extending fusion mode if normal fusion mode is on, and we are at
11940         -O2 or -O3.
11941         (fusion_gpr_load_p): New function, return true if we can fuse an
11942         addis instruction with a dependent load to a GPR.
11943         (emit_fusion_gpr_load): Emit the instructions for power8 load
11944         fusion to GPRs.
11946         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11947         (VSX load fusion peepholes): New peepholes to fuse together an
11948         addi instruction with a VSX load instruction.
11950         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11951         peepholes to fuse an addis instruction with a load to a GPR base
11952         register.  If we are supporting sign extending fusions, convert
11953         sign extending loads to zero extending loads and add an explicit
11954         sign extension.
11956 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11958         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11959         aarch-common-protos.h to extra_headers.
11960         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11961         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11962         (arm_early_store_addr_dep): Likewise.
11963         (arm_no_early_alu_shift_dep): Likewise.
11964         (arm_no_early_alu_shift_value_dep): Likewise.
11965         (arm_no_early_mul_dep): Likewise.
11966         (arm_no_early_store_addr_dep): Likewise.
11967         (arm_mac_accumulator_is_mul_result): Likewise.
11968         (arm_mac_accumulator_is_result): Likewise.
11969         * config/arm/aarch-common.c: ... here.  New file.
11970         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11971         here to ...
11972         (arm_early_store_addr_dep): Likewise.
11973         (arm_no_early_alu_shift_dep): Likewise.
11974         (arm_no_early_alu_shift_value_dep): Likewise.
11975         (arm_no_early_mul_dep): Likewise.
11976         (arm_no_early_store_addr_dep): Likewise.
11977         (arm_mac_accumulator_is_mul_result): Likewise.
11978         (arm_mac_accumulator_is_result): Likewise.
11979         * config/arm/aarch-common-protos.h: ... here.  New file.
11980         * config/arm/t-arm (aarch-common.o): Define.
11982 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11984         * config/arm/arm.md: Include new file "types.md".
11985         (define_attr "type"): Move from here to ...
11986         (define_attr "mul32"): Likewise.
11987         (define_attr "mul64"): Likewise.
11988         * config/arm/types.md: ... here.  New file.
11990 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11992         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11993         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11995 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11997         * gen-pass-instances.awk: Fix offset of substr().
11999 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
12001         * Makefile.in (pass-instances.def): New.
12002         (passes.o): Replace dependency on passes.def with one on
12003         pass-instances.def
12005         * gen-pass-instances.awk: New.
12007         * passes.c (pass_manager::pass_manager): Use pass-instances.def
12008         rather than passes.def, updating local definition of NEXT_PASS
12009         macro to add an extra NUM parameter (currently unused).
12011 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
12013         * Makefile.in (PASS_MANAGER_H): New.
12014         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
12015         (passes.o): Likewise.
12016         (statistics.o): Likewise.
12017         (cgraphunit.o): Likewise.
12018         (context.o): Depend on PASS_MANAGER_H.
12020         * pass_manager.h: New.
12022         * cgraphunit.c (cgraph_add_new_function): Update for moves
12023         of globals to fields of pass_manager.
12024         (analyze_function): Likewise.
12025         (expand_function): Likewise.
12026         (ipa_passes): Likewise.
12027         (compile): Likewise.
12029         * context.c (context::context): New.
12030         * context.h  (context::context): New.
12031         (context::get_passes): New.
12032         (context::passes_): New.
12034         * lto-cgraph.c (input_node): Update for moves of globals to
12035         fields of pass_manager.
12037         * passes.c (all_passes): Remove, in favor of a field of the
12038         same name within the new class pass_manager.
12039         (all_small_ipa_passes): Likewise.
12040         (all_lowering_passes): Likewise.
12041         (all_regular_ipa_passes): Likewise.
12042         (all_late_ipa_passes): Likewise.
12043         (all_lto_gen_passes): Likewise.
12044         (passes_by_id): Likewise.
12045         (passes_by_id_size): Likewise.
12046         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
12047         the new class pass_manager.
12048         (set_pass_for_id): Convert to...
12049         (pass_manager::set_pass_for_id): ...method.
12050         (get_pass_for_id): Convert to...
12051         (pass_manager::get_pass_for_id): ...method.
12052         (register_one_dump_file): Move body of implementation into...
12053         (pass_manager::register_one_dump_file): ...here.
12054         (register_dump_files_1): Convert to...
12055         (pass_manager::register_dump_files_1): ...method.
12056         (register_dump_files): Convert to...
12057         (pass_manager::register_dump_files): ...method.
12058         (create_pass_tab): Update for moves of globals to fields of
12059         pass_manager.
12060         (dump_passes): Move body of implementation into...
12061         (pass_manager::dump_passes): ...here.
12062         (register_pass): Move body of implementation into...
12063         (pass_manager::register_pass): ...here.
12064         (init_optimization_passes): Convert into...
12065         (pass_manager::pass_manager): ...constructor for new
12066         pass_manager class, and initialize the pass_lists array.
12067         (check_profile_consistency): Update for moves of globals to
12068         fields of pass_manager.
12069         (dump_profile_report): Move body of implementation into...
12070         (pass_manager::dump_profile_report): ...here.
12071         (ipa_write_summaries_1): Update for moves of pass lists from
12072         being globals to fields of pass_manager.
12073         (ipa_write_optimization_summaries): Likewise.
12074         (ipa_read_summaries):  Likewise.
12075         (ipa_read_optimization_summaries): Likewise.
12076         (execute_all_ipa_stmt_fixups): Likewise.
12078         * statistics.c (statistics_fini): Update for moves of globals to
12079         fields of pass_manager.
12081         * toplev.c (general_init): Replace call to
12082         init_optimization_passes with construction of the pass_manager
12083         instance.
12085         * tree-pass.h (all_passes): Remove, in favor of a field of the
12086         same name within the new class pass_manager.
12087         (all_small_ipa_passes): Likewise.
12088         (all_lowering_passes): Likewise.
12089         (all_regular_ipa_passes): Likewise.
12090         (all_lto_gen_passes): Likewise.
12091         (all_late_ipa_passes): Likewise.
12092         (passes_by_id): Likewise.
12093         (passes_by_id_size): Likewise.
12094         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
12095         the new class pass_manager.
12096         (get_pass_for_id): Remove.
12098 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
12100         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
12101         configs.
12103 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
12105         * arm.md (mulhi3): New expand pattern.
12107 2013-07-30  Jan Hubicka  <jh@suse.cz>
12108             Martin Liska  <marxin.liska@gmail.com>
12110         * profile.c (compute_value_histograms): Do not ICE when
12111         there is mismatch only on some counters.
12113 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
12115         PR rtl-optimization/57637
12116         * function.c (move_insn_for_shrink_wrap): Also check the
12117         GEN set of the LIVE problem for the liveness analysis
12118         if it exists, otherwise give up.
12120 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12122         PR tree-optimization/57993
12123         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
12124         replaced statement in the candidate table.
12125         (phi_add_costs): Return infinite cost when the hidden basis does
12126         not dominate all phis on which the candidate is dependent.
12127         (replace_one_candidate): Record replaced statement in the
12128         candidate table.
12130 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
12132         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
12133         (ashlv2si3): New expander.
12134         (*ashlv2si3_i): New define_insn_and_split.
12135         * predicates.md (float_operation): Allow patterns with three
12136         basic sub-patterns.
12138         PR rtl-optimization/58021
12139         * mode-switching.c (create_pre_exit): Always split off preceding
12140         insns if we are not at the basic block head.
12142 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
12144         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
12145         (UCLIBC_DYNAMIC_LINKER): New macro.
12146         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
12147         `-mnan=2008'.
12148         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
12149         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
12150         `-mnan=2008'.
12151         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
12152         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
12153         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
12154         for SF and DF modes.  Use ieee_quad_format for TF mode.
12155         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
12156         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
12157         (mips_option_override): Handle `-mnan=legacy'.
12158         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
12159         `-mabs=2008' and `-mnan=2008'.
12160         (OPTION_DEFAULT_SPECS): Add "nan" default.
12161         (ASM_SPEC): Handle `-mnan='.
12162         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
12163         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
12164         comment accordingly.
12165         (neg<mode>2): Likewise.
12166         * config/mips/mips.opt (mabs, mnan): New options.
12167         * doc/install.texi (Configuration): Document `--with-nan=' option.
12168         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
12169         `-mnan=' options.
12170         (MIPS Options): Document them.
12171         * config.gcc <mips*-*-*>: Handle `--with-nan='.
12172         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
12173         * configure: Regenerate.
12174         * config.in: Regenerate.
12176 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
12178         * config/i386/i386.md (float post-reload splitters): Do not check
12179         for subregs of SSE registers.
12181 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
12182             H.J. Lu  <hongjiu.lu@intel.com>
12184         PR target/57954
12185         PR target/57988
12186         * config/i386/i386.md (post-reload splitter
12187         to avoid partial SSE reg dependency stalls): New pattern.
12189 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
12191         * config/s390/s390.md ("movcc"): Swap load and store instructions.
12193 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
12195         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12196         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
12198 2013-07-26  Cary Coutant  <ccoutant@google.com>
12200         * dwarf2out.c (die_checksum_ordered): Don't include template
12201         instantiations in signature.
12202         (is_template_parameter): New function.
12203         (is_template_instantiation): New function.
12204         (generate_skeleton_bottom_up): Don't include template instantiations
12205         in type unit DIE.
12206         (generate_skeleton): Likewise.
12207         (break_out_comdat_types): Move recursive call to break out nested
12208         types earlier.
12209         (prune_unused_types_mark_generic_parms_dies): Call
12210         is_template_parameter.
12212 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
12214         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
12215         uses vector registers.
12216         * config/aarch64/iterators.md: Add attributes rtn and vas.
12218 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12219             Richard Earnshaw  <richard.earnshaw@arm.com>
12221         * combine.c (simplify_comparison): Re-canonicalize operands
12222         where appropriate.
12223         * config/arm/arm.md (movcond_addsi): New splitter.
12225 2013-07-25  Sterling Augustine  <saugustine@google.com>
12227         * dwarf2out.c (size_of_pubnames): Move code to...
12228         (include_pubname_in_output): ...here.  New.
12229         (want_pubnames): Rearrange.
12230         (output_pubnames): Call include_pubname_in_output.  Move assertion.
12232 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
12234         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
12236 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
12238         PR target/38836
12239         * doc/extend.texi: Remove obsolete builtins. Fix
12240         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
12242 2013-07-25  Jan Hubicka  <jh@suse.cz>
12244         * cgraph.c (release_function_body): Break out from ...
12245         (cgraph_release_function_body): ... this one; also release DECL_RESULT
12246         and DECL_ARGUMENTS.
12247         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
12248         old_tree in the map.
12249         (create_specialized_node): Update.
12250         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
12251         into index.
12252         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
12253         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
12254         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
12255         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
12256         DECL_RESULT.
12258 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12260         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
12261         addsi3_carryin_alt2_<optab>): Correct output template.
12263 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12265         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
12266         Adjust for arm_restrict_it.
12267         Remove trailing whitespace.
12269 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
12271         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
12272         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
12274         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
12276 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
12278         PR rtl-optimization/57960
12279         * lra-constraints.c (process_alt_operands): Use the right mode
12280         when checking strict_low.
12282 2013-07-25  Jan Hubicka  <jh@suse.cz>
12284         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
12285         * cgraph.c (cgraph_remove_node): Do not release function body
12286         when in cgraph streaming.
12287         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
12288         in other partitions are not considered reachable; fix handling of
12289         clones.
12291 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12293         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
12295 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12297         PR target/19599
12298         PR target/57731
12299         PR target/57837
12300         * config/arm/arm.md ("*sibcall_insn): Replace use of
12301         Ss with US. Adjust output for v5 and v4t.
12302         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
12304         * config/arm/constraints.md ("Ss"): Rename to US.
12306 2013-07-25  Terry Guo  <terry.guo@arm.com>
12308         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
12309         shift_add/shift_sub0/shift_sub1 RTXs.
12311 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
12312             Anton Blanchard  <anton@au1.ibm.com>
12314         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
12315         (altivec_vpks<VI_char>ss): Likewise.
12316         (altivec_vpks<VI_char>us): Likewise.
12317         (altivec_vpku<VI_char>us): Likewise.
12318         (altivec_vpku<VI_char>um): Likewise.
12320 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
12322         Introduce context class.
12324         * Makefile.in (CONTEXT_H): New.
12325         (OBJS): Add context.o.
12326         (toplev.o): Add CONTEXT_H to dependencies.
12327         (context.o): New.
12329         * toplev.c (general_init): Create the singleton gcc::context instance.
12331         * context.c: New.
12333         * context.h: New.
12335 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
12337         PR rtl-optimization/57968
12338         * mode-switching.c (create_pre_exit): Allow instructions that
12339         don't set a return register to need a non-exit mode.
12341 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12342             Anton Blanchard  <anton@au1.ibm.com>
12344         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
12345         operands to vperm for little endian.
12346         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
12347         of lvsl to create the control mask for a vperm for little endian.
12349 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12350             Anton Blanchard  <anton@au1.ibm.com>
12352         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
12353         two operands for little-endian.
12355 2013-07-23  Steve Ellcey  <sellcey@mips.com>
12357         * config/mips/mips.c (mips_case_values_threshold): New.
12358         (TARGET_CASE_VALUES_THRESHOLD): Define.
12360 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12361             Anton Blanchard  <anton@au1.ibm.com>
12363         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
12364         selection of field for vector splat in little endian mode.
12366 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
12368         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
12369         expanders to rs6000.md.
12370         (ior<mode>3): Likewise.
12371         (and<mode>3): Likewise.
12372         (one_cmpl<mode>2): Likewise.
12373         (nor<mode>3): Likewise.
12374         (andc<mode>3): Likewise.
12375         (eqv<mode>3): Likewise.
12376         (nand<mode>3): Likewise.
12377         (orc<mode>3): Likewise.
12379         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
12380         declaration.
12382         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
12383         to split multi-word logical operations.
12384         (rs6000_split_logical_di): Likewise.
12385         (rs6000_split_logical): Likewise.
12387         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
12388         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
12389         and allow TImode operations in 32-bit.
12390         (vsx_and<mode>3_64bit): Likewise.
12391         (vsx_ior<mode>3_32bit): Likewise.
12392         (vsx_ior<mode>3_64bit): Likewise.
12393         (vsx_xor<mode>3_32bit): Likewise.
12394         (vsx_xor<mode>3_64bit): Likewise.
12395         (vsx_one_cmpl<mode>2_32bit): Likewise.
12396         (vsx_one_cmpl<mode>2_64bit): Likewise.
12397         (vsx_nor<mode>3_32bit): Likewise.
12398         (vsx_nor<mode>3_64bit): Likewise.
12399         (vsx_andc<mode>3_32bit): Likewise.
12400         (vsx_andc<mode>3_64bit): Likewise.
12401         (vsx_eqv<mode>3_32bit): Likewise.
12402         (vsx_eqv<mode>3_64bit): Likewise.
12403         (vsx_nand<mode>3_32bit): Likewise.
12404         (vsx_nand<mode>3_64bit): Likewise.
12405         (vsx_orc<mode>3_32bit): Likewise.
12406         (vsx_orc<mode>3_64bit): Likewise.
12408         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
12409         logical types in GPRs.
12411         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
12412         logical insns to rs6000.md, and allow TImode operations in
12413         32-bit.
12414         (altivec_ior<mode>3): Likewise.
12415         (altivec_xor<mode>3): Likewise.
12416         (altivec_one_cmpl<mode>2): Likewise.
12417         (altivec_nor<mode>3): Likewise.
12418         (altivec_andc<mode>3): Likewise.
12420         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
12421         attributes for moving the 128-bit logical operations into
12422         rs6000.md.
12423         (BOOL_REGS_OUTPUT): Likewise.
12424         (BOOL_REGS_OP1): Likewise.
12425         (BOOL_REGS_OP2): Likewise.
12426         (BOOL_REGS_UNARY): Likewise.
12427         (BOOL_REGS_AND_CR0): Likewise.
12428         (one_cmpl<mode>2): Add support for DI logical operations on
12429         32-bit, splitting the operations to 32-bit.
12430         (anddi3): Likewise.
12431         (iordi3): Likewise.
12432         (xordi3): Likewise.
12433         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
12434         changes to combine the 32/64-bit code, allow logical operations on
12435         TI mode in 32-bit, and to use similar match_operator patterns like
12436         scalar mode uses.  Combine the Altivec and VSX code for logical
12437         operations, and move it here.
12438         (ior<mode>3, 128-bit types): Likewise.
12439         (xor<mode>3, 128-bit types): Likewise.
12440         (one_cmpl<mode>3, 128-bit types): Likewise.
12441         (nor<mode>3, 128-bit types): Likewise.
12442         (andc<mode>3, 128-bit types): Likewise.
12443         (eqv<mode>3, 128-bit types): Likewise.
12444         (nand<mode>3, 128-bit types): Likewise.
12445         (orc<mode>3, 128-bit types): Likewise.
12446         (and<mode>3_internal): Likewise.
12447         (bool<mode>3_internal): Likewise.
12448         (boolc<mode>3_internal1): Likewise.
12449         (boolc<mode>3_internal2): Likewise.
12450         (boolcc<mode>3_internal1): Likewise.
12451         (boolcc<mode>3_internal2): Likewise.
12452         (eqv<mode>3_internal1): Likewise.
12453         (eqv<mode>3_internal2): Likewise.
12454         (one_cmpl1<mode>3_internal): Likewise.
12456 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12458         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12459         Rename flag_stack_usage to flag_stack_usage_info.
12461 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12463         * config/microblaze/sync.md: New file.
12464         * config/microblaze/microblaze.md: Include sync.md
12465         * config/microblaze/microblaze.c: Add print_operand 'y'.
12466         * config/microblaze/constraints.md: Add memory_contraint
12467         'Q' which is a single register.
12469 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
12471         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
12473 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12475         * reload.c (find_reloads): Exit loop once we find this operand
12476         cannot be reloaded somehow for this alternative.
12478         * reload.c (find_reloads): Exit loop once we find a hard register.
12480         * rtlanal.c (computed_jump_p): Exit loop once we find label
12481         reference is used.
12483         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12485         * cfgloopmanip.c (remove_path): Exit loop after setting
12486         irred_invalidated.
12488         * gensupport.c (subst_dup): Avoid loop if code is not
12489         MATCH_DUP nor MATCH_OP_DUP.
12491 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12493         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12495 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12497         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12498         true for SP_REGNUM if mode == ptr_mode.
12499         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12500         with value R0_REGNUM + 31.
12502 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12504         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12505         pad pointer-typed argument downward.
12507 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12509         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12510         and __ILP32__ when the ILP32 model is in use.
12512 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12514         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12515         (aarch64_load_symref_appropriately): In the case of
12516         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12517         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12518         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12519         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12520         if 'mode' doesn't equal to 'ptr_mode'.
12521         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12522         'vcall_offset'; change to call aarch64_emit_move differently depending
12523         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12524         to calculate the upper bound of 'vcall_offset'.
12525         (aarch64_cannot_force_const_mem): Change to also return true if
12526         mode != ptr_mode.
12527         (aarch64_legitimize_reload_address): In the case of large
12528         displacements, add new local variable 'xmode' and an assertion
12529         based on it; change to use 'xmode' to generate the new rtx and
12530         reload.
12531         (aarch64_asm_trampoline_template): Change to generate the template
12532         differently depending on TARGET_ILP32 or not; change to use
12533         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12534         (aarch64_trampoline_size): Removed.
12535         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12536         and replace immediate literals with it.  Change to use 'ptr_mode'
12537         instead of 'DImode' and call convert_memory_address if the mode
12538         of 'fnaddr' doesn't equal to 'ptr_mode'.
12539         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12540         to output symbol.
12541         (aarch64_elf_asm_destructor): Likewise.
12542         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12543         on TARGET_ILP32 instead of aarch64_trampoline_size.
12544         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12545         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12546         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12547         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12548         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12549         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12550         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12551         depending on the value of 'mode'.
12552         (add_losym_<mode>): New.
12553         (ldr_got_small_<mode>): New, based on ldr_got_small.
12554         (ldr_got_small): Remove.
12555         (ldr_got_small_sidi): New.
12556         * config/aarch64/iterators.md (P): New.
12557         (PTR): Change to 'ptr_mode' in the condition.
12559 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12561         * config.gcc (aarch64*-*-*): Support --with-abi.
12562         (aarch64*-*-elf): Support --with-multilib-list.
12563         (aarch64*-*-linux*): Likewise.
12564         (supported_defaults): Add abi to aarch64*-*-*.
12565         * configure.ac: Mention AArch64 for --with-multilib-list.
12566         * configure: Re-generated.
12567         * config/aarch64/biarchilp32.h: New file.
12568         * config/aarch64/biarchlp64.h: New file.
12569         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12570         (ABI_SPEC): Ditto.
12571         (MULTILIB_DEFAULTS): Ditto.
12572         (DRIVER_SELF_SPECS): Ditto.
12573         (ASM_SPEC): Update to also substitute -mabi.
12574         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12575         file whose name depends on -mabi= and -mbig-endian.
12576         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12577         TARGET_ILP32.
12578         (POINTER_SIZE): New define.
12579         (POINTERS_EXTEND_UNSIGNED): Ditto.
12580         (enum aarch64_abi_type): New enumeration tag.
12581         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12582         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12583         (TARGET_ILP32): New define.
12584         * config/aarch64/aarch64.opt (mabi): New.
12585         (aarch64_abi): New.
12586         (ilp32, lp64): New values for -mabi.
12587         * config/aarch64/t-aarch64 (comma): New define.
12588         (MULTILIB_OPTIONS): Ditto.
12589         (MULTILIB_DIRNAMES): Ditto.
12590         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12591         * doc/invoke.texi: Document -mabi for AArch64.
12593 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12595         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12597 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12598             Anton Blanchard  <anton@au1.ibm.com>
12600         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12601         endianness when selecting field to splat.
12603 2013-07-22  Eric Christopher  <echristo@gmail.com>
12605         * dwarf2out.c (die_odr_checksum): New function to use
12606         CHECKSUM_ macros and ULEB128 for DIE tag.
12607         (generate_type_signature): Use.
12609 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12611         * config.gcc (sparc*-*-*): Accept leon3 processor.
12612         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12613         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12614         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12615         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12616         (mfix-ut699): Adjust comment.
12617         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12618         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12619         (CPP_CPU_SPEC): Likewise.
12620         (ASM_CPU_SPEC): Likewise.
12621         * config/sparc/sparc.c (leon3_cost): New constant.
12622         (sparc_option_override): Add leon3 support.
12623         (mem_ref): New function.
12624         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12625         (sparc_do_work_around_errata): Look into the instruction in the delay
12626         slot and adjust accordingly.  Add fix for the data cache nullify issues
12627         of the UT699.  Change insertion position for the NOP.
12628         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12629         (leon3_load): New reservation.
12630         (leon_store): Bump latency to 2.
12631         (grfpu): New automaton.
12632         (grfpu_alu): New unit.
12633         (grfpu_ds): Likewise.
12634         (leon_fp_alu): Adjust.
12635         (leon_fp_mult): Delete.
12636         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12637         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12638         * config/sparc/sparc.md (cpu): Add leon3.
12639         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12640         (swapsi): Likewise.
12641         (atomic_test_and_set): Likewise.
12642         (ldstub): Likewise.
12644 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12646         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12647         default for R5900 targets.
12648         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12649         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12650         * config/mips/mips.c (mips_option_override): Report an error for
12651         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12652         for -march=r5900 -mhard-float.
12654 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12656         * df-problems.c (can_move_insns_across): Exit loop once we
12657         find a non-fixed, non-global register.
12659         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12660         setting can_throw.
12662         * omega.c (omega_eliminate_red): Break after setting red_found.
12663         (omega_problem_has_red_equations): Similarly after setting found.
12664         (omega_query_variable): Similarly after setting coupled.
12666 2013-07-22  Marek Polacek  <polacek@redhat.com>
12668         * gimplify.c: Don't include gimple.h twice.
12670 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12672         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12673         instead of TARGET_THUMB1.
12674         (Pz): New constraint.
12675         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12676         encodings.
12677         (compare_negsi_si): Likewise.
12678         (compare_addsi2_op0): Likewise.
12679         (compare_addsi2_op1): Likewise.
12680         (addsi3_carryin_<optab>): Likewise.
12681         (addsi3_carryin_alt2_<optab>): Likewise.
12682         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12683         for arm_restrict_it.
12684         (subsi3_carryin): Likewise.
12685         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12686         (minmax_arithsi): Disable for arm_restrict_it.
12687         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12688         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12689         (satsi_<SAT:code>_shift): Likewise.
12690         (arm_shiftsi3): Add alternative for 16-bit encoding.
12691         (arm32_movhf): Disable for arm_restrict_it.
12692         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12693         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12695 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12697         * config/arm/arm.md (attribute "insn"): Delete.
12698         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12699         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12700         (not_shiftsi): Update for attribute change.
12701         (not_shiftsi_compare0): Likewise.
12702         (not_shiftsi_compare0_scratch): Likewise.
12703         (arm_one_cmplsi2): Likewise.
12704         (thumb1_one_cmplsi2): Likewise.
12705         (notsi_compare0): Likewise.
12706         (notsi_compare0_scratch): Likewise.
12707         (thumb1_movdi_insn): Likewise.
12708         (arm_movsi_insn): Likewise.
12709         (movhi_insn_arch4): Likewise.
12710         (movhi_bytes): Likewise.
12711         (arm_movqi_insn): Likewise.
12712         (thumb1_movqi_insn): Likewise.
12713         (arm32_movhf): Likewise.
12714         (thumb1_movhf): Likewise.
12715         (arm_movsf_soft_insn): Likewise.
12716         (thumb1_movsf_insn): Likewise.
12717         (thumb_movdf_insn): Likewise.
12718         (movsicc_insn): Likewise.
12719         (movsfcc_soft_insn): Likewise.
12720         (and_scc): Likewise.
12721         (cond_move): Likewise.
12722         (if_move_not): Likewise.
12723         (if_not_move): Likewise.
12724         (if_shift_move): Likewise.
12725         (if_move_shift): Likewise.
12726         (if_shift_shift): Likewise.
12727         (if_not_arith): Likewise.
12728         (if_arith_not): Likewise.
12729         (cond_move_not): Likewise.
12730         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12731         (neon_mov<mode>): Likewise.
12732         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12733         (thumb2_movsi_vfp): Likewise.
12734         (movsf_vfp): Likewise.
12735         (thumb2_movsf_vfp): Likewise.
12736         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12737         change.
12738         (cortexa7_older_only): Likewise.
12739         (cortexa7_younger): Likewise.
12740         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12741         (1020alu_shift_op): Likewise.
12742         (1020alu_shift_reg_op): Likewise.
12743         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12744         (alu_shift_op): Likewise.
12745         (alu_shift_reg_op): Likewise.
12746         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12747         (11_alu_shift_op): Likewise.
12748         (11_alu_shift_reg_op): Likewise.
12749         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12750         (9_alu_shift_reg_op): Likewise.
12751         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12752         change.
12753         (cortex_a15_alu_shift): Likewise.
12754         (cortex_a15_alu_shift_reg): Likewise.
12755         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12756         (cortex_a5_alu_shift): Likewise.
12757         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12758         change.
12759         (cortex_a53_alu_shift): Likewise.
12760         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12761         change.
12762         (cortex_a7_alu_reg): Likewise.
12763         (cortex_a7_alu_shift): Likewise.
12764         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12765         (cortex_a8_alu_shift): Likewise.
12766         (cortex_a8_alu_shift_reg): Likewise.
12767         (cortex_a8_mov): Likewise.
12768         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12769         (cortex_a9_dp_shift): Likewise.
12770         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12771         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12772         (cortex_r4_mov): Likewise.
12773         (cortex_r4_alu_shift): Likewise.
12774         (cortex_r4_alu_shift_reg): Likewise.
12775         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12776         (526_alu_shift_op): Likewise.
12777         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12778         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12779         (626te_alu_shift_op): Likewise.
12780         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12781         (726te_alu_op): Likewise.
12782         (726te_alu_shift_op): Likewise.
12783         (726te_alu_shift_reg_op): Likewise.
12784         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12785         (mp626_alu_shift_op): Likewise.
12786         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12787         (pj4_alu_e1_conds): Likewise.
12788         (pj4_alu): Likewise.
12789         (pj4_alu_conds): Likewise.
12790         (pj4_shift): Likewise.
12791         (pj4_shift_conds): Likewise.
12792         (pj4_alu_shift): Likewise.
12793         (pj4_alu_shift_conds): Likewise.
12795 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12797         * config/arm/predicates.md (shiftable_operator_strict_it):
12798         New predicate.
12799         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12800         Disable cond_exec version for arm_restrict_it.
12801         (thumb2_smaxsi3): Convert to generate cond_exec.
12802         (thumb2_sminsi3): Likewise.
12803         (thumb32_umaxsi3): Likewise.
12804         (thumb2_uminsi3): Likewise.
12805         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12806         (thumb2_neg_abssi2): Likewise.
12807         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12808         (thumb2_movsicc_insn): Adjust alternatives.
12809         (thumb2_mov_negscc): Disable for arm_restrict_it.
12810         (thumb2_mov_negscc_strict_it): New pattern.
12811         (thumb2_mov_notscc_strict_it): New pattern.
12812         (thumb2_mov_notscc): Disable for arm_restrict_it.
12813         (thumb2_ior_scc): Likewise.
12814         (thumb2_ior_scc_strict_it): New pattern.
12815         (thumb2_cond_move): Adjust for arm_restrict_it.
12816         (thumb2_cond_arith): Disable for arm_restrict_it.
12817         (thumb2_cond_arith_strict_it): New pattern.
12818         (thumb2_cond_sub): Adjust for arm_restrict_it.
12819         (thumb2_movcond): Likewise.
12820         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12821         (thumb2_zero_extendhisi2_v6): Likewise.
12822         (thumb2_zero_extendqisi2_v6): Likewise.
12823         (orsi_notsi_si): Likewise.
12824         (orsi_not_shiftsi_si): Likewise.
12826 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12828         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12829         instruction sequence is 1 byte shorter.
12831 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12833         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12834         it is not needed after split.
12836 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12838         PR target/51784
12839         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12840         second label for nonlocal goto receivers. Don't output pic base labels
12841         unless we're producing PIC; mark that action unreachable().
12842         (ix86_save_reg): If the function contains a nonlocal label, save the
12843         PIC base reg.
12844         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12845         * config/darwin.c (emitted_pic_label_num): New GTY.
12846         (update_pic_label_number_if_needed): New.
12847         (machopic_output_function_base_name): Adjust for nonlocal receiver
12848         case.
12849         (machopic_should_output_picbase_label): New.
12850         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12851         (nonlocal_goto_receiver): New insn and split.
12853 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12855         * config/aarch64/aarch64-builtins.c
12856         (aarch64_fold_builtin): Fold abs in all modes.
12857         * config/aarch64/aarch64-simd-builtins.def
12858         (abs): Enable for all modes.
12859         * config/aarch64/arm_neon.h
12860         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12861         (vabs_f64): Add missing intrinsic.
12863 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12865         * config/aarch64/arm_neon.h (vabs_s64): New function
12867 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12869         PR target/57516
12870         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12871         * config/avr/avr.md (adjust_len): Add `round'.
12872         * config/avr/avr-protos.h (avr_out_round): New prototype.
12873         (avr_out_plus): Add `out_label' argument.
12874         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12875         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12876         Handle the case where `insn' is just a pattern.
12877         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12878         (avr_out_round): New function.
12879         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12881 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12883         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12884         Add check for flag_stack_usage to handle -fstack-usage support
12886 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12888         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12889         interaction for new Power8 flags and VSX.
12891 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12893         PR middle-end/57698
12894         * tree-inline.c (expand_call_inline): Emit errors during
12895         early_inlining only if optimization is not turned on.
12897 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12899         * passes.def: New.
12901         * passes.c (init_optimization_passes): Move the construction of
12902         the pass hierarchy into a new passes.def file.
12904         * Makefile.in (passes.o): Add dependency on passes.def.
12906 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12908         * passes.c (init_optimization_passes): Introduce macros for
12909         constructing the tree of passes (INSERT_PASSES_AFTER,
12910         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12911         TERMINATE_PASS_LIST).
12913 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12914             Wei Mi  <wmi@google.com>
12916         PR rtl-optimization/57878
12917         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12918         top.
12919         (reload_pseudo_compare_func): Check nregs first for reload
12920         pseudos.
12922 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12924         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12926 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12928         * read-rtl.c (validate_const_int): Once an invalid character is
12929         seen, quit the loop.
12931         * gengtype.c (write_roots): Similarly once we find the "deletable"
12932         or "if_marked" option.
12934 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12936         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12937         "xtab" and "sat".  Move value "clz" from here to ...
12938         (attriubte "type"): ... here.
12939         (satsi_<SAT:code>): Delete "insn" attribute.
12940         (satsi_<SAT:code>_shift): Likewise.
12941         (arm_zero_extendqisi2addsi): Likewise.
12942         (arm_extendqisi2addsi): Likewise.
12943         (clzsi2): Update for attribute changes.
12944         (rbitsi2): Likewise.
12945         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12946         attribute.
12947         (arm_usatsihi): Likewise.
12948         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12950 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12952         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12953         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12954         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12955         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12956         in alphabetical order.
12957         (attribute "core_cycles"): Update for attribute changes.
12958         (arm_addsi3): Likewise.
12959         (addsi3_compare0): Likewise.
12960         (addsi3_compare0_scratch): Likewise.
12961         (addsi3_compare_op1): Likewise.
12962         (addsi3_compare_op2): Likewise.
12963         (compare_addsi2_op0): Likewise.
12964         (compare_addsi2_op1): Likewise.
12965         (addsi3_carryin_shift_<optab>): Likewise.
12966         (subsi3_carryin_shift): Likewise.
12967         (rsbsi3_carryin_shift): Likewise.
12968         (arm_subsi3_insn): Likewise.
12969         (subsi3_compare0): Likewise.
12970         (subsi3_compare): Likewise.
12971         (arm_andsi3_insn): Likewise.
12972         (thumb1_andsi3_insn): Likewise.
12973         (andsi3_compare0): Likewise.
12974         (andsi3_compare0_scratch): Likewise.
12975         (zeroextractsi_compare0_scratch
12976         (andsi_not_shiftsi_si): Likewise.
12977         (iorsi3_insn): Likewise.
12978         (iorsi3_compare0): Likewise.
12979         (iorsi3_compare0_scratch): Likewise.
12980         (arm_xorsi3): Likewise.
12981         (thumb1_xorsi3_insn): Likewise.
12982         (xorsi3_compare0): Likewise.
12983         (xorsi3_compare0_scratch): Likewise.
12984         (satsi_<SAT:code>_shift): Likewise.
12985         (rrx): Likewise.
12986         (arm_shiftsi3): Likewise.
12987         (shiftsi3_compare0): Likewise.
12988         (not_shiftsi): Likewise.
12989         (not_shiftsi_compare0): Likewise.
12990         (not_shiftsi_compare0_scratch): Likewise.
12991         (arm_one_cmplsi2): Likewise.
12992         (thumb_one_complsi2): Likewise.
12993         (notsi_compare0): Likewise.
12994         (notsi_compare0_scratch): Likewise.
12995         (thumb1_zero_extendhisi2): Likewise.
12996         (arm_zero_extendhisi2): Likewise.
12997         (arm_zero_extendhisi2_v6): Likewise.
12998         (arm_zero_extendhisi2addsi): Likewise.
12999         (thumb1_zero_extendqisi2): Likewise.
13000         (thumb1_zero_extendqisi2_v6): Likewise.
13001         (arm_zero_extendqisi2): Likewise.
13002         (arm_zero_extendqisi2_v6): Likewise.
13003         (arm_zero_extendqisi2addsi): Likewise.
13004         (thumb1_extendhisi2): Likewise.
13005         (arm_extendhisi2): Likewise.
13006         (arm_extendhisi2_v6): Likewise.
13007         (arm_extendqisi): Likewise.
13008         (arm_extendqisi_v6): Likewise.
13009         (arm_extendqisi2addsi): Likewise.
13010         (thumb1_extendqisi2): Likewise.
13011         (thumb1_movdi_insn): Likewise.
13012         (arm_movsi_insn): Likewise.
13013         (movsi_compare0): Likewise.
13014         (movhi_insn_arch4): Likewise.
13015         (movhi_bytes): Likewise.
13016         (arm_movqi_insn): Likewise.
13017         (thumb1_movqi_insn): Likewise.
13018         (arm32_movhf): Likewise.
13019         (thumb1_movhf): Likewise.
13020         (arm_movsf_soft_insn): Likewise.
13021         (thumb1_movsf_insn): Likewise.
13022         (movdf_soft_insn): Likewise.
13023         (thumb_movdf_insn): Likewise.
13024         (arm_cmpsi_insn): Likewise.
13025         (cmpsi_shiftsi): Likewise.
13026         (cmpsi_shiftsi_swp): Likewise.
13027         (arm_cmpsi_negshiftsi_si): Likewise.
13028         (movsicc_insn): Likewise.
13029         (movsfcc_soft_insn): Likewise.
13030         (arith_shiftsi): Likewise.
13031         (arith_shiftsi_compare0
13032         (arith_shiftsi_compare0_scratch
13033         (sub_shiftsi): Likewise.
13034         (sub_shiftsi_compare0
13035         (sub_shiftsi_compare0_scratch
13036         (and_scc): Likewise.
13037         (cond_move): Likewise.
13038         (if_plus_move): Likewise.
13039         (if_move_plus): Likewise.
13040         (if_move_not): Likewise.
13041         (if_not_move): Likewise.
13042         (if_shift_move): Likewise.
13043         (if_move_shift): Likewise.
13044         (if_shift_shift): Likewise.
13045         (if_not_arith): Likewise.
13046         (if_arith_not): Likewise.
13047         (cond_move_not): Likewise.
13048         (thumb1_ashlsi3): Set type attribute.
13049         (thumb1_ashrsi3): Likewise.
13050         (thumb1_lshrsi3): Likewise.
13051         (thumb1_rotrsi3): Likewise.
13052         (shiftsi3_compare0_scratch): Likewise.
13053         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
13054         (neon_mov<mode>): Likewise.
13055         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
13056         attribute changes.
13057         (thumb2_movsi_insn): Likewise.
13058         (thumb2_cmpsi_neg_shiftsi): Likewise.
13059         (thumb2_extendqisi_v6): Likewise.
13060         (thumb2_zero_extendhisi2_v6): Likewise.
13061         (thumb2_zero_extendqisi2_v6): Likewise.
13062         (thumb2_shiftsi3_short): Likewise.
13063         (thumb2_addsi3_compare0_scratch): Likewise.
13064         (orsi_not_shiftsi_si): Likewise.
13065         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
13066         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
13067         changes.
13068         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
13069         (1020alu_shift_op): Likewise.
13070         (1020alu_shift_reg_op): Likewise.
13071         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
13072         (alu_shift_op): Likewise.
13073         (alu_shift_reg_op): Likewise.
13074         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
13075         (11_alu_shift_op): Likewise.
13076         (11_alu_shift_reg_op): Likewise.
13077         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
13078         (9_alu_shift_reg_op): Likewise.
13079         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
13080         changes.
13081         (cortex_a15_alu_shift): Likewise.
13082         (cortex_a15_alu_shift_reg): Likewise.
13083         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
13084         changes.
13085         (cortex_a5_alu_shift): Likewise.
13086         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
13087         changes.
13088         (cortex_a53_alu_shift): Likewise.
13089         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
13090         changes.
13091         (cortex_a7_alu_reg): Likewise.
13092         (cortex_a7_alu_shift): Likewise.
13093         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
13094         changes.
13095         (cortex_a8_alu_shift): Likewise.
13096         (cortex_a8_alu_shift_reg): Likewise.
13097         (cortex_a8_mov): Likewise.
13098         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
13099         (cortex_a9_dp_shift): Likewise.
13100         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
13101         changes.
13102         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
13103         changes.
13104         (cortex_r4_mov): Likewise.
13105         (cortex_r4_alu_shift): Likewise.
13106         (cortex_r4_alu_shift_reg): Likewise.
13107         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
13108         (526_alu_shift_op): Likewise.
13109         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
13110         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
13111         (626te_alu_shift_op): Likewise.
13112         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
13113         (726te_alu_op): Likewise.
13114         (726te_alu_shift_op): Likewise.
13115         (726te_alu_shift_reg_op): Likewise.
13116         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
13117         (mp626_alu_shift_op): Likewise.
13118         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
13119         (pj4_alu_e1_conds): Likewise.
13120         (pj4_alu): Likewise.
13121         (pj4_alu_conds): Likewise.
13122         (pj4_shift): Likewise.
13123         (pj4_shift_conds): Likewise.
13124         (pj4_alu_shift): Likewise.
13125         (pj4_alu_shift_conds): Likewise.
13126         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
13127         changes.
13128         (cortexa7_older_only): Likewise.
13129         (cortexa7_younger): Likewise.
13131 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
13133         * ipa-pure-const.c (generate_summary): Rename to...
13134         (pure_const_generate_summary): ... this.
13136 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
13138         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
13140 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
13142         PR target/57909
13143         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
13144         usage in HI mode.
13146 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13148         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
13149         enabled without -march=zEC12.
13150         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
13151         flags to be set.
13153 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
13155         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
13156         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
13157         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
13158         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
13159         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
13160         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
13161         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
13162         ISA_HAS_FP4.
13163         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
13164         and ISA_HAS_NMADD3_NMSUB3.
13165         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
13166         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
13167         (nmsub4<mode>, nmsub3<mode>): Likewise.
13168         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
13170 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
13172         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
13173         TARGET_MIPS5400 checking.
13175 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
13176             Peter Bergner  <bergner@vnet.ibm.com>
13178         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
13179         registers in the comment.
13180         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
13181         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
13182         rather than FIRST_PSEUDO_REGISTERS.
13184 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
13186         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
13187         enable extra ISA flags with TARGET_HTM.
13189 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
13191         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
13192         Fix comment typos.
13194 2013-07-15  Cong Hou  <congh@google.com>
13196         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
13197         in compare function for sorting.
13199 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
13201         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
13202         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
13203         * config/rs6000/rs6000.opt: Add -mhtm option.
13204         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
13205         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
13206         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
13207         __HTM__ if the HTM instructions are available.
13208         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
13209         htm_spr_reg_operand): New define_predicates.
13210         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
13211         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
13212         Include htm.md.
13213         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
13214         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
13215         HTM builtin functions.
13216         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
13217         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
13218         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
13219         (rs6000_builtin_mask_calculate): Likewise.
13220         (rs6000_option_override_internal): Likewise.
13221         (bdesc_htm): Add new HTM builtin support.
13222         (htm_spr_num): New function.
13223         (htm_spr_regno): Likewise.
13224         (rs6000_htm_spr_icode): Likewise.
13225         (htm_expand_builtin): Likewise.
13226         (htm_init_builtins): Likewise.
13227         (rs6000_expand_builtin): Add support for HTM builtin functions.
13228         (rs6000_init_builtins): Likewise.
13229         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
13230         option.
13231         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
13232         (TARGET_HTM, MASK_HTM): Define macros.
13233         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
13234         (FIXED_REGISTERS): Likewise.
13235         (CALL_USED_REGISTERS): Likewise.
13236         (CALL_REALLY_USED_REGISTERS): Likewise.
13237         (REG_ALLOC_ORDER): Likewise.
13238         (enum reg_class): Likewise.
13239         (REG_CLASS_NAMES): Likewise.
13240         (REG_CLASS_CONTENTS): Likewise.
13241         (REGISTER_NAMES): Likewise.
13242         (ADDITIONAL_REGISTER_NAMES): Likewise.
13243         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
13244         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
13245         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
13246         * config/rs6000/htm.md: New file.
13247         * config/rs6000/htmintrin.h: New file.
13248         * config/rs6000/htmxlintrin.h: New file.
13250 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13252         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13253         Define SYMBOL_TINY_GOT, update comment.
13254         * config/aarch64/aarch64.c
13255         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
13256         (aarch64_expand_mov_immediate): Likewise.
13257         (aarch64_print_operand): Likewise.
13258         (aarch64_classify_symbol): Likewise.
13259         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
13260         (ldr_got_tiny): Define.
13262 2013-07-13  Tobias Grosser  <tobias@grosser.es>
13264         PR tree-optimization/54094
13265         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
13266           scheduling dimension for the parallelism check from the polyhedral
13267           information in the AST.
13268         * graphite-dependences.c (carries_deps): Do not assume the schedule is
13269           in 2D + 1 form.
13271 2013-07-13  Jason Merrill  <jason@redhat.com>
13273         * print-tree.c (debug_vec_tree): Use debug_raw.
13274         (debug_raw (vec<tree, va_gc> &)): New.
13275         (debug_raw (vec<tree, va_gc> *)): New.
13276         * tree.h: Declare them.
13278 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
13280         * ifcvt.c (ifcvt_after_combine): New static variable.
13281         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
13282         for size.
13283         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
13284         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
13285         rest_of_handle_if_after_reload): Pass new argument for if_convert.
13287 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
13289         * config/mips/mips.c (mips_expand_call): Remove empty statement.
13291 2013-07-12  Michael Matz  <matz@suse.de>
13293         PR middle-end/55771
13294         * convert.c (convert_to_real): Reject non-float inner types.
13296 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
13298         * config/aarch64/aarch64-protos.h
13299         (aarch64_simd_immediate_valid_for_move): Remove.
13300         * config/aarch64/aarch64.c (simd_immediate_info): New member.
13301         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
13302         cases.
13303         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
13305 2013-07-11  Steve Ellcey  <sellcey@mips.com>
13307         * config/mips/mips.c (mips_conditional_register_usage): Do not
13308         use t[0-7] registers in MIPS16 mode when optimizing for size.
13310 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
13312         * config/i386/i386.c (dispatch_function_versions): Fix array
13313         indexing of function_version_info to match actual_versions.
13315 2013-07-11  Teresa Johnson  <tejohnson@google.com>
13317         * vec.h (struct va_gc): Move release out-of-line.
13318         (va_gc::release): Call ggc_free on released vec.
13320 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13322         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
13323         Require GOT register as additional operand in UNSPEC.
13324         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
13325         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
13326         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
13327         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
13328         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
13329         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
13330         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
13332 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
13334         PR target/57631
13335         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
13336         name seen by assembler/linker rather if available.
13338 2013-07-11  Andreas Schwab  <schwab@suse.de>
13340         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
13342 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
13344         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
13346 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
13348         * doc/tm.texi.in: Move hook documentation to ....
13349         * target.def: ... here.
13351         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
13352         text on @hook line.
13353         * doc/tm.texi: Regenerate.
13355 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
13357         PR c++/57869
13358         * doc/invoke.texi: Document -Wconditionally-supported.
13360 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13362         PR target/57844
13363         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
13364         of my_fp.
13366 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13368         PR target/57506
13369         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
13370         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
13371         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
13372         Remove duplicate devices.
13373         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
13374         * config/avr/t-multilib: Regenerate.
13375         * config/avr/avr-tables.opt: Regenerate.
13376         * doc/avr-mmcu.texi: Regenerate.
13378 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13380         PR target/56987
13381         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
13383 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
13385         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
13386         the cost of MULT when optimizing for size.
13388 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
13390         * config/cr16/cr16-protos.h: Don't include target.h.
13392 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
13394         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
13395         adjust register size for TDmode and TFmode for VSX registers.
13397 2013-07-08  Kai Tietz  <ktietz@redhat.com>
13399         PR target/56892
13400         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
13401         hook_bool_const_tree_true.
13403 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13405         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
13406         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
13407         * config/s390/s390.md: Define FPR*_REGNUM constants.
13408         Fix FPR2_REGNUM constant (18 -> 17).
13409         ("*trunc<BFP:mode><DFP_ALL:mode>2")
13410         ("*trunc<DFP_ALL:mode><BFP:mode>2")
13411         ("trunc<BFP:mode><DFP_ALL:mode>2")
13412         ("trunc<DFP_ALL:mode><BFP:mode>2")
13413         ("*extend<BFP:mode><DFP_ALL:mode>2")
13414         ("*extend<DFP_ALL:mode><BFP:mode>2")
13415         ("extend<BFP:mode><DFP_ALL:mode>2")
13416         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
13417         FPR4_REGNUM.
13419 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
13421         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
13423 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13425         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
13426         and cfun_fpr_bit_p to cfun_fpr_save_p.
13427         (s390_frame_area, s390_register_info, s390_frame_info)
13428         (s390_emit_prologue, s390_emit_epilogue)
13429         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
13430         register numbers.
13431         * config/s390/s390.h: Define *_REGNUM macros for floating point
13432         register numbers.
13434 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
13436         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
13438 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
13440         PR rtl-optimization/57786
13441         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
13442         and break out of the loop when it is set to false.
13444 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
13446         PR target/57819
13447         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
13448         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
13449         (const_int 63)) 0)).
13450         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
13451         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
13452         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
13454         PR rtl-optimization/57829
13455         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
13456         mask bits outside of mode are just sign-extension from mode to HWI.
13458 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
13460         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
13461         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
13462         adjust_address instead of change_address to keep info about alignment.
13463         (emit_strmov): Remove.
13464         (emit_memmov): New function.
13465         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
13466         (expand_movmem_epilogue): Likewise and return updated rtx for
13467         destination.
13468         (expand_constant_movmem_prologue): Likewise and return updated rtx for
13469         destination and source.
13470         (decide_alignment): Refactor, handle vector_loop.
13471         (ix86_expand_movmem): Likewise.
13472         (ix86_expand_setmem): Likewise.
13473         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
13475 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13477         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13478         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13480 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13482         * config/i386/sse.md (sse_movlhps): Change alternative 3
13483         of operand 2 to "m".
13485 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13487         PR target/57807
13488         * config/i386/sse.md (iptr): New mode attribute.
13489         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13490         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13491         (<sse>_vmmul<mode>3): Ditto.
13492         (<sse>_vmdiv<mode>3): Ditto.
13493         (sse_vmrcpv4sf2): Ditto.
13494         (<sse>_vmsqrt<mode>2): Ditto.
13495         (sse_vmrsqrtv4sf2): Ditto.
13496         (<sse>_vm<code><mode>3): Ditto.
13497         (avx_vmcmp<mode>3): Ditto.
13498         (<sse>_vmmaskcmp<mode>3): Ditto.
13499         (<sse>_comi): Ditto.
13500         (<sse>_ucomi): Ditto.
13501         (*xop_vmfrcz_<mode>): Ditto.
13502         (*fmai_fmadd_<mode>): Ditto.
13503         (*fmai_fmsub_<mode>): Ditto.
13504         (*fmai_fnmadd_<mode>): Ditto.
13505         (*fmai_fnmsub_<mode>): Ditto.
13506         (*fma4i_vmfmadd_<mode>): Ditto.
13507         (*fma4i_vmfmsub_<mode>): Ditto.
13508         (*fma4i_vmfnmadd_<mode>): Ditto.
13509         (*fma4i_vmfnmsub_<mode>): Ditto.
13510         (*xop_vmfrcz_<mode>): Ditto.
13511         (sse_cvtps2pi): Ditto.
13512         (sse_cvttps2pi): Ditto.
13513         (sse_cvtss2si): Ditto.
13514         (sse_cvtss2si_2): Ditto.
13515         (sse_cvtss2siq_2): Ditto.
13516         (sse_cvttss2si): Ditto.
13517         (sse_cvttss2siq): Ditto.
13518         (sse_cvtsd2si): Ditto.
13519         (sse_cvtsd2si_2): Ditto.
13520         (sse_cvtsd2siq_2): Ditto.
13521         (sse_cvttsd2si): Ditto.
13522         (sse_cvttsd2siq): Ditto.
13523         (sse_cvtsd2ss): Ditto.
13524         (sse_cvtss2sd): Ditto.
13525         (avx2_pbroadcast<mode>): Ditto.
13526         (avx2_pbroadcast<mode>_1): Ditto.
13527         (*avx_vperm_broadcast_v4sf): Ditto.
13529         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13530         (sse_movlhps): Ditto.
13531         (sse_storehps): Ditto.
13532         (sse_loadhps): Ditto.
13533         (sse_storelps): Ditto.
13534         (sse_loadlps): Ditto.
13535         (*vec_concatv4sf): Ditto.
13536         (*vec_interleave_highv2df): Ditto.
13537         (*vec_interleave_lowv2df): Ditto.
13538         (*vec_extractv2df_1_sse): Ditto.
13539         (*vec_extractv2df_0_sse): Ditto.
13540         (sse2_storelpd): Ditto.
13541         (sse2_loadlpd): Ditto.
13542         (sse2_movsd): Ditto.
13543         (*vec_concatv4si): Ditto.
13544         (vec_concatv2di): Ditto.
13546         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13547         for Intel asm dialect.
13548         (mmx_punpcklwd): Ditto.
13549         (mmx_punpckldq): Ditto.
13551         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13552         for intel assembler dialect.
13554 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13556         PR target/29776
13557         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13558         for BUILT_IN_C{LZ,LRSB}*.
13559         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13560         * tree-vrp.c (extract_range_basic): Handle
13561         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13562         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13563         fall thru to code calling set_value*.
13564         * builtins.c (expand_builtin): Remove *IMAX cases.
13565         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13566         if width is bigger than 2*HWI.
13568 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13570         PR rtl-optimization/55342
13571         * lra-int.h (lra_subreg_reload_pseudos): New.
13572         * lra.c: Add undoing optional reloads to the block diagram.
13573         (lra_subreg_reload_pseudos): New.
13574         (lra_optional_reload_pseudos): Change comments.
13575         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13576         lra_optional_reload_pseudos after undo transformations.
13577         * lra-assigns.c (pseudo_prefix_title): New.
13578         (lra_setup_reg_renumber): Use it.
13579         (spill_for): Ditto.  Check subreg reload pseudos too.
13580         (assign_by_spills): Consider subreg reload pseudos too.
13581         * lra-constraints.c (simplify_operand_subreg): Use
13582         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13583         (curr_insn_transform): Recognize and do optional reloads.
13584         (undo_optional_reloads): New.
13585         (lra_undo_inheritance): Call undo_optional_reloads.
13587 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13589         * tree-complex.c (expand_complex_operations_1): Fix typo.
13591 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13593         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13594         (tune_params): New member 'const vec_costs'.
13595         * config/aarch64/aarch64.c (generic_vector_cost): New.
13596         (generic_tunings): New member 'generic_vector_cost'.
13597         (aarch64_builtin_vectorization_cost): New.
13598         (aarch64_add_stmt_cost): New.
13599         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13600         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13602 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13604         PR target/57777
13605         * config/i386/predicates.md (vsib_address_operand): Disallow
13606         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13608 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13610         PR middle-end/55030
13611         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13612         expand_builtin_setjmp_receiver.
13613         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13614         with NULL for the label parameter.
13615         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13616         the frame-pointer.  Adjust comments.
13617         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13618         only if LABEL is non-NULL.
13620 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13622         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13623         (ARM_ABI_AAPCS64): Ditto.
13624         (arm_abi): Ditto.
13625         (ARM_DEFAULT_ABI): Ditto.
13627 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13629         * config/aarch64/aarch64-builtins.c
13630         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13631         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13632         (st1): Likewise.
13633         * config/aarch64/aarch64-simd.md
13634         (aarch64_ld1<VALL:mode>): New.
13635         (aarch64_st1<VALL:mode>): Likewise.
13636         * config/aarch64/arm_neon.h
13637         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13639 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13641         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13642         target ISA is AVX.
13643         (ix86_option_override_internal):Turn on all -mavx target flags by
13644         default as they are dependent on AVX anyway.
13646 2013-07-02  Cary Coutant  <ccoutant@google.com>
13648         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13649         deterministic hash.
13650         (loc_checksum_ordered): Likewise.
13651         (hash_loc_operands): Remove inline keyword.
13653 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13655         PR tree-optimization/57741
13656         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13657         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13658         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13659         Allow REAL_CST step_exprs if flag_associative_math.
13660         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13662 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13664         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13666 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13668         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13670 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13672         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13673         encoding.
13674         (iorsi3_insn): Likewise.
13675         (arm_xorsi3): Likewise.
13677 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13679         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13680         to ...
13681         (attribute "type"): ... here, and prefix with "wmmx_".
13682         (attribute "core_cycles"): Update for attribute changes.
13683         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13684         (tbcstv4hi): Likewise.
13685         (tbcstv2si): Likewise.
13686         (iwmmxt_iordi3): Likewise.
13687         (iwmmxt_xordi3): Likewise.
13688         (iwmmxt_anddi3): Likewise.
13689         (iwmmxt_nanddi3): Likewise.
13690         (iwmmxt_arm_movdi): Likewise.
13691         (iwmmxt_movsi_insn): Likewise.
13692         (mov<mode>_internal): Likewise.
13693         (and<mode>3_iwmmxt): Likewise.
13694         (ior<mode>3_iwmmxt): Likewise.
13695         (xor<mode>3_iwmmxt): Likewise.
13696         (add<mode>3_iwmmxt): Likewise.
13697         (ssaddv8qi3): Likewise.
13698         (ssaddv4hi3): Likewise.
13699         (ssaddv2si3): Likewise.
13700         (usaddv8qi3): Likewise.
13701         (usaddv4hi3): Likewise.
13702         (usaddv2si3): Likewise.
13703         (sub<mode>3_iwmmxt): Likewise.
13704         (sssubv8qi3): Likewise.
13705         (sssubv4hi3): Likewise.
13706         (sssubv2si3): Likewise.
13707         (ussubv8qi3): Likewise.
13708         (ussubv4hi3): Likewise.
13709         (ussubv2si3): Likewise.
13710         (mulv4hi3_iwmmxt): Likewise.
13711         (smulv4hi3_highpart): Likewise.
13712         (umulv4hi3_highpart): Likewise.
13713         (iwmmxt_wmacs): Likewise.
13714         (iwmmxt_wmacsz): Likewise.
13715         (iwmmxt_wmacu): Likewise.
13716         (iwmmxt_wmacuz): Likewise.
13717         (iwmmxt_clrdi): Likewise.
13718         (iwmmxt_clrv8qi): Likewise.
13719         (iwmmxt_clr4hi): Likewise.
13720         (iwmmxt_clr2si): Likewise.
13721         (iwmmxt_uavgrndv8qi3): Likewise.
13722         (iwmmxt_uavgrndv4hi3): Likewise.
13723         (iwmmxt_uavgv8qi3): Likewise.
13724         (iwmmxt_uavgv4hi3): Likewise.
13725         (iwmmxt_tinsrb): Likewise.
13726         (iwmmxt_tinsrh): Likewise.
13727         (iwmmxt_tinsrw): Likewise.
13728         (iwmmxt_textrmub): Likewise.
13729         (iwmmxt_textrmsb): Likewise.
13730         (iwmmxt_textrmuh): Likewise.
13731         (iwmmxt_textrmsh): Likewise.
13732         (iwmmxt_textrmw): Likewise.
13733         (iwmxxt_wshufh): Likewise.
13734         (eqv8qi3): Likewise.
13735         (eqv4hi3): Likewise.
13736         (eqv2si3): Likewise.
13737         (gtuv8qi3): Likewise.
13738         (gtuv4hi3): Likewise.
13739         (gtuv2si3): Likewise.
13740         (gtv8qi3): Likewise.
13741         (gtv4hi3): Likewise.
13742         (gtv2si3): Likewise.
13743         (smax<mode>3_iwmmxt): Likewise.
13744         (umax<mode>3_iwmmxt): Likewise.
13745         (smin<mode>3_iwmmxt): Likewise.
13746         (umin<mode>3_iwmmxt): Likewise.
13747         (iwmmxt_wpackhss): Likewise.
13748         (iwmmxt_wpackwss): Likewise.
13749         (iwmmxt_wpackdss): Likewise.
13750         (iwmmxt_wpackhus): Likewise.
13751         (iwmmxt_wpackwus): Likewise.
13752         (iwmmxt_wpackdus): Likewise.
13753         (iwmmxt_wunpckihb): Likewise.
13754         (iwmmxt_wunpckihh): Likewise.
13755         (iwmmxt_wunpckihw): Likewise.
13756         (iwmmxt_wunpckilb): Likewise.
13757         (iwmmxt_wunpckilh): Likewise.
13758         (iwmmxt_wunpckilw): Likewise.
13759         (iwmmxt_wunpckehub): Likewise.
13760         (iwmmxt_wunpckehuh): Likewise.
13761         (iwmmxt_wunpckehuw): Likewise.
13762         (iwmmxt_wunpckehsb): Likewise.
13763         (iwmmxt_wunpckehsh): Likewise.
13764         (iwmmxt_wunpckehsw): Likewise.
13765         (iwmmxt_wunpckelub): Likewise.
13766         (iwmmxt_wunpckeluh): Likewise.
13767         (iwmmxt_wunpckeluw): Likewise.
13768         (iwmmxt_wunpckelsb): Likewise.
13769         (iwmmxt_wunpckelsh): Likewise.
13770         (iwmmxt_wunpckelsw): Likewise.
13771         (ror<mode>3): Likewise.
13772         (ashr<mode>3_iwmmxt): Likewise.
13773         (lshr<mode>3_iwmmxt): Likewise.
13774         (ashl<mode>3_iwmmxt): Likewise.
13775         (ror<mode>3_di): Likewise.
13776         (ashr<mode>3_di): Likewise.
13777         (lshr<mode>3_di): Likewise.
13778         (ashl<mode>3_di): Likewise.
13779         (iwmmxt_wmadds): Likewise.
13780         (iwmmxt_wmaddu): Likewise.
13781         (iwmmxt_tmia): Likewise.
13782         (iwmmxt_tmiaph): Likewise.
13783         (iwmmxt_tmiabb): Likewise.
13784         (iwmmxt_tmiatb): Likewise.
13785         (iwmmxt_tmiabt): Likewise.
13786         (iwmmxt_tmiatt): Likewise.
13787         (iwmmxt_tmovmskb): Likewise.
13788         (iwmmxt_tmovmskh): Likewise.
13789         (iwmmxt_tmovmskw): Likewise.
13790         (iwmmxt_waccb): Likewise.
13791         (iwmmxt_wacch): Likewise.
13792         (iwmmxt_waccw): Likewise.
13793         (iwmmxt_waligni): Likewise.
13794         (iwmmxt_walignr): Likewise.
13795         (iwmmxt_walignr0): Likewise.
13796         (iwmmxt_walignr1): Likewise.
13797         (iwmmxt_walignr2): Likewise.
13798         (iwmmxt_walignr3): Likewise.
13799         (iwmmxt_wsadb): Likewise.
13800         (iwmmxt_wsadh): Likewise.
13801         (iwmmxt_wsadbz): Likewise.
13802         (iwmmxt_wsadhz): Likewise.
13803         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13804         (iwmmxt_wabsdiffb): Likewise.
13805         (iwmmxt_wabsdiffh): Likewise.
13806         (iwmmxt_wabsdiffw): Likewise.
13807         (iwmmxt_waddsubhx): Likewise
13808         (iwmmxt_wsubaddhx): Likewise.
13809         (addc<mode>3): Likewise.
13810         (iwmmxt_avg4): Likewise.
13811         (iwmmxt_avg4r): Likewise.
13812         (iwmmxt_wmaddsx): Likewise.
13813         (iwmmxt_wmaddux): Likewise.
13814         (iwmmxt_wmaddsn): Likewise.
13815         (iwmmxt_wmaddun): Likewise.
13816         (iwmmxt_wmulwsm): Likewise.
13817         (iwmmxt_wmulwum): Likewise.
13818         (iwmmxt_wmulsmr): Likewise.
13819         (iwmmxt_wmulumr): Likewise.
13820         (iwmmxt_wmulwsmr): Likewise.
13821         (iwmmxt_wmulwumr): Likewise.
13822         (iwmmxt_wmulwl): Likewise.
13823         (iwmmxt_wqmulm): Likewise.
13824         (iwmmxt_wqmulwm): Likewise.
13825         (iwmmxt_wqmulmr): Likewise.
13826         (iwmmxt_wqmulwmr): Likewise.
13827         (iwmmxt_waddbhusm): Likewise.
13828         (iwmmxt_waddbhusl): Likewise.
13829         (iwmmxt_wqmiabb): Likewise.
13830         (iwmmxt_wqmiabt): Likewise.
13831         (iwmmxt_wqmiatb): Likewise.
13832         (iwmmxt_wqmiatt): Likewise.
13833         (iwmmxt_wqmiabbn): Likewise.
13834         (iwmmxt_wqmiabtn): Likewise.
13835         (iwmmxt_wqmiatbn): Likewise.
13836         (iwmmxt_wqmiattn): Likewise.
13837         (iwmmxt_wmiabb): Likewise.
13838         (iwmmxt_wmiabt): Likewise.
13839         (iwmmxt_wmiatb): Likewise.
13840         (iwmmxt_wmiatt): Likewise.
13841         (iwmmxt_wmiabbn): Likewise.
13842         (iwmmxt_wmiabtn): Likewise.
13843         (iwmmxt_wmiatbn): Likewise.
13844         (iwmmxt_wmiattn): Likewise.
13845         (iwmmxt_wmiawbb): Likewise.
13846         (iwmmxt_wmiawbt): Likewise.
13847         (iwmmxt_wmiawtb): Likewise.
13848         (iwmmxt_wmiawtt): Likewise.
13849         (iwmmxt_wmiawbbn): Likewise.
13850         (iwmmxt_wmiawbtn): Likewise.
13851         (iwmmxt_wmiawtbn): Likewise.
13852         (iwmmxt_wmiawttn): Likewise.
13853         (iwmmxt_wmerge): Likewise.
13854         (iwmmxt_tandc<mode>3): Likewise.
13855         (iwmmxt_torc<mode>3): Likewise.
13856         (iwmmxt_torvsc<mode>3): Likewise.
13857         (iwmmxt_textrc<mode>3): Likewise.
13858         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13859         (wmmxt_pack): Likewise.
13860         (wmmxt_mult_c1): Likewise.
13861         (wmmxt_mult_c2): Likewise.
13862         (wmmxt_alu_c1): Likewise.
13863         (wmmxt_alu_c2): Likewise.
13864         (wmmxt_alu_c3): Likewise.
13865         (wmmxt_transfer_c1): Likewise.
13866         (wmmxt_transfer_c2): Likewise.
13867         (wmmxt_transfer_c3): Likewise.
13868         (marvell_f_iwmmxt_wstr): Likewise.
13869         (marvell_f_iwmmxt_wldr): Likewise.
13871 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13873         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13875 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13877         Revert:
13878         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13879         * lra-constraints.c (need_for_split_p): Check call used hard regs
13880         living through calls.
13882         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13883         call used regs for call insn.
13885 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13887         PR target/57736
13888         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13889         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13891 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13893         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13895 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13897         * lra-constraints.c (need_for_split_p): Check call used hard regs
13898         living through calls.
13900 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13902         PR target/57744
13903         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13904         to tie with any other modes.  Eliminate Altivec vector mode tests,
13905         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13906         code, to return 0 if testing MODE2 for a condition, if we've
13907         already tested MODE1 for the same condition.
13909 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13911         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13912         layout.
13914 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13916         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13917         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13919 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13921         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13922         Define.
13923         (aarch64_symbolic_constant_p): Remove.
13924         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13925         static.  Fix line length and white space.
13926         (aarch64_symbolic_constant_p): Remove.
13927         * config/aarch64/predicates.md (aarch64_valid_symref):
13928         Use aarch64_classify_symbol_expression.
13930 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13932         * config/arm/constraints.md (Ts): New constraint.
13933         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13934         16-bit encodings.
13935         (compare_scc): Use "Ts" constraint for operand 0.
13936         (ior_scc_scc): Likewise.
13937         (and_scc_scc): Likewise.
13938         (and_scc_scc_nodom): Likewise.
13939         (ior_scc_scc_cmp): Likewise for operand 7.
13940         (and_scc_scc_cmp): Likewise.
13941         * config/arm/thumb2.md (thumb2_movsi_insn):
13942         Add alternatives for 16-bit encodings.
13943         (thumb2_movhi_insn): Likewise.
13944         (thumb2_movsicc_insn): Likewise.
13945         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13946         (thumb2_negscc): Use "Ts" constraint.
13947         Move mvn instruction outside cond_exec block.
13948         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13949         for 16-bit encodings.
13951 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13953         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13954         encoding.
13955         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13956         (mulsi3subsi): Likewise.
13957         (mulsidi3adddi): Likewise.
13958         (mulsidi3_v6): Likewise.
13959         (umulsidi3_v6): Likewise.
13960         (umulsidi3adddi_v6): Likewise.
13961         (smulsi3_highpart_v6): Likewise.
13962         (umulsi3_highpart_v6): Likewise.
13963         (mulhisi3tb): Likewise.
13964         (mulhisi3bt): Likewise.
13965         (mulhisi3tt): Likewise.
13966         (maddhisi4): Likewise.
13967         (maddhisi4tb): Likewise.
13968         (maddhisi4tt): Likewise.
13969         (maddhidi4): Likewise.
13970         (maddhidi4tb): Likewise.
13971         (maddhidi4tt): Likewise.
13972         (zeroextractsi_compare0_scratch): Likewise.
13973         (insv_zero): Likewise.
13974         (insv_t2): Likewise.
13975         (anddi_notzesidi_di): Likewise.
13976         (anddi_notsesidi_di): Likewise.
13977         (andsi_notsi_si): Likewise.
13978         (iordi_zesidi_di): Likewise.
13979         (xordi_zesidi_di): Likewise.
13980         (andsi_iorsi3_notsi): Likewise.
13981         (smax_0): Likewise.
13982         (smax_m1): Likewise.
13983         (smin_0): Likewise.
13984         (not_shiftsi): Likewise.
13985         (unaligned_loadsi): Likewise.
13986         (unaligned_loadhis): Likewise.
13987         (unaligned_loadhiu): Likewise.
13988         (unaligned_storesi): Likewise.
13989         (unaligned_storehi): Likewise.
13990         (extv_reg): Likewise.
13991         (extzv_t2): Likewise.
13992         (divsi3): Likewise.
13993         (udivsi3): Likewise.
13994         (arm_zero_extendhisi2addsi): Likewise.
13995         (arm_zero_extendqisi2addsi): Likewise.
13996         (compareqi_eq0): Likewise.
13997         (arm_extendhisi2_v6): Likewise.
13998         (arm_extendqisi2addsi): Likewise.
13999         (arm_movt): Likewise.
14000         (thumb2_ldrd): Likewise.
14001         (thumb2_ldrd_base): Likewise.
14002         (thumb2_ldrd_base_neg): Likewise.
14003         (thumb2_strd): Likewise.
14004         (thumb2_strd_base): Likewise.
14005         (thumb2_strd_base_neg): Likewise.
14006         (arm_negsi2): Add alternative for 16-bit encoding.
14007         (arm_one_cmplsi2): Likewise.
14009 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14011         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
14012         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
14013         (movdfcc): Likewise.
14014         * config/arm/vfp.md (*thumb2_movsf_vfp):
14015         Disable predication for arm_restrict_it.
14016         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
14017         (*thumb2_movdfcc_vfp): Likewise.
14018         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
14019         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
14020         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
14021         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
14022         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
14023         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
14024         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
14025         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
14026         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
14027         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
14028         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
14029         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
14030         Disable predication for arm_restrict_it.
14032 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
14034         * config/i386/bmiintrin.h (_bextr_u32): New.
14035         (_bextr_u64): Ditto.
14037 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
14039         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
14040         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
14041         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
14042         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
14043         * config/mips/n32-elf.h: ...this new file.
14045 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
14047         PR target/57224
14048         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
14049         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
14051 2013-06-27  Catherine Moore  <clm@codesourcery.com>
14053         * config/mips/mips-tables.opt: Regenerate.
14054         * config/mips/mips-cpus.def: Add m14ke and m14kec.
14055         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
14056         * doc/invoke.texi: Add -m14kc.
14058 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
14060         PR target/57623
14061         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
14062         constraints of operand 1 and 2.
14064         PR target/57623
14065         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
14066         to match RTL canonicalization.  Swap predicates and
14067         constraints of operand 1 and 2.
14069 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
14071         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
14072         Process OP_INOUT regs for splitting too.
14074 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
14076         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
14077         decl before the loop, initialize to NULL.
14078         (vectorizable_load): Initialize ptr_incr to NULL.
14080 2013-06-27  Martin Jambor  <mjambor@suse.cz>
14082         PR lto/57208
14083         * ipa-ref.h (ipa_maybe_record_reference): Declare.
14084         * ipa-ref.c (ipa_maybe_record_reference): New function.
14085         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
14086         * ipa-cp.c (create_specialized_node): Record potential references from
14087         aggvals.
14088         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
14090 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
14092         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
14093         parameter 'mode' of type 'enum machine_mode mode'; change to pass
14094         'mode' to force_reg.
14095         (aarch64_add_offset): Update calls to aarch64_force_temporary.
14096         (aarch64_expand_mov_immediate): Likewise.
14098 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
14100         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
14101         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
14103 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14105         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
14106         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
14107         (struct machine_function): Add tbegin_p.
14108         (s390_canonicalize_comparison): Fold CC mode compares to
14109         conditional jump if possible.
14110         (s390_emit_jump): Return the emitted jump.
14111         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
14112         Handle CCRAWmode compares.
14113         (s390_option_override): Default to -mhtm if available.
14114         (s390_reg_clobbered_rtx): Handle floating point regs as well.
14115         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
14116         FPRs instead of df_regs_ever_live_p.
14117         (s390_optimize_nonescaping_tx): New function.
14118         (s390_init_frame_layout): Extend clobbered_regs array to cover
14119         FPRs as well.
14120         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
14121         (s390_expand_tbegin): New function.
14122         (enum s390_builtin): New enum definition.
14123         (code_for_builtin): New array definition.
14124         (s390_init_builtins): New function.
14125         (s390_expand_builtin): New function.
14126         (TARGET_INIT_BUILTINS): Define.
14127         (TARGET_EXPAND_BUILTIN): Define.
14128         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
14129         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
14130         (s390_alc_comparison): Likewise.
14131         * config/s390/s390-modes.def: Add CCRAWmode.
14132         * config/s390/s390.h (processor_flags): Add PF_TX.
14133         (TARGET_CPU_HTM): Define macro.
14134         (TARGET_HTM): Define macro.
14135         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
14136         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
14137         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
14138         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
14139         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
14140         (TBEGIN_MASK, TBEGINC_MASK): New constants.
14141         ("*cc_to_int"): Move up.
14142         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
14143         constants other than 0.
14144         ("*ccraw_to_int"): New insn and splitter definition.
14145         ("tbegin", "tbegin_nofloat", "tbegin_retry")
14146         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
14147         ("tx_assist"): New expander.
14148         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
14149         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
14150         * config/s390/s390.opt: Add -mhtm option.
14151         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
14152         * config/s390/htmxlintrin.h: New file.
14153         * config/s390/htmintrin.h: New file.
14154         * config/s390/s390intrin.h: New file.
14155         * doc/extend.texi: Document htm builtins.
14156         * config.gcc: Add the new header files to extra_headers.
14158 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
14160         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
14161         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
14163 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
14164             Pat Haugen  <pthaugen@us.ibm.com>
14165             Peter Bergner  <bergner@vnet.ibm.com>
14167         * config/rs6000/power8.md: New.
14168         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
14169         setting for power8 entry.
14170         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
14171         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
14172         test for Power4/Power5 only.
14173         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
14174         support.
14175         (force_new_group): Adjust comment.
14176         * config/rs6000/rs6000.md: Include power8.md.
14178 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
14180         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
14181         * config/arm/arm-protos.h (arm_max_conditional_execute): New
14182         declaration.
14183         (tune_params): Update comment.
14184         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
14185         (arm_max_conditional_execute): New function.
14186         (thumb2_final_prescan_insn): Use max_insn_skipped and
14187         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
14189 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
14191         PR tree-optimization/57705
14192         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
14193         SSA_NAME step, provided that it is not defined inside the loop.
14194         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
14195         (get_initial_def_for_induction): Handle SSA_NAME IV step.
14197 2013-06-25  Martin Jambor  <mjambor@suse.cz>
14199         PR middle-end/57670
14200         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
14201         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
14202         calls in the dump.
14203         (ipa_note_param_call): Initialize member_ptr flag.
14204         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
14205         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
14206         (ipa_write_indirect_edge_info): Stream member_ptr flag.
14207         (ipa_read_indirect_edge_info): Likewise.
14209 2013-06-25  Richard Biener  <rguenther@suse.de>
14211         PR middle-end/56977
14212         * passes.c (init_optimization_passes): Move pass_fold_builtins
14213         and pass_dce earlier with -Og.
14215 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
14217         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
14218         <BIT_FIELD_REF>: Remove trailing TAB.
14219         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
14220         remove blank line.
14222 2013-06-24  Martin Jambor  <mjambor@suse.cz>
14224         PR tree-optimization/57358
14225         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
14226         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
14227         (ipa_analyze_params_uses): Generate pessimistic info when true.
14229 2013-06-24  Martin Jambor  <mjambor@suse.cz>
14231         PR tree-optimization/57539
14232         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
14233         global.inlined_to of the new node to it.  All callers changed.
14234         * ipa-inline-transform.c (clone_inlined_nodes): New variable
14235         inlining_into, pass it to cgraph_clone_node.
14236         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
14237         ipa_free_edge_args_substructures.
14238         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
14239         rdesc linked list.  Do not assert rdesc edges have inlined caller.
14240         Assert we have found an rdesc in the rdesc list.
14242 2013-06-24  Richard Biener  <rguenther@suse.de>
14244         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
14245         (pointer_set_lookup): Declare.
14246         (class pointer_map): New template class implementing a
14247         generic pointer to T map.
14248         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
14249         pointer_map<T>::contains, pointer_map<T>::insert,
14250         pointer_map<T>::traverse): New functions.
14251         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
14252         (pointer_set_lookup): New function.
14253         (pointer_set_contains): Use pointer_set_lookup.
14254         (pointer_set_insert): Likewise.
14255         (insert_aux): Remove.
14256         (struct pointer_map_t): Embed a pointer_set_t.
14257         (pointer_map_create): Adjust.
14258         (pointer_map_destroy): Likewise.
14259         (pointer_map_contains): Likewise.
14260         (pointer_map_insert): Likewise.
14261         (pointer_map_traverse): Likewise.
14262         * tree-streamer.h (struct streamer_tree_cache_d): Use a
14263         pointer_map<unsigned> instead of a pointer_map_t.
14264         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
14265         (streamer_tree_cache_lookup): Likewise.
14266         (streamer_tree_cache_create): Likewise.
14267         (streamer_tree_cache_delete): Likewise.
14268         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
14269         pointer_map<unsigned> instead of a pointer_map_t.
14270         (lto_init_tree_ref_encoder): Adjust.
14271         (lto_destroy_tree_ref_encoder): Likewise.
14272         * lto-section-out.c (lto_output_decl_index): Likewise.
14273         (lto_record_function_out_decl_state): Likewise.
14274         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
14276 2013-06-24  Richard Biener  <rguenther@suse.de>
14278         PR tree-optimization/57488
14279         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
14281 2013-06-24  Alan Modra  <amodra@gmail.com>
14283         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
14284         (gen_easy_altivec_constant): Likewise.
14285         * config/rs6000/predicates.md (easy_vector_constant_add_self,
14286         easy_vector_constant_msb): Likewise.
14288 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
14290         PR target/57688
14291         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
14292         add missing return true.
14294 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
14296         PR target/52483
14297         * config/sh/predicates.md (general_extend_operand): Invoke
14298         general_movsrc_operand for memory operands.
14299         (general_movsrc_operand): Allow reg+reg addressing, do not use
14300         general_operand for memory operands.
14302 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
14304         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
14305         when current target options does not apply.
14306         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
14307         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
14308         * config/i386/bmiintrin.h: Pass appropriate target
14309         attributes to header.
14310         * config/i386/mmintrin.h: Ditto.
14311         * config/i386/nmmintrin.h: Ditto.
14312         * config/i386/avx2intrin.h: Ditto.
14313         * config/i386/fxsrintrin.h: Ditto.
14314         * config/i386/tbmintrin.h: Ditto.
14315         * config/i386/xsaveintrin.h: Ditto.
14316         * config/i386/f16cintrin.h: Ditto.
14317         * config/i386/xtestintrin.h: Ditto.
14318         * config/i386/xsaveoptintrin.h: Ditto.
14319         * config/i386/bmi2intrin.h: Ditto.
14320         * config/i386/lzcntintrin.h: Ditto.
14321         * config/i386/smmintrin.h: Ditto.
14322         * config/i386/wmmintrin.h: Ditto.
14323         * config/i386/x86intrin.h: Remove all header include guards.
14324         * config/i386/prfchwintrin.h: Ditto.
14325         * config/i386/pmmintrin.h: Ditto.
14326         * config/i386/tmmintrin.h: Ditto.
14327         * config/i386/xmmintrin.h: Ditto.
14328         * config/i386/popcntintrin.h: Ditto.
14329         * config/i386/rdseedintrin.h: Ditto.
14330         * config/i386/ammintrin.h: Ditto.
14331         * config/i386/emmintrin.h: Ditto.
14332         * config/i386/immintrin.h: Remove all header include guards.
14333         * config/i386/fma4intrin.h: Ditto.
14334         * config/i386/lwpintrin.h: Ditto.
14335         * config/i386/xopintrin.h: Ditto.
14336         * config/i386/ia32intrin.h: Ditto.
14337         * config/i386/avxintrin.h: Ditto.
14338         * config/i386/rtmintrin.h: Ditto.
14339         * config/i386/fmaintrin.h: Ditto.
14340         * config/i386/mm3dnow.h: Ditto.
14342 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
14344         * common/config/i386/i386-common.c: Handle LZCNT.
14346 2013-06-22  Andi Kleen  <ak@linux.intel.com>
14348         * doc/extend.texi: Use __atomic_store_n instead of
14349         __atomic_store in HLE example.
14351 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
14353         * config/sh/sh.c: Remove <cstdlib> workaround.
14355 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14357         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
14359 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14361         * doc/extend.texi: Document that __atomic_clear and
14362           __atomic_test_and_set should only be used with bool.
14364 2013-06-20  Jan Hubicka  <jh@suse.cz>
14366         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
14367         types_same_for_odr.
14368         * tree.c (decls_same_for_odr): New function.
14369         (same_for_edr): New function.
14370         (types_same_for_odr): New function.
14371         (get_binfo_at_offset): Use it.
14372         * tree.h (types_same_for_odr): Declare.
14374 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
14375             Jason Merrill  <jason@redhat.com>
14377         * system.h: Include <cstdlib> as well as <stdlib.h>.
14379 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
14381         PR target/57655
14382         * config/i386/i386.c (construct_container): Report error if
14383         long double is used with disabled x87 float returns.
14385 2013-06-20  Jan Hubicka  <jh@suse.cz>
14387         * lto-cgraph.c (input_symtab): Do not set cgraph state.
14389 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
14391         PR rtl-optimization/57425
14392         PR rtl-optimization/57569
14393         * alias.c (write_dependence_p): Remove parameters mem_mode and
14394         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
14395         Changed all callers.
14396         (canon_anti_dependence): Get comments and semantics in sync.
14397         Add parameter mem_canonicalized.  Changed all callers.
14398         * rtl.h (canon_anti_dependence): Update prototype.
14400 2013-06-20  Richard Biener  <rguenther@suse.de>
14402         * data-streamer-in.c (streamer_read_uhwi): Optimize single
14403         byte case, inline streamer_read_uchar and defer section
14404         overrun check.
14406 2013-06-20  Richard Biener  <rguenther@suse.de>
14408         PR tree-optimization/57584
14409         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
14410         SSA names into the expanded expression that take part in
14411         abnormal coalescing.
14413 2013-06-19  Sharad Singhai  <singhai@google.com>
14415         * gcov.c (print_usage): Handle new option.
14416         (process_args): Ditto.
14417         (get_gcov_intermediate_filename): New function.
14418         (output_intermediate_file): New function.
14419         (output_gcov_file): New function
14420         (generate_results): Handle new option.
14421         (release_function): Relase demangled name.
14422         (read_graph_file): Handle demangled name.
14423         (output_lines): Ditto.
14424         * doc/gcov.texi: Document gcov intermediate format.
14426 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
14428         PR bootstrap/57604
14429         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
14430         (lra_emit_add): Use the functions.  Add comment about Y as an
14431         address segment.
14433 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
14435         PR driver/57652
14436         * collect2.c (collect_atexit): New.
14437         (collect_exit): Delete.
14438         (main): Register collect_atexit with atexit.
14439         (collect_wait): Change collect_exit to exit.
14440         (do_wait): Same.
14441         * collect2.h (collect_exit): Delete.
14442         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
14444 2013-06-19  Wei Mi  <wmi@google.com>
14446         PR rtl-optimization/57518
14447         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
14448         if regno is used in paradoxical subreg.
14449         (update_equiv_regs): Check pdx_subregs[regno] before
14450         set a reg to be equivalent with a mem.
14452 2013-06-19  Matthias Klose  <doko@ubuntu.com>
14454         PR driver/57651
14455         * file-find.h (find_a_file): Add a mode parameter.
14456         * file-find.c (find_a_file): Likewise.
14457         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
14458         with X_OK for the executables.
14459         * collect2.c (main): Call find_a_file with X_OK.
14461 2013-06-19  Steve Ellcey  <sellcey@mips.com>
14463         PR target/56942
14464         * config/mips/mips.md (casesi_internal_mips16_<mode>):
14465         Use NEXT_INSN instead of next_real_insn.
14467 2013-06-19  Jan Hubicka  <jh@suse.cz>
14469         * cgraph.h (const_value_known_p): Replace by ...
14470         (ctor_for_folding): .. this one.
14471         * cgraphunit.c (process_function_and_variable_attributes): Use it.
14472         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
14473         * expr.c (expand_expr_real_1): Likewise.
14474         (string_constant): Likewise.
14475         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14476         * ipa.c (process_references): Likewise.
14477         (symtab_remove_unreachable_nodes): Likewise.
14478         * ipa-inline-analysis.c (param_change_prob): Likewise.
14479         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14480         (get_base_constructor): Likwise.
14481         * varpool.c (varpool_remove_node): Likewise.
14482         (varpool_remove_initializer): LIkewise.
14483         (dump_varpool_node): LIkwise.
14484         (const_value_known_p): Rewrite to ...
14485         (ctor_for_folding): ... this one.
14487 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14489         PR driver/57651
14490         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14491         PERSONALITY in $PATH derived prefixes.
14493 2013-06-19  Jeff Law  <law@redhat.com>
14495         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14496         in comment.
14498         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14499         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14500         booleans.
14502 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14504         * config/arm/vfp.md: Move VFP instruction classification documentation
14505         to ...
14506         * config/arm/arm.md: ... here.  Update instruction classification
14507         documentation.
14509 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14511         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14512         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14513         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14515 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14517         PR target/57609
14518         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14519         with NEXT_INSN.  Use tablejump_p to check for jump table data
14520         insns.
14522 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14524         PR c++/56544
14525         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14526         that now in C++ the value is correct per the C++ standards.
14528 2013-06-19  Richard Biener  <rguenther@suse.de>
14530         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14531         for global context.
14533 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14535         Revert:
14536         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14538         PR target/57609
14539         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14540         with next_active_insn.
14542 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14544         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14545         functions are inlined during failures to flag an error.
14546         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14547         in early inline pass.
14549 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14551         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14552         in comments.
14554 2013-06-18  Julian Brown  <julian@codesourcery.com>
14556         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14557         Permit virtual register pre-reload if !strict.
14558         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14559         change.
14560         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14561         prototype.
14562         * config/arm/neon.md (movmisalign<mode>): Use
14563         neon_perm_struct_or_reg_operand instead of
14564         neon_struct_or_register_operand.
14565         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14566         neon_permissive_struct_operand instead of neon_struct_operand.
14567         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14568         neon_vector_mem_operand.
14569         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14570         neon_vector_mem_operand.
14571         (neon_permissive_struct_operand): New.
14572         (neon_struct_or_register_operand): Rename to...
14573         (neon_perm_struct_or_reg_operand): This. Adjust call to
14574         neon_vector_mem_operand.
14576 2013-06-18  Richard Biener  <rguenther@suse.de>
14578         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14579         * lto-streamer.h: Include pointer-set.h.
14580         (struct lto_decl_slot): Remove.
14581         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14582         Remove next_index entry.
14583         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14584         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14585         (lto_init_tree_ref_encoder): Adjust.
14586         (lto_destroy_tree_ref_encoder): Likewise.
14587         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14588         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14589         (lto_output_decl_index): Adjust.
14590         (lto_new_out_decl_state): Likewise.
14591         (lto_record_function_out_decl_state): Likewise.
14592         * lto-streamer-out.c (copy_function): Likewise.
14594 2013-06-18  Richard Biener  <rguenther@suse.de>
14596         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14597         * cgraphunit.c: Include cfgloop.h.
14598         (init_lowered_empty_function): Initialize the loop tree.
14599         (assemble_thunk): Insert new BBs into loops.
14601 2013-06-18  Richard Biener  <rguenther@suse.de>
14603         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14604         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14605         the map from cache entry to cache index optional.
14606         (streamer_tree_cache_replace_tree): Adjust accordingly.
14607         (streamer_tree_cache_append): Likewise.
14608         (streamer_tree_cache_delete): Likewise.
14609         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14610         streamer cache map from cache entry to cache index.
14611         * lto-streamer-out.c (create_output_block): Adjust.
14613 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14615         * config/arm/arm.md (attribute "insn"): Move multiplication and
14616         division attributes to...
14617         (attribute "type"): ... here.  Remove mult.
14618         (attribute "mul32"): New attribute.
14619         (attribute "mul64"): Add umaal.
14620         (*arm_mulsi3): Update attributes.
14621         (*arm_mulsi3_v6): Likewise.
14622         (*thumb_mulsi3): Likewise.
14623         (*thumb_mulsi3_v6): Likewise.
14624         (*mulsi3_compare0): Likewise.
14625         (*mulsi3_compare0_v6): Likewise.
14626         (*mulsi_compare0_scratch): Likewise.
14627         (*mulsi_compare0_scratch_v6): Likewise.
14628         (*mulsi3addsi): Likewise.
14629         (*mulsi3addsi_v6): Likewise.
14630         (*mulsi3addsi_compare0): Likewise.
14631         (*mulsi3addsi_compare0_v6): Likewise.
14632         (*mulsi3addsi_compare0_scratch): Likewise.
14633         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14634         (*mulsi3subsi): Likewise.
14635         (*mulsidi3adddi): Likewise.
14636         (*mulsi3addsi_v6): Likewise.
14637         (*mulsidi3adddi_v6): Likewise.
14638         (*mulsidi3_nov6): Likewise.
14639         (*mulsidi3_v6): Likewise.
14640         (*umulsidi3_nov6): Likewise.
14641         (*umulsidi3_v6): Likewise.
14642         (*umulsidi3adddi): Likewise.
14643         (*umulsidi3adddi_v6): Likewise.
14644         (*smulsi3_highpart_nov6): Likewise.
14645         (*smulsi3_highpart_v6): Likewise.
14646         (*umulsi3_highpart_nov6): Likewise.
14647         (*umulsi3_highpart_v6): Likewise.
14648         (mulhisi3): Likewise.
14649         (*mulhisi3tb): Likewise.
14650         (*mulhisi3bt): Likewise.
14651         (*mulhisi3tt): Likewise.
14652         (maddhisi4): Likewise.
14653         (*maddhisi4tb): Likewise.
14654         (*maddhisi4tt): Likewise.
14655         (maddhidi4): Likewise.
14656         (*maddhidi4tb): Likewise.
14657         (*maddhidi4tt): Likewise.
14658         (divsi3): Likewise.
14659         (udivsi3): Likewise.
14660         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14661         (thumb2_mulsi_short_compare0): Likewise.
14662         (thumb2_mulsi_short_compare0_scratch): Likewise.
14663         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14664         (1020mult2): Likewise.
14665         (1020mult3): Likewise.
14666         (1020mult4): Likewise.
14667         (1020mult5): Likewise.
14668         (1020mult6): Likewise.
14669         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14670         change.
14671         (cortex_a15_mult64): Likewise.
14672         (cortex_a15_sdiv): Likewise.
14673         (cortex_a15_udiv): Likewise.
14674         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14675         (mult2): Likewise.
14676         (mult3): Likewise.
14677         (mult4): Likewise.
14678         (mult5): Likewise.
14679         (mult6): Likewise.
14680         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14681         (pj4_ir_div): Likewise.
14682         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14683         (11_mult2): Likewise.
14684         (11_mult3): Likewise.
14685         (11_mult4): Likewise.
14686         (11_mult5): Likewise.
14687         (11_mult6): Likewise.
14688         (11_mult7): Likewise.
14689         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14690         (cortex_a8_mla): Likewise.
14691         (cortex_a8_mull): Likewise.
14692         (cortex_a8_smulwy): Likewise.
14693         (cortex_a8_smlald): Likewise.
14694         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14695         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14696         (cortex_r4_mul_3): Likewise.
14697         (cortex_r4_mla_4): Likewise.
14698         (cortex_r4_mla_3): Likewise.
14699         (cortex_r4_smlald): Likewise.
14700         (cortex_r4_mull): Likewise.
14701         (cortex_r4_sdiv): Likewise.
14702         (cortex_r4_udiv): Likewise.
14703         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14704         (cortex_a7_idiv): Likewise.
14705         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14706         (9_mult2): Likewise.
14707         (9_mult3): Likewise.
14708         (9_mult4): Likewise.
14709         (9_mult5): Likewise.
14710         (9_mult6): Likewise.
14711         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14712         (cortex_a53_sdiv): Likewise.
14713         (cortex_a53_udiv): Likewise.
14714         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14715         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14716         (mp626_mult2): Likewise.
14717         (mp626_mult3): Likewise.
14718         (mp626_mult4): Likewise.
14719         * config/arm/fa526.md (526_mult1): Update attribute change.
14720         (526_mult2): Likewise.
14721         * config/arm/arm-generic.md (mult): Update attribute change.
14722         (mult_ldsched_strongarm): Likewise.
14723         (mult_ldsched): Likewise.
14724         (multi_cycle): Likewise.
14725         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14726         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14727         (606te_mult2): Likewise.
14728         (606te_mult3): Likewise.
14729         (606te_mult4): Likewise.
14730         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14731         (cortex_a9_mac16): Likewise.
14732         (cortex_a9_multiply): Likewise.
14733         (cortex_a9_mac): Likewise.
14734         (cortex_a9_multiply_long): Likewise.
14735         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14736         (626te_mult2): Likewise.
14737         (626te_mult3): Likewise.
14738         (626te_mult4): Likewise.
14740 2013-06-18  Richard Biener  <rguenther@suse.de>
14742         PR lto/57334
14743         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14745 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14747         PR target/57609
14748         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14749         with next_active_insn.
14751 2013-06-18  Alan Modra  <amodra@gmail.com>
14753         * config/rs6000/rs6000.h (enum data_align): New.
14754         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14755         (DATA_ABI_ALIGNMENT): Define.
14756         (CONSTANT_ALIGNMENT): Correct comment.
14757         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14758         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14760 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14762         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14763         ATTRIBUTE_UNUSED marking.
14765 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14767         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14768         alternative and update.
14769         (aarch64_dup_lanedi): Delete.
14770         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14771         * config/aarch64/aarch64-simd-builtins.def: Update.
14773 2013-06-17  Richard Biener  <rguenther@suse.de>
14775         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14776         (lto_input_scc): Declare.
14777         (lto_input_tree_1): Likewise.
14778         (struct lto_stats_d): Add num_tree_bodies_output and
14779         num_pickle_refs_output.
14780         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14781         (lto_read_tree_1): Split out from ...
14782         (lto_read_tree): ... this.
14783         (lto_input_scc): New function.
14784         (lto_input_tree_1): Split out from ...
14785         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14786         (lto_data_in_create): Create the streamer cache without hashes.
14787         * lto-streamer-out.c (create_output_block): Create the streamer
14788         cache with hashes when not doing WPA.
14789         (lto_write_tree_1): Split out from ...
14790         (lto_write_tree): ... this.
14791         (get_symbol_initial_value): New function.
14792         (lto_output_tree_1): Split out from ...
14793         (lto_output_tree): ... this.  Write trees as series of SCCs
14794         using a DFS walk via DFS_write_tree.
14795         (struct sccs, struct scc_entry): New types.
14796         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14797         (DFS_write_tree_body): New function.
14798         (DFS_write_tree): Likewise.
14799         (hash_tree): Likewise.
14800         (scc_entry_compare): Likewise.
14801         (hash_scc): Likewise.
14802         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14803         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14804         TREE_CHAIN as regular reference.
14805         (streamer_read_integer_cst): Remove.
14806         (streamer_get_pickled_tree): Adjust.
14807         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14808         of DECL_EXTERNALs in BLOCK_VARS for now.
14809         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14810         reference.
14811         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14812         Add hash value argument and record that if hashes are recorded
14813         in the cache.
14814         (streamer_tree_cache_insert_1): Adjust.
14815         (streamer_tree_cache_insert): Likewise.
14816         (streamer_tree_cache_insert_at): Rename to ...
14817         (streamer_tree_cache_replace_tree): ... this and adjust.
14818         (streamer_tree_cache_append): Adjust.
14819         (record_common_node): Likewise.
14820         (streamer_tree_cache_create): Add argument whether to
14821         record hash values together with trees.
14822         (streamer_tree_cache_delete): Adjust.
14823         * tree-streamer.h (struct streamer_tree_cache_d): Add
14824         vector of hashes.
14825         (streamer_read_integer_cst): Remove.
14826         (streamer_tree_cache_insert): Adjust.
14827         (streamer_tree_cache_append): Likewise.
14828         (streamer_tree_cache_insert_at): Rename to ...
14829         (streamer_tree_cache_replace_tree): ... this and adjust.
14830         (streamer_tree_cache_create): Add argument whether to record hashes.
14831         (streamer_tree_cache_get): Rename to ...
14832         (streamer_tree_cache_get_tree): ... this.
14833         (streamer_tree_cache_get_hash): New function.
14834         * tree.c (cache_integer_cst): New function.
14835         * tree.h (cache_integer_cst): Declare.
14836         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14837         * lto-symtab.c (lto_varpool_replace_node): Only release
14838         DECL_INITIAL of non-prevailing decls.
14839         * varpool.c (varpool_remove_initializer): Do not release
14840         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14842 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14844         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14845         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14846         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14847         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14848         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14849         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14850         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14851         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14852         instead of TARGET_64BIT.
14853         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14854         Require ISA_HAS_<D>DIV.
14856 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14858         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14859         (mips*-*-linux*): Move default with_llsc setting to where other
14860         defaults are set.
14861         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14862         with_arch block.
14863         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14864         Likewise.  Remove default with_tune setting.  Move default float
14865         setting to its own block.  Handle with_llsc in the same block as above.
14867 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14869         PR rtl-optimization/57425
14870         PR rtl-optimization/57569
14871         * alias.c (write_dependence_p): Add new parameters mem_mode,
14872         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14873         Changed all callers.
14874         (canon_anti_dependence): New function.
14875         * cse.c (check_dependence): Use canon_anti_dependence.
14876         * cselib.c (cselib_invalidate_mem): Likewise.
14877         * rtl.h (canon_anti_dependence): Declare.
14879 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14881         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14882         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14883         ".set mips3" for 64-bit targets.
14885 2013-06-15  Dehao Chen  <dehao@google.com>
14887         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14888         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14889         (gimple_check_call_args): Likewise.
14890         * value-prof.c (check_ic_target): Likewise.
14891         * ipa-inline.c (early_inliner): Likewise.
14892         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14893         * cgraph.c (cgraph_create_edge_1): Likewise.
14894         (cgraph_make_edge_direct): Likewise.
14896 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14898         PR target/57615
14899         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14900         rs6000_output_move_128bit to handle emitting quad memory
14901         operations.  Set attribute length to 8 bytes.
14903 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14905         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14906         New pattern.
14907         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14908         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14909         (aarch64_<su>mlsl<mode>): Likewise.
14911 2013-06-14  Mike Stump  <mikestump@comcast.net>
14913         * Makefile.in (TARGET_H): Add insn-codes.h.
14915 2013-06-14  Alan Modra  <amodra@gmail.com>
14917         PR middle-end/57134
14918         PR middle-end/57586
14919         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14920         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14921         bitfield expansion when EXPAND_MEMORY.
14922         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14924 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14926         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14927         test for clearing quad memory on 32-bit later.
14929 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14931         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14932         (fold_negate_expr): Likewise.
14933         (fold_real_zero_addition_p): Handle vectors.
14934         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14936 2013-06-14  Alan Modra  <amodra@gmail.com>
14938         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14940 2013-06-13  Jan Hubicka  <jh@suse.cz>
14942         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14943         Local comdats are not externally visible.
14944         * symtab.c (dump_symtab_base): Dump externally visible.
14945         (verify_symtab_base): Verify back links in the symtab hash.
14947 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14949         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14950         CONVERT_EXPR as equal nodes.
14952 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14954         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14956 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14958         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14959         Generalize to complex and vector.
14960         * tree.c (build_all_ones_cst): New function.
14961         * tree.h (build_all_ones_cst): Declare it.
14963 2013-06-13  Alan Modra  <amodra@gmail.com>
14965         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14966         * config/rs6000/rs6000.md (signbittf2): New insn.
14967         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14968         (abstf2_internal, cmptf_internal2): Likewise.
14969         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14971 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14972             Pat Haugen  <pthaugen@us.ibm.com>
14973             Peter Bergner  <bergner@vnet.ibm.com>
14975         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14976         power8 byte, half-word, and quad-word atomic instructions.
14977         (emit_store_conditional): Likewise.
14978         (rs6000_expand_atomic_compare_and_swap): Likewise.
14979         (rs6000_expand_atomic_op): Likewise.
14981         * config/rs6000/sync.md (larx): Add new modes for power8.
14982         (stcx): Likewise.
14983         (AINT): New mode iterator to include TImode as well as normal
14984         integer modes on power8.
14985         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14986         that VSX registers are not considered.  Use AINT mode iterator
14987         instead of INT1 to allow inclusion of quad word atomic operations
14988         on power8.
14989         (load_locked<mode>): Likewise.
14990         (store_conditional<mode>): Likewise.
14991         (atomic_compare_and_swap<mode>): Likewise.
14992         (atomic_exchange<mode>): Likewise.
14993         (atomic_nand<mode>): Likewise.
14994         (atomic_fetch_<fetchop_name><mode>): Likewise.
14995         (atomic_nand_fetch<mode>): Likewise.
14996         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14997         each type.
14998         (ATOMIC): On power8, add QImode, HImode modes.
14999         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
15000         modes that promote to SImode.
15001         (load_lockedti): Convert TImode arguments to PTImode, so that we
15002         get a guaranteed even/odd register pair.
15003         (load_lockedpti): Likewise.
15004         (store_conditionalti): Likewise.
15005         (store_conditionalpti): Likewise.
15007         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
15008         atomic load/store instructions.
15009         (HSI): Likewise.
15011 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
15013         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
15014         loads.
15015         (insn_count): New attribute, with most cases extracted from...
15016         (length): ...here.  Redefine most cases in terms of insn_count.
15017         (single_insn): Delete.
15018         (can_delay): Use insn_count to check for single instructions.
15019         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
15020         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
15021         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
15022         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
15023         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
15024         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
15025         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
15026         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
15027         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
15028         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
15029         rather than "length".
15030         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
15031         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
15032         Use "insn_count" rather than "length".
15033         * config/mips/mips-dsp.md
15034         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
15035         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
15036         length attributes.
15038 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
15040         PR tree-optimization/57361
15041         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
15043 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
15045         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
15046         to split.
15047         (aarch64_simd_combine<mode>): New instruction expansion.
15048         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
15049         function prototype.
15050         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
15051         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
15053 2013-06-12  Jan Hubicka  <jh@suse.cz>
15055         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
15056         decl has when in streaming stage.
15057         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
15058         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
15060 2013-06-12  Roland Stigge  <stigge@antcom.de>
15062         PR target/57578
15063         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
15065 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
15067         PR tree-optimization/57537
15068         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
15069         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
15071 2013-06-12  Richard Biener  <rguenther@suse.de>
15073         * data-streamer.h (streamer_write_char_stream): CSE
15074         obs->current_pointer.
15075         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
15076         streamer_write_char_stream manually and optimize the resulting loop.
15077         (streamer_write_hwi_stream): Likewise.
15079 2013-06-12  Jan Hubicka  <jh@suse.cz>
15081         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
15082         * cgraph.h (varpool_create_empty_node): Declare.
15083         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
15084         duplicated nodes.
15085         * symtab.c (symtab_unregister_node): Be lax about missin entries
15086         in node hash.
15087         (symtab_get_node): Update comment.
15088         * varpool.c (varpool_create_empty_node): Break out from ...
15089         (varpool_node_for_decl): ... here.
15090         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
15092 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
15094         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
15095         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
15096         part.  Use straight-line flow at the end.
15097         <COMPONENT_REF>: Remove superfluous else.
15098         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
15100 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
15102         PR target/56564
15103         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
15104         target hook even for !TREE_PUBLIC decls.  If no resolution info
15105         is available, return false for common and external decls.
15107 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15109         * config/rl78/constraints.md (U): New constraint.
15110         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
15111         valloc attribute.
15113 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
15115         PR target/57589
15116         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
15117         to allow returning address to AT_PLATFORM name.
15119 2013-06-11  Jan Hubicka  <jh@suse.cz>
15121         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
15122         * cgraph.h (symtab_node_base): Add weakref flag.
15123         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
15124         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
15125         (output_weakrefs): Use weakref flag.
15126         * fold-const.c (simple_operand_p): Handle WEAK.
15127         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
15128         * ipa.c (varpool_externally_visible_p): Drop weakref.
15129         (function_and_variable_visibility): Update comment; fix weakref
15130         sanity checks; do not clear DECL_WEAK on them.
15131         * lto-cgraph.c (lto_output_node): update.
15132         (lto_output_varpool_node): Update.
15133         (input_overwrite_node): Update.
15134         (input_node): Update.
15135         (input_varpool_node): Update.
15136         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
15137         (lto_symtab_merge_symbols): Add sanity check.
15138         (lto_symtab_prevailing_decl): Do not special case weakrefs.
15139         * passes.c (rest_of_decl_compilation): Set static flag, too.
15140         * symtab.c (dump_symtab_base): Dump weakref.
15141         (verify_symtab_base): Sanity check weakrefs.
15142         (symtab_make_decl_local): Remove duplicated code.
15143         (symtab_alias_ultimate_target): Simplify.
15144         * varpool.c (varpool_create_variable_alias): Set weakref flag.
15146 2013-06-11  Tom de Vries  <tom@codesourcery.com>
15148         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
15149         sequence_vect == NULL.
15151 2013-06-11  DJ Delorie  <dj@redhat.com>
15153         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
15154         (rl78_unwind_word_mode): New.
15156 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
15158         * final.c (debug_prefix_maps): Make static.
15160 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
15162         * function.c (initial_trampoline): Remove stray copy.
15164 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
15166         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
15168 2013-06-11  Martin Jambor  <mjambor@suse.cz>
15170         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
15171         within bounds at the beginning of the function.
15173 2013-06-11  Alan Modra  <amodra@gmail.com>
15175         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
15176         reporting.
15177         (get_named_section): Don't NULL !DECL_P decl.
15179 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
15181         * doc/invoke.texi (core-avx2): Document.
15182         (slm): Likewise.
15183         (atom): Updated with MOVBE.
15185 2013-06-11  Richard Biener  <rguenther@suse.de>
15187         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
15189 2013-06-11  Anton Blanchard  <anton@samba.org>
15191         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
15192         correct shift value in little-endian mode.
15194 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
15196         PR target/56564
15197         * varasm.c (get_variable_align): Move #endif to the right place.
15199 2013-06-10  Cary Coutant  <ccoutant@google.com>
15201         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
15202         for hash so that hash table traversal order is deterministic.
15204 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
15205             Pat Haugen  <pthaugen@us.ibm.com>
15206             Peter Bergner  <bergner@vnet.ibm.com>
15208         * config/rs6000/vector.md (GPR move splitter): Do not split moves
15209         of vectors in GPRS if they are direct moves or quad word load or
15210         store moves.
15212         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
15213         declaration.
15214         (direct_move_p): Likewise.
15215         (quad_load_store_p): Likewise.
15217         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
15218         classes into bins based on the physical register type.
15219         (reg_class_to_reg_type): Likewise.
15220         (IS_STD_REG_TYPE): Likewise.
15221         (IS_FP_VECT_REG_TYPE): Likewise.
15222         (reload_fpr_gpr): Arrays to determine what insn to use if we can
15223         use direct move instructions.
15224         (reload_gpr_vsx): Likewise.
15225         (reload_vsx_gpr): Likewise.
15226         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
15227         information that is a simplification of register classes.  Also
15228         precalculate direct move reload helpers.
15229         (direct_move_p): New function to return true if the operation can
15230         be done as a direct move instruciton.
15231         (quad_load_store_p): New function to return true if the operation
15232         is a quad memory operation.
15233         (rs6000_legitimize_address): If quad memory, only allow register
15234         indirect for TImode addresses.
15235         (rs6000_legitimate_address_p): Likewise.
15236         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
15237         (rs6000_reload_register_type): Likewise.
15238         (register_to_reg_type): Return register type.
15239         (rs6000_secondary_reload_simple_move): New helper function for
15240         secondary reload and secondary memory needed to identify anything
15241         that is a simple move, and does not need reloading.
15242         (rs6000_secondary_reload_direct_move): New helper function for
15243         secondary reload to identify cases that can be done with several
15244         instructions via the direct move instructions.
15245         (rs6000_secondary_reload_move): New helper function for secondary
15246         reload to identify moves between register types that can be done.
15247         (rs6000_secondary_reload): Add support for quad memory operations
15248         and for direct move.
15249         (rs6000_secondary_memory_needed): Likewise.
15250         (rs6000_debug_secondary_memory_needed): Change argument names.
15251         (rs6000_output_move_128bit): New function to return the move to
15252         use for 128-bit moves, including knowing about the various
15253         limitations of quad memory operations.
15255         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
15256         memory operations.  call rs6000_output_move_128bit for the actual
15257         instruciton(s) to generate.
15258         (vsx_movti_64bit): Likewise.
15260         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
15261         (UNSPEC_P8V_MTVSRWZ): Likewise.
15262         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
15263         (UNSPEC_P8V_MTVSRD): Likewise.
15264         (UNSPEC_P8V_XXPERMDI): Likewise.
15265         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
15266         (UNSPEC_FUSION_GPR): Likewise.
15267         (FMOVE128_GPR): New iterator for direct move.
15268         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
15269         (f32_sv): Likewise.
15270         (f32_dm): Likewise.
15271         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
15272         loads and direct move instructions.
15273         (zero_extendsidi2_lfiwzx): Likewise.
15274         (extendsidi2_lfiwax): Likewise.
15275         (extendsidi2_nocell): Likewise.
15276         (floatsi<mode>2_lfiwax): Likewise.
15277         (lfiwax): Likewise.
15278         (floatunssi<mode>2_lfiwzx): Likewise.
15279         (lfiwzx): Likewise.
15280         (fix_trunc<mode>_stfiwx): Likewise.
15281         (fixuns_trunc<mode>_stfiwx): Likewise.
15282         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
15283         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
15284         (parity<mode>2_cmpb): Set length/type attr.
15285         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
15286         for 'mr.' to fast_compare.
15287         (bpermd_<mode>): Change type attr to popcnt.
15288         (p8_fmrgow_<mode>): New insns for power8 direct move support.
15289         (p8_mtvsrwz_1): Likewise.
15290         (p8_mtvsrwz_2): Likewise.
15291         (reload_fpr_from_gpr<mode>): Likewise.
15292         (p8_mtvsrd_1): Likewise.
15293         (p8_mtvsrd_2): Likewise.
15294         (p8_xxpermdi_<mode>): Likewise.
15295         (reload_vsx_from_gpr<mode>): Likewise.
15296         (reload_vsx_from_gprsf): Likewise.
15297         (p8_mfvsrd_3_<mode>): LIkewise.
15298         (reload_gpr_from_vsx<mode>): Likewise.
15299         (reload_gpr_from_vsxsf): Likewise.
15300         (p8_mfvsrd_4_disf): Likewise.
15301         (multi-word GPR splits): Do not split direct moves or quad memory
15302         operations.
15304 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
15306         * tree-into-ssa.c (interesting_blocks): Make static.
15308 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
15310         PR target/56564
15311         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
15312         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
15313         Use DATA_ABI_ALIGNMENT for that case instead if defined.
15314         (get_variable_align): New function.
15315         (get_variable_section, emit_bss, emit_common,
15316         assemble_variable_contents, place_block_symbol): Use
15317         get_variable_align instead of DECL_ALIGN.
15318         (assemble_noswitch_variable): Add align argument, use it
15319         instead of DECL_ALIGN.
15320         (assemble_variable): Adjust caller.  Use get_variable_align
15321         instead of DECL_ALIGN.
15322         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
15323         caller.
15324         (DATA_ABI_ALIGNMENT): Define.
15325         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
15326         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
15327         opt is false, only return the psABI mandated alignment increase.
15328         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
15329         (DATA_ABI_ALIGNMENT): ... this.
15330         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
15331         (DATA_ABI_ALIGNMENT): ... this.
15332         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
15333         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
15334         (DATA_ABI_ALIGNMENT): ... this.
15335         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
15336         * doc/tm.texi: Regenerated.
15338 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
15340         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
15341         cmp_code to construct REG_EQUAL note.
15343 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
15345         PR target/57568
15346         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
15347         that operands[2] doesn't overlap with operands[0].
15349 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
15350             Jan Hubicka  <jh@suse.cz>
15352         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
15353         hack to mark symbols as used.
15355 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
15357         PR rtl-optimization/57559
15358         * lra-constraints.c (process_alt_operands): Don't discourage
15359         memory with known offset for offsetable memory constraint.
15360         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
15362 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
15364         * varasm.c (struct oc_local_state): Reorder fields.
15365         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
15366         and adjust accordingly.
15367         (output_constructor): Reorder initialization code and adjust call to
15368         output_constructor_bitfield.
15370 2013-06-07  Jan Hubicka  <jh@suse.cz>
15372         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
15374 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
15376         * tree-object-size.c (unknown): Make const.
15378 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15380         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
15381         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
15382         for last alternative in the cpu_facility attribute.
15384 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15386         PR target/56315
15387         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
15388         (xordi3): Change operand 2 constraint to arm_xordi_operand.
15389         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
15390         * config/arm/constraints.md (Dg): New constraint.
15391         * config/arm/neon.md (xordi3_neon): Remove.
15392         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
15393         * config/arm/predicates.md (arm_xordi_operand): New predicate.
15395 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15397         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
15398         Clean up alternatives.
15400 2013-06-07  Alan Modra  <amodra@gmail.com>
15402         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
15403         va_list_gpr_size.
15405 2013-06-07  Alan Modra  <amodra@gmail.com>
15407         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
15409 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15411         * config/arm/constraints.md (Df): New constraint.
15412         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
15413         Correct length attribute for last two alternatives.
15415 2013-06-07  Alan Modra  <amodra@gmail.com>
15417         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
15418         override user -mfp-in-toc.
15419         (offsettable_ok_by_alignment): Consider just the current access
15420         rather than the whole object, unless BLKmode.  Handle
15421         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
15422         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
15423         for -mcmodel=medium.
15424         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
15425         override user -mfp-in-toc or -msum-in-toc.  Default to
15426         -mno-fp-in-toc for -mcmodel=medium.
15428 2013-06-06  DJ Delorie  <dj@redhat.com>
15430         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
15431         TARGET_VALID_POINTER_MODE.
15433 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
15434             Pat Haugen  <pthaugen@us.ibm.com>
15435             Peter Bergner  <bergner@vnet.ibm.com>
15437         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
15438         Document new power8 builtins.
15440         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
15441         condition code register, to allow 128-bit logical operations to be
15442         done in the VSX or GPR registers.
15443         (nor<mode>3): Use the canonical form for nor.
15444         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
15445         vclz*, and vpopcnt* vector instructions.
15446         (nand<mode>3): Likewise.
15447         (orc<mode>3): Likewise.
15448         (clz<mode>2): LIkewise.
15449         (popcount<mode>2): Likewise.
15451         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
15452         that only the GPRs are recognized.
15454         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15455         support for new power8 builtins.
15457         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
15458         builtin functions.
15459         (xscvdpspn): Likewise.
15460         (vclz): Likewise.
15461         (vclzb): Likewise.
15462         (vclzh): Likewise.
15463         (vclzw): Likewise.
15464         (vclzd): Likewise.
15465         (vpopcnt): Likewise.
15466         (vpopcntb): Likewise.
15467         (vpopcnth): Likewise.
15468         (vpopcntw): Likewise.
15469         (vpopcntd): Likewise.
15470         (vgbbd): Likewise.
15471         (vmrgew): Likewise.
15472         (vmrgow): Likewise.
15473         (eqv): Likewise.
15474         (eqv_v16qi3): Likewise.
15475         (eqv_v8hi3): Likewise.
15476         (eqv_v4si3): Likewise.
15477         (eqv_v2di3): Likewise.
15478         (eqv_v4sf3): Likewise.
15479         (eqv_v2df3): Likewise.
15480         (nand): Likewise.
15481         (nand_v16qi3): Likewise.
15482         (nand_v8hi3): Likewise.
15483         (nand_v4si3): Likewise.
15484         (nand_v2di3): Likewise.
15485         (nand_v4sf3): Likewise.
15486         (nand_v2df3): Likewise.
15487         (orc): Likewise.
15488         (orc_v16qi3): Likewise.
15489         (orc_v8hi3): Likewise.
15490         (orc_v4si3): Likewise.
15491         (orc_v2di3): Likewise.
15492         (orc_v4sf3): Likewise.
15493         (orc_v2df3): Likewise.
15495         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15496         allow power8 quad mode in 64-bit.
15497         (rs6000_builtin_vectorized_function): Add support to vectorize
15498         ISA 2.07 count leading zeros, population count builtins.
15499         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15500         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15501         (builtin_function_type): Add vgbbd builtin function which takes an
15502         unsigned argument.
15503         (altivec_expand_vec_perm_const): Add support for new power8 merge
15504         instructions.
15506         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15507         that does not include TImdoe for use with 32-bit.
15508         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15509         instructions.
15510         (UNSPEC_VSX_CVDPSPN): Likewise.
15511         (vsx_xscvdpspn): Likewise.
15512         (vsx_xscvspdpn): Likewise.
15513         (vsx_xscvdpspn_scalar): Likewise.
15514         (vsx_xscvspdpn_directmove): Likewise.
15515         (vsx_and<mode>3): Split logical operations into 32-bit and
15516         64-bit. Add support to do logical operations on TImode as well as
15517         VSX vector types.  Allow logical operations to be done in either
15518         VSX registers or in general purpose registers in 64-bit mode.  Add
15519         splitters if GPRs were used. For AND, add clobber of CCmode to
15520         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15521         encoding.
15522         (vsx_and<mode>3_32bit): Likewise.
15523         (vsx_and<mode>3_64bit): Likewise.
15524         (vsx_ior<mode>3): Likewise.
15525         (vsx_ior<mode>3_32bit): Likewise.
15526         (vsx_ior<mode>3_64bit): Likewise.
15527         (vsx_xor<mode>3): Likewise.
15528         (vsx_xor<mode>3_32bit): Likewise.
15529         (vsx_xor<mode>3_64bit): Likewise.
15530         (vsx_one_cmpl<mode>2): Likewise.
15531         (vsx_one_cmpl<mode>2_32bit): Likewise.
15532         (vsx_one_cmpl<mode>2_64bit): Likewise.
15533         (vsx_nor<mode>3): Likewise.
15534         (vsx_nor<mode>3_32bit): Likewise.
15535         (vsx_nor<mode>3_64bit): Likewise.
15536         (vsx_andc<mode>3): Likewise.
15537         (vsx_andc<mode>3_32bit): Likewise.
15538         (vsx_andc<mode>3_64bit): Likewise.
15539         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15540         and xxlorc instructions.
15541         (vsx_eqv<mode>3_64bit): Likewise.
15542         (vsx_nand<mode>3_32bit): Likewise.
15543         (vsx_nand<mode>3_64bit): Likewise.
15544         (vsx_orc<mode>3_32bit): Likewise.
15545         (vsx_orc<mode>3_64bit): Likewise.
15547         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15549         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15550         instruction.
15551         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15552         (p8_vmrgow): Likewise.
15553         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15554         GPRs to be split under VSX.
15555         (p8v_clz<mode>2): Add power8 count leading zero support.
15556         (p8v_popcount<mode>2): Add power8 population count support.
15557         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15558         support.
15560         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15561         instruction.
15563         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15564         builtin functions.
15565         (vec_nand): Likewise.
15566         (vec_vclz): Likewise.
15567         (vec_vclzb): Likewise.
15568         (vec_vclzd): Likewise.
15569         (vec_vclzh): Likewise.
15570         (vec_vclzw): Likewise.
15571         (vec_vgbbd): Likewise.
15572         (vec_vmrgew): Likewise.
15573         (vec_vmrgow): Likewise.
15574         (vec_vpopcnt): Likewise.
15575         (vec_vpopcntb): Likewise.
15576         (vec_vpopcntd): Likewise.
15577         (vec_vpopcnth): Likewise.
15578         (vec_vpopcntw): Likewise.
15580 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15582         PR rtl-optimization/57468
15583         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15584         spilled pseudos.
15586 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15588         PR rtl-optimization/57459
15589         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15590         type when setting live regs.
15592 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15594         * config/s390/s390.opt (mlra): New option.
15595         * config/s390/s390.c (s390_decompose_address): Check displacement
15596         for all registers for LRA.
15597         (s390_secondary_reload): Don't used secondary reloads for LRA.
15598         (s390_lra_p): New function.
15599         (TARGET_LRA_P): Define.
15600         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15601         of attribute cpu_facility to zarch for the last alternative.
15602         (*cmpmem_short): Ditto.
15604 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15606         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15607         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15608         (arm_expand_prologue): Likewise.
15610 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15612         PR c++/53743
15613         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15614         as this is now done by redirect_edge_and_branch_force.
15615         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15616         barriers, and fix interaction with splitting.
15617         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15618         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15619         reflect changes made in the routine.
15620         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15621         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15622         since this is called in cfglayout mode, and replace partition fixup
15623         with assert as that is now done by force_nonfallthru_and_redirect.
15624         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15625         already be marked with region crossing note.
15626         (insert_section_boundary_note): Make non-static, gate on flag
15627         has_bb_partition, rewrite to also check for multiple partitions.
15628         (rest_of_handle_reorder_blocks): Remove call to
15629         insert_section_boundary_note, now done later during free_cfg.
15630         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15631         * bb-reorder.h (insert_section_boundary_note): Declare.
15632         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15633         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15634         invoke insert_section_boundary_note.
15635         (try_redirect_by_replacing_jump): Remove unnecessary
15636         check for region crossing note.
15637         (fixup_partition_crossing): New function.
15638         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15639         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15640         in non-cfglayout mode.
15641         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15642         remove old code that tried to do this. Emit barrier correctly
15643         when we are in cfglayout mode.
15644         (last_bb_in_partition): New function.
15645         (rtl_split_edge): Correctly fixup partition boundaries.
15646         (commit_one_edge_insertion): Remove old code that tried to
15647         fixup region crossing edge since this is now handled in
15648         split_block, and set up insertion point correctly since
15649         block may now end in a jump.
15650         (verify_hot_cold_block_grouping): Guard against checking when not in
15651         linearized RTL mode.
15652         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15653         notes.
15654         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15655         rtl_verify_flow_info, so not called in cfglayout mode.
15656         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15657         (fixup_reorder_chain): Remove old code that attempted to fixup region
15658         crossing note as this is now handled in force_nonfallthru_and_redirect.
15659         (duplicate_insn_chain): Don't duplicate switch section notes.
15660         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15661         note.
15662         * basic-block.h (emit_barrier_after_bb): Declare.
15664 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15666         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15667         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15668         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15670 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15672         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15673         where appropriate.
15674         * config/arm/ldmstm.md: Regenerate.
15676 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15678         * config/arm/sync.md (atomic_loaddi_1):
15679         Disable predication for arm_restrict_it.
15680         (arm_load_exclusive<mode>): Likewise.
15681         (arm_load_exclusivesi): Likewise.
15682         (arm_load_exclusivedi): Likewise.
15683         (arm_load_acquire_exclusive<mode>): Likewise.
15684         (arm_load_acquire_exclusivesi): Likewise.
15685         (arm_load_acquire_exclusivedi): Likewise.
15686         (arm_store_exclusive<mode>): Likewise.
15687         (arm_store_exclusive<mode>): Likewise.
15688         (arm_store_release_exclusivedi): Likewise.
15689         (arm_store_release_exclusive<mode>): Likewise.
15691 2013-06-06  Richard Biener  <rguenther@suse.de>
15693         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15694         after LTO_null.
15695         (lto_tag_is_tree_code_p): Adjust.
15696         (lto_tag_is_gimple_code_p): Likewise.
15697         (lto_gimple_code_to_tag): Likewise.
15698         (lto_tag_to_gimple_code): Likewise.
15699         (lto_tree_code_to_tag): Likewise.
15700         (lto_tag_to_tree_code): Likewise.
15701         * data-streamer.h (streamer_write_hwi_in_range): Use
15702         uhwi streaming to stream the normalized range.
15703         (streamer_read_hwi_in_range): Likewise.
15705 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15707         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15708         (predicable_short_it): Likewise.
15709         (predicated): Likewise.
15710         (enabled): Handle above.
15711         (define_cond_exec): Set predicated attribute to yes.
15713 2013-06-05  Mike Stump  <mikestump@comcast.net>
15715         * gdbinit.in (__FUNCTION__): Add.
15717 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15719         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15720         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15722 2013-06-05  Jan Hubicka  <jh@suse.cz>
15724         * varasm.c (mark_decl_referenced): Revert the removal until targets
15725         are fixed.
15727 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15729         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15730         instead of mark_decl_referenced.
15732 2013-06-05  Jan Hubicka  <jh@suse.cz>
15734         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15735         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15736         and symtab_used_from_object_file_p.
15737         (cgraph_make_node_local_1): Clear forced_by_abi.
15738         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15739         * cgraph.h (symtab_node_base): Add forced_by_abi.
15740         (decide_is_variable_needed): Remove.
15741         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15742         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15743         (decide_is_symbol_needed): ... this one; handle symbols in general;
15744         always analyze virtuals; honnor forced_by_abi.
15745         (cgraph_finalize_function): Update.
15746         (varpool_finalize_decl): Update.
15747         (symbol_defined_and_needed): Remove.
15748         (analyze_functions): Update.
15749         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15750         output_refs, input_overwrite_node): Handle forced_by_abi.
15751         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15752         (address_taken_from_non_vtable_p): ... this one.
15753         (comdat_can_be_unshared_p_1): New function.
15754         (cgraph_comdat_can_be_unshared_p): Rename to ...
15755         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15756         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15757         (function_and_variable_visibility): Clear forced_by_abi as needed.
15758         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15759         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15760         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15761         * varpool.c (decide_is_variable_needed): Remove.
15763 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15765         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15766         (arm_option_override): Override arm_restrict_it where appropriate.
15767         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15768         * config/arm/arm.opt (mrestrict-it): New command-line option.
15769         * doc/invoke.texi: Document -mrestrict-it.
15771 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15773         * tsan.c (tsan_atomic_table): Make const.
15775 2013-06-05  Richard Biener  <rguenther@suse.de>
15777         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15778         index associated with the tree we are supposed to replace.
15779         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15780         TREE_ASM_WRITTEN as zero for everything but SSA names.
15782 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15784         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15786 2013-06-04  Jan Hubicka  <jh@suse.cz>
15788         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15789         (input_node, input_varpool_node): Handle correctly external same
15790         body aliases.
15791         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15792         nodes at ltrans stage.
15794 2013-06-04  Jan Hubicka  <jh@suse.cz>
15796         * ipa-inline.c (update_caller_keys): Fix availability test.
15797         (update_callee_keys): Likewise.
15798         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15799         to follow ELF standard.
15801 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15803         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15804         (mips64r5900el-*-elf*): New configurations.
15805         * config/mips/mips-cpus.def (r5900): New processor.
15806         * config/mips/mips-tables.opt: Regenerate.
15807         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15808         (mips_issue_rate): Handle PROCESSOR_R5900.
15809         (mips_reorg_process_insns): Force reorder mode for the R5900.
15810         * config/mips/mips.h (TARGET_MIPS5900): Define.
15811         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15812         TARGET_MIPS5900.
15813         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15814         TARGET_MIPS5900.
15815         * config/mips/mips.md (processor): Add r5900.
15816         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15818 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15820         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15821         into function to generate MOVI instruction.
15822         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15823         (aarch64_preferred_simd_mode): Turn into wrapper.
15824         (aarch64_output_scalar_simd_mov_immediate): New function.
15825         * config/aarch64/aarch64-protos.h: Add prototype for above.
15827 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15829         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15830         element_char member.
15831         (sizetochar): Return signed char.
15832         (aarch64_simd_valid_immediate): Remove elchar and other
15833         unnecessary variables.
15834         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15835         Calculate element_char as required.
15836         * config/aarch64/aarch64-protos.h: Update and move prototype
15837         for aarch64_output_simd_mov_immediate.
15838         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15839         Update arguments.
15841 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15843         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15844         information completed by aarch64_simd_valid_immediate.
15845         (aarch64_legitimate_constant_p): Update arguments.
15846         (aarch64_simd_valid_immediate): Work with struct rather than many
15847         pointers.
15848         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15849         (aarch64_simd_make_constant): Update arguments.
15850         (aarch64_output_simd_mov_immediate): Work with struct rather than
15851         many pointers.  Output immediate directly rather than as operand.
15852         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15853         Update prototype.
15854         * config/aarch64/constraints.md (Dn): Update arguments.
15856 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15858         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15859         longer static.
15860         (aarch64_simd_immediate_valid_for_move): Remove.
15861         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15862         (aarch64_simd_make_constant): Update call.
15863         (aarch64_output_simd_mov_immediate): Update call.
15864         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15865         Add prototype.
15866         * config/aarch64/constraints.md (Dn): Update call.
15868 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15870         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15871         return type to bool for prototype.
15872         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15873         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15874         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15876 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15878         * config/mips/mips.opt (meva): New.
15879         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15880         (ASM_SPEC): Handle -meva.
15881         * doc/invoke.texi (meva):  Document.
15883 2013-06-04  Alan Modra  <amodra@gmail.com>
15885         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15886         constant output.
15888 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15890         * rtl.def: Add extra fourth optional field to define_cond_exec.
15891         * gensupport.c (process_one_cond_exec): Process attributes from
15892         define_cond_exec.
15893         * doc/md.texi: Document fourth field in define_cond_exec.
15895 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15897         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15898         out the processing order as in store_bit_field_1.
15900 2013-06-04  Jan Hubicka  <jh@suse.cz>
15902         PR middle-end/57500
15903         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15904         non-VAR_DECL node if it does not exist yet.
15906 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15908         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15909         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15910         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15911         target_cpu_default setting.
15913 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15915         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15916         default to -fopt-info=optimized instead of all.
15917         * doc/invoke.texi: Ditto.
15918         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15919         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15920         (execute_vect_slp): Emit BB vectorization success under
15921         MSG_OPTIMIZED_LOCATIONS.
15922         * tree-vect-slp.c (vect_slp_transform_bb): Change
15923         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15924         * tree-vect-loop.c (vect_transform_loop): Ditto.
15926 2013-06-03  Jason Merrill  <jason@redhat.com>
15928         PR c++/57415
15929         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15930         Use TARGET_EXPR for C++.
15932 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15934         PR rtl-optimization/57268
15935         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15936         if DEBUG_INSN_P (insn).
15938         Reapply
15939         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15941         PR rtl-optimization/57268
15942         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15943         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15945 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15947         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15948         (ix86_avoid_lea_for_addr): Likewise.
15949         (exact_dependency_1): Likewise.
15950         (ix86_adjust_cost): Likewise.
15951         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15952         removed.
15953         (do_reorder_for_imul): Fix typo, formatting and
15954         !reload_completed check removed.
15955         (ix86_sched_reorder): Fix typo and formatting.
15956         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15957         list.
15959 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15961         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15963 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15965         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15966         <STRING_CST>: Likewise.
15967         <VECTOR_CST>: Likewise.
15969 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15970             Mikael Morin  <mikael@gcc.gnu.org>
15972         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15973         * config.in: Regenerated.
15974         * configure: Regenerated.
15976 2013-06-01  Jan Hubicka  <jh@suse.cz>
15978         PR middle-end/57366
15979         * cgraphunit.c (compile): When weakref is not supported,
15980         set up transparent aliases before final output pass.
15981         * varasm.c (assemble_alias): Do not try to do it here.
15983 2013-06-01  Jan Hubicka  <jh@suse.cz>
15985         PR middle-end/57467
15986         * passes.c (for_per_function): Skip unanalyzed functions.
15988 2013-06-01  Jan Hubicka  <jh@suse.cz>
15990         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15991         (lto_symtab_merge_symbols_1): ... this one.
15992         (lto_symtab_merge_cgraph_nodes): Rename to ...
15993         (lto_symtab_merge_symbols): ... this one; simplify.
15994         * cgraph.c (same_body_aliases_done): Rename to ...
15995         (cpp_implicit_aliases_done): ... this one.
15996         (cgraph_create_function_alias): Update.
15997         (cgraph_same_body_alias): Update.
15998         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15999         (verify_edge_corresponds_to_fndecl): Simplify.
16000         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
16001         (cgraph_node): Remove same_body_alias.
16002         (varpool_node): Remove alias_of and extra_name_alias.
16003         (same_body_aliases_done): Rename to ..
16004         (cpp_implicit_aliases_done): ... this one.
16005         (symtab_alias_ultimate_target): Add default parameter.
16006         (symtab_resolve_alias): New function.
16007         (fixup_same_cpp_alias_visibility): Declare.
16008         (cgraph_function_node): Add default parameter.
16009         (cgraph_node_asm_name): Likewise.
16010         (cgraph_function_or_thunk_node): Add default parameter; do
16011         not ICE when it is NULL.
16012         (varpool_variable_node): Likewise.
16013         * tree-emutls.c (create_emultls_var): Update.
16014         (ipa_lower_emutls): Update.
16015         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
16016         (cgraph_reset_node): Reset alias info.
16017         (cgraph_finalize_function): Update.
16018         (fixup_same_cpp_alias_visibility): Move to symtab.c.
16019         (analyze_function): Simplify.
16020         (cgraph_process_same_body_aliases): Simplify.
16021         (analyze_functions): Fixup same body aliases.
16022         (handle_alias_pairs): Simplify.
16023         (assemble_thunk): Update.
16024         (assemble_thunks_and_aliases): Update.
16025         (output_weakrefs): Rewrite.
16026         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
16027         (lto_output_varpool_node): Likewise.
16028         (compute_ltrans_boundary): Remve assert.
16029         (get_alias_symbol): New functoin.
16030         (input_node): Rewrite alias handling.
16031         (input_varpool_node): Likewise.
16032         * ipa-pure-const.c (propagate_pure_const): Fix formating.
16033         * ipa.c (process_references): Handle weakrefs correctly.
16034         (symtab_remove_unreachable_nodes): Likewise.
16035         * trans-mem.c (get_cg_data): Update.
16036         (ipa_tm_create_version_alias): Update.
16037         (ipa_tm_execute): Update.
16038         * symtab.c (dump_symtab_base): Dump aliases.
16039         (verify_symtab_base): Verify aliases.
16040         (symtab_node_availability): New function.
16041         (symtab_alias_ultimate_target): Simplify.
16042         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
16043         handle all the fixup cases.
16044         (symtab_resolve_alias): New function.
16045         * passes.c (ipa_write_summaries): Handle weakrefs.
16046         * varpool.c (varpool_analyze_node): Simplify.
16047         (assemble_aliases): Update.
16048         (varpool_create_variable_alias): Simplify.
16049         (varpool_extra_name_alias): Simplify.
16050         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
16051         (lto_symtab_merge_symbols): ... this one.
16053 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
16055         Revert
16056         PR rtl-optimization/57268
16057         * sched-deps.c (sched_analyze_2): Flush dependence lists if
16058         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
16060 2013-06-01  Tobias Burnus  <burnus@net-b.de>
16062         Partially reverted:
16063         2013-05-31  Tobias Burnus  <burnus@net-b.de>
16065         PR middle-end/57073
16066         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
16067         further up.
16069 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
16071         PR rtl-optimization/57268
16072         * sched-deps.c (sched_analyze_2): Flush dependence lists if
16073         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
16075 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
16077         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
16078         unordered comparison operators when -fno-trapping-math is in effect
16079         on the e500.
16080         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
16081         and implement unordered comparison operators properly on the e500.
16083 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
16085         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
16086         for constant scalar integers.
16087         (simplify_relational_operation_1): Likewise.
16089 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
16091         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
16092         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
16093         Fix comment.
16095 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
16096             Igor Zamyatin  <igor.zamyatin@intel.com>
16098         Silvermont (SLM) architecture performance tuning.
16099         * config/i386/i386.h (enum ix86_tune_indices): Add
16100         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
16101         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
16103         * config/i386/i386.c (initial_ix86_tune_features)
16104         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
16105         (ix86_lea_outperforms): Handle Silvermont tuning.
16106         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
16107         call.
16108         (ix86_use_lea_for_mov): Likewise.
16109         (ix86_avoid_lea_for_addr): Likewise.
16110         (ix86_lea_for_add_ok): Likewise.
16111         (exact_dependency_1): New function.
16112         (exact_store_load_dependency): Likewise.
16113         (ix86_adjust_cost): Handle Silvermont tuning.
16114         (do_reoder_for_imul): Likewise.
16115         (swap_top_of_ready_list): New function.
16116         (ix86_sched_reorder): Changed to handle Silvermont tuning.
16118         * config/i386/i386.md (peepholes that split memory operand in fp
16119         converts): New.
16121 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16123         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
16124         Remove un-necessary braces.
16126 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16128         * config/aarch64/aarch64.c (aarch64_classify_symbol):
16129         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
16131 2013-05-31  Tobias Burnus  <burnus@net-b.de>
16133         PR middle-end/57073
16134         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
16136 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16138         PR target/56315
16139         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
16140         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
16141         * config/arm/neon.md (iordi3_neon): Remove.
16142         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
16143         * config/arm/predicates.md (imm_for_neon_logic_operand):
16144         Move to earlier in the file.
16145         (neon_logic_op2): Likewise.
16146         (arm_iordi_operand_neon): New predicate.
16148 2013-05-31  Richard Biener  <rguenther@suse.de>
16150         PR tree-optimization/57478
16151         PR tree-optimization/57453
16152         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
16153         are life as well.
16155 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
16157         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
16158         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
16160 2013-05-30  Tobias Burnus  <burnus@net-b.de>
16161             Thomas Koenig  <tkoenig@gcc.gnu.org>
16163         PR middle-end/57073
16164         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
16165         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
16167 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
16169         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
16171 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
16173         * target.def (register_usage_leveling_p): New hook.
16174         * targhooks.c (default_register_usage_leveling_p): New.
16175         * targhooks.h (default_register_usage_leveling_p): New prototype.
16176         * lra-assigns.c (register_usage_leveling_p): Use the hook.
16177         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
16178         * doc/tm.texi: Update.
16179         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
16181 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
16183         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
16184         (*insv_reg<mode>): New define_insn.
16186 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
16188         PR rtl-optimization/57439
16189         * postreload.c (move2add_valid_value_p): Check that we have
16190         a zero subreg_regno_offset when accessing the register in
16191         the requested mode.
16193 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
16194             Igor Zamyatin  <igor.zamyatin@intel.com>
16196         Silvermont (SLM) architecture pipeline model, tuning and
16197         insn selection.
16198         * config.gcc: Add slm config options and target.
16200         * config/i386/slm.md: New.
16202         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
16204         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
16205         PROCESSOR_SLM.
16206         (ix86_target_macros_internal): Likewise.
16208         * config/i386/i386.c (slm_cost): New cost.
16209         (m_SLM): New macro flag.
16210         (initial_ix86_tune_features): Set m_SLM.
16211         (x86_accumulate_outgoing_args): Likewise.
16212         (x86_arch_always_fancy_math_387): Likewise.
16213         (processor_target_table): Add slm cost.
16214         (cpu_names): Add slm cpu name.
16215         (x86_option_override_internal): Set SLM ISA.
16216         (ix86_issue_rate): New case PROCESSOR_SLM.
16217         (ia32_multipass_dfa_lookahead): Likewise.
16218         (fold_builtin_cpu): Add slm.
16220         * config/i386/i386.h (TARGET_SLM): New target macro.
16221         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
16222         (processor_type): Add PROCESSOR_SLM.
16224         * config/i386/i386.md (cpu): Add new value "slm".
16225         (slm.md): Include slm.md.
16227 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
16228             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
16230         * config/arm/arm-protos.h: Add and update function protos.
16231         * config/arm/arm.c (use_simple_return_p): New added.
16232         (thumb2_expand_return): Check simple_return flag.
16233         * config/arm/arm.md: Add simple_return and conditional simple_return.
16234         * config/arm/iterators.md: Add iterator for return and simple_return.
16236 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
16238         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
16239         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
16240         (arm_emit_vfp_multi_reg_pop): Likewise.
16241         (thumb2_emit_ldrd_pop): Likewise.
16242         (arm_expand_epilogue): Add misc REG_CFA notes.
16243         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
16245 2013-05-29  Lawrence Crowl  <crowl@google.com>
16247         * config/arm/t-arm: Update for below.
16249         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
16250         Change type to hash_table.  Update dependent calls and types.
16252         * config/i386/t-cygming: Update for below.
16254         * config/i386/t-interix: Update for below.
16256         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
16257         Change type to hash_table.  Update dependent calls and types.
16258         (i386_find_on_wrapper_list::wrappers): Likewise.
16260         * config/ia64/t-ia64: Update for below.
16262         * config/ia64/ia64.c (bundle_state_table):
16263         Change type to hash_table.  Update dependent calls and types.
16265         * config/mips/mips.c (mips_reorg_process_insns::htab):
16266         Change type to hash_table.  Update dependent calls and types.
16268         * config/sol2.c (solaris_comdat_htab):
16269         Change type to hash_table.  Update dependent calls and types.
16271         * config/t-sol2: Update for above.
16273 2013-05-29  Teresa Johnson  <tejohnson@google.com>
16275         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
16276         functions are not yet marked as defined.
16278 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
16279             Pat Haugen  <pthaugen@us.ibm.com>
16280             Peter Bergner  <bergner@vnet.ibm.com>
16282         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
16283         instructions.
16284         (VEC_A): Likewise.
16285         (VEC_C): Likewise.
16286         (vrotl<mode>3): Likewise.
16287         (vashl<mode>3): Likewise.
16288         (vlshr<mode>3): Likewise.
16289         (vashr<mode>3): Likewise.
16291         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16292         support for power8 V2DI builtins.
16294         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
16295         power8 V2DI builtins.
16296         (vupkhsw): Likewise.
16297         (vupklsw): Likewise.
16298         (vaddudm): Likewise.
16299         (vminsd): Likewise.
16300         (vmaxsd): Likewise.
16301         (vminud): Likewise.
16302         (vmaxud): Likewise.
16303         (vpkudum): Likewise.
16304         (vpksdss): Likewise.
16305         (vpkudus): Likewise.
16306         (vpksdus): Likewise.
16307         (vrld): Likewise.
16308         (vsld): Likewise.
16309         (vsrd): Likewise.
16310         (vsrad): Likewise.
16311         (vsubudm): Likewise.
16312         (vcmpequd): Likewise.
16313         (vcmpgtsd): Likewise.
16314         (vcmpgtud): Likewise.
16315         (vcmpequd_p): Likewise.
16316         (vcmpgtsd_p): Likewise.
16317         (vcmpgtud_p): Likewise.
16318         (vupkhsw): Likewise.
16319         (vupklsw): Likewise.
16320         (vaddudm): Likewise.
16321         (vmaxsd): Likewise.
16322         (vmaxud): Likewise.
16323         (vminsd): Likewise.
16324         (vminud): Likewise.
16325         (vpksdss): Likewise.
16326         (vpksdus): Likewise.
16327         (vpkudum): Likewise.
16328         (vpkudus): Likewise.
16329         (vrld): Likewise.
16330         (vsld): Likewise.
16331         (vsrad): Likewise.
16332         (vsrd): Likewise.
16333         (vsubudm): Likewise.
16335         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
16336         support for power8 V2DI instructions.
16338         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
16339         power8 V2DI instructions.  Combine pack and unpack insns to use an
16340         iterator for each mode.  Check whether a particular mode supports
16341         Altivec instructions instead of just checking TARGET_ALTIVEC.
16342         (UNSPEC_VPKUWUM): Likewise.
16343         (UNSPEC_VPKSHSS): Likewise.
16344         (UNSPEC_VPKSWSS): Likewise.
16345         (UNSPEC_VPKUHUS): Likewise.
16346         (UNSPEC_VPKSHUS): Likewise.
16347         (UNSPEC_VPKUWUS): Likewise.
16348         (UNSPEC_VPKSWUS): Likewise.
16349         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
16350         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
16351         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
16352         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
16353         (UNSPEC_VUPKHSB): Likewise.
16354         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
16355         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
16356         (UNSPEC_VUPKHSH): Likewise.
16357         (UNSPEC_VUPKLSB): Likewise.
16358         (UNSPEC_VUPKLSH): Likewise.
16359         (VI2): Likewise.
16360         (VI_char): Likewise.
16361         (VI_scalar): Likewise.
16362         (VI_unit): Likewise.
16363         (VP): Likewise.
16364         (VP_small): Likewise.
16365         (VP_small_lc): Likewise.
16366         (VU_char): Likewise.
16367         (add<mode>3): Likewise.
16368         (altivec_vaddcuw): Likewise.
16369         (altivec_vaddu<VI_char>s): Likewise.
16370         (altivec_vadds<VI_char>s): Likewise.
16371         (sub<mode>3): Likewise.
16372         (altivec_vsubcuw): Likewise.
16373         (altivec_vsubu<VI_char>s): Likewise.
16374         (altivec_vsubs<VI_char>s): Likewise.
16375         (altivec_vavgs<VI_char>): Likewise.
16376         (altivec_vcmpbfp): Likewise.
16377         (altivec_eq<mode>): Likewise.
16378         (altivec_gt<mode>): Likewise.
16379         (altivec_gtu<mode>): Likewise.
16380         (umax<mode>3): Likewise.
16381         (smax<mode>3): Likewise.
16382         (umin<mode>3): Likewise.
16383         (smin<mode>3): Likewise.
16384         (altivec_vpkuhum): Likewise.
16385         (altivec_vpkuwum): Likewise.
16386         (altivec_vpkshss): Likewise.
16387         (altivec_vpkswss): Likewise.
16388         (altivec_vpkuhus): Likewise.
16389         (altivec_vpkshus): Likewise.
16390         (altivec_vpkuwus): Likewise.
16391         (altivec_vpkswus): Likewise.
16392         (altivec_vpks<VI_char>ss): Likewise.
16393         (altivec_vpks<VI_char>us): Likewise.
16394         (altivec_vpku<VI_char>us): Likewise.
16395         (altivec_vpku<VI_char>um): Likewise.
16396         (altivec_vrl<VI_char>): Likewise.
16397         (altivec_vsl<VI_char>): Likewise.
16398         (altivec_vsr<VI_char>): Likewise.
16399         (altivec_vsra<VI_char>): Likewise.
16400         (altivec_vsldoi_<mode>): Likewise.
16401         (altivec_vupkhsb): Likewise.
16402         (altivec_vupkhs<VU_char>): Likewise.
16403         (altivec_vupkls<VU_char>): Likewise.
16404         (altivec_vupkhsh): Likewise.
16405         (altivec_vupklsb): Likewise.
16406         (altivec_vupklsh): Likewise.
16407         (altivec_vcmpequ<VI_char>_p): Likewise.
16408         (altivec_vcmpgts<VI_char>_p): Likewise.
16409         (altivec_vcmpgtu<VI_char>_p): Likewise.
16410         (abs<mode>2): Likewise.
16411         (vec_unpacks_hi_v16qi): Likewise.
16412         (vec_unpacks_hi_v8hi): Likewise.
16413         (vec_unpacks_lo_v16qi): Likewise.
16414         (vec_unpacks_hi_<VP_small_lc>): Likewise.
16415         (vec_unpacks_lo_v8hi): Likewise.
16416         (vec_unpacks_lo_<VP_small_lc>): Likewise.
16417         (vec_pack_trunc_v8h): Likewise.
16418         (vec_pack_trunc_v4si): Likewise.
16419         (vec_pack_trunc_<mode>): Likewise.
16421         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
16422         V2DI builtins.
16423         (vec_vmaxsd): Likewise.
16424         (vec_vmaxud): Likewise.
16425         (vec_vminsd): Likewise.
16426         (vec_vminud): Likewise.
16427         (vec_vpksdss): Likewise.
16428         (vec_vpksdus): Likewise.
16429         (vec_vpkudum): Likewise.
16430         (vec_vpkudus): Likewise.
16431         (vec_vrld): Likewise.
16432         (vec_vsld): Likewise.
16433         (vec_vsrad): Likewise.
16434         (vec_vsrd): Likewise.
16435         (vec_vsubudm): Likewise.
16436         (vec_vupkhsw): Likewise.
16437         (vec_vupklsw): Likewise.
16439 2013-05-29  Jan Hubicka  <jh@suse.cz>
16441         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
16442         flags; reorder rest of fields in more consistent way.
16443         (varpool_node): Remove analyzed, finalized and alias.
16444         (cgraph_ndoe): Likewise.
16445         (symtab_alias_ultimate_target): New function.
16446         (cgraph_function_node): Move offline.
16447         (cgraph_reset_node): Declare.
16448         (cgraph_comdat_can_be_unshared_p): Remove.
16449         (varpool_remove_initializer): Declare.
16450         (varpool_first_defined_variable, varpool_next_defined_variable
16451         cgraph_first_defined_function, cgraph_next_defined_function): Update.
16452         (cgraph_function_with_gimple_body_p): Update.
16453         (varpool_all_refs_explicit_p): Update.
16454         (symtab_alias_target): New function.
16455         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
16456         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
16457         (cgraph_function_or_thunk_node): Simplify using
16458         symtab_alias_ultimate_target.
16459         (varpool_variable_node): Likewise.
16460         * cgraph.c (cgraph_create_function_alias): Update.
16461         (cgraph_add_thunk): Update.
16462         (cgraph_remove_node): Update.
16463         (dump_cgraph_node): Do not dump removed flags.
16464         (cgraph_function_body_availability): Update.
16465         (cgraph_propagate_frequency): Update.
16466         (verify_cgraph_node): Check sanity of local flag.
16467         (cgraph_function_node): Move here from cgraph.h; revamp for
16468         cgraph_function_or_thunk_node.
16469         * lto-symtab.c (lto_varpool_replace_node): Update.
16470         (lto_symtab_resolve_can_prevail_p): Update.
16471         (lto_symtab_merge_cgraph_nodes): Update.
16472         * ipa-cp.c (determine_versionability, initialize_node_lattices,
16473         propagate_constants_accross_call, devirtualization_time_bonus,
16474         ipcp_propagate_stage): Update.
16475         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16476         * ipa-inline-transform.c (clone_inlined_nodes,
16477         preserve_function_body_p): Update.
16478         * ipa-reference.c (propagate): Update.
16479         (write_node_summary_p): Update.
16480         * toplev.c (wrapup_global_declaration_2): Update.
16481         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16482         (analyze_function) ... this one.
16483         (cgraph_process_new_functions): Update.
16484         (cgraph_reset_node): Export.
16485         (cgraph_finalize_function): Update.
16486         (cgraph_add_new_function): Update.
16487         (process_function_and_variable_attributes): Update.
16488         (varpool_finalize_decl): Update.
16489         (symbol_finalized): Remove.
16490         (symbol_finalized_and_needed): Rename to ...
16491         (symbol_defined_and_needed): ... update.
16492         (cgraph_analyze_functions): Update.
16493         (handle_alias_pairs): Update.
16494         (mark_functions_to_output): Update.
16495         (assemble_thunk): Update.
16496         (output_in_order): Update.
16497         (output_weakrefs): Update.
16498         (finalize_compilation_unit): Update.
16499         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16500         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16501         input_node, input_varpool_node): Update.
16502         * dbxout.c (dbxout_expand_expr): Update.
16503         * cgraphclones.c (cgraph_clone_node): Update.
16504         (cgraph_copy_node_for_versioning): Update.
16505         (cgraph_materialize_clone): Update.
16506         (cgraph_materialize_all_clones): Update.
16507         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16508         propagate_pure_const, propagate_nothrow): Update.
16509         * lto-streamer-out.c (lto_output, write_symbol): Update.
16510         * ipa-utils.c (ipa_reverse_postorder): Update.
16511         * ipa-inline.c (can_inline_edge_p): Update.
16512         (update_caller_keys, ipa_inline): Update.
16513         * dwarf2out.c (reference_to_unused,
16514         premark_types_used_by_global_vars_helper): Update.
16515         * tree-eh.c (tree_could_trap_p): Update.
16516         * ipa-split.c (consider_split, execute_split_functions): Update.
16517         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16518         has_addr_references_p): Update; move ahead in file for better
16519         readability.
16520         (process_references): Simplify.
16521         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16522         bodies are removed.
16523         (cgraph_comdat_can_be_unshared_p): Make static.
16524         (cgraph_externally_visible_p): Update.
16525         (varpool_externally_visible_p): Update.
16526         (function_and_variable_visibility): Update.
16527         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16528         ipa_tm_mark_force_output_node): Update.
16529         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16530         estimate_edge_devirt_benefit, inline_generate_summary,
16531         inline_write_summary): Update.
16532         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16533         * ipa-prop.c (ipa_compute_jump_functions): Update.
16534         (ipa_print_node_params, ipa_prop_read_section,
16535         ipa_update_after_lto_read, read_replacements_section): Update.
16536         * varasm.c (mark_decl_referenced): Update.
16537         (assemble_alias, dump_tm_clone_pairs): Update.
16538         * tree-inline.c (copy_bb): Update.
16539         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16540         Update.
16541         * symtab.c (dump_symtab_base): Print new flags.
16542         (verify_symtab_base): Verify new flags.
16543         (symtab_alias_ultimate_target): New function.
16544         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16545         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16546         Update.
16547         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16548         Update.
16549         * i386.c (ix86_get_function_versions_dispatcher,
16550         ix86_generate_version_dispatcher_body): Update.
16551         (fold_builtin_cpu): Use varpool_add_new_variable.
16552         * varpool.c (varpool_remove_initializer): Break out from ...
16553         (varpool_remove_node): ... this one.
16554         (dump_varpool_node, varpool_node_for_asm,
16555         cgraph_variable_initializer_availability, varpool_analyze_node,
16556         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16557         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16558         Update.
16560 2013-05-29  Jan Hubicka  <jh@suse.cz>
16562         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16564 2013-05-29  Easwaran Raman  <eraman@google.com>
16566         PR tree-optimization/57442
16567         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16568         when control exits the main loop.
16570 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16572         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16573         and RX600.
16574         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16575         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16576         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16577         option.
16579 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16581         PR tree-optimization/57441
16582         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16583         Don't limit size of incr_vec to number of candidates.
16585 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16587         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16588         and mips16 directories.
16589         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16590         (MULTILIB_DIRNAMES): Ditto.
16591         (MULTILIB_EXCEPTIONS): Add new exceptions.
16592         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16593         (MULTILIB_DIRNAMES): Ditto.
16594         (MULTILIB_EXCEPTIONS): Add new exceptions.
16596 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16597             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16599         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16600         SYMBOL_TINY_ABSOLUTE.
16601         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16602         SYMBOL_TINY_ABSOLUTE.
16603         (aarch64_expand_mov_immediate): Likewise.
16604         (aarch64_classify_symbol): Likewise.
16605         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16606         Permit SYMBOL_TINY_ABSOLUTE.
16607         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16609 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16610             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16612         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16613         Refactor if/switch.  Replace gcc_assert with if.
16615 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16617         * config/i386/i386.c (initial_ix86_tune_features): Enable
16618         FP Reassociation for AMD bdver1 and bdver2.
16620 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16622         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16623         and IMAGPART_EXPR do not occur within other handled_components.
16625 2013-05-29  Richard Biener  <rguenther@suse.de>
16627         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16628         access on whether the use is in the BB we currently try to
16629         vectorize.
16630         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16631         vectorize to vect_bb_slp_scalar_cost.
16633 2013-05-29  Richard Biener  <rguenther@suse.de>
16635         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16636         computing scalar cost offsetted by stmts that are kept live
16637         by scalar uses.
16638         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16639         for computation of scalar cost.
16641 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16643         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16645 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16647         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16648         array notation built-in reduction functions.
16649         * doc/passes.texi (Passes): Added documentation about changes done
16650         for Cilk Plus.
16651         * doc/invoke.texi (C Dialect Options): Added documentation about
16652         the -fcilkplus flag.
16653         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16654         (BUILTINS_DEF): Depend on cilkplus.def.
16655         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16656         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16657         * cilkplus.def: New file.
16659 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16661         PR rtl-optimization/57439
16662         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16664 2013-05-28  Easwaran Raman  <eraman@google.com>
16666         PR tree-optimization/57337
16667         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16668         (find_insert_point): Correctly identify the insertion point
16669         when two statements with the same UID is compared.
16671 2013-05-28  Richard Biener  <rguenther@suse.de>
16673         PR tree-optimization/56787
16674         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16675         from the list of data references.
16676         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16677         clobbers.
16678         (vect_analyze_loop_operations): Likewise.
16679         (vect_transform_loop): Remove clobbers.
16681 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16683         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16684         and REALPART_EXPRs have scalar type.
16686 2013-05-28  Richard Biener  <rguenther@suse.de>
16688         PR tree-optimization/57411
16689         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16690         virtual operands.
16691         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16692         virtual operand propagation.
16694 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16696         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16697         destination register for bmasksi_vis.
16698         (vector_init_bshuffle): Likewise.
16699         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16701 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16703         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16704         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16705         mode if the instruction isn't available in the original mode.
16706         * config/sparc/sparc.opt (mfix-ut699): New option.
16707         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16708         (divdf3): Turn into expander.
16709         (divdf3_nofix): New insn.
16710         (divdf3_fix): Likewise.
16711         (divsf3): Disable if -mfix-ut699.
16712         (sqrtdf2): Turn into expander.
16713         (sqrtdf2_nofix): New insn.
16714         (sqrtdf2_fix): Likewise.
16715         (sqrtsf2): Disable if -mfix-ut699.
16717 2013-05-27  Richard Biener  <rguenther@suse.de>
16719         PR middle-end/57412
16720         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16721         block for the new loop.
16723 2013-05-27  Richard Biener  <rguenther@suse.de>
16725         PR tree-optimization/57343
16726         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16727         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16728         (number_of_iterations_cond): Do not build the folded tree.
16730 2013-05-27  Richard Biener  <rguenther@suse.de>
16732         Revert
16733         PR middle-end/57381
16734         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16735         OEP_CONSTANT_ADDRESS_OF retained.
16737         PR tree-optimization/57417
16738         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16739         for unchanged base.
16740         (set_ssa_val_to): Compare addresses using
16741         get_addr_base_and_unit_offset.
16743 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16745         PR rtl-optimization/56833
16746         * postreload.c (move2add_record_mode): New function.
16747         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16748         (move2add_use_add2_insn): Use move2add_record_sym_value.
16749         (move2add_use_add3_insn): Likewise.
16750         (reload_cse_move2add): Use move2add_valid_value_p and
16751         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16752         affected regs by setting reg_mode to VOIDmode.
16753         (move2add_note_store): Don't pretend the inside of a SUBREG is
16754         the actual destination.  Invalidate single/leading registers by
16755         setting reg_mode to VOIDmode.
16756         Use move2add_record_sym_value, move2add_valid_value_p and
16757         move2add_record_mode.
16759 2013-05-27  Richard Biener  <rguenther@suse.de>
16761         PR tree-optimization/57396
16762         * tree-affine.c (double_int_constant_multiple_p): Properly
16763         return false for val == 0 and div != 0.
16765 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16767         * config/mips/mips.h: Use #elif in preprocessor conditions.
16769 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16771         PR target/53916
16772         * config/mips/constraints.md (kl): New constraint.
16773         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16774         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16775         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16776         from LO for MIPS16.
16777         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16779 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16781         PR target/55777
16782         * config/mips/mips.c (mips_can_inline_p): New function.
16783         (TARGET_CAN_INLINE_P): Define.
16785 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16787         * sched-int.h (ds_t, dw_t): Make unsigned int.
16788         Fix documentation that describes how all the ds_t bits are used.
16789         Reserve the last bit for delayed-branch scheduling.
16790         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16791         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16792         (gen_dep_weak_1): Remove prototype.
16793         * sched-deps.c (get_dep_weak_1): Make static.
16794         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16795         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16796         * doc/tm.texi: Regenerate.
16797         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16799 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16801         PR debug/56950
16802         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16804 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16805             Sandra Loosemore  <sandra@codesourcery.com>
16807         * config.gcc (powerpc-*): Allow native for with-cpu.
16809 2013-05-24  Jeff Law  <law@redhat.com>
16811         PR tree-optimization/57124
16812         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16813         conversion feeding a condition if the range has an overflow
16814         if -fstrict-overflow.  Add warnings for when we do make the
16815         transformation.
16817 2013-05-24  Dehao Chen  <dehao@google.com>
16819         * tree-cfg.c (locus_discrim_map): Fix the typo.
16820         (locus_discrim_hasher): Likewise.
16821         (locus_discrim_hasher::hash): Likewise.
16822         (locus_discrim_hasher::equal): Likewise.
16824 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16826         PR tree-optimization/57294
16827         * cgraph.h (ipa_record_stmt_references): Declare.
16828         * cgraphbuild.c (ipa_record_stmt_references): New function.
16829         (build_cgraph_edges): Use ipa_record_stmt_references.
16830         (rebuild_cgraph_edges): Likewise.
16831         (cgraph_rebuild_references): Likewise.
16832         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16833         associated with the old statement and build references from the
16834         newly built statements.
16835         * ipa-ref.c (ipa_remove_stmt_references): New function.
16836         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16838 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16840         * lra-constraints.c (emit_spill_move): Use smaller mode for
16841         mem-mem moves.
16842         (check_and_process_move): Consider mem-reg moves for secondary
16843         too.
16844         (curr_insn_transform): Don't lose insns emitted before for
16845         secondary memory moves.
16846         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16847         reg set up in the current insn.
16849 2013-05-24  Dehao Chen  <dehao@google.com>
16851         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16852         hash function.
16853         (locus_descrim_hasher::equal): Likewise.
16854         (build_gimple_cfg): New discriminator assignment algorithm.
16855         (make_edges): Likewise.
16856         (next_discriminator_for_locus): Likewise.
16857         (same_line_p): Likewise.
16858         (assign_discriminators): Likewise.
16859         (make_cond_expr_edges): Likewise.
16860         (make_gimple_switch_edges): Likewise.
16861         (make_goto_expr_edges): Likewise.
16862         (make_gimple_asm_edges): Likewise.
16864 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16866         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16867         X format specifier to only display bottom 16 bits.
16868         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16869         immediate to match for operand 2, since it will be masked.
16871 2013-05-24  Richard Biener  <rguenther@suse.de>
16873         PR tree-optimization/57287
16874         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16875         all SSA names that occur in abnormal PHIs.
16877 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16879         PR tree-ssa/57385
16880         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16881         that index is not negative.
16883 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16885         PR rtl-optimization/55177
16886         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16887         (simplify_byte_swapping_operation): New.
16888         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16889         (simplify_relational_operation_1): Deal with BSWAP.
16891 2013-05-23  Richard Henderson  <rth@redhat.com>
16893         PR target/56742
16894         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16895         (ix86_reorg): Call it.
16897 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16899         PR target/57379
16900         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16901         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16902         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16904 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16906         PR debug/57351
16907         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16909 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16910             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16912         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16913         * config/aarch64/constraints.md (Usa): Remove.
16914         * doc/md.texi (AArch64 Usa): Remove.
16916 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16917             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16919         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16920         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16921         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16922         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16924 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16926         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16927         instruction (AdvSIMD).
16928         * config/aarch64/aarch64-builtins.c
16929         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16930         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16932 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16934         PR middle-end/57347
16935         * tree.h (contains_bitfld_component_ref_p): Declare.
16936         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16937         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16938         caller.
16939         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16940         not access a bit-field.  Assert all final offsets are byte-aligned.
16942 2013-05-23  Richard Biener  <rguenther@suse.de>
16944         PR tree-optimization/57380
16945         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16946         least one invariant or re-used load.
16947         * passes.c (init_optimization_passes): Move pass_phiprop before
16948         pass_forwprop.
16950 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16952         * config/aarch64/aarch64-simd.md
16953         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16955 2013-05-23  Richard Biener  <rguenther@suse.de>
16957         PR middle-end/57381
16958         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16959         OEP_CONSTANT_ADDRESS_OF retained.
16961 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16963         PR middle-end/57344
16964         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16965         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16967 2013-05-23  Richard Biener  <rguenther@suse.de>
16969         PR rtl-optimization/57341
16970         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16971         instead of true_dependence.
16973 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16975         * bb-reorder.c (branch_threshold): Make const.
16976         (exec_threshold): Ditto.
16978 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16979             Pat Haugen  <pthaugen@us.ibm.com>
16980             Peter Bergner  <bergner@vnet.ibm.com>
16982         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16983         documentation for the power8 crypto builtins.
16985         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16987         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16988         macros for defining power8 builtin functions.
16989         (BU_P8V_AV_2): Likewise.
16990         (BU_P8V_AV_P): Likewise.
16991         (BU_P8V_VSX_1): Likewise.
16992         (BU_P8V_OVERLOAD_1): Likewise.
16993         (BU_P8V_OVERLOAD_2): Likewise.
16994         (BU_CRYPTO_1): Likewise.
16995         (BU_CRYPTO_2): Likewise.
16996         (BU_CRYPTO_3): Likewise.
16997         (BU_CRYPTO_OVERLOAD_1): Likewise.
16998         (BU_CRYPTO_OVERLOAD_2): Likewise.
16999         (XSCVSPDP): Fix typo, point to the correct instruction.
17000         (VCIPHER): Add power8 crypto builtins.
17001         (VCIPHERLAST): Likewise.
17002         (VNCIPHER): Likewise.
17003         (VNCIPHERLAST): Likewise.
17004         (VPMSUMB): Likewise.
17005         (VPMSUMH): Likewise.
17006         (VPMSUMW): Likewise.
17007         (VPERMXOR_V2DI): Likewise.
17008         (VPERMXOR_V4SI: Likewise.
17009         (VPERMXOR_V8HI: Likewise.
17010         (VPERMXOR_V16QI: Likewise.
17011         (VSHASIGMAW): Likewise.
17012         (VSHASIGMAD): Likewise.
17013         (VPMSUM): Likewise.
17014         (VPERMXOR): Likewise.
17015         (VSHASIGMA): Likewise.
17017         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
17018         __CRYPTO__ if the crypto instructions are available.
17019         (altivec_overloaded_builtins): Add support for overloaded power8
17020         builtins.
17022         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
17023         support for power8 crypto builtins.
17024         (builtin_function_type): Likewise.
17025         (altivec_init_builtins): Add support for builtins that take vector
17026         long long (V2DI) arguments.
17028         * config/rs6000/crypto.md: New file, define power8 crypto
17029         instructions.
17031 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
17032             Pat Haugen  <pthaugen@us.ibm.com>
17033             Peter Bergner  <bergner@vnet.ibm.com>
17035         * doc/invoke.texi (Option Summary): Add power8 options.
17036         (RS/6000 and PowerPC Options): Likewise.
17038         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
17039         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
17040         wm, wn, wr documentation.
17042         * config/rs6000/constraints.md (wm): New constraint for VSX
17043         registers if direct move instructions are enabled.
17044         (wn): New constraint for no registers.
17045         (wq): New constraint for quad word even GPR registers.
17046         (wr): New constraint if 64-bit instructions are enabled.
17047         (wv): New constraint if power8 vector instructions are enabled.
17048         (wQ): New constraint for quad word memory locations.
17050         * config/rs6000/predicates.md (const_0_to_15_operand): New
17051         constraint for 0..15 for crypto instructions.
17052         (gpc_reg_operand): If VSX allow registers in VSX registers as well
17053         as GPR and floating point registers.
17054         (int_reg_operand): New predicate to match only GPR registers.
17055         (base_reg_operand): New predicate to match base registers.
17056         (quad_int_reg_operand): New predicate to match even GPR registers
17057         for quad memory operations.
17058         (vsx_reg_or_cint_operand): New predicate to allow vector logical
17059         operations in both GPR and VSX registers.
17060         (quad_memory_operand): New predicate for quad memory operations.
17061         (reg_or_indexed_operand): New predicate for direct move support.
17063         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
17064         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
17065         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
17066         (POWERPC_MASKS): Add power8 options.
17067         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
17068         various options.
17070         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
17071         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
17073         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
17074         (-mpower8-fusion): New power8 options.
17075         (-mpower8-fusion-sign): Likewise.
17076         (-mpower8-vector): Likewise.
17077         (-mcrypto): Likewise.
17078         (-mdirect-move): Likewise.
17079         (-mquad-memory): Likewise.
17081         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
17082         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
17083         registers.
17084         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
17085         (rs6000_debug_vector_unit): Add p8_vector.
17086         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
17087         definitions.  Also print fusion state.
17088         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
17089         (rs6000_builtin_mask_calculate): Add power8 builtin support.
17090         (rs6000_option_override_internal): Add support for power8.
17091         (rs6000_common_init_builtins): Add debugging for skipped builtins
17092         if -mdebug=builtin.
17093         (rs6000_adjust_cost): Add power8 support.
17094         (rs6000_issue_rate): Likewise.
17095         (insn_must_be_first_in_group): Likewise.
17096         (insn_must_be_last_in_group): Likewise.
17097         (force_new_group): Likewise.
17098         (rs6000_register_move_cost): Likewise.
17099         (rs6000_opt_masks): Likewise.
17101         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
17102         power8 capable assembler, default to power7 options.
17103         (TARGET_DIRECT_MOVE): Likewise.
17104         (TARGET_CRYPTO): Likewise.
17105         (TARGET_P8_VECTOR): Likewise.
17106         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
17107         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
17108         (VECTOR_MEM_P8_VECTOR_P): Likewise.
17109         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
17110         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
17111         (TARGET_XSCVDPSPN): Likewise.
17112         (TARGET_XSCVSPDPN): Likewsie.
17113         (TARGET_SYNC_HI_QI): Likewise.
17114         (TARGET_SYNC_TI): Likewise.
17115         (MASK_CRYPTO): Likewise.
17116         (MASK_DIRECT_MOVE): Likewise.
17117         (MASK_P8_FUSION): Likewise.
17118         (MASK_P8_VECTOR): Likewise.
17119         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
17120         temporary used by some of the direct move instructions to get two FP
17121         temporary registers does not force creation of a stack frame.
17122         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
17123         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
17124         that any VSX registers are tieable, even if they are also an
17125         Altivec vector mode.
17126         (r6000_reg_class_enum): Add wm, wr, wv constraints.
17127         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
17128         (RS6000_BTM_CRYPTO): Likewise.
17129         (RS6000_BTM_COMMON): Likewise.
17131         * config/rs6000/rs6000.md (cpu attribute): Add power8.
17132         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
17133         (enum rs6000_vector): Add power8 vector support.
17135 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17137         PR target/19599
17138         PR target/57340
17139         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
17140         (any_sibcall_could_use_r3): this and handle indirect calls.
17141         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
17143 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17145         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
17147 2013-05-22  Richard Biener  <rguenther@suse.de>
17149         PR middle-end/57349
17150         * profile.c (branch_prob): Do not split blocks that are
17151         abnormally receiving from ECF_RETURNS_TWICE functions.
17153 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
17155         * recog.c (offsettable_address_addr_space_p): Fix calculation of
17156         address mode.  Move pointer mode initialization to the same place.
17158 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
17160         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
17161         while it has any effect.
17163 2013-05-21  Easwaran Raman  <eraman@google.com>
17165         PR tree-optimization/57322
17166         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
17167         UID of the statement added to the BB to be 1.
17169 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
17171         PR tree-optimization/57331
17172         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
17173         of conversion from pointer type to integral type with integer.
17175 2013-05-21  Martin Jambor  <mjambor@suse.cz>
17177         PR lto/57289
17178         * ipa-prop.c (ipa_read_node_info): Process param_used and
17179         controlled_uses in the same order as when writing.
17181 2013-05-21  Magnus Granberg  <baldrick@free.fr>
17183         PR plugins/56754
17184         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
17186 2013-05-21  Richard Biener  <rguenther@suse.de>
17188         PR tree-optimization/57318
17189         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
17190         estimate stmts with side-effects as likely eliminated.
17192 2013-05-21  Richard Biener  <rguenther@suse.de>
17194         PR tree-optimization/57330
17195         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
17196         preserve the call stmts fntype.
17198 2013-05-21  Richard Biener  <rguenther@suse.de>
17200         PR tree-optimization/57303
17201         * tree-ssa-sink.c (statement_sink_location): Improve killing
17202         stmt detection and properly handle self-assignments.
17204 2013-05-21  Christian Bruel  <christian.bruel@st.com>
17206         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
17207         spanning registers. LEAF_REG_REMAP is supported only for contiguous
17208         registers. Set register size out of the PARALLEL loop.
17210 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
17212         PR target/56547
17213         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
17214         (*fmasf4, *fmasf4_media): New insns.
17216 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
17218         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
17219         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
17220         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
17221         (mips_idiv_insns): Update the comments to say that the returned
17222         instruction counts are in units of BASE_INSN_LENGTH.
17223         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
17224         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
17225         using 2 rather than 4 as the length of indirect MIPS16 and
17226         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
17227         length of a NOP.  Don't divide MIPS16 lengths by 2.
17228         (mips16_split_long_branches): Assume a branch is long if the
17229         length is greater than 4 rather than 8.
17230         * config/mips/mips.md (length): Give MIPS16 lengths directly,
17231         rather than multiplying them by 2.  Multiply instruction counts
17232         by BASE_INSN_LENGTH rather than 4.
17233         (*jump_mips16, tls_get_tp_mips16_<mode>)
17234         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
17236 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
17238         * config/mips/mips.md (extended_mips16): Remove branch case.
17239         (length): Remove duplicated extended_mips16 test.
17241 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
17243         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
17245 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
17247         * recog.h (Recog_data): Rename to...
17248         (recog_data_d): ...this.
17249         (recog_data): Update accordingly.
17250         * recog.c (recog_data): Likewise.
17251         * reload.c (save_recog_data): Likewise.
17252         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
17253         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
17255 2013-05-17  Julian Brown  <julian@codesourcery.com>
17257         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
17258         found in a REG_EQUAL note, invalidate it.
17260 2013-05-17  Easwaran Raman  <eraman@google.com>
17262         * tree-ssa-reassoc.c (find_insert_point): New function.
17263         (insert_stmt_after): Likewise.
17264         (get_def_stmt): Likewise.
17265         (ensure_ops_are_available): Likewise.
17266         (not_dominated_by): Likewise.
17267         (rewrite_expr_tree): Do not move statements beyond what is
17268         necessary. Remove call to swap_ops_for_binary_stmt...
17269         (reassociate_bb): ... and move it here.
17270         (build_and_add_sum): Assign UIDs for new statements.
17271         (linearize_expr): Likewise.
17272         (do_reassoc): Renumber gimple statement UIDs.
17274 2013-05-17  Jan Hubicka  <jh@suse.cz>
17276         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
17277         weakrefs.
17278         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
17279         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
17280         weakrefs.
17281         (output_weakrefs): Update.
17283 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
17284             Martin Jambor  <mjambor@suse.cz>
17286         PR middle-end/57276
17287         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
17288         value that corresponds to the given aggval is found in values vector.
17290 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
17292         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
17293         sse, sse2, sse3, ssse3 and sse4a flags to options.
17295 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
17297         * gengtype-state.c: (s_expr_writer): New class, to handle
17298         prettifying of output layout of s-expressions.
17299         (state_writer): New class, to write out gtype.state.
17300         (state_written_type_count): Move this variable into member data of
17301         state_writer.
17302         (s_expr_writer::s_expr_writer): New code: constructor for new class
17303         (state_writer::state_writer(): ditto
17304         (s_expr_writer::write_new_line): New function
17305         (s_expr_writer::write_any_indent): ditto
17306         (s_expr_writer::begin_s_expr): ditto
17307         (s_expr_writer::end_s_expr): ditto
17308         (write_state_fileloc): convert to method of state_writer...
17309         (state_writer:: write_state_fileloc): ...and use methods of
17310         s_expr_writer to write indentation into the gtype.state output file
17311         to visually represent the hierarchical structure of the list
17312         structures
17313         (write_state_fields): ditto, renaming to...
17314         (state_writer::write_state_fields)
17315         (write_state_a_string): ditto, renaming to...
17316         (state_writer::write_state_a_string)
17317         (write_state_string_option): ditto, renaming to...
17318         (state_writer::write_state_string_option)
17319         (write_state_type_option): ditto, renaming to...
17320         (state_writer::write_state_type_option)
17321         (write_state_nested_option): ditto, renaming to...
17322         (state_writer::write_state_nested_option)
17323         (write_state_option): ditto, renaming to...
17324         (state_writer::write_state_option)
17325         (write_state_options): ditto, renaming to...
17326         (state_writer::write_state_options)
17327         (write_state_lang_bitmap): ditto, renaming to...
17328         (state_writer::write_state_lang_bitmap)
17329         (write_state_version): ditto, renaming to...
17330         (state_writer::write_state_version)
17331         (write_state_scalar_type): ditto, renaming to...
17332         (state_writer::write_state_scalar_type)
17333         (write_state_string_type): ditto, renaming to...
17334         (state_writer::write_state_string_type)
17335         (write_state_undefined_type): ditto, renaming to...
17336         (state_writer::write_state_undefined_type)
17337         (write_state_struct_union_type): ditto, renaming to...
17338         (state_writer::write_state_struct_union_type)
17339         (write_state_struct_type): ditto, renaming to...
17340         (state_writer::write_state_struct_type)
17341         (write_state_user_struct_type): ditto, renaming to...
17342         (state_writer::write_state_user_struct_type)
17343         (write_state_lang_struct_type): ditto, renaming to...
17344         (state_writer::write_state_lang_struct_type)
17345         (write_state_param_struct_type): ditto, renaming to...
17346         (state_writer::write_state_param_struct_type)
17347         (write_state_pointer_type): ditto, renaming to...
17348         (state_writer::write_state_pointer_type)
17349         (write_state_array_type): ditto, renaming to...
17350         (state_writer::write_state_array_type)
17351         (write_state_gc_used): ditto, renaming to...
17352         (state_writer::write_state_gc_used)
17353         (write_state_common_type_content): ditto, renaming to...
17354         (state_writer::write_state_common_type_content)
17355         (write_state_type): ditto, renaming to...
17356         (state_writer::write_state_type)
17357         (write_state_pair_list): ditto, renaming to...
17358         (state_writer::write_state_pair_list)
17359         (write_state_pair): ditto, renaming to...
17360         (state_writer::write_state_pair)
17361         (write_state_typedefs): ditto, renaming to...
17362         (state_writer::write_state_typedefs)
17363         (write_state_structures): ditto, renaming to...
17364         (state_writer::write_state_structures)
17365         (write_state_param_structs): ditto, renaming to...
17366         (state_writer::write_state_param_structs)
17367         (write_state_variables): ditto, renaming to...
17368         (state_writer::write_state_variables)
17369         (write_state_srcdir): ditto, renaming to...
17370         (state_writer::write_state_srcdir)
17371         (write_state_files_list): ditto, renaming to...
17372         (state_writer::write_state_files_list)
17373         (write_state_languages): ditto, renaming to...
17374         (state_writer::write_state_languages)
17375         (write_state): create a state_writer instance and use it when
17376         writing out the state file
17378 2013-05-17  Mike Stump  <mikestump@comcast.net>
17380         PR rtl-optimization/57304
17381         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
17382         accessing DF_REF_REAL_LOC.
17384 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17386         PR rtl-optimization/57281
17387         PR rtl-optimization/57300
17388         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
17389         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
17390         what the other splitter did if the registers are dead.
17392 2013-05-17  Richard Biener  <rguenther@suse.de>
17394         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
17395         MEM_REF offsets.
17397 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17399         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
17400         linking.
17402 2013-05-17  Marek Polacek  <polacek@redhat.com>
17404         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
17405         length when doing non-zero store of storing '\0' to '\0'.
17407 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17409         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
17410         vect_external_def oprnd1 with loop_vinfo, try to emit
17411         optional cast, negation and and stmts on the loop preheader
17412         edge instead of into the pattern def seq.
17414         PR tree-optimization/57051
17415         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
17416         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
17418 2013-05-16  Nick Clifton  <nickc@redhat.com>
17420         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
17421         (rl78_is_naked_func): New function.
17422         (rl78_expand_prologue): Skip prologue generation for naked functions.
17423         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
17424         * doc/extend.texi (naked): Add RL78 to the list of processors
17425         that supports this attribute.
17427 2013-05-16  Jeff Law  <law@redhat.com>
17429         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
17431 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17433         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
17434         cache parameters using detect_caches_amd also for CYRIX,
17435         NSC and TM2 signatures.
17437 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17438             Dzianis Kahanovich  <mahatma@eu.by>
17440         PR target/45359
17441         PR target/46396
17442         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17443         VIA/Centaur processors and determine their cache parameters
17444         using detect_caches_amd.
17446 2013-05-16  Teresa Johnson  <tejohnson@google.com>
17448         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
17449         (rtl_verify_edges): New function.
17450         (rtl_verify_bb_insns): Ditto.
17451         (rtl_verify_bb_pointers): Ditto.
17452         (rtl_verify_bb_insn_chain): Ditto.
17453         (rtl_verify_fallthru): Ditto.
17454         (rtl_verify_bb_layout): Ditto.
17455         (rtl_verify_flow_info_1): Outline checks into new functions.
17456         (rtl_verify_flow_info): Ditto.
17458 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
17460         * cfghooks.c (copy_bbs): Add update_dominance argument.
17461         * cfghooks.h (copy_bbs): Update prototype.
17462         * tree-cfg.c (gimple_duplicate_sese_region):
17463         Add update_dominance argument.
17464         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
17465         * tree-ssa-loop-ch.c (copy_loop_headers): Update
17466         gimple_duplicate_sese_region call.
17467         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
17468         Update copy_bbs call.
17469         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
17470         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
17472 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17474         * tree-vectorizer.h (NUM_PATTERNS): Increment.
17475         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17476         vect_recog_rotate_pattern.
17477         (vect_recog_rotate_pattern): New function.
17479 2013-05-16  Jason Merrill  <jason@redhat.com>
17481         * Makefile.in (LLINKER): New variable.
17482         (mostlyclean): Remove link mutex.
17483         * configure.ac: Handle --enable-link-mutex.
17484         * lock-and-run.sh: New script.
17486 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17488         PR target/19599
17489         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17490         for NULL decl.
17492 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17494         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17496 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17498         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17499         * config/arm/arm.c (next_consecutive_mem): New function.
17500         (gen_movmem_ldrd_strd): Likewise.
17501         * config/arm/arm.md (movmemqi): Update condition and code.
17502         (unaligned_loaddi, unaligned_storedi): New patterns.
17504 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17506         * config.gcc: Obsolete *-*-solaris2.9*.
17507         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17509 2013-05-16  Richard Biener  <rguenther@suse.de>
17511         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17512         earlier, after GRAPHITE transforms and IV canonicalization.
17514 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17516         * omp-low.c (extract_omp_for_data): For collapsed loops,
17517         if at least one of the loops is known at compile time to
17518         iterate zero times, set count to 0.
17519         (expand_omp_regimplify_p): New function.
17520         (expand_omp_for_generic): For collapsed loops, if at least
17521         one of the loops isn't known to iterate at least once,
17522         add runtime check with setting count to 0.
17523         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17524         For unsigned types if it isn't known at compile time that
17525         the loop will iterate at least once, add runtime check to bypass
17526         the whole loop if initial condition isn't true.
17528 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17530         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17532 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17534         PR middle-end/57286
17535         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17536         transformations to avoid an infinite loop.
17538 2013-05-16  Marek Polacek  <polacek@redhat.com>
17540         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17542 2013-05-15  Leif Ekblad  <leif@rdos.net>
17544         * config/i386/i386.c (ix86_decompose_address): Use
17545         DEFAULT_TLS_SEG_REG to access TLS segment register.
17546         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17547         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17548         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17550 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17552         PR target/57260
17553         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17554         sibling calls to functions that would normally be lazily bound,
17555         unless $gp is call-clobbered.
17557 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17559         * config/i386/i386.c (ix86_option_override_internal): Update
17560         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17561         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17562         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17563         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17564         of TARGET_3DNOW.
17565         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17567 2013-05-15  Andreas Schwab  <schwab@suse.de>
17569         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17570         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17571         third operand.
17573 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17575         * loop-unroll.c (report_unroll_peel): Check decision before
17576         emitting unroll/peel message.
17578 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17580         * function.h (has_bb_partition): New rtl_data flag.
17581         (bb_reorder_complete): Ditto.
17582         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17583         instead of flag_reorder_blocks_and_partition.
17584         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17585         with some enhancements.
17586         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17587         * bb-reorder.c (connect_traces): Check for has_bb_partition
17588         instead of flag_reorder_blocks_and_partition.
17589         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17590         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17591         verify_hot_cold_block_grouping.
17592         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17594 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17596         PR target/19599
17597         * config/arm/predicates.md (call_insn_operand): New predicate.
17598         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17599         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17600         if insn is not a tail call.
17601         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17602         registers.
17603         * config/arm/arm.h (enum reg_class): New caller save register class.
17604         (REG_CLASS_NAMES): Likewise.
17605         (REG_CLASS_CONTENTS): Likewise.
17606         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17607         without decls.
17609 2013-05-15  Richard Biener  <rguenther@suse.de>
17611         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17612         of MSG_OPTIMIZED_LOCATIONS.
17613         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17614         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17615         message.
17616         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17617         of MSG_OPTIMIZED_LOCATIONS.
17618         (execute_vect_slp): Likewise.
17619         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17620         (vect_create_cond_for_alias_checks): Likewise.
17621         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17622         (vect_recog_widen_mult_pattern): Likewise.
17623         (vect_recog_widen_sum_pattern): Likewise.
17624         (vect_recog_over_widening_pattern): Likewise.
17625         (vect_recog_widen_shift_pattern): Likewise.
17626         (vect_recog_vector_vector_shift_pattern): Likewise.
17627         (vect_recog_divmod_pattern): Likewise.
17628         (vect_recog_mixed_size_cond_pattern): Likewise.
17629         (vect_recog_bool_pattern): Likewise.
17630         (vect_pattern_recog_1): Likewise.
17632 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17634         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17635         non-functions to builtin_unreachable.
17636         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17637         correct when new direct edges were discovered.
17639 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17641         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17642         header, print symbol order instead of node uid, print more information
17643         about indirect edge targets.
17644         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17645         uids.
17646         (ipa_make_edge_direct_to_target): Likewise.
17647         (remove_described_reference): Likewise.
17648         (propagate_controlled_uses): Likewise.
17649         (ipa_print_node_params): Also print symbol order.
17650         (ipcp_transform_function): Print symbol order instead of node uids.
17651         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17652         (cgraph_get_create_real_symbol_node): Likewise.
17653         * ipa-cp.c (print_lattice): Likewise.
17654         (print_all_lattices): Likewise.
17655         (determine_versionability): Likewise.
17656         (initialize_node_lattices): Likewise.
17657         (estimate_local_effects): Likewise.
17658         (update_profiling_info): Likewise.
17659         (create_specialized_node): Likewise.
17660         (perhaps_add_new_callers): Likewise.
17661         (decide_about_value): Likewise.
17662         (decide_whether_version_node): Likewise.
17663         (identify_dead_nodes): Likewise.
17664         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17665         (dump_inline_summary): Likewise.
17666         (estimate_node_size_and_time): Likewise.
17667         (inline_analyze_function): Likewise.
17668         * ipa-inline.c (report_inline_failed_reason): Likewise.
17669         (want_early_inline_function_p): Likewise.
17670         (edge_badness): Likewise.
17671         (update_edge_key): Likewise.
17672         (inline_small_functions): Likewise.  Add dumping of order to two other
17673         dumps.
17674         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17675         instead of node uids.
17676         (propagate_pure_const): Likewise.
17677         (propagate_pure_const): Likewise.
17678         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17679         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17680         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17681         of node uids.
17682         * tree-pretty-print.c (dump_function_header): Likewise.
17683         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17684         Print symbol order instead of node uids.
17686 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17688         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17689         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17691 2013-05-15  Richard Biener  <rguenther@suse.de>
17693         PR tree-optimization/57275
17694         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17695         return value for fail to do runtime alias checks for gather loads.
17697 2013-05-15  Jan Hubicka  <jh@suse.cz>
17699         PR lto/57038
17700         PR lto/47375
17701         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17702         weakrefs are not external.
17703         (lto_symtab_merge_decls): Fix thinko when dealing with
17704         non-lto_symtab decls.
17705         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17706         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17707         * varpool.c (dump_varpool_node): Dump more flags.
17709 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17711         * config/i386/i386.c (processor_alias_table): Add instruction
17712         FSGSBASE for AMD bdver3 architecture.
17714 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17716         * tree.c (warn_deprecated_use): Print file:line using locus color.
17717         * diagnostic.c (diagnostic_report_current_module): Print file:line
17718         and file:line:column using locus color.
17720 2013-05-14  Mike Stump  <mikestump@comcast.net>
17722         * gdbinit.in: Add __null.
17724 2013-05-14  Mike Stump  <mikestump@comcast.net>
17726         * recog.h: Rename struct recog_data to Recog_data.
17727         * recog.c: Likewise.
17728         * reload.c (can_reload_into): Likewise.
17729         * config/picochip/picochip.c: Likewise.
17731 2013-05-14  Mike Stump  <mikestump@comcast.net>
17733         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17735 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17737         * resource.h (struct resources): Remove unch_memory member.
17738         (CLEAR_RESOURCE): Don't clear unch_memory.
17739         * resource.c (mark_referenced_resources): Don't set it.
17740         (mark_set_resources): Likewise.
17741         (mark_target_live_regs): Don't clear it.
17742         (init_resource_info): Likewise.
17743         * reorg.c (resource_conflicts_p): Don't compare it.
17744         (redundant_insn): Don't set it.
17746         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17747         Remove prototypes.
17748         * emit-rtl.c (next_label): Remove unused function.
17749         (skip_consecutive_labels, link_cc0_insns): Move to ...
17750         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17751         only place where these functions are used, and make them static.
17753 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17755         * fold-const.c (fold_negate_expr): Handle vectors.
17756         (fold_truth_not_expr): Make it static.
17757         (fold_invert_truthvalue): New static function.
17758         (invert_truthvalue_loc): Handle vectors. Do not call
17759         fold_truth_not_expr directly.
17760         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17761         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17762         (fold_comparison): Handle vector constants.
17763         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17764         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17765         * tree.h (fold_truth_not_expr): Remove declaration.
17767 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17769         * config/aarch64/aarch64-simd.md
17770         (aarch64_vcond_internal<mode>): Rename to...
17771         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17772         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17773         float modes. Clarify all iterator modes.
17774         (vcond<mode><mode>): Use new name for vcond expanders.
17775         (vcond<v_cmp_result><mode>): Likewise.
17776         (vcondu<mode><mode>: Likewise.
17777         * config/aarch64/iterators.md (VDQF_COND): New.
17779 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17781         PR bootstrap/57266
17782         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17783         variable for the shift amount. Check that we shift by non-negative
17784         amounts.
17786 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17788         PR target/42017
17789         * config/arm/arm.h (EPILOGUE_USES): Only return true
17790         for LR_REGNUM after epilogue_completed.
17792 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17794         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17795         is error_mark_node.
17797 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17799         PR target/57261
17800         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17801         and Solaris 11+/x86 with gld.
17802         * configure: Regenerate.
17804 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17806         * expmed.c (expand_shift_1): Canonicalize rotates by
17807         constant bitsize / 2 to bitsize - 1.
17808         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17809         case ROTATERT>: Likewise.
17811         Revert:
17812         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17814         * config/i386/i386.md (rotateinv): New code attr.
17815         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17816         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17817         roll $31, %eax, etc.
17819 2013-05-14  Richard Biener  <rguenther@suse.de>
17821         PR middle-end/57235
17822         * tree-eh.c (sink_clobbers): Give up for successors with
17823         multiple predecessors and no virtual uses.
17825 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17827         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17828         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17830 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17832         PR middle-end/57251
17833         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17834         the case when both op0 and op1 have VOIDmode.
17836 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17838         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17839         in multiply-accumulate mode.
17841 2013-05-13  Guozhi Wei  <carrot@google.com>
17843         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17845 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17847         PR target/56975
17848         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17849         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17850         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17851         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17852         (ix86_expand_prologue): Likewise.
17853         (ix86_expand_split_stack_prologue): Likewise.
17854         (legitimate_pic_address_disp_p): Likewise.
17855         (legitimize_pic_address): Likewise.
17856         (legitimize_tls_address): Likewise.
17857         (legitimize_pe_coff_symbol): Likewise.
17858         (output_pic_addr_const): Likewise.
17859         (construct_plt_address): Likewise.
17860         (ix86_expand_call): Likewise.
17861         (x86_output_mi_thunk): Likewise.
17862         (x86_function_profiler): Likewise.
17864 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17866         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17867         similar switch cases.
17868         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17869         (aarch64_simd_mov_to_<mode>low): Delete.
17870         (aarch64_simd_mov_to_<mode>high): Delete.
17871         (move_lo_quad_<mode>): Add w<-r alternative.
17872         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17873         (aarch64_simd_mov_from_*): Update type attribute.
17874         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17875         statement.
17877 2013-05-13  Jan Hubicka  <jh@suse.cz>
17879         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17880         * config/i386/i386.c (ix86_compute_frame_layout,
17881         ix86_expand_epilogue, emit_i387_cw_initialization,
17882         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17883         ix86_local_alignment): Fix use of size/speed predicates.
17885 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17887         PR tree-optimization/45216
17888         PR tree-optimization/57157
17889         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17890         the (-Y) & (B - 1) variant if OP is |.
17891         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17892         return shifted.  Use (-op1) & (prec - 1) as other_amount
17893         instead of prec - op1.
17895 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17897         PR middle-end/42371
17898         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17899         (ipa_constant_data): New type.
17900         (ipa_jump_func): Use ipa_constant_data to hold information about
17901         constant jump functions.
17902         (ipa_get_jf_constant): Adjust to jump function type changes.
17903         (ipa_get_jf_constant_rdesc): New function.
17904         (ipa_param_descriptor): New field controlled_uses.
17905         (ipa_get_controlled_uses): New function.
17906         (ipa_set_controlled_uses): Likewise.
17907         * ipa-ref.h (ipa_find_reference): Declare.
17908         * ipa-prop.c (ipa_cst_ref_desc): New type.
17909         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17910         changes.
17911         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17912         New parameter cs.  Adjust all callers.
17913         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17914         (remove_described_reference): New function.
17915         (jfunc_rdesc_usable): Likewise.
17916         (try_make_edge_direct_simple_call): Decrement controlled use count,
17917         attempt to remove reference if it hits zero.
17918         (combine_controlled_uses_counters): New function.
17919         (propagate_controlled_uses): Likewise.
17920         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17921         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17922         (ipa_print_node_params): Print described use counter.
17923         (ipa_write_jump_function): Adjust to jump function type changes.
17924         (ipa_read_jump_function): New parameter CS, pass it to
17925         ipa_set_jf_constant.  Adjust caller.
17926         (ipa_write_node_info): Stream controlled use count
17927         (ipa_read_node_info): Likewise.
17928         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17929         asserting.
17930         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17931         count.  Remove cloning-added reference if it reaches zero.
17932         * ipa-ref.c (ipa_find_reference): New function.
17934 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17936         * config/i386/i386.c (processor_target_table): Modified default
17937         alignment values for AMD BD and BT architectures.
17939 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17941         * tree-vect-generic.c (uniform_vector_p): Move ...
17942         * tree.c (uniform_vector_p): ... here.
17943         * tree.h (uniform_vector_p): Declare it.
17944         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17945         into a scalar.
17947 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17949         PR tree-optimization/57230
17950         * tree-ssa-strlen.c (handle_char_store): Record length for
17951         array store from STRING_CST.
17953         PR tree-optimization/57230
17954         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17955         check.
17957 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17959         * config/epiphany/epiphany.c (epiphany_init): Check size of
17960         NUM_MODES_FOR_MODE_SWITCHING.
17961         (epiphany_expand_prologue):
17962         Remove CONFIG_REGNUM initial value handling code.
17963         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17964         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17965         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17966         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17967         Don't return 1 for FP_MODE_NONE.
17968         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17969         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17970         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17971         * config/epiphany/epiphany.md (save_config): New pattern.
17973 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17975         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17977 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17979         * config/i386/i386.md (memory): Handle sseishft1.
17980         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17981         (*vec_extractv2di_1): Ditto.
17983 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17985         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17986         saved registers.
17988 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17990         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17991         Add mthumb/march=armv7-a multilib.
17992         Add mthumb/march=armv7-r multilib.
17993         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17995 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17997         * config/v850/t-rtems: Add more multilibs.
17999 2013-05-10  Richard Biener  <rguenther@suse.de>
18001         PR tree-optimization/57214
18002         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
18003         not propagate from SSA names that occur in abnormal PHI nodes.
18005 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
18007         * stor-layout.c (element_precision): New function.
18008         * machmode.h (element_precision): Declare it.
18009         * tree.c (build_minus_one_cst): New function.
18010         (element_precision): Likewise.
18011         * tree.h (build_minus_one_cst): Declare new function.
18012         (element_precision): Likewise.
18013         * fold-const.c (operand_equal_p): Use element_precision.
18014         (fold_binary_loc): Handle vector types.
18015         * convert.c (convert_to_integer): Use element_precision.
18016         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
18017         separately.
18019 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
18021         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
18022         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
18023         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
18024         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
18025         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
18026         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
18027         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
18028         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
18029         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
18030         (Uuw8): New constraints.
18031         (Usb4): Move into alphabetical order.
18032         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
18033         (sd8_operand, ub8_operand, uw8_operand): New predicates.
18034         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
18035         previously unnamed patterns.
18036         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
18037         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
18038         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
18039         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
18040         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
18041         of specific lengths.
18043 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
18045         * config/i386/i386.md (rotateinv): New code attr.
18046         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
18047         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
18048         roll $31, %eax, etc.
18050         PR tree-optimization/45216
18051         PR tree-optimization/57157
18052         * tree-ssa-forwprop.c (simplify_rotate): New function.
18053         (ssa_forward_propagate_and_combine): Call it.
18055 2013-05-10  Richard Biener  <rguenther@suse.de>
18057         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
18058         disable peeling when we version for aliasing.
18059         (vector_alignment_reachable_p): Honor explicit user alignment.
18060         (vect_supportable_dr_alignment): Likewise.
18061         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
18062         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
18063         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
18064         then peeling to arrange for the cost-model check to come first.
18066 2013-05-10  Alan Modra  <amodra@gmail.com>
18068         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
18069         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
18070         * configure: Regenerate.
18072 2013-05-10  Alan Modra  <amodra@gmail.com>
18074         PR target/55033
18075         * varasm.c (default_elf_select_section): Move !DECL_P check..
18076         (get_named_section): ..to here before calling get_section_name.
18077         Adjust assertion.
18078         (default_section_type_flags): Add DECL_P check.
18079         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
18080         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
18082 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
18084         * config/epiphany/epiphany.c (epiphany_expand_prologue):
18085         When using gen_stack_adjust_str with a register offset, add a
18086         REG_FRAME_RELATED_EXPR note.
18088 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
18090         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
18091         (*vec_extractv4si_zext_mem): Ditto.
18092         (*vec_extractv2di): Add 0->x and x->x alternatives.
18093         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
18094         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
18096 2013-05-09  Jason Merrill  <jason@redhat.com>
18098         N3639 C++1y VLA support
18099         * gimplify.c (gimplify_vla_decl): Don't touch an existing
18100         DECL_VALUE_EXPR.
18102         * tree.c (build_constructor_va): New.
18103         * tree.h: Declare it.
18105 2013-05-09  Martin Jambor  <mjambor@suse.cz>
18107         PR lto/57084
18108         * gimple-fold.c (canonicalize_constructor_val): Call
18109         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
18111 2013-05-09  Jan Hubicka  <jh@suse.cz>
18112             Richard Biener  <rguenther@suse.de>
18114         PR lto/54095
18115         * symtab.c (symtab_make_decl_local): Do not add private names.
18117 2013-05-09  Jan Hubicka  <jh@suse.cz>
18119         PR lto/54095
18120         * symtab.c (insert_to_assembler_name_hash): Handle clones.
18121         (unlink_from_assembler_name_hash): Likewise.
18122         (symtab_prevail_in_asm_name_hash, symtab_register_node,
18123         symtab_unregister_node, symtab_initialize_asm_name_hash,
18124         change_decl_assembler_name): Update.
18126 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
18128         * config/aarch64/aarch64.md: New movtf split.
18129         (*movtf_aarch64): Update.
18130         (aarch64_movdi_tilow): Handle TF modes and rename to
18131         aarch64_movdi_<mode>low.
18132         (aarch64_movdi_tihigh): Handle TF modes and rename to
18133         aarch64_movdi_<mode>high
18134         (aarch64_movtihigh_di): Handle TF modes and rename to
18135         aarch64_mov<mode>high_di
18136         (aarch64_movtilow_di): Handle TF modes and rename to
18137         aarch64_mov<mode>low_di
18138         (aarch64_movtilow_tilow): Remove spurious whitespace.
18139         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
18140         splits.
18141         (aarch64_print_operand): Update.
18143 2013-05-09  Alan Modra  <amodra@gmail.com>
18145         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
18146         powerpc64le.
18147         * configure: Regenerate.
18149 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
18151         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
18152         splitter preparation statements.
18153         * config/i386/sse.md (*vec_extract* splitters): Ditto.
18154         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
18155         adjust_address_nv.
18157 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18159         * gimple-ssa-strength-reduction.c (count_candidates): Change
18160         return value to int.
18161         (analyze_candidates_and_replace): Change type of length to int.
18163 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
18165         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
18166         (*vec_extract<mode>): Use VI12_128 mode iterator.
18167         (*vec_extract<mode>_mem): Ditto.
18168         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
18169         attribute.
18171 2013-05-08  Diego Novillo  <dnovillo@google.com>
18173         PR bootstrap/54659
18175         Revert:
18176         2012-08-17  Diego Novillo  <dnovillo@google.com>
18178         PR bootstrap/54281
18179         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
18180         * config.in: Regenerate.
18181         * configure: Regenerate.
18182         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
18184 2013-05-08  Jan Hubicka  <jh@suse.cz>
18186         PR lto/54095
18187         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
18188         * cgraph.h (symtab_node_base): Add unique_name.
18189         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
18190         input_overwrite_node, input_varpool_node): Stream unique_name.
18191         * cgraphclones.c (cgraph_create_virtual_clone,
18192         cgraph_function_versioning): Set unique_name.
18193         * ipa.c (function_and_variable_visibility): Set unique_name.
18195 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18197         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
18198         (alloc_cand_and_find_basis): Restrict conditional candidate
18199         processing to CAND_MULTs.
18201 2013-05-08  Jan Hubicka  <jh@suse.cz>
18203         PR lto/54095
18204         lto-symtab.c (lto_symtab_symbol_p): New function.
18205         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
18206         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
18207         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
18208         Skip static symbols.
18210 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
18212         PR tree-optimization/57200
18213         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
18214         Only call inform if the preceding warning_at returns true.
18216 2013-05-07  Han Shen  <shenhan@google.com>
18218         * cfgexpand.c (record_or_union_type_has_array_p): New function.
18219         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
18220         * common.opt (fstack-protector-strong): New option.
18221         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
18222         * doc/invoke.texi (Optimization Options): Document
18223         "-fstack-protector-strong".
18224         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
18226 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18228         * config/mips/mips.c (mips_machine_reorg2): Return 0.
18230 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
18232         * ira.c (update_equiv_regs): Add insn having equiv memory even if
18233         it is not lhs of the insn.
18234         (setup_reg_equiv): Remove insn having equiv memory which it is not
18235         lhs of the insn.
18236         * lra-constraints.c (process_address): Try to improve generation
18237         code for address base + disp.
18238         (lra_constraints): Make correct the code for checking insn setting
18239         up backward equivalence.  Remove insn only if it is in the init
18240         insn list.
18241         * lra-eliminations.c (update_reg_eliminate): Change return value.
18242         (lra_eliminate): Use the result.
18244 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
18246         * config/i386/sse.md (ssescalarnummask): New mode attribute.
18247         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
18248         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
18249         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
18250         register target operands.
18251         (*vec_extractv8hi_sse2): New pattern.
18252         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
18253         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
18254         (*vec_extract<mode>_mem): New insn and split pattern.
18256 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
18258         * config/arm/arm.c (arm_asan_shadow_offset): New function.
18259         (TARGET_ASAN_SHADOW_OFFSET): Define.
18260         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
18261         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
18263 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18265         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
18266         (incr_vec_index): Return -1 if increment not found.
18267         (create_add_on_incoming_edge): Assert if increment not found.
18268         (record_increment): Limit number of increments recorded.
18269         (all_phi_incrs_profitable): Return false if an increment not found.
18270         (replace_profitable_candidates): Don't process increments that were
18271         not recorded.
18272         (analyze_candidates_and_replace): Limit size of incr_vec.
18274 2013-05-07  Richard Biener  <rguenther@suse.de>
18276         * calls.c (special_function_p): setjmp-like functions are leaf.
18277         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
18278         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
18280 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
18282         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
18283         (aarch64_simd_mov<mode>): New expander.
18284         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
18285         (aarch64_simd_mov_to_<mode>high): Likewise.
18286         (aarch64_simd_mov_from_<mode>low): Likewise.
18287         (aarch64_simd_mov_from_<mode>high): Likewise.
18288         (aarch64_dup_lane<mode>): Update.
18289         (aarch64_dup_lanedi): New instruction pattern.
18290         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
18291         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
18293 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18295         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
18296         (replace_mult_candidate): Remove unnecessary argument; remove
18297         unnecessary parameter from call to introduce_cast_before_cand.
18298         (replace_unconditional_candidate): Remove unnecessary parameter
18299         from call to replace_mult_candidate.
18300         (replace_conditional_candidate): Likewise.
18301         (insert_initializers): Use make_temp_ssa_name.
18302         (introduce_cast_before_cand): Remove unnecessary argument; use
18303         make_temp_ssa_name.
18304         (replace_one_candidate): Remove unnecessary argument; remove
18305         unnecessary parameter from calls to introduce_cast_before_cand.
18306         (replace_profitable_candidates): Remove unnecessary parameters
18307         from calls to replace_one_candidate.
18309 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18311         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
18312         phi def as possibly hiding a basis for a CAND_ADD whose operands
18313         have been commuted in the analysis.
18314         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
18316 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18318         * config/aarch64/aarch64.md
18319         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
18320         shift value between 0-4.
18322 2013-05-07  Richard Biener  <rguenther@suse.de>
18324         * double-int.h (rshift): New overload.
18325         * double-int.c (rshift): New function.
18326         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
18327         (create_reference_ops_from_ref): Remove.
18328         (vn_reference_insert): Use shared ops for constructing the
18329         reference and copy it.
18331 2013-05-07  Richard Biener  <rguenther@suse.de>
18333         PR middle-end/57190
18334         * tree-eh.c (sink_clobbers): Properly propagate
18335         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
18337 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
18339         PR tree-optimization/57149
18340         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
18341         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
18342         collect_phi_def_edges, execute_late_warn_uninitialized): Use
18343         uninit_undefined_value_p instead of ssa_undefined_value_p.
18345         PR debug/57184
18346         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
18347         for modifier == EXPAND_INITIALIZER.
18349 2013-05-07  Anton Blanchard  <anton@samba.org>
18351         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
18352         for powerpc64 little endian.
18353         * configure: Regenerate.
18355 2013-05-06  Graham Stott  <grahams@btinternet.com>
18357         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
18358         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
18359         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
18360         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
18362 2013-05-06  Graham Stott  <grahams@btinternet.com>
18364         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
18365         codes which allow non-lvalues.
18367 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
18369         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
18370         components are all 1s.
18371         (integer_minus_onep): New function.
18372         * tree.h (integer_minus_onep): Declare it.
18373         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
18374         integer_minus_onep instead of integer_all_onesp.
18376 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18378         PR target/52933
18379         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
18380         variations of these patterns.
18382 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18384         * config/i386/i386.md (isa): Add x64_sse4 member.
18385         (enabled): Handle x64_sse4.
18386         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
18387         instruction for 64bit SSE4_1 targets.  Update insn attributes.
18388         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
18389         instruction for SSE4_1 targets.  Update insn attributes.
18390         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18391         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
18392         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
18393         const_1 selector.
18394         (*vec_extractv4si): Rename from *sse4_1_pextrd.
18395         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
18396         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
18398 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18400         PR target/57108
18401         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
18403 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
18405         * final.c (do_assembler_dialects): Don't handle curly braces and
18406         vertical bar escaped by % as dialect delimiters.
18407         (output_asm_insn): Print curly braces and vertical bar if escaped
18408         by % and ASSEMBLER_DIALECT defined.
18409         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
18410         * doc/tm.texi: Regenerated.
18412 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18414         * config/mips/mips.c: Include tree-pass.h.
18415         (mips_reorg): Split in pre- and post-dbr_schedule parts.
18416         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
18417         (pass_mips_machine_reorg2): New machine specific pass.
18418         (insert_pass_mips_machine_reorg2): New pass plugin definition.
18419         (mips_option_override): Register the new pass.
18420         * rtl.h (cleanup_barriers): Remove prototype.
18421         (dbr_schedule): Likewise.
18422         * jump.c (cleanup_barriers): Make static.
18423         * reorg.c (dbr_schedule): Likewise.
18425 2013-05-06  Richard Biener  <rguenther@suse.de>
18427         PR tree-optimization/57185
18428         * tree-parloops.c (add_field_for_reduction): Handle anonymous
18429         SSA names properly.
18431 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18433         PR target/57106
18434         * config/i386/i386.c (add_parameter_dependencies): Add dependence
18435         between "first_arg" and "insn", not "last" and "insn".
18437 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18439         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
18440         (find_candidates_in_block): Re-enable slsr_process_phi.
18441         (create_phi_basis): Fix double counting of candidate adjustment.
18443 2013-05-06  Richard Biener  <rguenther@suse.de>
18445         PR middle-end/57147
18446         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
18447         the edge is also fallthru, preserve it and just clear the
18448         abnormal flag.
18449         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
18450         also complex, preserve that and just clear the fallthru flag.
18451         * tree-inline.c (update_ssa_across_abnormal_edges): Also
18452         update virtual operands.
18454 2013-05-06  Alan Modra  <amodra@gmail.com>
18456         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
18457         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
18458         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
18459         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
18460         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18461         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
18463 2013-05-06  Alan Modra  <amodra@gmail.com>
18465         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
18466         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
18467         (DEFAULT_ASM_ENDIAN): Define.
18468         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18469         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
18470         Update -K PIC clause from sysv4.h.
18471         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
18472         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
18474 2013-05-06  Alan Modra  <amodra@gmail.com>
18476         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18477         twice for little-endian.
18478         (ashrdi3_no_power, ashrdi3): Support little-endian.
18480 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18482         PR target/55303
18483         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18484         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18485         related expanders.
18486         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18487         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18488         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18489         New predicates.
18491 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18492             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18494         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18495         * config/pa/pa.opt: Make mbig-switch a no-op.
18496         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18497         (CASE_VECTOR_MODE): Always return SImode.
18498         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18499         for the !TARGET_BIG_SWITCH case.
18500         * config/pa/pa-linux.h: Likewise.
18501         * config/pa/pa-openbsd.h: Likewise.
18502         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18503         * config/pa/pa.md (short_jump): Remove define_insn.
18504         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18505         (casesi0): Remove define_insn.
18506         (type): Remove btable_branch.
18507         (pa_combine_type): Likewise.
18508         (in_nullified_branch_delay): Likewise.
18509         (in_call_delay): Likewise.
18510         (define_delay): Likewise.
18511         (define_insn_reservation "Z3"): Likewise.
18512         (define_insn_reservation "Z4"): Likewise.
18513         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18514         (pa_adjust_insn_length): Remove adjustment for btable branches.
18515         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18516         and mno-big-switch
18518 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18520         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18521         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18522         Add m->r,x alternatives.
18523         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18524         splitters using SWI48x mode iterator.
18525         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18526         TARGET_64BIT.  Add m->x alternative.
18527         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18528         Add o->x alternative.  Enable for TARGET_SSE.
18529         (sse_storeq): Remove expander.
18530         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18531         with memory input operand.
18532         (*vec_extractv2di_1 splitter): New.
18533         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18534         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18536 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18538         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18539         (INT_LOWPART): Delete.
18540         (extract_MB): Adjust.
18541         (extract_ME): Adjust.
18542         (print_operand): Adjust.
18544 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18546         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18547         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18548         (reg_or_logical_cint_operand, easy_fp_constant,
18549         logical_const_operand): Delete "CONST_DOUBLE" case.
18550         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18551         "HOST_BITS_PER_WIDE_INT == 64" test.
18552         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18553         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18554         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18555         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18556         test.
18557         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18558         CONST_DOUBLE DImode/VOIDmode case.
18559         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18560         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18561         CONST_DOUBLE VOIDmode case.
18562         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18563         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18564         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18565         Delete CONST_DOUBLE case.
18566         (splitters for mov FMOVE64 const_double): Delete
18567         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18568         "HOST_BITS_PER_WIDE_INT >= 64" test.
18569         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18570         case.
18571         (mov DI const_double): Delete.
18573 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18575         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18576         on op shows all bits zero in mode of a lowpart subreg, return zero.
18578 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18580         PR target/57150
18581         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18582         to save TFmode registers and DImode to save TImode registers for
18583         caller save operations.
18584         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18585         mark being partially clobbered since they only use the first
18586         double word.
18588         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18589         and TDmode only use the upper 64-bits of each VSX register.
18591 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18593         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18594         (find_candidates_in_block): Disable slsr_process_phi.
18596 2013-05-03  Guozhi Wei  <carrot@google.com>
18598         * coverage.c (coverage_obj_init): Move the construction of gcov
18599         constructor to ...
18600         (build_init_ctor): ... here.
18602 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18604         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18605         (slsr_cand_d): Redefine def_phi.
18606         (stride_status, phi_adjust_status, count_phis_status): New enums.
18607         (find_phi_def): New.
18608         (find_basis_for_base_expr): New.
18609         (find_basis_for_candidate): Handle hidden bases.
18610         (alloc_cand_and_find_basis): Handle phi candidates.
18611         (slsr_process_phi): New.
18612         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18613         (create_mul_imm_cand): Likewise.
18614         (create_add_ssa_cand): Exclude phi base candidates.
18615         (create_add_imm_cand): Likewise.
18616         (slsr_process_cast): Likewise.
18617         (slsr_process_copy): Likewise.
18618         (find_candidates_in_block): Handle phi candidates.
18619         (dump_candidate): Likewise.
18620         (unconditional_cands): Delete.
18621         (unconditional_cands_with_known_stride_p): Delete.
18622         (phi_dependent_cand_p): New.
18623         (cand_increment): Handle phi-dependent candidates.
18624         (replace_dependent): Delete.
18625         (replace_mult_candidate): New.
18626         (replace_unconditional_candidate): New.
18627         (incr_vec_index): Move to avoid forward reference.
18628         (create_add_on_incoming_edge): New.
18629         (create_phi_basis): New.
18630         (replace_dependents): Delete.
18631         (replace_conditional_candidate): New.
18632         (phi_add_costs): New.
18633         (replace_uncond_cands_and_profitable_phis): New.
18634         (record_increment): Handle phi adjustments.
18635         (record_phi_increments): New.
18636         (record_increments): Handle phi adjustments.
18637         (phi_incr_cost): New.
18638         (lowest_cost_path): Handle phis.
18639         (total_savings): Likewise.
18640         (analyze_increments): Likewise.
18641         (ncd_with_phi): New.
18642         (ncd_of_cand_and_phis): New.
18643         (nearest_common_dominator_for_cands): Handle phi increments.
18644         (all_phi_incrs_profitable): New.
18645         (replace_profitable_candidates): Handle phi-dependent candidates.
18646         (analyze_candidates_and_replace): Likewise.
18648 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18650         PR bootstrap/57154
18651         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18652         do not exceed REG_BR_PROB_BASE.
18654 2013-05-03  Jeff Law  <law@redhat.com>
18656         PR tree-optimization/57144
18657         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18658         operand of the condition will bit into the new type when eliminating
18659         a cast feeding a condition.
18661 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18663         PR rtl-optimization/57130
18664         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18665         of COMPARE as in_code to the recursive call if needed.
18667 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18669         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18670         (enabled): Handle new members.
18671         * config/i386/sse.md (*vec_concatv2si): Merge from
18672         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18673         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18675 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18677         PR tree-optimization/57027
18678         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18679         for fnms opportunity, check we got the prerequisite kind
18680         of tree / gimple before using accessor functions.
18682 2013-05-03  Richard Biener  <rguenther@suse.de>
18684         * double-int.h (lshift): New overload without precision
18685         and arith argument.
18686         (operator *=, operator +=, operator -=): Move ...
18687         * double-int.c (operator *=, operator +=, operator -=): ... here
18688         and implement more efficiently.
18689         (mul_double_with_sign): Remove.
18690         (lshift_double): Adjust to take unsinged shift argument, push
18691         dispatching code to callers.
18692         (mul_double_wide_with_sign): Add early out for callers that
18693         are not interested in high parts or overflow.
18694         (lshift): New function.
18695         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18696         dispatch code here.
18697         (lrotate, rrotate): Use logical shifts.
18698         * expr.c (get_inner_reference): Use lshift.
18699         * fixed-value.c (do_fixed_divide): Likewise.
18700         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18701         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18702         (indirect_refs_may_alias_p): Likewise.
18703         (stmt_kills_ref_p_1): Likewise.
18705 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18707         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18709 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18711         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18712         scalar form of FABD instruction.
18714 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18716         * lra-constraints.c (process_alt_operands): Add checking alt
18717         number to choose the best alternative.
18719 2013-05-02  Richard Biener  <rguenther@suse.de>
18721         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18722         bitmap and its handling.
18723         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18725 2013-05-02  Richard Biener  <rguenther@suse.de>
18727         PR middle-end/57140
18728         * tree-inline.c (copy_loops): Properly handle removed loops.
18729         (copy_cfg_body): Mark destination loops for fixup if source
18730         loops needed fixup.
18732 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18734         PR target/56732
18735         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18736         generating simple_return for naked functions.
18738 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18740         PR middle-end/56988
18741         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18742         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18743         flags match.
18744         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18745         ipa_agg_replacement_value structures.
18746         (known_aggs_to_agg_replacement_list): Likewise.
18747         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18748         (read_agg_replacement_chain): Likewise.
18749         (ipcp_transform_function): Also check that by_ref flags match.
18751 2013-05-02  Richard Biener  <rguenther@suse.de>
18753         * graphds.h (struct graph): Add obstack member.
18754         * graphds.c (new_graph): Initialize obstack and allocate
18755         vertices from it.
18756         (add_edge): Allocate edge from the obstack.
18757         (free_graph): Free the obstack instead of all edges and vertices.
18759 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18761         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18762         divides.
18763         * cfg.c (update_bb_profile_for_threading): Ditto.
18764         * tree-inline.c (copy_bb): Ditto.
18765         (copy_edges_for_bb): Ditto.
18766         (initialize_cfun): Ditto.
18767         (copy_cfg_body): Ditto.
18768         (expand_call_inline): Ditto.
18769         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18770         (estimate_node_size_and_time): Ditto.
18771         (inline_merge_summary): Ditto.
18772         * cgraphclones.c (cgraph_clone_edge): Ditto.
18773         (cgraph_clone_node): Ditto.
18774         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18775         (compute_trg_info): Ditto.
18777 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18779         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18780         S reg when fp attribute set.
18781         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18783 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18785         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18786         New pattern.
18787         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18788         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18789         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18791 2013-05-02  Richard Biener  <rguenther@suse.de>
18793         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18794         (struct instantiate_cache_entry): New type.
18795         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18796         (struct instantiate_cache_type): New type.
18797         (set_instantiated_value, get_instantiated_value): Remove.
18798         (get_instantiated_value_entry): New function.
18799         (instantiate_scev_name): Use the new cache and adjust.
18800         (instantiate_scev_poly): Adjust.
18801         (instantiate_scev_binary): Likewise.
18802         (instantiate_array_ref): Likewise.
18803         (instantiate_scev_convert): Likewise.
18804         (instantiate_scev_not): Likewise.
18805         (instantiate_scev_3): Likewise.
18806         (instantiate_scev_2): Likewise.
18807         (instantiate_scev_r): Likewise.
18808         (instantiate_scev): Likewise.
18809         (resolve_mixers): Likewise.
18811 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18813         PR target/57091
18814         * lra-constraints.c (best_small_class_operands_num): Remove.
18815         (process_alt_operands): Remove small_class_operands_num.  Take
18816         small classes operands into losers and only if the operand is not
18817         matched.  Modify debugging output.
18818         (curr_insn_transform): Remove best_small_class_operands_num.
18819         Print insn name.
18821 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18823         * config/aarch64/aarch64-builtins.c
18824         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18825         * config/aarch64/aarch64-simd-builtins.def
18826         (reduc_splus_): Add new modes.
18827         (reduc_uplus_): New.
18828         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18829         (reduc_uplus_v4sf): Likewise.
18830         (reduc_splus_v4sf): Likewise.
18831         (aarch64_addv<mode>): Likewise.
18832         (reduc_uplus_<mode>): Likewise.
18833         (reduc_splus_<mode>): Likewise.
18834         (aarch64_addvv2di): Likewise.
18835         (reduc_uplus_v2di): Likewise.
18836         (reduc_splus_v2di): Likewise.
18837         (aarch64_addvv2si): Likewise.
18838         (reduc_uplus_v2si): Likewise.
18839         (reduc_splus_v2si): Likewise.
18840         (reduc_<sur>plus_<mode>): New.
18841         (reduc_<sur>plus_v2di): Likewise.
18842         (reduc_<sur>plus_v2si): Likewise.
18843         (reduc_<sur>plus_v4sf): Likewise.
18844         (aarch64_addpv4sf): Likewise.
18845         * config/aarch64/arm_neon.h
18846         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18847         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18848         add UNSPEC_SADDV, UNSPEC_UADDV.
18849         (SUADDV): New.
18850         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18852 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18854         * config/aarch64/arm_neon.h
18855         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18857 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18859         * config/aarch64/aarch64-builtins
18860         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18862 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18864         * config/aarch64/aarch64-simd-builtins.def
18865         (reduc_smax_): New.
18866         (reduc_smin_): Likewise.
18867         (reduc_umax_): Likewise.
18868         (reduc_umin_): Likewise.
18869         (reduc_smax_nan_): Likewise.
18870         (reduc_smin_nan_): Likewise.
18871         (fmax): Remove.
18872         (fmin): Likewise.
18873         (smax): Update for V2SF, V4SF and V2DF modes.
18874         (smin): Likewise.
18875         (smax_nan): New.
18876         (smin_nan): Likewise.
18877         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18878         (<su><maxmin><mode>3): ...This, refactor.
18879         (s<maxmin><mode>3): New.
18880         (<maxmin_uns><mode>3): Likewise.
18881         (reduc_<maxmin_uns>_<mode>): Refactor.
18882         (reduc_<maxmin_uns>_v4sf): Likewise.
18883         (reduc_<maxmin_uns>_v2si): Likewise.
18884         (aarch64_<fmaxmin><mode>: Remove.
18885         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18886         new builtin names.
18887         (vmin<q>_f<32,64>): Likewise.
18888         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18889         (FMAXMIN): New.
18890         (su): Add mappings for smax, smin, umax, umin.
18891         (maxmin): New.
18892         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18893         (FMAXMIN): Rename as...
18894         (FMAXMIN_UNS): ...This.
18895         (maxminv): Remove.
18896         (fmaxminv): Likewise.
18897         (fmaxmin): Likewise.
18898         (maxmin_uns): New.
18899         (maxmin_uns_op): Likewise.
18901 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18903         * config/aarch64/arm_neon.h
18904         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18905         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18906         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18908 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18910         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18911         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18913 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18915         * config/aarch64/aarch64-simd.md
18916         (vcond<mode>_internal): Handle special cases for constant masks.
18917         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18918         (vcondu<mode><mode>): Likewise.
18919         (vcond<v_cmp_result><mode>): New.
18921 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18923         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18924         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18925         * config/aarch64/aarch64-simd-builtins.def
18926         (cmeq): Update to BUILTIN_VALLDI.
18927         (cmgt): Likewise.
18928         (cmge): Likewise.
18929         (cmle): Likewise.
18930         (cmlt): Likewise.
18931         * config/aarch64/arm_neon.h
18932         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18933         to builtins or C as appropriate.
18935 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18937         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18938         (cmgeu): ...This.
18939         (cmhi): Rename to...
18940         (cmgtu): ...This.
18941         * config/aarch64/aarch64-simd.md
18942         (simd_mode): Add SF.
18943         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18944         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18945         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18946         (cstore<mode>_neg): ...This.
18947         * config/aarch64/iterators.md
18948         (VALLF): new.
18949         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18950         (COMPARISONS): New.
18951         (UCOMPARISONS): Likewise.
18952         (optab): Add missing comparisons.
18953         (n_optab): New.
18954         (cmp_1): Likewise.
18955         (cmp_2): Likewise.
18956         (CMP): Likewise.
18957         (cmp): Remove.
18958         (VCMP_S): Likewise.
18959         (VCMP_U): Likewise.
18960         (V_cmp_result): Add DF, SF modes.
18961         (v_cmp_result): Likewise.
18962         (v): Likewise.
18963         (vmtype): Likewise.
18964         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18966 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18968         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18969         define_insn to define_insn_and_split.
18970         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18971         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18972         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18973         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18974         (thumb2_negscc): Likewise.
18976 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18978         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18980 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18982         * config/arm/thumb2.md: Remove trailing whitespaces.
18984 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18986         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18987         Use gen_int_mode rather than GEN_INT.
18989 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18991         * value-prof.c (stream_in_histogram_value): Remove the strayed
18992         debug_gimple_stmt.
18994 2013-04-30  Richard Biener  <rguenther@suse.de>
18996         PR middle-end/57122
18997         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18999 2013-04-30  Richard Biener  <rguenther@suse.de>
19001         PR middle-end/57107
19002         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
19004 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
19006         PR rtl-optimization/56957
19007         PR rtl-optimization/57105
19008         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
19009         variable.  Use just INSN_UID for determining whether an insn
19010         should be only disconnected from the insn stream.
19011         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
19013 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
19015         PR tree-optimization/57104
19016         * tsan.c (instrument_expr): Don't instrument accesses to
19017         DECL_HARD_REGISTER VAR_DECLs.
19019 2013-04-30  Richard Biener  <rguenther@suse.de>
19021         * function.h (loops_for_fn): New inline function.
19022         (set_loops_for_fn): Likewise.
19023         * cfgloop.h (place_new_loop): Add struct function parameter.
19024         (get_loop): Likewise.
19025         (get_loops): Likewise.
19026         (number_of_loops): Likewise.
19027         (fel_next): Adjust.
19028         (fel_init): Likewise.
19029         * cfg.c (get_loop_copy): Adjust.
19030         * cfgloop.c (flow_loops_dump): Likewise.
19031         (record_loop_exits): Likewise.
19032         (verify_loop_structure): Likewise.
19033         * cfgloopanal.c (mark_irreducible_loops): Likewise.
19034         (estimate_reg_pressure_cost): Likewise.
19035         (mark_loop_exit_edges): Likewise.
19036         * cfgloopmanip.c (place_new_loop): Likewise.
19037         (add_loop): Likewise.
19038         (duplicate_loop): Likewise.
19039         * graph.c (draw_cfg_nodes): Likewise.
19040         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
19041         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
19042         (extract_affine_chrec): Likewise.
19043         (build_scop_iteration_domain): Likewise.
19044         * graphite.c (graphite_initialize): Likewise.
19045         * ira-build.c (create_loop_tree_nodes): Likewise.
19046         (more_one_region_p): Likewise.
19047         (rebuild_regno_allocno_maps): Likewise.
19048         (mark_loops_for_removal): Likewise.
19049         (mark_all_loops_for_removal): Likewise.
19050         (remove_unnecessary_regions): Likewise.
19051         (ira_build): Likewise.
19052         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19053         * loop-init.c (fix_loop_structure): Likewise.
19054         (gate_rtl_move_loop_invariants): Likewise.
19055         (gate_rtl_unswitch): Likewise.
19056         (gate_rtl_unroll_and_peel_loops): Likewise.
19057         (rtl_doloop): Likewise.
19058         * lto-streamer-in.c (input_cfg): Likewise.
19059         * lto-streamer-out.c (output_cfg): Likewise.
19060         * modulo-sched.c (sms_schedule): Likewise.
19061         * predict.c (tree_estimate_probability): Likewise.
19062         (tree_estimate_probability_driver): Likewise.
19063         (estimate_loops): Likewise.
19064         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
19065         (move_sese_region_to_fn): Likewise.
19066         (debug_loop_num): Likewise.
19067         * tree-chrec.c (chrec_evaluate): Likewise.
19068         (hide_evolution_in_other_loops_than_loop): Likewise.
19069         (chrec_component_in_loop_num): Likewise.
19070         (reset_evolution_in_loop): Likewise.
19071         (evolution_function_is_invariant_rec_p): Likewise.
19072         * tree-if-conv.c (main_tree_if_conversion): Likewise.
19073         * tree-inline.c (copy_loops): Likewise.
19074         (copy_cfg_body): Likewise.
19075         (tree_function_versioning): Likewise.
19076         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
19077         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
19078         Likewise.
19079         (add_to_evolution_1): Likewise.
19080         (scev_const_prop): Likewise.
19081         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
19082         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
19083         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
19084         (tree_ssa_lim_initialize): Likewise.
19085         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
19086         (verify_loop_closed_ssa): Likewise.
19087         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
19088         (tree_ssa_loop_im): Likewise.
19089         (tree_ssa_loop_unswitch): Likewise.
19090         (tree_vectorize): Likewise.
19091         (check_data_deps): Likewise.
19092         (tree_ssa_loop_ivcanon): Likewise.
19093         (tree_ssa_loop_bounds): Likewise.
19094         (tree_complete_unroll): Likewise.
19095         (tree_complete_unroll_inner): Likewise.
19096         (tree_parallelize_loops): Likewise.
19097         (tree_ssa_loop_prefetch): Likewise.
19098         (tree_ssa_loop_ivopts): Likewise.
19099         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
19100         * tree-vectorizer.c (vectorize_loops): Likewise.
19102 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
19104         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
19105         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
19106         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
19107         with EABI_LINK_SPEC.
19109 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
19111         PR target/44578
19112         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
19113         alternative.
19115 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
19117         PR target/57097
19118         * lra-constraints.c (process_alt_operands): Discourage a bit more
19119         using memory for pseudos.  Print cost dump for alternatives.
19120         Modify cost values for conflicts with early clobbers.
19121         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
19123 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
19125         PR target/57098
19126         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
19128 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
19130         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
19131         from/to S register.
19132         (movdi_aarch64): Support LDR/STR from/to D register.
19134 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
19136         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
19137         or higher by default.
19139 2013-04-29  Richard Biener  <rguenther@suse.de>
19141         PR middle-end/57075
19142         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
19143         even if not adding abnormal edges for calls that can make
19144         abnormal gotos.
19146 2013-04-29  Richard Biener  <rguenther@suse.de>
19148         PR middle-end/57103
19149         * tree-cfg.c (move_stmt_op): Fix condition under which to update
19150         TREE_BLOCK.
19151         (move_stmt_r): Remove redundant checking.
19153 2013-04-29  Teresa Johnson  <tejohnson@google.com>
19155         PR bootstrap/57077
19156         * basic-block.h (apply_scale): New function.
19157         (apply_probability): Use apply_scale.
19158         * gimple-streamer-in.c (input_bb): Ditto.
19159         * lto-streamer-in.c (input_cfg): Ditto.
19160         * lto-cgraph.c (merge_profile_summaries): Ditto.
19161         * tree-optimize.c (execute_fixup_cfg): Ditto.
19162         * tree-inline.c (copy_bb): Update comment to use apply_scale.
19163         (copy_edges_for_bb): Ditto.
19164         (copy_cfg_body): Ditto.
19166 2013-04-29  Tom de Vries  <tom@codesourcery.com>
19168         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
19169         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
19170         (tail_merge_optimize): Handle current_loops == NULL.
19172 2013-04-26  Jeff Law  <law@redhat.com>
19174         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
19175         (simplify_cond_using_ranges): Generalize code to simplify
19176         COND_EXPRs where one argument is a constant and the other
19177         is an SSA_NAME created by an integral type conversion.
19179 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19181         * config/arm/arm.md (store_minmaxsi): Use only when
19182         optimize_insn_for_size_p.
19184 2013-04-29  Christian Bruel  <christian.bruel@st.com>
19186         PR target/57108
19187         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
19189 2013-04-29  Richard Biener  <rguenther@suse.de>
19191         PR middle-end/57089
19192         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
19193         loop tree make sure to schedule a fixup for the child as well.
19194         (expand_omp_for_generic): Properly add loops.
19195         (expand_omp_for_static_nochunk): Likewise.
19196         (expand_omp_for_static_chunk): Likewise.
19197         (expand_omp_for): For the degenerate case fixup loops.
19198         (expand_omp_sections): Fix default bb placement in loops.
19199         (expand_omp_atomic_pipeline): Properly add loops.
19201 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19203         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
19205 2013-04-29  Tom de Vries  <tom@codesourcery.com>
19207         * tree-ssa-tail-merge.c: Update header comment.
19209 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19211         * config/aarch64/arm_neon.h
19212         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
19213         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
19214         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
19215         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
19216         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
19217         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
19218         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
19219         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
19221 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19223         * config/aarch64/aarch64-simd.md
19224         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
19225         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
19226         fix_trunc, fixuns_trunc.
19227         (ftrunc<VDQF:mode>2): New.
19228         * config/aarch64/iterators.md (optab): Add fix, fixuns.
19229         (fix_trunc_optab): New.
19231 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19233         * config/aarch64/aarch64-builtins.c
19234         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
19235         iceilf, lround, iroundf.
19237 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
19239         PR target/54349
19240         * config/i386/i386.h (enum ix86_tune_indices)
19241         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
19242         New, split from X86_TUNE_INTER_UNIT_MOVES.
19243         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
19244         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
19245         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
19246         (TARGET_INTER_UNIT_MOVES): Remove.
19247         * config/i386/i386.c (initial_ix86_tune_features): Update.
19248         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
19249         (ix86_expand_convert_uns_didf_sse): Use
19250         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
19251         (ix86_expand_vector_init_one_nonzero): Ditto.
19252         (ix86_expand_vector_init_interleave): Ditto.
19253         (inline_secondary_memory_needed): Return true for moves from SSE class
19254         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
19255         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19256         * config/i386/constraints.md (Yi, Ym): Depend on
19257         TARGET_INTER_UNIT_MOVES_TO_VEC.
19258         (Yj, Yn): New constraints.
19259         * config/i386/i386.md (*movdi_internal): Change constraints of
19260         operand 1 from Yi to Yj and from Ym to Yn.
19261         (*movsi_internal): Ditto.
19262         (*movdf_internal): Ditto.
19263         (*movsf_internal): Ditto.
19264         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
19265         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
19266         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
19267         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
19268         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
19269         * config/i386/sse.md (movdi_to_sse): Ditto.
19270         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
19271         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
19272         TARGET_INTER_UNIT_MOVES.
19273         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
19274         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
19275         instead of TARGET_INTER_UNIT_MOVES.
19276         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
19277         operand 1 from Yi to Yj and from Ym to Yn.
19279 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19281         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
19282         (float_truncate_hi_): Likewise.
19283         (float_extend_lo_): Likewise.
19284         (float_truncate_lo_): Likewise.
19285         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
19286         (aarch64_float_extend_lo_v2df): Likewise.
19287         (vec_unpacks_hi_v4sf): Likewise.
19288         (aarch64_float_truncate_lo_v2sf): Likewise.
19289         (aarch64_float_truncate_hi_v4sf): Likewise.
19290         (vec_pack_trunc_v2df): Likewise.
19291         (vec_pack_trunc_df): Likewise.
19293 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19295         * config/aarch64/aarch64-builtins.c
19296         (aarch64_fold_builtin): Fold float conversions.
19297         * config/aarch64/aarch64-simd-builtins.def
19298         (floatv2si, floatv4si, floatv2di): New.
19299         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
19300         * config/aarch64/aarch64-simd.md
19301         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
19302         * config/aarch64/iterators.md (FLOATUORS): New.
19303         (optab): Add float, floatuns.
19304         (su_optab): Likewise.
19306 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19308         * config/aarch64/aarch64-builtins.c
19309         (aarch64_builtin_vectorized_function): Use new names for
19310         fcvt builtins.
19311         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
19312         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
19313         (fcvtzu): Split as...
19314         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
19315         (fcvtas): Split as...
19316         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
19317         (fcvtau): Split as...
19318         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
19319         (fcvtps): Split as...
19320         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
19321         (fcvtpu): Split as...
19322         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
19323         (fcvtms): Split as...
19324         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
19325         (fcvtmu): Split as...
19326         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
19327         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
19328         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
19329         (lfrintnusf, lfrintnudf): Likewise.
19330         * config/aarch64/aarch64-simd.md
19331         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
19332         define_insn.
19333         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
19334         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
19335         (fcvt_pattern): Likewise.
19337 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19339         * config/aarch64/aarch64-simd.md
19340         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
19341         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
19343 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19345         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
19346         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
19347         (vrnd<a,m,n,p>_f32): Implement using builtins.
19348         (vrnd<i,x><q>_f<32, 64>): New.
19350 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19352         * config/aarch64/aarch64-builtins.c
19353         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
19354         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
19355         (frintz): Rename to...
19356         (btrunc): ...this.
19357         (frintp): Rename to...
19358         (ceil): ...this.
19359         (frintm): Rename to...
19360         (floor): ...this.
19361         (frinti): Rename to...
19362         (nearbyint): ...this.
19363         (frintx): Rename to...
19364         (rint): ...this.
19365         (frinta): Rename to...
19366         (round): ...this.
19367         * config/aarch64/aarch64-simd.md
19368         (aarch64_frint<frint_suffix><mode>): Delete.
19369         (<frint_pattern><mode>2): Convert to insn.
19370         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
19371         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
19372         (frint_pattern): Likewise.
19373         (frint_suffix): Likewise.
19375 2013-04-29  Richard Biener  <rguenther@suse.de>
19377         PR tree-optimization/57081
19378         * loop-init.c: Include tree-flow.h.
19379         (loop_optimizer_finalize): Free number of iteration estimates.
19380         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
19382 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
19384         PR tree-optimization/57083
19385         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
19386         non-singleton shift count range, zero extend low_bound for uns case.
19388         * config/i386/predicates.md (general_vector_operand): New predicate.
19389         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
19390         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
19391         if they aren't nonimmediate operands.  If their original values
19392         satisfy const_vector_equal_evenodd_p, don't shift them.
19393         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
19394         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
19395         if not nonimmediate_operand.
19396         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
19397         instead of register_operand.
19398         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
19400 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
19402         * stor-layout.c (finalize_size_functions): Allocate a structure and
19403         reset cfun before dumping the functions.
19405 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
19407         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
19409         PR target/56866
19410         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
19411         use xop_pmacsdqh if uns_p.
19412         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
19413         the immediate rotate count.
19415 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
19417         * rtl.h (struct rtx_def): Add comment for field jump.
19418         (LRA_SUBREG_P): New macro.
19419         * recog.c (register_operand): Check LRA_SUBREG_P.
19420         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19421         stack frame.
19422         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19423         (lra_final_code_change): Skip subreg change for operators.
19424         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19425         if there are no operand changes.
19426         * lra-constraints.c (curr_insn_set): New.
19427         (match_reload): Set LRA_SUBREG_P.
19428         (emit_spill_move): Ditto.
19429         (check_and_process_move): Use curr_insn_set. Process only single
19430         set insns.  Don't initialize sec_mem_p and change_p.
19431         (simplify_operand_subreg): Use LRA_SUBREG_P.
19432         (reg_in_class_p): New function.
19433         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19434         of #ifdef.  Add code to remove cycling.
19435         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19436         non-null disp.  Reload inner instead of disp when base and index
19437         are null.  Try to put lo_sum into register.
19438         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19439         (check_and_process_move): Move code for move cost check to
19440         simple_move_p.  Remove equiv_substitution.
19441         (simple_move_p): New function.
19442         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19443         curr_insn_set.  Call check_and_process_move only for single set
19444         insns.  Use the new function.  Move call of check_and_process_move
19445         after operand equiv substitution and address process.
19447 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19449         PR go/57045
19450         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
19451         with nonlocal goto receivers or returns twice calls, ignore
19452         unininitialized values from abnormal edges to nl goto receiver
19453         or returns twice call.
19455 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19457         PR tree-optimization/57051
19458         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
19459         and VEC_RSHIFT_EXPR if shift count is a multiple of element
19460         bitsize.
19462 2013-04-26  Richard Biener  <rguenther@suse.de>
19464         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
19465         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
19466         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
19467         (fixup_loop_arrays_after_move): New function.
19468         (move_sese_region_to_fn): Properly outline the loop tree parts
19469         of the SESE region.
19471 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
19473         * config/i386/i386.md (type, unit): Fix long lines.
19475 2013-04-26  Richard Biener  <rguenther@suse.de>
19477         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19478         (lto-streamer-out.o): Likewise.
19479         * cfgloop.c (init_loops_structure): Export, add struct function
19480         argument and adjust.
19481         (flow_loops_find): Adjust.
19482         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19483         (init_loops_structure): Declare.
19484         * lto-streamer-in.c: Include cfgloop.h.
19485         (input_cfg): Input the loop tree.
19486         * lto-streamer-out.c: Include cfgloop.h.
19487         (output_cfg): Output the loop tree.
19488         (output_struct_function_base): Do not drop PROP_loops.
19490 2013-03-26  Richard Biener  <rguenther@suse.de>
19492         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19493         (pass_build_cfg): Provide PROP_loops.
19494         (move_sese_region_to_fn): Remove loops that are outlined into fn
19495         for now.
19496         * tree-inline.c: Include cfgloop.h.
19497         (initialize_cfun): Do not drop PROP_loops.
19498         (copy_loops): New function.
19499         (copy_cfg_body): Copy loop structure.
19500         (tree_function_versioning): Initialize destination loop tree.
19501         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19502         (pass_parallelize_loops): Do IL verification.
19503         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19504         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19505         the CFG make sure we fixup loops as well.
19506         * tree-ssa-tail-merge.c: Include cfgloop.h.
19507         (replace_block_by): When merging loop latches mark loops for fixup.
19508         * lto-streamer-out.c (output_struct_function_base): Drop
19509         PROP_loops for now.
19510         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19511         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19512         * ipa-split.c: Include cfgloop.h.
19513         (split_function): Add the new return block to the loop tree root.
19514         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19515         whether we have removed the forwarder block.
19516         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19517         * cfgloop.h (place_new_loop): Declare.
19518         * cfgloopmanip.c (place_new_loop): Export.
19519         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19520         (tree-switch-conversion.o): Likewise.
19521         (tree-complex.o): Likewise.
19522         (tree-inline.o): Likewise.
19523         (tree-ssa-tailmerge.o): Likewise.
19524         (ipa-split.o): Likewise.
19525         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19526         (tree-ssa-copy.o): Likewise.
19527         * tree-switch-conversion.c: Include cfgloop.h
19528         (process_switch): If we emit a bit-test cascade, schedule loops
19529         for fixup.
19530         * tree-complex.c: Include cfgloop.h.
19531         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19532         * asan.c: Include cfgloop.h.
19533         (create_cond_insert_point): Properly add new basic-blocks to
19534         loops, schedule loop fixup.
19535         * cfgloop.c (verify_loop_structure): Check that looks are not
19536         marked for fixup.
19537         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19538         to loops.
19539         (expand_omp_for_generic): Likewise.
19540         (expand_omp_sections): Likewise.
19541         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19542         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19543         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19544         is initialized, not when loops are present.
19545         * tree-parloops.c (parallelize_loops): Remove checking here.
19546         * passes.c (init_optimization_passes): Schedule a copy-propagation
19547         pass before complete unrolling of inner loops.
19549 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19551         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19552         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19553         (colorize_init): Add argument to _WIN32 version.
19554         * toplev.c: Include diagnostic-color.h.
19555         (process_options): Default to -fdiagnostics-color=auto if
19556         GCC_COLORS env var is in the environment.
19557         * common.opt (fdiagnostics-color=): Add Var and Init.
19558         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19559         env var is in the environment, the default is auto rather than never.
19561         * diagnostic.h (file_name_as_prefix): Add context argument.
19562         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19563         the string as locus.
19564         * langhooks.c (lhd_print_error_function): Adjust caller.
19566 2013-04-25  Lawrence Crowl  <crowl@google.com>
19568         * var-tracking.c (shared_hash_def::htab):
19569         Change type to hash_table.  Update dependent calls and types.
19571 2013-04-25  Lawrence Crowl  <crowl@google.com>
19573         * Makefile.in: Update as needed below.
19575         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19576         Move declaration to after the type's method definitons.
19578         * attribs.c (htab_t scoped_attributes::attribute_hash):
19579         Change type to hash_table.  Update dependent calls and types.
19581         * bitmap.c (htab_t bitmap_desc_hash):
19582         Change type to hash_table.  Update dependent calls and types.
19584         * cselib.c (htab_t cselib_hash_table):
19585         Change type to hash_table.  Update dependent calls and types.
19587         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19588         (hash_string_slot_node): Move implementation into lto-streamer.h
19589         struct string_slot_hasher.
19590         (eq_string_slot_node): Likewise.
19592         * data-streamer-out.c: Update output_block::string_hash_table
19593         dependent calls and types.
19595         * dwarf2cfi.c (htab_t trace_index):
19596         Change type to hash_table.  Update dependent calls and types.
19598         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19599         Change type to hash_table.  Update dependent calls and types.
19600         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19601         (htab_t optimize_external_refs::map): Likewise.
19602         (htab_t output_comp_unit::extern_map): Likewise.
19603         (htab_t output_comdat_type_unit::extern_map): Likewise.
19604         (htab_t output_macinfo::macinfo_htab): Likewise.
19605         (htab_t optimize_location_lists::htab): Likewise.
19606         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19608         * except.c (htab_t ehspec_hash_type):
19609         Change type to hash_table.  Update dependent calls and types.
19610         (assign_filter_values::ttypes): Likewise.
19611         (assign_filter_values::ehspec): Likewise.
19612         (sjlj_assign_call_site_values::ar_hash): Likewise.
19613         (convert_to_eh_region_ranges::ar_hash): Likewise.
19615         * gcse.c (htab_t pre_ldst_table):
19616         Change type to hash_table.  Update dependent calls and types.
19618         * ggc-common.c (htab_t saving_htab):
19619         Change type to hash_table.  Update dependent calls and types.
19620         (htab_t loc_hash): Likewise.
19621         (htab_t ptr_hash): Likewise.
19622         (call_count): Rename ggc_call_count.
19623         (call_alloc): Rename ggc_call_alloc.
19624         (loc_descriptor): Rename make_loc_descriptor.
19625         (add_statistics): Rename ggc_add_statistics.
19627         * ggc-common.c (saving_htab):
19628         Change type to hash_table.  Update dependent calls and types.
19630         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19631         (push_gimplify_context): Likewise.
19632         (pop_gimplify_context): Likewise.
19633         (struct gimple_temp_hash_elt): Added.
19634         (struct gimplify_hasher): Likewise.
19635         (struct gimplify_ctx.temp_htab):
19636         Change type to hash_table.  Update dependent calls and types.
19638         * gimple-fold.c: Include gimplify-ctx.h.
19640         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19641         Change type to hash_table.  Update dependent calls and types.
19642         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19643         avoid potential global name collision.
19645         * gimplify.c: Include gimplify-ctx.h.
19646         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19647         (htab_t gimplify_ctx::temp_htab):
19648         Update dependent calls and types for new type hash_table.
19649         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19650         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19652         * gimplify-ctx.h: New.
19653         (struct gimple_temp_hash_elt): Move from gimplify.c.
19654         (class gimplify_hasher): New.
19655         (struct gimplify_ctx): Move from gimple.h.
19656         (htab_t gimplify_ctx::temp_htab):
19657         Change type to hash_table.  Update dependent calls and types.
19659         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19660         (htab_t ivs_params::newivs_index):
19661         Change type to hash_table.  Update dependent calls and types.
19662         (htab_t ivs_params::params_index): Likewise.
19663         (htab_t print_generated_program::params_index): Likewise.
19664         (htab_t gloog::newivs_index): Likewise.
19665         (htab_t gloog::params_index): Likewise.
19667         * graphite.c: Include graphite-htab.h.
19668         4htab_t graphite_transform_loops::bb_pbb_mapping):
19669         Change type to hash_table.  Update dependent calls and types.
19671         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19672         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19673         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19675         * graphite-dependences.c: Include graphite-htab.h.
19676         (loop_is_parallel_p): Change hash table type of parameter.
19678         * graphite-htab.h: New.
19679         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19680         (extern find_pbb_via_hash): Move from graphite-poly.h.
19681         (extern loop_is_parallel_p): Move from graphite-poly.h.
19682         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19684         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19685         (extern loop_is_parallel_p): Move to graphite-htab.h.
19686         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19688         * haifa-sched.c (htab_t delay_htab):
19689         Change type to hash_table.  Update dependent calls and types.
19690         (htab_t delay_htab_i2): Likewise.
19692         * ira-color.c (htab_t allocno_hard_regs_htab):
19693         Change type to hash_table.  Update dependent calls and types.
19695         * ira-costs.c (htab_t cost_classes_htab):
19696         Change type to hash_table.  Update dependent calls and types.
19698         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19699         Change type to hash_table.  Update dependent calls and types.
19701         * loop-iv.c (htab_t bivs):
19702         Change type to hash_table.  Update dependent calls and types.
19704         * loop-unroll.c (htab_t opt_info::insns_to_split):
19705         Change type to hash_table.  Update dependent calls and types.
19706         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19708         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19709         (struct string_slot_hasher): New.
19710         (htab_t output_block::string_hash_table):
19711         Change type to hash_table.  Update dependent calls and types.
19713         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19714         (htab_t file_name_hash_table):
19715         Change type to hash_table.  Update dependent calls and types.
19717         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19718         calls and types.
19720         * lto-streamer.c (htab_t tree_htab):
19721         Change type to hash_table.  Update dependent calls and types.
19723         * omp-low.c: Include gimplify-ctx.h.
19725         * passes.c (htab_t name_to_pass_map):
19726         Change type to hash_table.  Update dependent calls and types.
19727         (pass_traverse): Rename to passes_pass_traverse.
19729         * plugin.c (htab_t event_tab):
19730         Change type to hash_table.  Update dependent calls and types.
19732         * postreload-gcse.c (htab_t expr_table):
19733         Change type to hash_table.  Update dependent calls and types.
19734         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19736         * sese.c (debug_rename_map_1): Make extern.
19737         (htab_t copy_bb_and_scalar_dependences::rename_map):
19738         Change type to hash_table.  Update dependent calls and types.
19740         * sese.h (extern debug_rename_map): Move to .c file.
19742         * store-motion.c (htab_t store_motion_mems_table):
19743         Change type to hash_table.  Update dependent calls and types.
19745         * trans-mem.c (htab_t tm_new_mem_hash):
19746         Change type to hash_table.  Update dependent calls and types.
19748         * tree-browser.c (htab_t TB_up_ht):
19749         Change type to hash_table.  Update dependent calls and types.
19751         * tree-cfg.c (htab_t discriminator_per_locus):
19752         Change type to hash_table.  Update dependent calls and types.
19754         * tree-complex.c: Include tree-hasher.h
19755         (htab_t complex_variable_components):
19756         Change type to hash_table.  Update dependent calls and types.
19758         * tree-eh.c (htab_t finally_tree):
19759         Change type to hash_table.  Update dependent calls and types.
19761         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19762         struct int_tree_hasher.
19763         (extern int_tree_map_eq): Likewise.
19764         (uid_decl_map_hash): Removed.
19765         (extern decl_tree_map_eq): Likewise.
19767         * tree-hasher.h: New.
19768         (struct int_tree_hasher): New.
19769         (typedef int_tree_htab_type): New.
19771         * tree-inline.c: Include gimplify-ctx.h.
19773         * tree-mudflap.c: Include gimplify-ctx.h.
19775         * tree-parloops.c: Include tree-hasher.h.
19776         (htab_t eliminate_local_variables_stmt::decl_address):
19777         Change type to hash_table.  Update dependent calls and types.
19778         (htab_t separate_decls_in_region::decl_copies): Likewise.
19780         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19781         Change type to hash_table.  Update dependent calls and types.
19783         * tree-sra.c (candidates):
19784         Change type to hash_table.  Update dependent calls and types.
19786         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19787         in tree-flow.h.
19788         (int_tree_map_hash): Likewise.
19790         * tree-ssa-dom.c (htab_t avail_exprs):
19791         Change type to hash_table.  Update dependent calls and types.
19793         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19794         Change type to hash_table.  Update dependent calls and types.
19796         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19797         Change type to hash_table.  Update dependent calls and types.
19799         * tree-ssa-phiopt.c (seen_ssa_names):
19800         Change type to hash_table.  Update dependent calls and types.
19802         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19803         Change type to hash_table.  Update dependent calls and types.
19805         * tree-ssa-uncprop.c (equiv):
19806         Change type to hash_table.  Update dependent calls and types.
19808 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19810         PR rtl-optimization/57003
19811         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19812         call note_stores with kill_clobbered_value callback again after
19813         killing regs_invalidated_by_call.
19815 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19817         * config/aarch64/aarch64-simd.md
19818         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19819         (aarch64_simd_bsl<mode>): Likewise.
19820         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19822 2013-04-25  Marek Polacek  <polacek@redhat.com>
19824         PR tree-optimization/57066
19825         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19827 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19829         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19831 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19833         * config/aarch64/aarch64-builtins.c
19834         (aarch64_fold_builtin): New.
19835         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19836         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19837         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19838         * config/aarch64/arm_neon.h
19839         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19841 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19842             Tejas Belagod  <tejas.belagod@arm.com>
19844         * config/aarch64/aarch64-builtins.c
19845         (aarch64_gimple_fold_builtin): New.
19846         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19847         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19848         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19849         (addvv4sf): Update.
19850         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19852 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19854         * config/aarch64/aarch64.md
19855         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19857 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19859         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19860         (*ngcsi_uxtw): New pattern.
19862 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19863             Julian Brown  <julian@codesourcery.com>
19865         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19866         (TB_DREG): Add T_V4HF.
19867         (v4hf_UP): New macro.
19868         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19869         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19870         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19871         built-ins.
19872         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19873         (arm_vector_mode_supported_p): Handle V4HF.
19874         (arm_mangle_map): Handle V4HFmode.
19875         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19876         * config/arm/arm_neon_builtins.def: Add entries for
19877         vcvtv4hfv4sf, vcvtv4sfv4hf.
19878         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19879         (neon_vcvtv4hfv4sf): Likewise.
19880         * config/arm/neon-gen.ml: Handle half-precision floating point
19881         features.
19882         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19883         * config/arm/arm_neon.h: Regenerate.
19884         * config/arm/neon.ml (type elts): Add F16.
19885         (type vectype): Add T_float16x4, T_floatHF.
19886         (type vecmode): Add V4HF.
19887         (type features): Add Requires_FP_bit feature.
19888         (elt_width): Handle F16.
19889         (elt_class): Likewise.
19890         (elt_of_class_width): Likewise.
19891         (mode_of_elt): Refactor.
19892         (type_for_elt): Handle F16, fix error messages.
19893         (vectype_size): Handle T_float16x4.
19894         (vcvt_sh): New function.
19895         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19896         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19897         (string_of_mode): Handle V4HF.
19898         * doc/arm-neon-intrinsics.texi: Regenerate.
19900 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19902         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19903         format specifier in 'X' case.
19905 2013-04-25  Alan Modra  <amodra@gmail.com>
19907         PR target/57052
19908         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19909         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19910         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19911         Repeat for many other rotate/shift and mask patterns using subregs.
19912         Name lshiftrt insns.
19913         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19914         on WORDS_BIG_ENDIAN.
19916 2013-04-25  Alan Modra  <amodra@gmail.com>
19918         * config.gcc: Support little-endian powerpc-linux targets.
19919         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19920         (LINK_OS_LINUX_SPEC): Define.
19921         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19922         Preserve MASK_LITTLE_ENDIAN.
19923         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19924         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19925         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19926         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19927         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19928         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19929         Correct fp word order for little-endian.  Don't shift toc entries
19930         smaller than a word for little-endian.
19931         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19932         (bswapdi2 splits): Correct low-part subreg for little-endian.
19933         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19934         low/high where such is correct only for be.
19935         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19936         little-endian for -mcall-aixdesc.
19938 2013-04-25  Alan Modra  <amodra@gmail.com>
19940         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19941         replace_equiv_address_nv.
19943 2013-04-25  Alan Modra  <amodra@gmail.com>
19945         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19947 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19949         Revert:
19950         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19951         * rtl.h (struct rtx_def): ...
19953 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19955         PR rtl-optimizations/57046
19956         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19957         for multi-reg splits.
19959 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19961         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19963 2013-04-24  Sterling Augustine  <saugustine@google.com>
19965         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19966         (comp_dir_string, debug_str_dwo_section): New.
19967         (DEBUG_STR_DWO_SECTION): Rename to ...
19968         (DEBUG_DWO_STR_SECTION): ... this.
19969         (DEBUG_NORM_STR_SECTION): Delete.
19970         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19971         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19972         (find_AT_string): Move most logic to ...
19973         (find_AT_string_in_table): ... here.  New.
19974         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19975         add_skeleton_AT_string.  Delete logic.
19976         (output_skeleton_debug_sections): Remove call to
19977         add_top_level_skeleton_die_attrs.
19978         (add_comp_dir_attribute): Move logic to comp_dir_string.
19979         (dwarf2out_init): Initialize debug_str_dwo_section.
19980         (output_indirect_string): Call find_string_form.
19981         (output_indirect_strings): Rewrite.
19982         (prune_unused_types): Empty skeleton_debug_str_hash.
19983         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19984         (dwarf2out_finish):  Call output_indirect_strings.
19986 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19988         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19990 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19992         * rtl.h (struct rtx_def): Add comment for field jump.
19993         (LRA_SUBREG_P): New macro.
19994         * recog.c (register_operand): Check LRA_SUBREG_P.
19995         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19996         stack frame.
19997         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19998         (lra_final_code_change): Skip subreg change for operators.
19999         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
20000         if there are no operand changes.
20001         * lra-constraints.c (curr_insn_set): New.
20002         (match_reload): Set LRA_SUBREG_P.
20003         (emit_spill_move): Ditto.
20004         (check_and_process_move): Use curr_insn_set. Process only single
20005         set insns.  Don't initialize sec_mem_p and change_p.
20006         (simplify_operand_subreg): Use LRA_SUBREG_P.
20007         (reg_in_class_p): New function.
20008         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
20009         of #ifdef.  Add code to remove cycling.
20010         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
20011         non-null disp.  Reload inner instead of disp when base and index
20012         are null.  Try to put lo_sum into register.
20013         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
20014         (check_and_process_move): Move code for move cost check to
20015         simple_move_p.  Remove equiv_substitution.
20016         (simple_move_p): New function.
20017         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
20018         curr_insn_set.  Call check_and_process_move only for single set
20019         insns.  Use the new function.  Move call of check_and_process_move
20020         after operand equiv substitution and address process.
20022 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
20024         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
20025         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
20026         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
20028 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
20030         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
20032 2013-04-24  Marek Polacek  <polacek@redhat.com>
20034         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
20035         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
20036         (select_loops_exit_conditions): Likewise.
20037         (number_of_iterations_for_all_loops): Likewise.
20038         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
20039         (scev_analysis): Likewise.
20041 2013-04-02  Catherine Moore  <clm@codesourcery.com>
20042             Chao-ying Fu  <fu@mips.com>
20044         * config/mips/micromips.md (jraddiusp): New pattern.
20045         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
20046         instruction if possible.
20048 2013-04-24  Alan Modra  <amodra@gmail.com>
20050         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
20052 2013-04-24  Julian Brown  <julian@codesourcery.com>
20053             Chung-Lin Tang  <cltang@codesourcery.com>
20055         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
20056         dependency behavior in enumeration type DIE generation. Add TODO note
20057         to comments about future DW_FORM_sdata/udata re-work of related code.
20059 2013-04-23  Lawrence Crowl  <crowl@google.com>
20061         * Makefile.in: Update as needed below.
20063         * hash-table.h (class hash_table):
20064         Correct many methods with parameter types compare_type to the correct
20065         value_type.  (Correct code was unlikely to notice the change.)
20066         (hash_table::elements_with_deleted) New.
20067         (class hashtable::iterator): New.
20068         (hashtable::begin()): New.
20069         (hashtable::end()): New.
20070         (FOR_EACH_HASH_TABLE_ELEMENT): New.
20072         * statistics.c (statistics_hashes):
20073         Change type to hash_table.  Update dependent calls and types.
20075         * tree-into-ssa.c (var_infos):
20076         Change type to hash_table.  Update dependent calls and types.
20078         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
20079         Change type to hash_table.  Update dependent calls and types.
20081         * tree-ssa-loop-im.c (struct mem_ref.refs):
20082         Change type to hash_table.  Update dependent calls and types.
20084         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
20085         Change type to hash_table.  Update dependent calls and types.
20087         * tree-ssa-sccvn.c (vn_tables_s::nary):
20088         Change type to hash_table.  Update dependent calls and types.
20089         (vn_tables_s::phis): Likewise.
20090         (vn_tables_s::references): Likewise.
20092         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
20093         (vn_reference_eq): Update parameter and return types.
20095         * tree-ssa-structalias.c (pointer_equiv_class_table):
20096         Change type to hash_table.  Update dependent calls and types.
20097         (location_equiv_class_table): Likewise.
20099         * tree-vect-data-refs.c: Consequential changes for making
20100         peeling a hash_table.
20102         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
20103         (destroy_loop_vec_info): Dependent hash_table update.
20105         * tree-vectorizer.h (peeling_htab):
20106         Change type to hash_table.  Update dependent calls and types.
20108 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
20110         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
20111         to check the register content is equal or not.
20112         * lra-constraints.c (match_reload): Use lra_assign_reg_val
20113         to assign register content record.
20114         * lra-eliminations.c (update_reg_eliminate): Use
20115         lra_update_reg_val_offset to update register content offset.
20116         * lra-int.h (struct lra_reg): Add offset member.
20117         (lra_reg_val_equal_p): New static inline function.
20118         (lra_update_reg_val_offset): New static inline function.
20119         (lra_assign_reg_val): New static inline function.
20120         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
20121         to assign register content record.
20122         (initialize_lra_reg_info_element): Initial offset to zero.
20124 2013-04-23  Catherine Moore  <clm@codesourcery.com>
20126         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
20127         operands.  Record compression.
20129 2013-04-23  Xinliang David Li  <davidxl@google.com>
20131         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
20133 2013-04-23  Richard Biener  <rguenther@suse.de>
20135         PR middle-end/57036
20136         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
20137         parameter, only add abnormal goto edges from the copied body
20138         if the call could perform abnormal gotos.
20139         (copy_cfg_body): Adjust.
20141 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
20143         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
20145 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
20147         * coretypes.h (gimple_stmt_iterator): Add struct to make
20148         compatible with C.
20150 2013-04-23  Richard Biener  <rguenther@suse.de>
20152         PR tree-optimization/57026
20153         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
20154         from SSA names occuring in abnormal PHI nodes.
20156 2013-04-22  Andi Kleen  <ak@linux.intel.com>
20158         * lto/lto.c (print_lto_report_1): Fix LTO report names.
20160 2013-04-22  Andi Kleen  <ak@linux.intel.com>
20162         * lto/lto.c (print_lto_report_1): Declare early.
20163         (read_cgraph_and_symbols): Call print_lto_report_1 early.
20165 2013-04-22  Andi Kleen  <ak@linux.intel.com>
20167         * common.opt (-flto-report-wpa): Add.
20168         * doc/invoke.texi (-flto-report-wpa): Add.
20169         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
20170         (lto_main): dito.
20172 2013-04-22  Xinliang David Li  <davidxl@google.com>
20174         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
20175         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
20176         * Makefile.in: New dependency
20178             David Daney <ddaney.cavm@gmail.com>
20180         * configure.ac (gcc_cv_as_micromips_support):  Use the
20181         --fatal-warnings option.
20182         * configure: Regenerate.
20184 2013-04-22  Marek Polacek  <polacek@redhat.com>
20186         PR sanitizer/56990
20187         * tsan.c (instrument_expr): Don't instrument expression
20188         in case its size is zero.
20190 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
20192         PR target/57032
20193         Revert:
20194         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
20196         * config/alpha/alpha.c (TARGET_LRA_P): New define.
20198 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
20200         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
20201         (gimple_stmt_iterator): New typedef.
20202         * gimple.h (gimple_stmt_iterator): Rename to...
20203         (gimple_stmt_iterator_d): ... This.
20204         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
20205         trees be valid for GIMPLE and GENERIC.
20206         (TARGET_GIMPLE_FOLD_BUILTIN): New.
20207         * gimple-fold.c (gimple_fold_call): Call target hook
20208         gimple_fold_builtin.
20209         * hooks.c (hook_bool_gsiptr_false): New.
20210         * hooks.h (hook_bool_gsiptr_false): New.
20211         * target.def (fold_stmt): New.
20212         * doc/tm.texi: Regenerate.
20214 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
20216         PR target/57018
20217         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
20218         a set sp if no stack realignment.
20220 2013-04-22  Nick Clifton  <nickc@redhat.com>
20222         * config.gcc (tilegx-linux): Extend extra_objs rather than
20223         overwriting it.
20224         (tilepro-linux): Likewise.
20226 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
20228         * config/aarch64/aarch64-builtins.c
20229         (CF): Remove.
20230         (CF0, CF1, CF2, CF3, CF4, CF10): New.
20231         (VAR<1-12>): Add MAP parameter.
20232         (BUILTIN_*): Likewise.
20233         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
20234         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
20235         (aarch64_ushl_n<mode>): Likewise.
20236         (aarch64_sshr_n<mode>): Likewise.
20237         (aarch64_ushr_n<mode>): Likewise.
20238         (aarch64_<maxmin><mode>): Likewise.
20239         (aarch64_sqrt<mode>): Likewise.
20240         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
20241         (vshr<q>_n_*): Likewise.
20243 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
20245         * config/aarch64/aarch64-builtins.c
20246         (aarch64_simd_builtin_type_mode): Handle SF types.
20247         (sf_UP): Define.
20248         (BUILTIN_GPF): Define.
20249         (aarch64_init_simd_builtins): Handle SF types.
20250         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
20251         (frecps): Likewise.
20252         (frecpx): Likewise.
20253         * config/aarch64/aarch64-simd.md
20254         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
20255         (aarch64_frecpe<mode>): New.
20256         (aarch64_frecps<mode>): Likewise.
20257         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
20258         (v8type): Add frecp<esx>.
20259         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
20260         (aarch64_frecps<mode>): Likewise.
20261         * config/aarch64/iterators.md (FRECP): New.
20262         (frecp_suffix): Likewise.
20263         * config/aarch64/arm_neon.h
20264         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
20266 2013-04-22  Christian Bruel  <christian.bruel@st.com>
20268         PR target/56995
20269         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
20270         (REG_CLASS_NAMES): Idem.
20271         (REG_CLASS_CONTENTS): Idem.
20272         (REGCLASS_HAS_FP_REG): Idem.
20273         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
20274         (sh_conditional_register_usage): Idem.
20276 2013-04-21  Jeff Law  <law@redhat.com>
20278         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
20279         (ssa_forward_propagate_and_combine): Use it.
20281 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
20283         * lra.c: Update the flow chart diagram.
20285 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
20287         PR rtl-optimization/56847
20288         * lra-constraints.c (process_alt_operands): Discourage alternative
20289         with non-matche doffsettable memory constraint fro memory with
20290         known offset.
20292 2013-04-19  Richard Biener  <rguenther@suse.de>
20294         PR tree-optimization/56982
20295         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
20296         function.
20297         * gimplify.c (gimplify_call_expr): Notice special calls.
20298         (gimplify_modify_expr): Likewise.
20299         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
20300         abnormal control flow receivers.
20301         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
20302         in the same way as cfun->has_nonlocal_labels.
20303         (gimple_purge_dead_abnormal_call_edges): Likewise.
20304         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
20305         receivers start a basic-block.
20307 2013-04-19  Richard Biener  <rguenther@suse.de>
20309         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
20310         member ...
20311         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
20312         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
20313         (SLP_TREE_LOAD_PERMUTATION): Add.
20314         (vect_transform_slp_perm_load): Adjust prototype.
20315         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
20316         (vect_free_slp_instance): Likewise.
20317         (vect_create_new_slp_node): Likewise.
20318         (vect_supported_slp_permutation_p): Remove.
20319         (vect_slp_rearrange_stmts): Adjust.
20320         (vect_supported_load_permutation_p): Likewise.  Inline
20321         vect_supported_slp_permutation_p here.
20322         (vect_analyze_slp_instance): Compute load permutations per
20323         slp node instead of per instance.
20324         (vect_get_slp_defs): Adjust.
20325         (vect_transform_slp_perm_load): Likewise.
20326         (vect_schedule_slp_instance): Remove redundant code.
20327         (vect_schedule_slp): Remove hack for PR56270, add it ...
20328         * tree-vect-stmts.c (vectorizable_load): ... here, do not
20329         CSE loads for SLP.  Adjust.
20331 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
20333         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
20334         spelling in two comments.
20336 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
20338         PR target/56797
20339         * config/arm/arm.c (load_multiple_sequence): Require SP
20340         as base register for loads if SP is in the register list.
20342 2013-04-19  Martin Jambor  <mjambor@suse.cz>
20344         PR tree-optimization/56718
20345         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
20346         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
20347         and made public.  Adjusted all callers.
20348         (ipa_intraprocedural_devirtualization): New function.
20349         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
20350         (ipa_intraprocedural_devirtualization): Likewise.
20351         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
20353 2013-04-19  Richard Biener  <rguenther@suse.de>
20355         PR tree-optimization/57000
20356         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
20358 2013-04-19  Terry Guo  <terry.guo@arm.com>
20360         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
20361         Replace with ...
20362         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
20363         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
20364         (cortex_m4_fmacs): Use new reservations.
20365         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
20367 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
20369         PR rtl-optimization/56999
20370         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
20371         related code.
20372         (lra_coalesce): Remove split_origin_bitmap and related code.
20373         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
20374         ranges if necessary.
20376 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
20378         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
20379         New array.
20380         (ix86_expand_call): Remove clobbered_registers array and use
20381         x86_64_ms_sysv_extra_clobbered_registers instead.
20382         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
20383         Declare here.
20384         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
20385         predicate.
20386         * config/i386/i386.md (*call_rex64_ms_sysv): Use
20387         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
20388         (*call_value_rex64_ms_sysv): Ditto.
20390 2013-04-18  Cary Coutant  <ccoutant@google.com>
20392         * dwarf2out.c (output_pubnames): Check die_perennial_p of
20393         parent instead of die_mark.
20395 2013-04-18  Diego Novillo  <dnovillo@google.com>
20397         * gimple.c (create_gimple_tmp): New.
20398         (get_expr_type): New.
20399         (build_assign): New.
20400         (build_type_cast): New.
20401         * gimple.h (enum ssa_mode): Define.
20402         (gimple_seq_set_location): New.
20403         * asan.c (build_check_stmt): Change some gimple_build_* calls
20404         to use build_assign and build_type_cast.
20406 2013-04-18  Richard Biener  <rguenther@suse.de>
20408         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
20409         handle negative step.  Remove redundant checks.
20410         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
20411         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
20412         for negative step and grouped loads fail to vectorize.
20414 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20416         * emit-rtl.c (reset_insn_used_flags): New function.
20417         (reset_all_used_flags): Use it.
20418         (verify_insn_sharing): New function.
20419         (verify_rtl_sharing): Fix verification for SEQUENCEs.
20421 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
20423         PR tree-optimization/56984
20424         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
20425         and (x >> M) >= N don't register any assertion if N << M is the
20426         minimum value.
20428 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20430         * lower-subreg.c (resolve_simple_move): If called self-recursive,
20431         do not delete_insn insns that have not yet been emitted, only
20432         unlink them with remove_insn.
20433         * df-scan.c (df_insn_delete): Revert r197492.
20435 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
20437         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
20438         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
20440 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20442         * config/arm/arm.md (movsicc_insn): Convert define_insn into
20443         define_insn_and_split.
20444         (and_scc,ior_scc,negscc): Likewise.
20445         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
20447 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20449         * config/arm/arm.c (use_return_insn): Return 0 for targets that
20450         can benefit from using a sequence of LDRD instructions in epilogue
20451         instead of a single LDM instruction.
20453 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20455         PR 45688
20456         * doc/extend.texi: Fix typo.
20458 2013-04-17  Richard Biener  <rguenther@suse.de>
20460         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
20461         (vect_build_slp_tree): ... here.
20462         (vect_build_slp_tree_1): Compute which stmts of the SLP group
20463         match.  Remove special-casing of mismatched complex loads.
20464         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
20465         re-try the match with swapped commutative operands.
20466         (vect_supported_load_permutation_p): Remove special-casing of
20467         mismatched complex loads.
20468         (vect_analyze_slp_instance): Adjust.
20470 2013-04-17  Richard Biener  <rguenther@suse.de>
20472         PR rtl-optimization/56921
20473         * cfgloop.h (struct loop): Add simple_loop_desc member.
20474         (struct niter_desc): Mark with GTY(()).
20475         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20476         * loop-iv.c (get_simple_loop_desc): Likewise.
20477         (free_simple_loop_desc): Likewise.
20479         Revert
20480         2013-04-16  Richard Biener  <rguenther@suse.de>
20482         PR rtl-optimization/56921
20483         * loop-init.c (pass_rtl_move_loop_invariants): Add
20484         TODO_do_not_ggc_collect to todo_flags_finish.
20485         (pass_rtl_unswitch): Same.
20486         (pass_rtl_unroll_and_peel_loops): Same.
20487         (pass_rtl_doloop): Same.
20489 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20491         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20492         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20493         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20494         references.
20495         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20496         * tree-streamer.c (record_common_node): Adjust reference in comment.
20498 2013-04-17  Terry Guo  <terry.guo@arm.com>
20500         * config/arm/cortex-m4.md: Add a new bypass.
20502 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20504         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20505         New pattern.
20506         (*subs_<optab><mode>_multp2): New pattern.
20507         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20508         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20510 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20512         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20513         (*subs_mul_imm_<mode>): New pattern.
20515 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20517         PR target/56948
20518         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20519         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20520         (vsx_movti_32bit): Change j->wa to O->wa.
20522 2013-04-16  Richard Biener  <rguenther@suse.de>
20524         PR rtl-optimization/56921
20525         * loop-init.c (pass_rtl_move_loop_invariants): Add
20526         TODO_do_not_ggc_collect to todo_flags_finish.
20527         (pass_rtl_unswitch): Same.
20528         (pass_rtl_unroll_and_peel_loops): Same.
20529         (pass_rtl_doloop): Same.
20531 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20533         * config/arm/arm.c (emit_multi_reg_push): New declaration
20534         for an existing function.
20535         (arm_emit_strd_push): New function.
20536         (arm_expand_prologue): Used here.
20537         (arm_emit_ldrd_pop): New function.
20538         (arm_expand_epilogue): Used here.
20539         (arm_get_frame_offsets): Update condition.
20540         (arm_emit_multi_reg_pop): Add a special case for load of a single
20541         register with writeback.
20543 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20545         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20546         description.
20548 2013-04-16  Richard Biener  <rguenther@suse.de>
20550         PR tree-optimization/56756
20551         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20552         (first_mem_ref_loc): New.
20553         (execute_sm): Place the load temporarily before a previous
20554         access instead of in the latch edge to ensure its SSA dependencies
20555         are defined at points dominating the load.
20557 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20559         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20560         correct fix by moving header and footer insn to the footer of
20561         the merged basic block.  Clear BB_END of the merged-away block.
20563         PR middle-end/43631
20564         * emit-rtl.c (make_note_raw): New function.
20565         (link_insn_into_chain): New static inline function.
20566         (add_insn): Use it.
20567         (add_insn_before, add_insn_after): Factor insn chain linking code...
20568         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20569         using link_insn_into_chain.
20570         (note_outside_basic_block_p): New helper function for emit_note_after
20571         and emit_note_before.
20572         (emit_note_after): Use nobb variant of add_insn_after if the note
20573         should not be contained in a basic block.
20574         (emit_note_before): Use nobb variant of add_insn_before if the note
20575         should not be contained in a basic block.
20576         (emit_note_copy): Use make_note_raw.
20577         (emit_note): Likewise.
20578         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20579         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20580         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20581         the moved barrier the tail of the basic block it follows.
20582         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20584 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20586         PR tree-optimization/56962
20587         * gimple-ssa-strength-reduction.c (record_increment): Only set
20588         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20589         either rhs1 or rhs2 is equal to c->base_expr.
20591 2013-04-15  Richard Biener  <rguenther@suse.de>
20593         PR tree-optimization/56933
20594         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20595         member.
20596         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20597         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20598         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20599         dependence check ...
20600         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20601         ... here.
20602         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20603         GROUP_READ_WRITE_DEPENDENCE.
20605 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20607         * emit-rtl.c (reset_all_used_flags): New function.
20608         (verify_rtl_sharing): Call reset_all_used_flags before and after
20609         performing the checks.
20611 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20613         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20614         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20615         * config/arm/constraints.md (De): New constraint.
20616         * config/arm/neon.md (anddi3_neon): Delete.
20617         (neon_vand<mode>): Expand to standard anddi3 pattern.
20618         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20619         Move earlier in the file.
20620         (neon_inv_logic_op2): Likewise.
20621         (arm_anddi_operand_neon): New predicate.
20623 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20625         * configure.ac (gcc_cv_ld_as_needed): Set
20626         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20627         Use -z ignore, -z record on *-*-solaris2*.
20628         (HAVE_LD_AS_NEEDED): Update comment.
20629         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20630         * configure: Regenerate.
20631         * config.in: Regenerate.
20632         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20633         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20634         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20635         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20636         equivalents.  Fix markup.
20637         * doc/tm.texi: Regenerate.
20639 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20641         * config/i386/i386.opt: New option mstack-protector-guard=.
20642         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20643         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20644         TARGET_SSP_TLS_GUARD.
20645         * config/i386/i386.c (ix86_option_override_internal): Set
20646         ix86_stack_protector_guard.
20647         * config/i386/i386.md (stack_protect_set): Enable for
20648         TARGET_SSP_TLS_GUARD only.
20649         (stack_protect_set_<mode>): Ditto.
20650         (stack_protect_test): Ditto.
20651         (stack_protect_test_<mode>): Ditto.
20652         * doc/invoke.texi (i386 Option): Document.
20654 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20656         PR target/56890
20657         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20658         (S_MODES): Set H_MODE bit.
20659         (SF_MODES): Set only S_MODE and SF_MODE bits.
20660         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20661         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20662         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20663         <MODE_FLOAT>: Likewise.
20665 2013-04-15  Joey Ye  <joey.ye@arm.com>
20667         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20669 2013-04-15  Joey Ye  <joey.ye@arm.com>
20671         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20672         for real far jump.
20673         (thumb_far_jump_used_p): Count instruction size and set
20674         far_jump_used.
20676 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20678         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20679         * resource.c (mark_target_live_regs): Reformat conditional block.
20681 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20683         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20684         notes, they are emitted only just before final.
20685         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20687 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20689         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20690         * cfgrtl.c (delete_insn): Call it here instead.
20691         * lra-spills.c (lra_final_code_change): Use delete_insn.
20692         * haifa-sched.c (sched_remove_insn): Likewise.
20693         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20694         returning to the nop pool.
20695         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20696         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20698 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20700         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20701         * doc/tm.texi: Regenerated.
20703 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20705         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20706         QImode checks.
20708 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20710         * df-core.c (df_find_def): Compare register numbers.
20711         (df_find_use): Likewise.
20713 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20715         PR target/56903
20716         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20717         lra_in_progress for return.
20719 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20721         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20722         define_insn into define_insn_and_split and emit movsicc patterns.
20724 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20726         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20728 2013-04-12  Richard Biener  <rguenther@suse.de>
20730         * tree-pass.h (TODO_do_not_ggc_collect): New.
20731         * passes.c (execute_one_ipa_transform_pass): Honor
20732         TODO_do_not_ggc_collect.
20733         (execute_one_pass): Likewise.
20735         Revert
20736         2013-04-10  Richard Biener  <rguenther@suse.de>
20738         * passes.c (init_optimization_passes): Remove reload pass.
20739         * ira.c (do_reload): Merge into ...
20740         (ira): ... this.
20741         (rest_of_handle_reload): Remove.
20742         (pass_reload): Likewise.
20743         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20744         of reload for vzeroupper pass placement.
20746 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20748         PR tree-optimization/56918
20749         PR tree-optimization/56920
20750         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20751         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20752         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20753         use wide_mul_with_sign method.
20755 2013-04-12  Richard Biener  <rguenther@suse.de>
20757         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20758         not be considered a gimple constant.
20760 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20762         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20763         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20764         operations.
20766 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20767             Jakub Jelinek  <jakub@redhat.com>
20769         * opts.c: Include diagnostic-color.h.
20770         (common_handle_option): Handle OPT_fdiagnostics_color_.
20771         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20772         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20773         (diagnostic-color.o): New.
20774         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20775         (diagnostic_color_rule): New enum.
20776         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20777         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20778         the location string.
20779         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20780         either NULL, or color kind.
20781         * diagnostic-color.c: New file.
20782         * diagnostic-color.h: New file.
20783         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20784         arguments.
20785         * doc/invoke.texi (-fdiagnostics-color): Document.
20786         * pretty-print.h (pp_show_color): Define.
20787         (struct pretty_print_info): Add show_color field.
20788         * diagnostic.c: Include diagnostic-color.h.
20789         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20790         macros.  Colorize error:, warning: etc. strings and also the location
20791         string.
20792         (diagnostic_show_locus): Colorize the caret line.
20793         * pretty-print.c: Include diagnostic-color.h.
20794         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20795         inside of %< %> quotes or quoted through q format modifier.
20797 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20799         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20801 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20803         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20804         code in CC_NZ mode.
20805         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20806         pattern.
20808 2013-04-11  Marek Polacek  <polacek@redhat.com>
20810         PR tree-optimization/48184
20811         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20813 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20815         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20816         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20817         (skip_simple_arithmetic): Tidy up.
20818         * tree.h (skip_simple_constant_arithmetic): Declare.
20820 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20822         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20824 2013-04-11  Richard Biener  <rguenther@suse.de>
20826         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20827         generate vector constants.
20829 2013-04-11  Richard Biener  <rguenther@suse.de>
20831         PR tree-optimization/56878
20832         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20833         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20834         New function.
20835         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20836         Prefer to align the DR with the most invariant base address.
20838 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20840         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20841         comment.
20843 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20845         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20846         floating-point vector comparisons against 0.
20848 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20850         PR tree-optimization/56899
20851         * fold-const.c (extract_muldiv_1): Apply distributive law
20852         only if TYPE_OVERFLOW_WRAPS (ctype).
20854 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20856         PR target/56124
20857         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20858         loading has side effect.
20860 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20862         * config/sparc/sparc.c: Include tree-pass.h.
20863         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20864         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20865         head of file.  Change return type.  Split off gate function.
20866         (sparc_gate_work_around_errata): New function.
20867         (pass_work_around_errata): New pass definition.
20868         (insert_pass_work_around_errata) New pass insert definition to
20869         insert pass_work_around_errata just after delayed-branch scheduling.
20870         (sparc_option_override): Insert the pass.
20871         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20873 2013-04-10  David S. Miller  <davem@davemloft.net>
20875         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20876         or -mcpu=hypersparc.
20878         * target.def (cstore_mode): New hook.
20879         * target.h: Include insn-codes.h
20880         * targhooks.c: Likewise.
20881         (default_cstore_mode): New function.
20882         * targhooks.h: Declare it.
20883         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20884         * doc/tm.texi: Rebuild.
20885         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20886         target hook, rather than inspecting the insn_data.
20887         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20888         (TARGET_CSTORE_MODE): Redefine.
20889         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20890         result patterns.
20891         * config/sparc/predicates.md (cstore_result_operand): New special
20892         predicate.
20893         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20894         Use it for operand 0.
20895         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20896         (*snesi_special): Likewise.
20897         (*snesi_zero): Likewise.
20898         (*seqsi_zero): Likewise.
20899         (*sltu_insn): Likewise.
20900         (*sgeu_insn): Likewise.
20901         (*seqdi_special): Make operand 0 and comparison operation be of
20902         DImode.
20903         (*snedi_special): Likewise.
20904         (*snedi_special_vis3): Likewise.
20905         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20906         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20907         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20908         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20909         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20910         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20911         (*sltu_extend_sp64): Likewise.
20912         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20913         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20914         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20915         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20916         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20918 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20920         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20921         (aarch64_start_file): Use the new function.
20923 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20924             Jason Merrill  <jason@redhat.com>
20926         * common.opt: Add -gdwarf.
20927         * opts.c (common_handle_option): Handle it.
20928         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20930 2013-04-10  Richard Biener  <rguenther@suse.de>
20932         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20933         (execute_one_ipa_transform_pass): But unconditionally here.
20934         (execute_one_pass): And here.
20935         (init_optimization_passes): Remove reload pass.
20936         * tree-pass.h (TODO_ggc_collect): Remove.
20937         (pass_reload): Likewise.
20938         * ira.c (do_reload): Merge into ...
20939         (ira): ... this.
20940         (rest_of_handle_reload): Remove.
20941         (pass_reload): Likewise.
20942         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20943         of reload for vzeroupper pass placement.
20944         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20945         and todo_flags_finish of all passes.
20947 2013-04-10  Richard Biener  <rguenther@suse.de>
20949         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20950         first_const_oprnd field, rename first_def_type to first_op_type.
20951         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20952         (vect_get_and_check_slp_defs): Always use the type of the
20953         operand.  Allow mixed vect_external_def, vect_constant_def types.
20954         (vect_get_constant_vectors): Handle mixed vect_external_def,
20955         vect_constant_def types.
20957 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20959         PR tree-optimization/55524
20960         * tree-ssa-math-opts.c
20961         (convert_mult_to_fma): Don't use an fms construct
20962         when we don't have an fms operation, but fnma, and it looks
20963         likely that we'll be able to use the latter.
20965 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20967         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20968         function.
20969         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20970         inline fail caused by overwritable functions.
20972 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20974         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20975         unnecessary bits in the constant power of two case.
20977 2013-04-10  Richard Biener  <rguenther@suse.de>
20979         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20980         broken code swapping operands.
20981         (vect_build_slp_tree): Do not compute load permutations here.
20982         (vect_analyze_slp_instance): Compute load permutations here,
20983         after building the SLP tree.
20985 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20987         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20988         of next/prev_real_insn.
20990 2013-04-09  Jan Hubicka  <jh@suse.cz>
20992         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20993         Drop aliased parameter.
20994         (function_and_variable_visibility): Do not handle alias pairs.
20995         * cgraph.c (varpool_externally_visible_p): Update prototype.
20996         * varpool.c (varpool_add_new_variable): Update.
20998 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21000         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
21002 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
21004         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
21006         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
21008 2013-04-09  Marek Polacek  <polacek@redhat.com>
21010         PR tree-optimization/48762
21011         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
21013 2013-04-09  Richard Biener  <rguenther@suse.de>
21015         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
21016         dealing with cost.
21017         (vect_build_slp_tree): Likewise.
21018         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
21019         calculating the cost of a SLP instance.
21020         (vect_analyze_slp_instance): Use it from here, after building
21021         the SLP tree.
21023 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
21025         PR middle-end/56883
21026         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
21027         expand_omp_for_static_chunk): Use simple_p = true in
21028         force_gimple_operand_gsi calls when assigning to addressable decls.
21030 2013-04-09  Jeff Law  <law@redhat.com>
21032         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
21033         when the boolean was created by converting a wider object which
21034         had a boolean range.
21036 2013-04-09  Richard Biener  <rguenther@suse.de>
21038         * tree-vectorizer.h (slp_void_p): Remove.
21039         (slp_tree): Typedef before _slp_tree declaration.
21040         (struct _slp_tree): Use a vector of slp_tree as children.
21041         (vect_get_place_in_interleaving_chain): Remove.
21042         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
21043         Move ...
21044         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
21045         and make static.
21046         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
21047         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
21048         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
21049         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
21050         Use slp_node instead of slp_void_p and adjust.
21052 2013-04-09  Richard Biener  <rguenther@suse.de>
21054         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
21055         work that is not necessary.
21057 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
21059         PR tree-optimization/56854
21060         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
21061         forward into clobber stmts if it would change MEM_REF lhs into
21062         non-MEM_REF.
21064 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
21066         * tree.c (type_hash_lookup, type_hash_add): Make static.
21067         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
21069 2013-04-09  Richard Biener  <rguenther@suse.de>
21071         * tree.h (unsave_expr_now): Remove.
21072         * tree-inline.c (mark_local_for_remap_r): Remove.
21073         (unsave_expr_1): Likewise.
21074         (unsave_r): Likewise.
21075         (unsave_expr_now): Likewise.
21076         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
21077         (propagate_tree_value): Likewise.
21079 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
21081         * doc/rtl.texi (sequence): Rewrite documentation to match the
21082         current use of SEQUENCE rtl objects.
21083         * rtl.def (SEQUENCE): Likewise.
21085         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
21086         Update documentation.
21087         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
21088         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
21090         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
21092 2013-04-08  Teresa Johnson  <tejohnson@google.com>
21094         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
21095         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
21096         methods.
21097         (estimate_edge_size_and_time): Add comment to suggest using rounding
21098         methods.
21099         (estimate_node_size_and_time): Ditto.
21100         (remap_edge_change_prob): Use helper rounding divide methods.
21101         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
21102         (gimple_mod_pow2_value_transform): Ditto.
21103         (gimple_mod_subtract_transform): Ditto.
21104         (gimple_ic_transform): Ditto.
21105         (gimple_stringops_transform): Ditto.
21106         * stmt.c (conditional_probability): Ditto.
21107         (emit_case_dispatch_table): Ditto.
21108         * lto-cgraph.c (merge_profile_summaries): Ditto.
21109         * tree-optimize.c (execute_fixup_cfg): Ditto.
21110         * cfgcleanup.c (try_forward_edges): Ditto.
21111         * cfgloopmanip.c (scale_loop_profile): Ditto.
21112         (loopify): Ditto.
21113         (duplicate_loop_to_header_edge): Ditto.
21114         (lv_adjust_loop_entry_edge): Ditto.
21115         * tree-vect-loop.c (vect_transform_loop): Ditto.
21116         * profile.c (compute_branch_probabilities): Ditto.
21117         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
21118         * lto-streamer-in.c (input_cfg): Ditto.
21119         * gimple-streamer-in.c (input_bb): Ditto.
21120         * ipa-cp.c (update_profiling_info): Ditto.
21121         (update_specialized_profile): Ditto.
21122         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
21123         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
21124         rounding methods.
21125         * sched-rgn.c (compute_dom_prob_ps): Ditto.
21126         (compute_trg_info): Ditto.
21127         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
21128         (purge_dead_edges): Ditto.
21129         * loop-unswitch.c (unswitch_loop): Ditto.
21130         * cgraphclones.c (cgraph_clone_edge): Ditto.
21131         (cgraph_clone_node): Ditto.
21132         * tree-inline.c (copy_bb): Ditto.
21133         (copy_edges_for_bb): Ditto.
21134         (initialize_cfun): Ditto.
21135         (copy_cfg_body): Ditto.
21136         (expand_call_inline): Ditto.
21138 2013-04-08  Kai Tietz  <ktietz@redhat.com>
21140         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
21141         TARGET_CYGWIN64 by TARGET_64BIT.
21143 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
21145         * config/epiphany/epiphany.md (GPR_1): New constant.
21146         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
21147         * config/epiphany/epiphany.c (gen_compare_reg):
21148         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
21149         is already in place.
21150         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
21151         Don't require being called during rtl expansion; If y operlaps r0,
21152         return 0.
21153         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
21154         (epiphany_expand_epilogue): Likewise.
21156         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
21157         Don't use CC_FPmode for ORDERED / UNORDERED.
21158         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
21160         * config/epiphany/constraints.md (CnL): New constraint.
21161         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
21162         * config/epiphany/predicates.md (add_operand): Allow 1024.
21164         * config/epiphany/epiphany.md (logical_op): New code iterator.
21165         (op_mnc): New code attribute.
21166         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
21167         (mov_f+1, mov_f+2): New peephole2 patterns.
21169         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
21170         (cstoresi4): Also allow re-use of zero result when doing a NE
21171         comparison to a non-zero operand.
21172         Use (clobber (scratch)) for first insn if the gpr output is not needed.
21174         * config/epiphany/epiphany.md (<insn_opname>v2si3):
21175         Use gen_addsi3_i / gen_subsi3_i.
21177 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
21179         PR c++/34949
21180         PR c++/50243
21181         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
21182         contain anything but clobbers, at most one __builtin_stack_restore,
21183         optionally debug stmts and final resx, and if it has at least one
21184         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
21185         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
21186         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
21187         which isn't defaut definition, remove them.
21188         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
21189         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
21190         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
21191         with MEM_REF LHS with SSA_NAME address.
21193 2013-04-08  Jeff Law  <law@redhat.com>
21195         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
21197 2013-04-08  Richard Biener  <rguenther@suse.de>
21199         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
21200         extra newline.
21201         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
21202         determined vector type.
21203         (vect_analyze_data_refs): Likewise.
21204         (vect_get_new_vect_var): Adjust.
21205         (vect_create_destination_var): Preserve SSA name versions.
21206         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
21207         not dump anything here.
21209 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
21211         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
21212         Add member lr_slot_known.
21213         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
21214         if necessary.
21215         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
21216         Remove code that sets lr_slot_offset according to what a previous
21217         version of epiphany_emit_save_restore used to do.
21218         (epiphany_emit_save_restore): When doing an lr save or restore,
21219         set/verify lr_slot_known and lr_slot_offset.
21221 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
21223         PR target/54338
21224         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
21225         in ALL_REGS.
21227 2013-04-08  Richard Biener  <rguenther@suse.de>
21229         * alias.c (find_base_term): Fix thinko in previous change.
21231 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
21233         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
21234         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
21235         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
21236         if possible to compute val.
21237         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
21238         For QImode integers don't require anything about precision.  Use
21239         const_with_all_bytes_same to find out if the constant doesn't have
21240         repeated bytes in it.
21242 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21244         * config/s390/s390.c (s390_expand_insv): Only accept insertions
21245         within mode size.
21247 2013-04-08  Marek Polacek  <polacek@redhat.com>
21249         PR rtl-optimization/48182
21250         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
21251         value to 1.
21253 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21255         PR target/55487
21256         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
21257         nuses, make sure we have a label.
21259 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
21261         PR target/56843
21262         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
21263         (rs6000_emit_swdiv_low_precision): Remove.
21264         (rs6000_emit_swdiv): Rewrite to handle between one and four
21265         iterations of Newton-Raphson generally; modify required number of
21266         iterations for some cases.
21267         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
21269 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
21271         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
21272         set-but-unused variable.
21274         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
21275         basic blocks of released function bodies garbage-collectable.
21277         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
21278         (struct rtl_opt_pass): Add TODO_df_finish.
21280         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
21282 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21284         * config/arm/constraints.md (q): New constraint.
21285         * config/arm/ldrdstrd.md: New file.
21286         * config/arm/arm.md (ldrdstrd.md) New include.
21287         (arm_movdi): Use "q" instead of "r" constraint
21288         for double-word memory access.
21289         (movdf_soft_insn): Likewise.
21290         * config/arm/vfp.md (movdi_vfp): Likewise.
21291         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
21292         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
21293         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
21294         (mem_ok_for_ldrd_strd): Likewise.
21295         (output_move_double): Update assertion.
21297 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21299         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
21301 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21303         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
21304         define_insn_and_split.
21305         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
21307 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21309         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
21310         define_insn_and_split.
21311         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
21312         (shiftsi3_compare): New pattern.
21313         (rrx): New pattern.
21314         * config/arm/unspecs.md (UNSPEC_RRX): New.
21316 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21318         * config/arm/arm.md (negdi_extendsidi): New pattern.
21319         (negdi_zero_extendsidi): Likewise.
21321 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21323         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
21324         define_insn_and_split.
21325         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
21326         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
21328 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21330         * config/arm/arm.md (arm_subdi3): Convert define_insn into
21331         define_insn_and_split.
21332         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
21333         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
21335 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21337         * config/arm/arm.md (subsi3_carryin): New pattern.
21338         (subsi3_carryin_const): Likewise.
21339         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
21340         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
21342 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21344         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
21346 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21348         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
21349         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
21351 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21353         * config/arm/arm.c (arm_expand_builtin): Change fcode
21354         type to unsigned int.
21356 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21358         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
21360 2013-04-04  Ian Lance Taylor  <iant@google.com>
21362         * doc/standards.texi (Standards): The Go frontend supports the Go 1
21363         language standard.
21365 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
21367         PR middle-end/56729
21368         * df-scan.c (df_insn_delete): Disable failing assert.
21370 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21372         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
21373         New function prototype.
21374         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
21375         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
21376         (arm_builtin_vectorized_function): New function.
21378 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21380         * config/arm/arm_neon_builtins.def: New file.
21381         * config/arm/arm.c (neon_builtin_data): Move contents to
21382         arm_neon_builtins.def.
21383         (enum arm_builtins): Include neon builtin definitions.
21384         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
21385         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
21387 2013-04-04  Marek Polacek  <polacek@redhat.com>
21389         PR tree-optimization/48186
21390         * predict.c (maybe_hot_frequency_p): Return false if
21391         HOT_BB_FREQUENCY_FRACTION is 0.
21392         (cgraph_maybe_hot_edge_p): Likewise.
21394 2013-04-04  Richard Biener  <rguenther@suse.de>
21396         PR tree-optimization/56826
21397         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
21398         more accurately.
21400 2013-04-04  Richard Biener  <rguenther@suse.de>
21402         PR tree-optimization/56213
21403         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
21404         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
21406 2013-04-04  Richard Biener  <rguenther@suse.de>
21408         PR tree-optimization/56837
21409         * tree-loop-distribution.c (classify_partition): For non-zero
21410         values require that the value has the same precision as its
21411         mode to be useful as memset value.
21413 2013-04-03  Nick Clifton  <nickc@redhat.com>
21415         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
21416         (fmssf4): Use fmsf.s on E3V5 architectures.
21417         (fnmasf4): Use fnmaf.s on E3V5 architectures.
21418         (fnmssf4): Use fnmsf.s on E3V5 architectures.
21420 2013-04-03  Jeff Law  <law@redhat.com>
21422         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
21423         (lra-eliminations.o): Likewise.
21425 2013-04-03  Teresa Johnson  <tejohnson@google.com>
21427         * gcov-io.c (compute_working_sets): Moved most of body of old
21428         compute_working_sets here from profile.c.
21429         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
21430         (gcov_working_set_t): Moved typedef here from basic-block.h
21431         (compute_working_set): Declare.
21432         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
21433         (get_working_sets): Renamed from compute_working_set,
21434         replace most of body with call to new compute_working_sets.
21435         (get_exec_counts): Replace call to compute_working_sets
21436         to get_working_sets.
21437         * profile.h (get_working_sets): Renamed from compute_working_set.
21438         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
21439         to get_working_sets.
21440         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
21441         * gcov-dump.c (dump_working_sets): New function.
21443 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
21445         * hwint.c (sext_hwi, zext_hwi): New functions.
21446         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
21447         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
21448         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
21449         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
21450         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
21451         (sext_hwi, zext_hwi): New functions.
21453 2013-04-03  Jeff Law  <law@redhat.com>
21455         PR tree-optimization/56799
21456         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
21457         back test for widening conversion erroneously dropped in prior change.
21459 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21461         PR target/56809
21462         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
21463         instead of next_real_insn.
21465 2013-04-03  Marek Polacek  <polacek@redhat.com>
21467         PR sanitizer/55702
21468         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
21470 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21472         PR target/56809
21473         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
21474         next_real_insn.
21475         (thumb1_output_casesi): Likewise.
21476         (thumb2_output_casesi): Likewise.
21478 2013-04-03  Richard Biener  <rguenther@suse.de>
21480         PR tree-optimization/56817
21481         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21482         Split out ...
21483         (tree_unroll_loops_completely_1): ... new function to manually
21484         walk the loop tree, properly defering outer loops of unrolled
21485         loops to later iterations.
21487 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21489         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21490         (vectorizable_load): Likewise.
21491         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21492         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21494 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21496         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21497         BIT_FIELD_REF.
21499 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21501         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21503 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21505         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21507 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21509         PR tree-optimization/56790
21510         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21511         folding.
21513 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21515         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21516         Handle VEC_MERGE.
21517         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21518         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21519         equal arguments.
21521 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21523         PR c/19449
21524         * tree.h (force_folding_builtin_constant_p): New decl.
21525         * builtins.c (force_folding_builtin_constant_p): New variable.
21526         (fold_builtin_constant_p): Fold immediately also if
21527         force_folding_builtin_constant_p.
21529 2013-04-03  Richard Biener  <rguenther@suse.de>
21531         PR tree-optimization/56812
21532         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21533         DRs of the same interleaving chain are independent.
21535 2013-04-02  Jason Merrill  <jason@redhat.com>
21537         * gdbinit.in (pbb): Use debug fn.
21539 2013-04-02  Lawrence Crowl  <crowl@google.com>
21541         * sese.h (struct ivtype_map_elt_s): Remove unused.
21542         (extern debug_ivtype_map): Remove unused.
21543         (extern eq_ivtype_map_elts): Remove unused.
21544         * sese.c (debug_ivtype_map): Removed unused.
21545         (debug_ivtype_map_1): Removed unused.
21546         (debug_ivtype_elt): Remove unused.
21547         (eq_ivtype_map_elts): Remove unused.
21550 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21552         PR target/52790
21553         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21554         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21555         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21556         function.
21557         (legitimize_pe_coff_symbol): Likewise.
21558         (is_imported_p): New helper-function.
21559         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21560         for Windows x64 targets.
21561         (ix86_expand_prologue): Optimize for pe-coff targets.
21562         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21563         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21564         medium/large code-model.
21565         (legitimize_pic_address): Likewise.
21566         (legitimize_tls_address): Likewise.
21567         (ix86_expand_call): Likewise.
21568         (x86_output_mi_thunk): Likewise.
21569         (get_dllimport_decl): Add new beimport argument.
21570         (construct_plt_address): Don't assert for x64 pe-coff targets.
21571         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21572         targets.
21573         (SYMBOL_FLAG_STUBVAR): New macro.
21574         (SYMBOL_REF_STUBVAR_P): Likewise.
21575         * config/i386/winnt.c (stub_list): New structure.
21576         (stub_head): New local variable.
21577         (i386_pe_record_stub): New function.
21578         (i386_pe_file_end): Emit refptr-stubs.
21580 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21582         PR rtl-optimization/56745
21583         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21584         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21586         PR c++/34949
21587         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21588         and both of them are MEM_REFs, just compare first argument for
21589         equality and attempt to deal even with differing offsets.
21591         PR c++/34949
21592         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21593         of gimple_clobber_p to be MEM_REF.
21594         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21595         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21596         after gimplification.
21597         * asan.c (get_mem_ref_of_assignment): Don't instrument
21598         gimple_clobber_p stmts.
21599         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21600         gimple_clobber_p stmt if they have MEM_REF lhs and
21601         are dead because of another gimple_clobber_p stmt.
21602         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21603         gimple_clobber_p stmts like debug stmts.
21604         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21605         that refer to unused VAR_DECLs or uninitialized values.
21606         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21607         gimple_clobber_p stmts if they refer to removed parameters.
21608         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21609         formatting.
21611 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21613         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21614         using SWI48 mode attribute.
21616 2013-04-02  Wei Mi  <wmi@google.com>
21618         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21619         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21620         *<rotate_insn><mode>3_mask in i386.md.
21622 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21624         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21626 2013-04-02  Richard Biener  <rguenther@suse.de>
21628         PR tree-optimization/56778
21629         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21630         Runtime alias tests are not supported for gather loads.
21631         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21632         stmts referenced from SSA operands before updating SSA form.
21634 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21635             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21637         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21638         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21639         * config/arm/cortex-a53.md: New file.
21640         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21641         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21642         * config/arm/arm.c (arm_issue_rate): Likewise.
21643         * config/arm/arm-tune.md: Regenerate
21644         * config/arm/arm-tables.opt: Regenerate.
21645         * config/arm/arm-cores.def: Add cortex-a53.
21647 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21649         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21650         non-static link.
21652 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21654         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21655         scalar load/store operations using B/H registers.
21656         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21658 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21660         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21661         scalar move.
21662         * config/aarch64/aarch64.c
21663         (aarch64_simd_scalar_immediate_valid_for_move): New.
21664         * config/aarch64/aarch64-protos.h
21665         (aarch64_simd_scalar_immediate_valid_for_move): New.
21666         * config/aarch64/constraints.md (Dh, Dq): New.
21667         * config/aarch64/iterators.md (hq): New.
21669 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21671         * reorg.c (get_branch_condition): Deal with conditional returns.
21672         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21674 2013-04-01  Wei Mi  <wmi@google.com>
21676         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21677         Truncate operand 2 using %b asm operand modifier.
21678         (*<shift_insn><mode>3_mask): Ditto.
21679         (*<rotate_insn><mode>3_mask): Ditto.
21681 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21683         PR middle-end/56798
21684         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21686 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21688         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21689         of next_real_insn.
21690         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21692 2013-03-30  Lawrence Crowl  <crowl@google.com>
21694         * dse.c (clear_alias_sets): Remove never set.
21695         (disqualified_clear_alias_sets): Remove never set.
21696         (clear_alias_mode_pool): Remove never set.
21697         (dse_step0): Remove condition that is never true.
21698         (canon_address): Remove condition that is never true.
21699         (dse_step7): Remove condition that is never true.
21700         (rest_of_handle_dse): Remove condition that is never true.
21701         (rest_of_handle_dse::did_global): Remove never read from above.
21702         (dse_step2_spill): Remove never called from above.
21703         (dse_step5_spill): Remove never called from above.
21705 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21707         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21708         JUMP_TABLE_DATA changes.
21709         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21710         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21711         (Insns) <jump_table_data>: New entry.
21712         * doc/tm.texi: Regenerate.
21714         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21716         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21717         for table jump at the end of a basic block using tablejump_p.
21718         * targhooks.c (default_invalid_within_doloop): Likewise.
21719         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21720         target hook implementation that is identical to the default hook.
21721         (rs6000_invalid_within_doloop): Remove.
21723         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21724         unused variable from tablejump_p call.
21726         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21727         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21728         (INSN_DELETED_P): Likewise.
21729         (emit_jump_table_data): New prototype.
21730         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21731         after 4th as unused.
21732         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21733         * sched-vis.c (print_insn): Likewise.
21734         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21735         insn for compatibility with back ends that use next_active_insn to
21736         identify jump table data.
21737         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21738         (remove_insn): Likewise.
21739         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21740         to be emitted.
21741         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21742         (emit_jump_table_data): New function.
21744         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21745         basic block, a JUMP_TABLE_DATA never is.
21746         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21747         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21748         off from code handling real insns.
21749         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21750         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21751         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21752         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21753         is not a NONDEBUG_INSN_P.
21754         * ira-costs.c (scan_one_insn): Likewise.
21755         * jump.c (mark_all_labels): Likewise.
21756         (mark_jump_label_1): Likewise.
21757         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21758         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21759         a basic block.
21760         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21761         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21762         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21763         JUMP_TABLE_DATA_P insns.
21764         (calculate_elim_costs_all_insns): Likewise.
21765         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21766         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21767         (delete_output_reload): Code style fixups.
21768         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21769         insn flags on this non-insn.
21770         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21771         as scheduling barriers, for pre-change compatibility.
21772         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21773         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21775         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21776         redundant JUMP_TABLE_DATA_P test.
21777         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21778         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21779         (frv_for_each_packet): Likewise.
21780         * config/i386/i386.c (min_insn_size): Likewise.
21781         (ix86_avoid_jump_mispredicts): Likewise.
21782         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21783         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21784         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21785         (mips16_insn_length): Robustify.
21786         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21787         (mips16_split_long_branches): Likewise.
21788         * config/pa/pa.c (pa_combine_instructions): Likewise.
21789         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21790         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21791         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21792         as contributing to pool range lengths.
21793         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21794         Remove redundant JUMP_TABLE_DATA_P test.
21795         (sh_loop_align): Likewise.
21796         (split_branches): Likewise.
21797         (sh_insn_length_adjustment): Likewise.
21798         * config/spu/spu.c (get_branch_target): Likewise.
21800 2013-03-29  Jan Hubicka  <jh@suse.cz>
21802         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21803         gcov streaming; stream hot bb threshold to ltrans.
21804         * predict.c (get_hot_bb_threshold): Break out from ....
21805         (maybe_hot_count_p): ... here.
21806         (set_hot_bb_threshold): New function.
21807         * lto-section-in.c (lto_section_name): Add profile.
21808         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21809         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21810         and data-streamer.h
21811         (histogram_entry): New structure.
21812         (histogram, histogram_pool): New global vars.
21813         (histogram_hash): New structure.
21814         (histogram_hash::hash): New method.
21815         (histogram_hash::equal): Likewise.
21816         (account_time_size): New function.
21817         (cmp_counts): New function.
21818         (dump_histogram): New function.
21819         (ipa_profile_generate_summary): New function.
21820         (ipa_profile_write_summary): New function.
21821         (ipa_profile_read_summary): New function.
21822         (ipa_profile): Decide on threshold.
21823         (pass_ipa_profile): Add ipa_profile_write_summary and
21824         ipa_profile_read_summary.
21825         * Makefile.in (ipa.o): Update dependencies.
21826         * lto-streamer.h (LTO_section_ipa_profile): New section.
21828 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21830         * tree.h (VAR_P): New.
21832 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21834         PR lto/56777
21835         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21837 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21839         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21840         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21841         (control_flow_insn_p): Likewise.
21842         * cfgrtl.c (duplicate_insn_chain): Likewise.
21843         * final.c (get_attr_length_1): Likewise.
21844         (shorten_branches): Likewise.
21845         (final_scan_insn): Likewise.
21846         * function.c (instantiate_virtual_regs): Likewise.
21847         * gcse.c (insert_insn_end_basic_block): Likewise.
21848         * ira-costs.c (scan_one_insn): Likewise.
21849         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21850         * lra.c (check_rtl): Likewise.
21851         * reload1.c (elimination_costs_in_insn): Likewise.
21852         * reorg.c (follow_jumps): Likewise.
21854         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21855         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21856         (thumb_far_jump_used_p): Likewise.
21857         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21858         (workaround_speculation): Likewise.
21859         (add_sched_insns_for_speculation): Likewise.
21860         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21861         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21862         (frv_for_each_packet): Likewise.
21863         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21864         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21865         (final_emit_insn_group_barriers): Likewise.
21866         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21867         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21868         (mips16_insn_length): Likewise.
21869         * config/pa/pa.c (pa_reorg): Likewise.
21870         (pa_combine_instructions): Likewise.
21871         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21872         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21873         (sh_reorg): Likewise.
21874         (split_branches): Likewise.
21875         * config/spu/spu.c (get_branch_target): Likewise.
21877         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21878         JUMP_TABLE_DATA_P.
21880 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21882         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21883         Fix declaration name.
21885 2013-03-28  Lawrence Crowl  <crowl@google.com>
21887         * graphds.h (struct graph.indicies): Remove unused.
21888         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21889         (SCOP_ORIGINAL_PDDRS): Remove unused.
21890         * sese.h (extern insert_loop_close_phis): Removed unused.
21891         (extern insert_guard_phis): Removed unused.
21892         (extern ivtype_map_elt_info): Removed unused.
21893         (new_ivtype_map_elt): Removed unused.
21894         * sese.c (ivtype_map_elt_info): Removed unused.
21896 2013-03-28  Lawrence Crowl  <crowl@google.com>
21898         * Makefile.in: Add several missing include dependences.
21899         (DUMPFILE_H): New.
21900         (test-dump.o): New.  This object is not added to any executable,
21901         but is present for ad-hoc testing.
21902         * bitmap.c
21903         (debug (const bitmap_head_def &)): New.
21904         (debug (const bitmap_head_def *)): New.
21905         * bitmap.h
21906         (extern debug (const bitmap_head_def &)): New.
21907         (extern debug (const bitmap_head_def *)): New.
21908         * cfg.c
21909         (debug (edge_def &)): New.
21910         (debug (edge_def *)): New.
21911         * cfghooks.c
21912         (debug (basic_block_def &)): New.
21913         (debug (basic_block_def *)): New.
21914         * dumpfile.h
21915         (dump_node (const_tree, int, FILE *)): Correct source file.
21916         * dwarf2out.c
21917         (debug (die_struct &)): New.
21918         (debug (die_struct *)): New.
21919         * dwarf2out.h
21920         (extern debug (die_struct &)): New.
21921         (extern debug (die_struct *)): New.
21922         * gimple-pretty-print.c
21923         (debug (gimple_statement_d &)): New.
21924         (debug (gimple_statement_d *)): New.
21925         * gimple-pretty-print.h
21926         (extern debug (gimple_statement_d &)): New.
21927         (extern debug (gimple_statement_d *)): New.
21928         * ira-build.c
21929         (debug (ira_allocno_copy &)): New.
21930         (debug (ira_allocno_copy *)): New.
21931         (debug (ira_allocno &)): New.
21932         (debug (ira_allocno *)): New.
21933         * ira-int.h
21934         (extern debug (ira_allocno_copy &)): New.
21935         (extern debug (ira_allocno_copy *)): New.
21936         (extern debug (ira_allocno &)): New.
21937         (extern debug (ira_allocno *)): New.
21938         * ira-lives.c
21939         (debug (live_range &)): New.
21940         (debug (live_range *)): New.
21941         * lra-int.h
21942         (debug (lra_live_range &)): New.
21943         (debug (lra_live_range *)): New.
21944         * lra-lives.c
21945         (debug (lra_live_range &)): New.
21946         (debug (lra_live_range *)): New.
21947         * omega.c
21948         (debug (omega_pb_d &)): New.
21949         (debug (omega_pb_d *)): New.
21950         * omega.h
21951         (extern debug (omega_pb_d &)): New.
21952         (extern debug (omega_pb_d *)): New.
21953         * print-rtl.c
21954         (debug (const rtx_def &)): New.
21955         (debug (const rtx_def *)): New.
21956         * print-tree.c
21957         (debug_tree (tree): Move within file.
21958         (debug_raw (const tree_node &)): New.
21959         (debug_raw (const tree_node *)): New.
21960         (dump_tree_via_hooks (const tree_node *, int)): New.
21961         (debug (const tree_node &)): New.
21962         (debug (const tree_node *)): New.
21963         (debug_verbose (const tree_node &)): New.
21964         (debug_verbose (const tree_node *)): New.
21965         (debug_head (const tree_node &)): New.
21966         (debug_head (const tree_node *)): New.
21967         (debug_body (const tree_node &)): New.
21968         (debug_body (const tree_node *)): New.
21969         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21970         (debug (vec<tree, va_gc> &)): New.
21971         (debug (vec<tree, va_gc> *)): New.
21972         * rtl.h
21973         (extern debug (const rtx_def &)): New.
21974         (extern debug (const rtx_def *)): New.
21975         * sbitmap.c
21976         (debug_raw (simple_bitmap_def &)): New.
21977         (debug_raw (simple_bitmap_def *)): New.
21978         (debug (simple_bitmap_def &)): New.
21979         (debug (simple_bitmap_def *)): New.
21980         * sbitmap.h
21981         (extern debug (simple_bitmap_def &)): New.
21982         (extern debug (simple_bitmap_def *)): New.
21983         (extern debug_raw (simple_bitmap_def &)): New.
21984         (extern debug_raw (simple_bitmap_def *)): New.
21985         * sel-sched-dump.c
21986         (debug (vinsn_def &)): New.
21987         (debug (vinsn_def *)): New.
21988         (debug_verbose (vinsn_def &)): New.
21989         (debug_verbose (vinsn_def *)): New.
21990         (debug (expr_def &)): New.
21991         (debug (expr_def *)): New.
21992         (debug_verbose (expr_def &)): New.
21993         (debug_verbose (expr_def *)): New.
21994         (debug (vec<rtx> &)): New.
21995         (debug (vec<rtx> *)): New.
21996         * sel-sched-dump.h
21997         (extern debug (vinsn_def &)): New.
21998         (extern debug (vinsn_def *)): New.
21999         (extern debug_verbose (vinsn_def &)): New.
22000         (extern debug_verbose (vinsn_def *)): New.
22001         (extern debug (expr_def &)): New.
22002         (extern debug (expr_def *)): New.
22003         (extern debug_verbose (expr_def &)): New.
22004         (extern debug_verbose (expr_def *)): New.
22005         (extern debug (vec<rtx> &)): New.
22006         (extern debug (vec<rtx> *)): New.
22007         * sel-sched-ir.h
22008         (_list_iter_cond_expr): Make inline instead of static.
22009         * sreal.c
22010         (debug (sreal &)): New.
22011         (debug (sreal *)): New.
22012         * sreal.h
22013         (extern debug (sreal &)): New.
22014         (extern debug (sreal *)): New.
22015         * tree.h
22016         (extern debug_raw (const tree_node &)): New.
22017         (extern debug_raw (const tree_node *)): New.
22018         (extern debug (const tree_node &)): New.
22019         (extern debug (const tree_node *)): New.
22020         (extern debug_verbose (const tree_node &)): New.
22021         (extern debug_verbose (const tree_node *)): New.
22022         (extern debug_head (const tree_node &)): New.
22023         (extern debug_head (const tree_node *)): New.
22024         (extern debug_body (const tree_node &)): New.
22025         (extern debug_body (const tree_node *)): New.
22026         (extern debug (vec<tree, va_gc> &)): New.
22027         (extern debug (vec<tree, va_gc> *)): New.
22028         * tree-cfg.c
22029         (debug (struct loop &)): New.
22030         (debug (struct loop *)): New.
22031         (debug_verbose (struct loop &)): New.
22032         (debug_verbose (struct loop *)): New.
22033         * tree-dump.c: Add header dependence.
22034         * tree-flow.h
22035         (extern debug (struct loop &)): New.
22036         (extern debug (struct loop *)): New.
22037         (extern debug_verbose (struct loop &)): New.
22038         (extern debug_verbose (struct loop *)): New.
22039         * tree-data-ref.c
22040         (debug (data_reference &)): New.
22041         (debug (data_reference *)): New.
22042         (debug (vec<data_reference_p> &)): New.
22043         (debug (vec<data_reference_p> *)): New.
22044         (debug (vec<ddr_p> &)): New.
22045         (debug (vec<ddr_p> *)): New.
22046         * tree-data-ref.h
22047         (extern debug (data_reference &)): New.
22048         (extern debug (data_reference *)): New.
22049         (extern debug (vec<data_reference_p> &)): New.
22050         (extern debug (vec<data_reference_p> *)): New.
22051         (extern debug (vec<ddr_p> &)): New.
22052         (extern debug (vec<ddr_p> *)): New.
22053         * tree-ssa-alias.c
22054         (debug (pt_solution &)): New.
22055         (debug (pt_solution *)): New.
22056         * tree-ssa-alias.h
22057         (extern debug (pt_solution &)): New.
22058         (extern debug (pt_solution *)): New.
22059         * tree-ssa-alias.c
22060         (debug (_var_map &)): New.
22061         (debug (_var_map *)): New.
22062         (debug (tree_live_info_d &)): New.
22063         (debug (tree_live_info_d *)): New.
22064         * tree-ssa-alias.h
22065         (extern debug (_var_map &)): New.
22066         (extern debug (_var_map *)): New.
22067         (extern debug (tree_live_info_d &)): New.
22068         (extern debug (tree_live_info_d *)): New.
22070 2013-03-28  Jan Hubicka  <jh@suse.cz>
22072         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
22074 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
22076         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
22077         record only when desired or required.
22079 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
22081         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
22082         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
22084 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22086         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
22087         (*andsi3_compare0_uxtw): New pattern.
22088         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
22089         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
22091 2013-03-28  Jan Hubicka  <jh@suse.cz>
22093         * data-streamer-in.c (streamer_read_gcov_count): New function.
22094         * gimple-streamer-out.c: Include value-prof.h.
22095         (output_gimple_stmt): Output histogram.
22096         (output_bb): Use streamer_write_gcov_count.
22097         * value-prof.c: Include data-streamer.h
22098         (dump_histogram_value): Add HIST_TYPE_MAX.
22099         (stream_out_histogram_value): New function.
22100         (stream_in_histogram_value): New function.
22101         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
22102         (stream_out_histogram_value, stream_in_histogram_value): Declare.
22103         * data-streamer-out.c (streamer_write_gcov_count): New function.
22104         (streamer_write_gcov_count_stream): New function.
22105         * lto-cgraph.c (lto_output_edge): Update counter streaming.
22106         (lto_output_node): Likewise.
22107         (input_node, input_edge): Likewise.
22108         * lto-streamer-out.c (output_cfg): Update streaming.
22109         * lto-streamer-in.c (input_cfg): Likewise.
22110         * data-streamer.h (streamer_write_gcov_count,
22111         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
22112         * gimple-streamer-in.c: Include value-prof.h
22113         (input_gimple_stmt): Input histograms.
22114         (input_bb): Update profile streaming.
22116 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
22118         * genmodes.c (emit_max_int): New function.
22119         (emit_insn_modes_h): Added call to emit_max_function.
22120         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
22121         Added doc.
22122         * machmode.def: Fixed comment.
22124 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
22126         * combine.c (try_combine): Removed useless assert.
22127         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
22129 2013-03-28  Marek Polacek  <polacek@redhat.com>
22130             Richard Biener  <rguenther@suse.de>
22132         PR tree-optimization/56695
22133         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
22134         build signed result of a vector comparison.
22135         * tree-cfg.c (verify_gimple_comparison): Check that a result
22136         of a vector comparison has signed type.
22138 2013-03-28  Richard Biener  <rguenther@suse.de>
22140         PR tree-optimization/37021
22141         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
22142         do not restrict gaps between groups.
22143         * tree-vect-stmts.c (vectorizable_load): Properly account for
22144         a gap between groups.
22146 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
22148         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
22149         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
22150         is not enabled.
22152 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
22154         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
22155         * doc/extend.texi (Named Address Spaces): Ditto.
22156         (Variable Attributes): Ditto.
22158 2013-03-27  Kai Tietz  <ktietz@redhat.com>
22160         * config.build: Add support for cygwin x64 target.
22161         * config.gcc: Likewise.
22162         * config.host: Likewise.
22163         * configure.ac: Likewise
22164         * configure: Regenerated.
22166 2013-03-27  Kai Tietz  <ktietz@redhat.com>
22168         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
22169         * config/i386/t-cygwin-w64: New file.
22170         * config/i386/cygwin-w64.h: New file.
22171         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
22172         and add support for x64-cygwin target.
22173         (CPP_SPEC): Likewise.
22174         (CXX_WRAP_SPEC_LIST): Undefine before define.
22175         (LIBGCJ_SONAME): Use 15 as version.
22177 2013-03-27  Richard Biener  <rguenther@suse.de>
22179         PR tree-optimization/56716
22180         * tree-ssa-structalias.c (perform_var_substitution): Adjust
22181         dumping for ref nodes.
22183 2013-03-27  Martin Jambor  <mjambor@suse.cz>
22185         PR tree-optimization/55334
22186         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
22187         restricted pointers to arrays.
22189 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
22191         * Makefile.in (.SUFFIXES): Add .cc.
22192         (.c.o): Apply same recipe for implicit rule .cc.o.
22194 2013-03-27  Richard Biener  <rguenther@suse.de>
22196         PR tree-optimization/37021
22197         * tree-vect-data-refs.c (vect_check_strided_load): Allow
22198         REALPART/IMAGPART_EXPRs around the supported refs.
22199         * tree-ssa-structalias.c (find_func_aliases): Assume that
22200         floating-point values are not used to transfer pointers.
22202 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
22204         * target.def (TARGET_HAS_IFUNC_P): New target hook.
22205         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
22206         * doc/tm.texi: Regenerate.
22207         * targhooks.h (default_has_ifunc_p): New.
22208         * targhooks.c (default_has_ifunc_p): Ditto.
22209         * config/linux-protos.h: New file.
22210         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
22211         hook for linux which disables support of indirect functions in android.
22212         * config/linux-android.c: New file.
22213         * config/t-linux-android.c: Ditto.
22214         * config.gcc: Added new object file linux-android.o.
22215         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
22216         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
22217         * varasm.c (do_assemble_alias): Likewise.
22218         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
22219         doesn't support indirect functions.
22220         * configure: Regenerate.
22222 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
22224         PR target/56102
22225         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
22226         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
22227         mult-word mode.
22229 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22231         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
22233 2013-03-27  Terry Guo  <terry.guo@arm.com>
22235         * config/arm/arm-cores.def: Added core cortex-r7.
22236         * config/arm/arm-tune.md: Regenerated.
22237         * config/arm/arm-tables.opt: Regenerated.
22238         * doc/invoke.texi: Added entry for core cortex-r7.
22240 2013-03-27  Walter Lee  <walt@tilera.com>
22242         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
22243         double-decrement of next_scratch_regno.
22245 2013-03-27  Walter Lee  <walt@tilera.com>
22247         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
22248         input operands.
22249         (insn_v1mulus): Ditto.
22250         (insn_v2muls): Ditto.
22252 2013-03-27  Walter Lee  <walt@tilera.com>
22254         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
22255         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
22257 2013-03-27  Walter Lee  <walt@tilera.com>
22259         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
22260         (*sibcall_value): Ditto.
22262 2013-03-27  Walter Lee  <walt@tilera.com>
22264         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
22265         (insn_mnz_v8qi): ... this ...
22266         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
22267         vector equivalent.
22268         (insn_v<n>mnz): Replaced by ...
22269         (insn_v1mnz): ... this ...
22270         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
22271         equivalent.
22272         (insn_mz_<mode>): Replaced by ...
22273         (insn_mz_v8qi): ... this ...
22274         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
22275         vector equivalent.
22276         (insn_v<n>mz): Replaced by ...
22277         (insn_v1mz): ... this ...
22278         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
22279         equivalent.
22281 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
22283         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
22285 2013-03-26  Roland McGrath  <mcgrathr@google.com>
22287         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
22288         than fprintf with a non-constant, non-format string.
22290 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
22292         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
22293         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
22294         operand 0 predicate.
22295         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
22296         attribute.  Use general_x64nomem_operand as operand 1 predicate.
22297         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
22298         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
22299         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
22300         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
22301         (mov<mode>_insv_1): Remove expander.  Merge insn with
22302         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
22303         Use general_x64nomem_operand as operand 1 predicate.
22304         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
22305         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
22306         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
22307         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
22308         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
22309         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
22310         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
22311         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
22312         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
22313         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
22314         (general_x64nomem_operand): Ditto.
22316 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
22318         * config/rtems.opt: Add -pthread option.
22320 2013-03-26  Richard Biener  <rguenther@suse.de>
22322         * alias.c (find_base_term): Avoid redundant and not used recursion.
22323         (base_alias_check): Get the initial base term from the caller.
22324         (true_dependence_1): Compute and pass base terms to base_alias_check.
22325         (write_dependence_p): Likewise.
22326         (may_alias_p): Likewise.
22328 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
22330         * config/aarch64/aarch64.c (aarch64_classify_address): Support
22331         PC-relative load in SI modes and above only.
22333 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
22335         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
22336         * config/arm/iwmmxt.md (WCGR0): Update.
22337         (WCGR1, WCGR2, WCGR3): Likewise.
22339 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
22341         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
22342         Use x64 and nox64 isa attributes.
22344 2013-03-26  Richard Biener  <rguenther@suse.de>
22346         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
22347         alignment computations and rely on get_object_alignment_1
22348         for the !TYPE_P case.
22349         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
22351 2013-03-26  Walter Lee  <walt@tilera.com>
22353         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
22354         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
22356 2013-03-25  Jeff Law  <law@redhat.com>
22358         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
22359         check for INTEGRAL_TYPE_P that was missing due to checking in
22360         wrong version of prior patch.
22362 2013-03-25  Walter Lee  <walt@tilera.com>
22364         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
22365         TILEGX_INSN_SHUFFLEBYTES1.
22366         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
22367         shufflebytes1.
22368         (tilegx_builtins): Ditto.
22369         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
22371 2013-03-25  Walter Lee  <walt@tilera.com>
22373         * config/tilegx/tilegx.md (floatsisf2): New pattern.
22374         (floatunssisf2): New pattern.
22375         (floatsidf2): New pattern.
22376         (floatunssidf2): New pattern.
22378 2013-03-25  Walter Lee  <walt@tilera.com>
22380         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
22381         tests for constraint J, K, N, P.
22383 2013-03-25  Walter Lee  <walt@tilera.com>
22385         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
22386         Use indirect/pcrel encoding.
22387         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
22388         Ditto.
22390 2013-03-25  Steve Ellcey  <sellcey@mips.com>
22392         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
22393         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
22394         * config/mips/mips.c (mips_option_override): Set IMADD default.
22395         * config/mips/mips.h (PTF_AVOID_IMADD): New.
22396         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
22397         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
22398         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
22399         * doc/invoke.texi (-mimadd/-mno-imadd): New.
22401 2013-03-25  Jeff Law  <law@redhat.com>
22403         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
22404         slightly to avoid creating and folding useless trees.  Simplify
22405         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
22407 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
22409         * config/i386/i386.md (*zero_extendsidi2): Merge with
22410         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
22411         * config/i386/predicates.md (x86_64_zext_operand): Rename from
22412         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
22413         targets.  Clarify comment.
22415 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22417         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
22418         pass-through jump functions differently.
22419         (ipa_read_jump_function): Likewise.  Also use setter functions to set
22420         up jump functions.
22422 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22424         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
22425         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
22426         process it.
22427         (ipa_get_indirect_edge_target): New function.
22428         (devirtualization_time_bonus): New parameter known_aggs, pass it to
22429         ipa_get_indirect_edge_target.  Update all callers.
22430         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
22431         ipa_get_indirect_edge_target_1 instead of calling
22432         ipa_get_indirect_edge_target.
22433         (create_specialized_node): Pass aggvlas to
22434         ipcp_discover_new_direct_edges.
22436 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22438         * config/arm/arm.md (f_sels, f_seld): New types.
22439         (*cmov<mode>): New pattern.
22440         * config/arm/predicates.md (arm_vsel_comparison_operator): New
22441         predicate.
22443 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22445         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
22446         POSIX-printf for mingw-hosted  builds.
22448 2013-03-25  Richard Biener  <rguenther@suse.de>
22450         PR middle-end/56694
22451         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
22452         must-not-throw stmt location.
22454 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22456         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
22457         Emit load-acquire versions when acq is true.
22458         (arm_emit_store_exclusive): Add rel parameter.
22459         Emit store-release versions when rel is true.
22460         (arm_split_compare_and_swap): Use acquire-release instructions
22461         instead.
22462         of barriers when appropriate.
22463         (arm_split_atomic_op): Likewise.
22464         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
22465         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
22466         (VUNSPEC_SLX): Likewise.
22467         (VUNSPEC_LDA): Likewise.
22468         (VUNSPEC_STL): Likewise.
22469         * config/arm/sync.md (atomic_load<mode>): New pattern.
22470         (atomic_store<mode>): Likewise.
22471         (arm_load_acquire_exclusive<mode>): Likewise.
22472         (arm_load_acquire_exclusivesi): Likewise.
22473         (arm_load_acquire_exclusivedi): Likewise.
22474         (arm_store_release_exclusive<mode>): Likewise.
22476 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22478         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22479         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22480         * config/mip/predicates.md (lwsp_swsp_operand,
22481         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22482         sb16_operand, db4_operand, db7_operand, ib3_operand,
22483         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22484         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22485         andi16_operand): New predicates.
22486         * config/mips/mips.md (compression): New attribute.
22487         (enabled): New attribute.
22488         (length): Consider compression in computing length.
22489         (shift_compression): New code attribute.
22490         (*add<mode>3): New operands. Record compression.
22491         (sub<mode>3): Likewise.
22492         (one_cmpl<mode>2): Likewise.
22493         (*and<mode>3): Likewise.
22494         (*ior<mode>3): Likewise.
22495         (unnamed pattern for xor): Likewise.
22496         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22497         (*<optab><mode>3): Likewise.
22498         (*mov<mode>_internal: Likewise.
22499         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22500         (mips_unsigned_immediate_p): New.
22501         (umips_lwsp_swsp_address_p): New.
22502         (m16_based_address_p): New.
22503         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22504         (mips_unsigned_immediate_p): New prototype.
22505         (lwsp_swsp_address_p): New prototype.
22506         (m16_based_address_p): New prototype.
22507         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22508         (mips_signed_immediate_p): New function.
22509         (m16_based_address_p): New function.
22510         (lwsp_swsp_address_p): New function.
22511         (mips_print_operand_punctuation): Recognize short delay slot insns
22512         for microMIPS.add<mode>3"
22514 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22516         PR target/56720
22517         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22518         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22520 2013-03-25  Richard Biener  <rguenther@suse.de>
22522         PR tree-optimization/56689
22523         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22524         any edge.
22526 2013-03-25  Richard Biener  <rguenther@suse.de>
22528         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22529         of bitmap.
22530         (memory_references): Likewise.
22531         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22532         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22533         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22534         (gather_mem_refs_in_loops): Fold into ...
22535         (analyze_memory_references): ... this.  Move initialization
22536         to tree_ssa_lim_initialize.
22537         (fill_always_executed_in): Rename to ...
22538         (fill_always_executed_in_1): ... this.
22539         (fill_always_executed_in): Move contains_call computation to
22540         this new function from ...
22541         (tree_ssa_lim_initialize): ... here.
22542         (tree_ssa_lim): Call fill_always_executed_in.
22544 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22546         * postreload.c (reload_combine): Fix code detecting returns.
22548 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22550         * function.c (emit_use_return_register_into_block): On cc0 targets,
22551         do not emit the sequence between cc0 setter and user.
22553 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22555         * config/i386/predicates.md (local_symbolic_operand): Interpret
22556         dll-imported symbols as none-local.
22558 2013-03-25  Richard Biener  <rguenther@suse.de>
22560         * tree-ssa-loop-im.c (struct depend): Remove.
22561         (struct lim_aux_data): Make depends a vec of gimples.
22562         (free_lim_aux_data): Adjust.
22563         (add_dependency): Likewise.
22564         (set_level): Likewise.
22566 2013-03-25  Richard Biener  <rguenther@suse.de>
22568         PR middle-end/56434
22569         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22570         the pointer returned by calls with ECF_MALLOC set.
22572 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22574         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22576 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22578         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22579         using MMXMODE mode iterator.
22580         (*move<mode>_internal): Merge with *movv2sf_internal and
22581         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22583 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22585         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22586         (record_last_mem_set_info): Likewise.
22588         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22589         of XNEWVEC followed by memset.
22590         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22592 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22594         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22595         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22596         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22597         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22598         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22599         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22600         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22601         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22602         BARRIER_P instead of GET_CODE.
22604 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22606         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22607         inaccuracy in the probing code.
22609         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22610         (ctrapdi4): Likewise.
22612 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22614         * calls.c (expand_call): Add missing guard to code handling return
22615         of non-BLKmode structures in MSB.
22616         * function.c (expand_function_end): Likewise.
22618 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22620         * combine.c (try_combine): Adjust comment.  Do not add the set of
22621         insn #0 if the destination indirectly is set or dies in insn #2.
22622         Tidy up code to distribute a new note.
22624 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22626         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22627         also for alternatives 16 and 17.
22629 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22631         * config/i386/sse.md (*mov<mode>_internal): Merge with
22632         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22633         Emit insn template depending on type attribute.  Use
22634         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22635         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22636         attribute calculation.  Remove unit attribute calculation.
22637         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22638         Set prefix_data16 attribute for DImode ssemov types.
22639         Use Ym instead of y for SSE-MMX conversion alternatives.
22640         Reorder operand constraints.
22642 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22644         * df.h (df_insn_delete): Adjust prototype.
22645         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22646         and let it decide whether mark the basic block dirty.
22647         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22648         * df-scan.c (df_insn_info_delete): New helper function, split
22649         off from df_insn_delete.
22650         (df_scan_free_bb_info): Use it.
22651         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22652         Likewise.
22653         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22654         that the insn is actually an insn and it has a non-NULL basic block.
22655         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22657 2013-03-22  Richard Biener  <rguenther@suse.de>
22659         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22660         dep_ref members.
22661         (mem_ref_alloc): Do not allocate them.
22662         (refs_independent_p): Do not query or maintain a cache.
22664 2013-03-22  Richard Biener  <rguenther@suse.de>
22666         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22667         (gather_mem_refs_in_loops): Do not compute it.
22668         (analyze_memory_references): Do not allocate it.
22669         (tree_ssa_lim_finalize): Do not free it.
22670         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22672 2013-03-22  Richard Biener  <rguenther@suse.de>
22674         * is-a.h (as_a): Use gcc_checking_assert.
22676 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22678         * config/aarch64/aarch64.c (aarch64_print_operand): New
22679         format specifier for printing a constant in hex.
22680         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22681         format specifier for printing second operand.
22683 2013-03-22  Richard Biener  <rguenther@suse.de>
22685         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22686         bitmaps.
22687         (gather_mem_refs_in_loops): Perform store accumulation here.
22688         (create_vop_ref_mapping_loop): Remove.
22689         (create_vop_ref_mapping): Likewise.
22690         (analyze_memory_references): Initialize refs_stored_in_loop.
22691         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22692         (record_indep_loop): Remove.
22693         (record_dep_loop): New function.
22694         (ref_indep_loop_p_1): Adjust to only walk over references
22695         in the loop, not its subloops.
22696         (ref_indep_loop_p): Rename to ...
22697         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22698         maintaining a more fine-grained cache.
22699         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22700         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22702 2013-03-22  Richard Biener  <rguenther@suse.de>
22704         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22705         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22706         aggregate mem_ref_loc.
22707         (free_mem_ref_locs): Inline into ...
22708         (memref_free): ... this and adjust.
22709         (mem_ref_alloc): Adjust.
22710         (mem_ref_locs_alloc): Remove.
22711         (record_mem_ref_loc): Adjust.
22712         (get_all_locs_in_loop): Rewrite into ...
22713         (for_all_locs_in_loop): ... this iterator.
22714         (rewrite_mem_ref_loc): New functor.
22715         (rewrite_mem_refs): Use for_all_locs_in_loop.
22716         (sm_set_flag_if_changed): New functor.
22717         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22718         (ref_always_accessed): New functor.
22719         (ref_always_accessed_p): Use for_all_locs_in_loop.
22721 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22723         * tree-pass.h (PROP_gimple_lvec): New.
22724         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22725         (init_optimization_passes): Move pass_lower_vector.
22726         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22727         PROP_gimple_lvec.
22728         (pass_lower_vector): Provide PROP_gimple_lvec.
22729         (pass_lower_vector_ssa): Likewise.
22730         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22732 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22734         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22736 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22738         * config/i386/i386.md (*movdi_internal): Disparage slightly
22739         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22740         conversion alternatives.
22742 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22744         PR middle-end/48087
22745         * diagnostic.def (DK_WERROR): New kind.
22746         * diagnostic.h (werrorcount): Define.
22747         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22748         promoted to DK_ERROR, increment DK_WERROR counter instead of
22749         DK_ERROR counter.
22750         * toplev.c (toplev_main): Call print_ignored_options even if
22751         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22752         even if just werrorcount is non-zero.
22754         PR debug/55608
22755         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22756         on failure.
22757         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22758         (string_cst_pool_decl): New function.
22759         (optimize_one_addr_into_implicit_ptr): New function.
22760         (resolve_addr_in_expr): Optimize DWARF location expression
22761         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22762         which doesn't live in memory, but has DW_AT_location or
22763         DW_AT_const_value, or refers to a string literal, into
22764         DW_OP_GNU_implicit_pointer.
22765         (optimize_location_into_implicit_ptr): New function.
22766         (resolve_addr): If removing DW_AT_location of a variable because
22767         it was DW_OP_addr of address of the variable, but the variable doesn't
22768         live in memory, try to emit const value attribute for the initializer.
22770 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22772         * tree.h (VECTOR_TYPE_P): New macro.
22773         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22774         TYPE_MODE): Use it.
22775         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22776         VEC_COND_EXPR cannot be lvalues.
22777         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22779 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22781         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22782         Restrict the transformation to equal modes.
22784 2013-03-21  Richard Biener  <rguenther@suse.de>
22786         PR tree-optimization/39326
22787         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22788         (MEM_ANALYZABLE): Adjust.
22789         (record_mem_ref_loc): Move bitmap ops ...
22790         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22791         unanalyzable refs, do not record locations for it.
22792         (analyze_memory_references): Allocate ref zero as shared
22793         unanalyzable ref.
22794         (refs_independent_p): Do not test for unanalyzed mems here.
22795         (ref_indep_loop_p_1): Special-case disambiguation against
22796         the unanalyzed ref.
22797         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22799 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22801         * config/arm/arm-protos.h (tune_params): Add
22802         prefer_neon_for_64bits field.
22803         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22804         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22805         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22806         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22807         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22808         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22809         (arm_option_override): Handle -mneon-for-64bits new option.
22810         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22811         (prefer_neon_for_64bits): Declare new variable.
22812         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22813         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22814         (arch_enabled): Handle new arch types. Remove support for onlya8
22815         and nota8.
22816         (one_cmpldi2): Use new arch names.
22817         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22818         * config/arm/arm.opt (mneon-for-64bits): Add option.
22819         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22820         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22821         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22822         of onlya8.
22823         * doc/invoke.texi (-mneon-for-64bits): Document.
22825 2013-03-21  Richard Biener  <rguenther@suse.de>
22827         PR tree-optimization/39326
22828         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22829         (sort_bbs_in_loop_postorder_cmp): New function.
22830         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22832 2013-03-21  Richard Biener  <rguenther@suse.de>
22834         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22835         (vect_insert_into_interleaving_chain): Likewise.
22836         (vect_drs_dependent_in_basic_block): Inline ...
22837         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22838         split out from ...
22839         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22840         (vect_check_interleaving): Simplify.
22841         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22842         (vect_slp_analyze_data_ref_dependences): ... this new function.
22843         (dr_group_sort_cmp): New function.
22844         (vect_analyze_data_ref_accesses): Compute data-reference groups
22845         here instead of in vect_analyze_data_ref_dependence.  Use
22846         a more efficient algorithm.
22847         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22848         vect_slp_analyze_data_ref_dependences.  Call
22849         vect_analyze_data_ref_accesses earlier.
22850         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22851         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22852         (vect_slp_analyze_data_ref_dependences): New prototype.
22854 2013-03-21  Richard Biener  <rguenther@suse.de>
22856         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22857         ref is stored in the loop.
22858         (find_refs_for_sm): Walk only over all stores.
22859         (store_motion_loop): Allocate from lim_bitmap_obstack.
22860         (store_motion): Likewise.
22862 2013-03-21  Richard Biener  <rguenther@suse.de>
22864         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22865         Update virtual SSA form.
22867 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22869         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22870         * configure: Regenerate.
22871         * config.in: Regenerate.
22872         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22873         if !HAVE_LD_EH_FRAME_CIEV3.
22875 2013-03-21  Richard Biener  <rguenther@suse.de>
22877         * tree-cfg.c (verify_expr_no_block): New function.
22878         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22879         nor DECL_VALUE_EXPR have locations with associated blocks.
22880         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22881         (clear_unused_block_pointer): Remove code dealing with
22882         blocks in DECL_DEBUG_EXPR locations.
22884 2013-03-21  Richard Biener  <rguenther@suse.de>
22886         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22887         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22888         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22889         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22890         instead of DECL_DEBUG_EXPR_IS_FROM.
22891         * gimplify.c (gimplify_modify_expr): Likewise.
22892         * tree-cfg.c (verify_expr_location_1): Likewise.
22893         * tree-complex.c (create_one_component_var): Likewise.
22894         * tree-sra.c (create_access_replacement): Likewise.
22895         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22896         (clear_unused_block_pointer): Likewise.
22897         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22898         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22899         * var-tracking.c (var_debug_decl): Likewise.
22900         (track_expr_p): Likewise.
22901         * tree-inline.c (add_local_variables): Likewise.  Set
22902         DECL_HAS_DEBUG_EXPR_P after copying it.
22903         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22904         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22906 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22908         PR bootstrap/56656
22909         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22910         * configure: Regenerate.
22911         * config.in: Regenerate.
22912         * config/i386/i386.md (*movdf_internal): Use
22913         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22914         movd instead of movq mnemonic for interunit moves.
22915         (*movdi_internal): Ditto.
22917 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22919         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22920         (abd<mode>_3): New pattern.
22921         (aba<mode>_3): New pattern.
22922         (fabd<mode>_3): New pattern.
22924 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22926         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22927         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22928         occurrence of REGISTER_PREFIX as its empty string.
22930 2013-03-20  Jeff Law  <law@redhat.com>
22932         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22933         addititional equivalences for equality comparisons between an SSA_NAME
22934         and a constant where the SSA_NAME was set from a widening conversion.
22936 2013-03-20  Walter Lee  <walt@tilera.com>
22938         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22940 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22942         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22943         depending on type attribute.
22944         (*movti_internal): Ditto.
22945         (*movtf_internal): Ditto.
22946         (*movxf_internal): Ditto.
22947         (*movdf_internal): Ditto.
22948         (*movsf_internal): Ditto.
22950 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22952         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22953         maybe_vex for sselog1 and ssemov types.
22954         (*movdi_internal): Reorder operand constraints.
22955         (*movsi_internal): Ditto.  Set prefix attribute to
22956         maybe_vex for sselog1 and ssemov types.
22957         (*movtf_internal): Set prefix attribute to maybe_vex
22958         for sselog1 and ssemov types.
22959         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22960         DImode ssemov types.  Reorder operand constraints.
22961         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22962         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22963         attribute for SImode ssemov types.  Reorder operand constraints.
22965 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22967         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22968         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22970 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22972         * config/rs6000/predicates.md (indexed_address, update_address_mem
22973         update_indexed_address_mem): New predicates.
22974         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22975         attribute for load/store instructions.
22976         * config/rs6000/dfp.md (movsd_store): Likewise.
22977         (movsd_load): Likewise.
22978         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22979         (unnamed HI->DI extend define_insn): Likewise.
22980         (unnamed SI->DI extend define_insn): Likewise.
22981         (unnamed QI->SI extend define_insn): Likewise.
22982         (unnamed QI->HI extend define_insn): Likewise.
22983         (unnamed HI->SI extend define_insn): Likewise.
22984         (unnamed HI->SI extend define_insn): Likewise.
22985         (extendsfdf2_fpr): Likewise.
22986         (movsi_internal1): Likewise.
22987         (movsi_internal1_single): Likewise.
22988         (movhi_internal): Likewise.
22989         (movqi_internal): Likewise.
22990         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22991         attribute for load/store instructions.
22992         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22993         instructions.
22994         (mov<mode>_softfloat): Likewise.
22995         (mov<mode>_hardfloat32): Likewise.
22996         (mov<mode>_hardfloat64): Likewise.
22997         (mov<mode>_softfloat64): Likewise.
22998         (movdi_internal32): Likewise.
22999         (movdi_internal64): Likewise.
23000         (probe_stack_<mode>): Likewise.
23002 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
23004         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
23005         floating point, and decimal floating point to reload iterator.
23007         * config/rs6000/constraints.md (wl constraint): New constraints to
23008         return FLOAT_REGS if certain options are used to reduce the number
23009         of separate patterns that exist in the file.
23010         (wx constraint): Likewise.
23011         (wz constraint): Likewise.
23013         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
23014         -mdebug=reg, print wg, wl, wx, and wz constraints.
23015         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
23016         Initialize the reload functions for 64-bit binary/decimal floating
23017         point types.
23018         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
23019         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
23020         create the buffer on the stack to overcome not having a 32-bit
23021         load and store.
23022         (rs6000_emit_move): Likewise.
23023         (rs6000_secondary_memory_needed_rtx): Likewise.
23024         (rs6000_alloc_sdmode_stack_slot): Likewise.
23025         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
23026         via xxlxor, just like DFmode 0.0.
23028         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
23029         define as 1 if we are running on a power7 or newer.
23030         (enum r6000_reg_class_enum): Add new constraints.
23032         * config/rs6000/dfp.md (movsd): Delete, combine with binary
23033         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
23034         with other moves by using conditional constraits (wg).  Use LFIWZX
23035         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
23036         (movsd splitter): Likewise.
23037         (movsd_hardfloat): Likewise.
23038         (movsd_softfloat): Likewise.
23040         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
23041         binary and decimal floating point moves.
23042         (fmove_ok): New attributes to combine binary and decimal floating
23043         point moves, and to combine power6x (mfpgpr) moves along normal
23044         floating moves.
23045         (real_value_to_target): Likewise.
23046         (f32_lr): Likewise.
23047         (f32_lm): Likewise.
23048         (f32_li): Likewise.
23049         (f32_sr): Likewise.
23050         (f32_sm): Likewise.
23051         (f32_si): Likewise.
23052         (movsf): Combine binary and decimal floating point moves.  Combine
23053         power6x (mfpgpr) moves with other moves by using conditional
23054         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
23055         (mov<mode> for SFmode/SDmode); Likewise.
23056         (SFmode/SDmode splitters): Likewise.
23057         (movsf_hardfloat): Likewise.
23058         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
23059         (movsf_softfloat): Likewise.
23060         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
23062         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
23063         wx and wz constraints.
23065         * config/rs6000/constraints.md (wg constraint): New constraint to
23066         return FLOAT_REGS if -mmfpgpr (power6x) was used.
23068         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
23069         constraint.
23071         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
23072         -mdebug=reg, print wg, wl, wx, and wz constraints.
23073         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
23074         Initialize the reload functions for 64-bit binary/decimal floating
23075         point types.
23076         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
23077         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
23078         create the buffer on the stack to overcome not having a 32-bit
23079         load and store.
23080         (rs6000_emit_move): Likewise.
23081         (rs6000_secondary_memory_needed_rtx): Likewise.
23082         (rs6000_alloc_sdmode_stack_slot): Likewise.
23083         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
23084         via xxlxor, just like DFmode 0.0.
23086         * config/rs6000/dfp.md (movdd): Delete, combine with binary
23087         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
23088         with other moves by using conditional constraits (wg).  Use LFIWZX
23089         and STFIWX for loading SDmode on power7.
23090         (movdd splitters): Likewise.
23091         (movdd_hardfloat32): Likewise.
23092         (movdd_softfloat32): Likewise.
23093         (movdd_hardfloat64_mfpgpr): Likewise.
23094         (movdd_hardfloat64): Likewise.
23095         (movdd_softfloat64): Likewise.
23097         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
23098         64-bit binary and decimal floating point moves.
23099         (FMOVE64X): Likewise.
23100         (movdf): Combine 64-bit binary and decimal floating point moves.
23101         Combine power6x (mfpgpr) moves with other moves by using
23102         conditional constraits (wg).
23103         (mov<mode> for DFmode/DDmode): Likewise.
23104         (DFmode/DDmode splitters): Likewise.
23105         (movdf_hardfloat32): Likewise.
23106         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
23107         (movdf_softfloat32): Likewise.
23108         (movdf_hardfloat64_mfpgpr): Likewise.
23109         (movdf_hardfloat64): Likewise.
23110         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
23111         (movdf_softfloat64): Likewise.
23112         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
23113         (reload_<mode>_load): Move to later in the file so they aren't in
23114         the middle of the floating point move insns.
23115         (reload_<mode>_store): Likewise.
23117         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
23118         constraint.
23120         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
23121         constraint if -mdebug=reg.
23122         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
23123         Enable using dd reload support if needed.
23125         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
23126         binary and decimal floating point moves in rs6000.md.
23127         (movtd_internal): Likewise.
23129         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
23130         decimal floating point moves.
23131         (movtf): Likewise.
23132         (movtf_internal): Likewise.
23133         (mov<mode>_internal, TDmode/TFmode): Likewise.
23134         (movtf_softfloat): Likewise.
23135         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
23137         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
23138         movdi_internal64, using wg constraint for move direct operations.
23139         (movdi_internal64): Likewise.
23141         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
23142         MODES_TIEABLE_P for selected modes.  Print the numerical value of
23143         the various virtual registers. Use GPR/FPR first/last values,
23144         instead of hard coding the register numbers.  Print which modes
23145         have reload functions registered.
23146         (rs6000_option_override_internal): If -mdebug=reg, trace the options
23147         settings before/after setting cpu, target and subtarget settings.
23148         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
23149         and for secondary reload failures in rs6000_secondary_reload_inner.
23150         (rs6000_secondary_reload_fail): Likewise.
23151         (rs6000_secondary_reload_inner): Likewise.
23153         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
23154         macros for first/last GPR and FPR registers.
23155         (LAST_GPR_REGNO): Likewise.
23156         (FIRST_FPR_REGNO): Likewise.
23157         (LAST_FPR_REGNO): Likewise.
23159         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
23160         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
23161         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
23162         (vcond<mode><mode>): Likewise.
23163         (vcondu<mode><mode>): Likewise.
23164         (vector_gtu<mode>): Likewise.
23165         (vector_gte<mode>): Likewise.
23166         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
23167         to prevent the compiler from converting DImode operations to TImode.
23168         (ior<mode>3): Likewise.
23169         (and<mode>3): Likewise.
23170         (one_cmpl<mode>2): Likewise.
23171         (nor<mode>3): Likewise.
23172         (andc<mode>3): Likewise.
23174         * config/rs6000/constraints.md (wt constraint): New constraint
23175         that returns VSX_REGS if TImode is allowed in VSX registers.
23177         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
23178         constant under VSX.
23180         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
23181         similar to TImode, but it is restricted to being in the GPRs.
23183         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
23184         TImode to occupy a single VSX register.
23186         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
23187         -mvsx-timode for power7/power8.
23188         (power7 cpu): Likewise.
23189         (power8 cpu): Likewise.
23191         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
23192         sure that TFmode/TDmode take up two registers if they are ever
23193         allowed in the upper VSX registers.
23194         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
23195         registers.
23196         (rs6000_init_hard_regno_mode_ok): Likewise.
23197         (rs6000_debug_reg_global): Add debugging for PTImode and wt
23198         constraint.  Print if LRA is turned on.
23199         (rs6000_option_override_internal): Give an error if -mvsx-timode
23200         and VSX is not enabled.
23201         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
23202         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
23203         to reg+offset addressing.  Use PTImode when checking offset
23204         addresses for validity.
23205         (reg_offset_addressing_ok_p): Likewise.
23206         (rs6000_legitimate_offset_address_p): Likewise.
23207         (rs6000_legitimize_address): Likewise.
23208         (rs6000_legitimize_reload_address): Likewise.
23209         (rs6000_legitimate_address_p): Likewise.
23210         (rs6000_eliminate_indexed_memrefs): Likewise.
23211         (rs6000_emit_move): Likewise.
23212         (rs6000_secondary_reload): Likewise.
23213         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
23214         reloads to fpr registers to continue to use reg+offset addressing,
23215         but 64-bit reloads to altivec registers need reg+reg addressing.
23216         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
23217         it.  Treat LO_SUM like a PLUS operation.
23218         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
23219         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
23220         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
23221         registers to share a register with a smaller sized type, since VSX
23222         puts scalars in the upper 64-bits.
23223         (print_operand): Add support for PTImode.
23224         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
23225         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
23226         registers, but don't have arithmetic support.
23227         (rs6000_memory_move_cost): Add test for VSX.
23228         (rs6000_opt_masks): Add -mvsx-timode.
23230         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
23231         for TImode.
23232         (VSs): Likewise.
23233         (VSr): Use wt constraint for TImode.
23234         (VSv): Drop TImode support.
23235         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
23236         (vsx_movti_64bit): Likewise.
23237         (vsx_movti_32bit): Likewise.
23238         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
23239         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
23240         one '?' on the appropriate output constraint.  Do not allow TImode
23241         logical operations on 32-bit systems.
23242         (vsx_ior<mode>3): Likewise.
23243         (vsx_xor<mode>3): Likewise.
23244         (vsx_one_cmpl<mode>2): Likewise.
23245         (vsx_nor<mode>3): Likewise.
23246         (vsx_andc<mode>3): Likewise.
23247         (vsx_concat_<mode>): Likewise.
23248         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
23250         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
23251         OPTION_MASK_VSX_TIMODE.
23252         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
23253         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
23255         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
23256         (TI2 iterator): New iterator for TImode, PTImode.
23257         (wd mode attribute): Add values for vector types.
23258         (movti_string): Replace TI move operations with operations for TImode
23259         and PTImode.  Add support for TImode being allowed in VSX registers.
23260         (mov<mode>_string, TImode/PTImode): Likewise.
23261         (movti_ppc64): Likewise.
23262         (mov<mode>_ppc64, TImode/PTImode): Likewise.
23263         (TI mode splitters): Likewise.
23265         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
23266         constraint.
23268 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
23270         PR tree-optimization/56355
23271         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
23272         Also handle integers with undefined overflow.
23274 2013-03-20  Catherine Moore  <clm@codesourcery.com>
23275             Maciej W. Rozycki  <macro@codesourcery.com>
23276             Tom de Vries  <tom@codesourcery.com>
23277             Nathan Sidwell  <nathan@codesourcery.com>
23278             Iain Sandoe  <iain@codesourcery.com>
23279             Nathan Froyd  <froydnj@codesourcery.com>
23280             Chao-ying Fu  <fu@mips.com>
23282         * doc/extend.texi: (micromips, nomicromips, nocompression):
23283         Document new function attributes.
23284         * doc/invoke.texi (minterlink-compressed, mmicromips,
23285         m14k, m14ke, m14kec): Document new options.
23286         (minterlink-mips16): Update documentation.
23287         * doc/md.texi (ZC, ZD): Document new constraints.
23288         * configure.ac (gcc_cv_as_micromips): Check if linker
23289         supports the .set micromips directive.
23290         * configure: Regenerate.
23291         * config.in: Regenerate.
23292         * config/mips/mips-tables.opt: Regenerate.
23293         * config/mips/micromips.md: New file.
23294         * constraints.md (ZC, ZD): New constraints.
23295         * config/mips/predicates.md (movep_src_register): New predicate.
23296         (movep_src_operand): New predicate.
23297         (non_volatile_mem_operand): New predicate.
23298         * config/mips/mips.md (multimem): New type.
23299         (length): Differentiate between 17-bit and 18-bit branch offsets.
23300         (MOVEP1, MOVEP2): New mode iterator.
23301         (mov_<load>l): Use ZC constraint.
23302         (mov_<load>r): Likewise.
23303         (mov_<store>l): Likewise.
23304         (mov_<store>r): Likewise.
23305         (*branch_equality<mode>_inverted): Add microMIPS support.
23306         (*branch_equality<mode>): Likewise.
23307         (*jump_absolute): Likewise.
23308         (indirect_jump_<mode>): Likewise.
23309         (tablejump_<mode>): Likewise.
23310         (<optab>_internal): Likewise.
23311         (sibcall_internal): Likewise.
23312         (sibcall_value_internal): Likewise.
23313         (prefetch): Use constraint ZD.
23314         * config/mips/mips.opt (minterlink-compressed): New option.
23315         (minterlink-mips16): Now an alias for minterlink-compressed.
23316         (mmicromips): New option.
23317         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
23318         (compare_and_swap_12): Likewise.
23319         (sync_add<mode>): Likewise.
23320         (sync_<optab>_12): Likewise.
23321         (sync_old_<optab>_12): Likewise.
23322         (sync_new_<optab>_12): Likewise.
23323         (sync_nand_12): Likewise.
23324         (sync_old_nand_12): Likewise.
23325         (sync_new_nand_12): Likewise.
23326         (sync_sub<mode>): Likewise.
23327         (sync_old_add<mode>): Likewise.
23328         (sync_old_sub<mode>): Likewise.
23329         (sync_new_add<mode>): Likewise.
23330         (sync_new_sub<mode>): Likewise.
23331         (sync_<optab><mode>): Likewise.
23332         (sync_old_<optab><mode>): Likewise.
23333         (sync_new_<optab><mode>): Likewise.
23334         (sync_nand<mode>): Likewise.
23335         (sync_old_nand<mode>): Likewise.
23336         (sync_new_nand<mode>): Likewise.
23337         (sync_lock_test_and_set<mode>): Likewise.
23338         (test_and_set_12): Likewise.
23339         (atomic_compare_and_swap<mode>): Likewise.
23340         (atomic_exchange<mode>_llsc): Likewise.
23341         (atomic_fetch_add<mode>_llsc): Likewise.
23342         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
23343         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
23344         (umips_save_restore_pattern_p): Likewise.
23345         (umips_load_store_pair_p): Likewise.
23346         (umips_output_load_store_pair): Likewise.
23347         (umips_movep_target_p): Likewise.
23348         (umips_12bit_offset_address_p): Likewise.
23349         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
23350         (mips_base_mips16): Rename this...
23351         (mips_base_compression_flags): ...to this. Update all uses.
23352         (mips_attribute_table): Add micromips, nomicromips and nocompression.
23353         (mips_mips16_decl_p): Delete.
23354         (mips_nomips16_decl_p): Delete.
23355         (mips_get_compress_on_flags): New function.
23356         (mips_get_compress_off_flags): New function.
23357         (mips_get_compress_mode): New function.
23358         (mips_get_compress_on_name): New function.
23359         (mips_get_compress_off_name): New function.
23360         (mips_insert_attributes): Support multiple compression types.
23361         (mips_merge_decl_attributes): Likewise.
23362         (umips_12bit_offset_address_p): New function.
23363         (mips_start_function_definition): Emit .set micromips directive.
23364         (mips_call_may_need_jalx_p): New function.
23365         (mips_function_ok_for_sibcall): Add microMIPS support.
23366         (mips_print_operand_punctuation): Support short delay slots and
23367         compact jumps.
23368         (umips_swm_mask, umips_swm_encoding): New.
23369         (umips_build_save_restore): New function.
23370         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
23371         (was_mips16_p): Remove.
23372         (old_compression_mode): New.
23373         (mips_set_compression_mode): New function.
23374         (mips_set_current_function): Add microMIPS support.
23375         (mips_option_override): Likewise.
23376         (umips_save_restore_pattern_p): New function.
23377         (umips_output_save_restore): New function.
23378         (umips_load_store_pair_p_1): New function.
23379         (umips_load_store_pair_p): New function.
23380         (umips_output_load_store_pair_1): New function.
23381         (umips_output_load_store_pair): New function.
23382         (umips_movep_target_p) New function.
23383         (mips_prepare_pch_save): Add microMIPS support.
23384         * config/mips/mips.h (TARGET_COMPRESSION): New.
23385         (TARGET_CPU_CPP_BUILTINS): Update macro
23386         to use new compression flags and to support microMIPS.
23387         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
23388         (MIPS_ARCH_FLOAT_SPEC): Likewise.
23389         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
23390         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
23391         (ASM_SPEC): Support mmicromips and mno-micromips.
23392         (M16STORE_REG_P): New macro.
23393         (MIPS_CALL): Support TARGET_MICROMIPS.
23394         (MICROMIPS_J): New macro.
23395         (mips_base_mips16): Rename this...
23396         (mips_base_compression_flags): ...to this.
23397         (UMIPS_12BIT_OFFSET_P): New macro.
23398         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
23399         (MULTILIB_DIRNAMES): Likewise.
23400 2013-03-20  Richard Biener  <rguenther@suse.de>
23402         PR tree-optimization/56661
23403         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
23404         the result does not have to be distinct.
23406 2013-03-20  Richard Biener  <rguenther@suse.de>
23408         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
23409         remap_gimple_op_r.
23411 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23412             Steven Bosscher  <steven@gcc.gnu.org>
23414         PR rtl-optimization/56605
23415         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
23417 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23419         PR bootstrap/56656
23420         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
23421         that require movd instead of movq.
23423 2013-03-20  Richard Biener  <rguenther@suse.de>
23425         * tree-ssa-structalias.c (struct variable_info): Add pointer
23426         to the first field of an aggregate with sub-vars.  Make
23427         this and the pointer to the next subfield its ID.
23428         (vi_next): New function.
23429         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
23430         storedanything_id, integer_id): Increment by one.
23431         (new_var_info, get_call_vi, lookup_call_clobber_vi,
23432         get_call_clobber_vi): Adjust.
23433         (solution_set_expand): Simplify and speedup.
23434         (solution_set_add): Inline into ...
23435         (set_union_with_increment): ... this.  Adjust accordingly.
23436         (do_sd_constraint): Likewise.
23437         (do_ds_constraint): Likewise.
23438         (do_complex_constraint): Simplify.
23439         (build_pred_graph): Adjust.
23440         (solve_graph): Likewise.  Simplify and speedup.
23441         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
23442         get_constraint_for_component_ref, get_constraint_for_1,
23443         first_vi_for_offset, first_or_preceding_vi_for_offset,
23444         create_function_info_for, create_variable_info_for_1,
23445         create_variable_info_for, intra_create_variable_infos): Adjust.
23446         (init_base_vars): Push NULL for ID zero.
23447         (compute_points_to_sets): Adjust.
23449 2013-03-20  Richard Biener  <rguenther@suse.de>
23451         * cfgloop.c (verify_loop_structure): Streamline and avoid
23452         ICEing on corrupt loop tree.
23453         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
23454         loop tree.
23456 2013-03-20  Richard Biener  <rguenther@suse.de>
23458         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
23459         check whether an SSA update is needed.
23461 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
23463         * config/mips/constraints.md (T): Rename to...
23464         (Yf): ...this.
23465         (U): Rename to...
23466         (Yd): ...this.
23467         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
23468         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
23470 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23472         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
23473         (*subsi3_carryin_uxtw): Likewise.
23475 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23477         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23478         (*rorsi3_insn_uxtw): Likewise.
23480 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23482         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23483         (*extrsi5_insn_uxtw): Likewise.
23485 2013-03-19  Richard Biener  <rguenther@suse.de>
23487         PR tree-optimization/56273
23488         * passes.c (init_optimization_passes): Move second VRP after DOM.
23490 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23492         * config/i386/i386.md (*movti_internal): Merge from
23493         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23494         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23495         nox64 isa attributes.
23497 2013-03-18  Richard Biener  <rguenther@suse.de>
23499         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23500         (unite): Likewise.
23501         (merge_node_constraints): Likewise.
23502         (build_succ_graph): Likewise.
23503         (valid_graph_edge): Inline into single caller.
23504         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23505         and cache varinfo.
23506         (scc_visit): Fix formatting and variable use.
23507         (do_sd_constraint): Use gcc_checking_assert.
23508         (do_ds_constraint): Likewise.
23509         (do_complex_constraint): Likewise.
23510         (condense_visit): Likewise.  Cleanup.
23511         (dump_pred_graph): New function.
23512         (perform_var_substitution): Dump the pred-graph before
23513         variable substitution.
23514         (find_equivalent_node): Use gcc_checking_assert.
23515         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23517 2013-03-18  Richard Biener  <rguenther@suse.de>
23519         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23520         Remove cond_expr_stmt_list argument and do not gimplify the
23521         built expression.
23522         (vect_loop_versioning): Adjust.
23523         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23524         Cleanup to use less temporaries.
23525         (vect_create_data_ref_ptr): Cleanup.
23527 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23529         PR tree-optimization/56635
23530         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23531         require types_compatible_p types.
23533 2013-03-18  Nick Clifton  <nickc@redhat.com>
23535         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23536         spurious backslash.
23538         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23539         Add missing line to comment describing function.
23541 2013-03-18  Richard Biener  <rguenther@suse.de>
23543         PR tree-optimization/56210
23544         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23545         Handle string / character search functions.
23546         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23548 2013-03-18  Richard Biener  <rguenther@suse.de>
23550         PR middle-end/56483
23551         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23552         and implement properly.
23553         * gimple.h (gimple_cond_single_var_p): Remove.
23555 2013-03-18  Richard Biener  <rguenther@suse.de>
23557         * tree-data-ref.h (find_data_references_in_loop): Declare.
23558         * tree-data-ref.c (get_references_in_stmt): Use a stack
23559         vector pre-allocated in the callers.
23560         (find_data_references_in_stmt): Adjust.
23561         (graphite_find_data_references_in_stmt): Likewise.
23562         (create_rdg_vertices): Likewise.
23563         (find_data_references_in_loop): Export.
23564         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23565         Compute dependences here...
23566         (vect_analyze_data_refs): ...not here.  When we encounter
23567         a non-vectorizable data reference in basic-block vectorization
23568         truncate the data reference vector.  Do not bother to
23569         fixup data-dependence information for gather loads.
23570         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23571         of data references, as reported.
23573 2013-03-18  Richard Biener  <rguenther@suse.de>
23575         PR tree-optimization/3713
23576         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23577         has_constants and expr.
23578         (stmt_has_constants): Properly valueize SSA names when deciding
23579         whether the stmt has constants.
23581 2013-03-18  Richard Biener  <rguenther@suse.de>
23583         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23584         whole function when there is nothing to do.
23585         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23586         * tree-vectorizer.c (vectorize_loops): Update virtual and
23587         loop-closed SSA once.
23588         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23590 2013-03-18  Richard Biener  <rguenther@suse.de>
23592         PR middle-end/56113
23593         * domwalk.c (bb_postorder): New global static.
23594         (cmp_bb_postorder): New function.
23595         (walk_dominator_tree): Replace scheme imposing an order for
23596         visiting dominator sons by one sorting them at the time they
23597         are pushed on the stack.
23599 2013-03-18  Richard Biener  <rguenther@suse.de>
23601         PR tree-optimization/39326
23602         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23603         (struct mem_ref): Replace mem member with ao_ref typed member.
23604         (MEM_ANALYZABLE): Adjust.
23605         (memref_eq): Likewise.
23606         (mem_ref_alloc): Likewise.
23607         (gather_mem_refs_stmt): Likewise.
23608         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23609         (execute_sm_if_changed_flag_set): Adjust.
23610         (execute_sm): Likewise.
23611         (ref_always_accessed_p): Likewise.
23612         (refs_independent_p): Likewise.
23613         (can_sm_ref_p): Likewise.
23615 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23617         PR c/56566
23618         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23619         return 1 even for !unsignedp.
23621 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23623         * config/i386/i386.md (isa): Add x64 and nox64.
23624         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23625         (*pushtf): Enable *roF alternative for x64 isa only.
23626         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23627         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23628         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23629         (*movtf_internal): Merge from *movtf_internal_rex64 and
23630         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23631         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23632         nox64 isa attributes.
23633         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23634         nox64 isa attributes.
23635         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23637 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23639         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23641 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23643         PR target/56640
23644         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23645         class names.  Remove trailing comma after "ALL_REGS".
23647 2013-03-16  Jan Hubicka  <jh@suse.cz>
23649         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23650         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23651         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23652         of cgraph_get_create_node.
23653         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23655 2013-03-16  Jason Merrill  <jason@redhat.com>
23657         PR debug/49090
23658         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23659         with DW_AT_default_value.
23661 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23663         * BASE-VER: Set to 4.9.0.
23665 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23667         PR target/56619
23668         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23669         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23670         Document _x* TSX intrinsics.
23672 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23673             David Holsgrove  <david.holsgrove@xilinx.com>
23675         * configure.ac: Add MicroBlaze TLS support detection.
23676         * configure: Regenerate.
23677         * config/microblaze/microblaze-protos.h
23678         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23679         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23680         * config/microblaze/microblaze.c (microblaze_address_type): Add
23681         ADDRESS_TLS and tls_reloc address types.
23682         (microblaze_address_info): Add tls_reloc.
23683         (TARGET_HAVE_TLS): Define.
23684         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23685          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23686          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23687         load_tls_operand,  microblaze_call_tls_get_addr,
23688         microblaze_legitimize_tls_address): New functions.
23689         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23690         (get_base_reg): Use microblaze_tls_symbol_p.
23691         (microblaze_classify_address): Handle TLS.
23692         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23693         label_mentioned_p and microblaze_tls_referenced_p.
23694         (microblaze_legitimize_address): Handle TLS.
23695         (microblaze_address_insns): Handle ADDRESS_TLS.
23696         (pic_address_needs_scratch): Handle TLS.
23697         (print_operand_address): Handle TLS.
23698         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23699         (microblaze_expand_move): Handle TLS.
23700         (microblaze_legitimate_constant_p): Check
23701         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23702         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23703         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23704         (PIC_OFFSET_TABLE_REGNUM): Set.
23705         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23706         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23707         (addsi3, movsi_internal2, movdf_internal): Update constraints
23708         * config/microblaze/predicates.md (arith_plus_operand): Define
23709         (move_operand): Redefine as move_src_operand,
23710         check microblaze_tls_referenced_p.
23712 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23714         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23715         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23717 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23719         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23720         CC mode for AND.
23722 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23724         PR tree-optimization/53265
23725         * common.opt (Waggressive-loop-optimizations): New option.
23726         * tree-ssa-loop-niter.c: Include tree-pass.h.
23727         (do_warn_aggressive_loop_optimizations): New function.
23728         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23729         if number_of_latch_executions returned constant.
23730         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23731         early.  If number_of_latch_executions returned constant, set
23732         nb_iterations_upper_bound back to it.
23733         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23734         field.
23735         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23736         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23738         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23739         (MULTILIB_OSDIRNAMES): Set.
23740         * genmultilib: If defaultosdirname doesn't start with :: , set
23741         defaultosdirname2 instead, clear it and emit two . multilib_raw
23742         entries instead of just one.
23744 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23746         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23747         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23748         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23749         (SUBTARGET_OVERRIDE_OPTIONS): New.
23751 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23753         PR target/49880
23754         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23755         (musermode): Convert to Var(TARGET_USERMODE).
23756         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23757         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23758         * config/sh/sh.c (sh_option_override): Use
23759         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23760         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23761         condition.
23762         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23763         TARGET_SH4.
23764         (udivsi3_i4_single, divsi3_i4_single): Use
23765         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23767 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
23769         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23770         default setting.
23772 2013-03-13  Richard Biener  <rguenther@suse.de>
23774         PR tree-optimization/56608
23775         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23776         calls when vectorizing basic-blocks.
23778 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23780         PR plugins/45078
23781         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23782         tm_file.
23784 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23786         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23788 2013-03-11  Jan Hubicka  <jh@suse.cz>
23790         PR lto/56557
23791         * lto-streamer-out.c (output_symbol_p): Skip references from
23792         constructors of external variables.
23794 2013-03-11  Jan Hubicka  <jh@suse.cz>
23796         PR middle-end/56571
23797         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23798         from pseudos.
23799         * emit-rtl.c (verify_rtx_sharing): Likewise.
23800         (copy_insn_1): Likewise.
23801         * rtl.c (copy_rtx): Likewise.
23803 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23805         PR target/56591
23806         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23807         output_operand_lossage message.
23809 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23811         PR target/56470
23812         * arm.c (shift_op): Validate RTL pattern on the fly.
23813         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23814         the RTL.
23816 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23818         PR target/56347
23819         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23820         new call patterns that clobber %fr12.
23821         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23822         split and postreload patterns.
23823         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23824         registers %fr12 and %fr12R as call used.
23826 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23828         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23829         (canon_address, record_store, replace_read, check_mem_read_rtx,
23830         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23831         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23832         rest_of_handle_dse): Likewise.
23834 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23836         PR middle-end/56524
23837         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23838         Add base_optabs.
23839         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23840         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23841         (save_optabs_if_changed): Replace with...
23842         (init_tree_optimization_optabs): ...this.
23843         * optabs.c (save_optabs_if_changed): Rename to...
23844         (init_tree_optimization_optabs): ...this.  Take the optimization node
23845         as argument.  Do nothing if the base optabs are already correct.
23846         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23847         to recompute optabs.
23848         * function.h (function): Remove optabs field.
23849         * function.c (invoke_set_current_function_hook): Call
23850         init_tree_optimization_optabs.  Use the result to initialize
23851         this_fn_optabs.
23853 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23855         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23856         if GTMA_HAS_NO_INSTRUMENTATION.
23857         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23858         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23859         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23860         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23861         GTMA_HAS_NO_INSTRUMENTATION.
23863 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23865         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23866         libasan_preinit.o.
23868 2013-03-08  Marek Polacek  <polacek@redhat.com>
23869             Jakub Jelinek  <jakub@redhat.com>
23871         PR tree-optimization/56478
23872         * predict.c (is_comparison_with_loop_invariant_p): Change the
23873         type of loop_step to tree.
23874         (predict_loops): Adjust.
23875         (predict_iv_comparison): Perform the computations on double_ints.
23877 2013-03-08  Richard Biener  <rguenther@suse.de>
23879         PR tree-optimization/56570
23880         * tree-cfg.c (verify_expr_location_1): Verify locations for
23881         DECL_DEBUG_EXPR.
23882         * tree-sra.c (create_access_replacement): Strip locations
23883         from DECL_DEBUG_EXPRs.
23885 2013-03-08  Richard Biener  <rguenther@suse.de>
23887         * tree-inline.c (expand_call_inline): Do not associate
23888         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23889         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23891 2013-03-08  Richard Biener  <rguenther@suse.de>
23893         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23894         or block changes with -Og.  Fix for location / block encoding
23895         changes and PHI arguments with locations.
23897 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23899         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23900         for all counters.
23901         (struct output_info): Likewise.
23902         (register_overhead): Remove bad gcc_assert.
23903         (bitmap_find_bit): If there is only a single bitmap element, do not
23904         count a miss as a search.
23905         (print_statistics): Update for counter type changes.
23906         (dump_bitmap_statistics): Likewise.  Print headers such that they
23907         are properly lined up with the printed counters.
23909 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23911         PR tree-optimization/56559
23912         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23913         check that it has only a single use.
23915 2013-03-07  Richard Biener  <rguenther@suse.de>
23917         * doc/invoke.texi (fwhole-program): Discourage use in combination
23918         with -flto.
23920 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23922         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23924         PR tree-optimization/56539
23925         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23926         instead of GSI_CONTINUE_LINKING as last argument to
23927         force_gimple_operand_gsi.  Adjust function comment.
23929         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23930         aarch64-cores.def.
23932         PR middle-end/56548
23933         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23934         promoted mode, convert the result back to the original mode.
23936 2013-03-06  Richard Biener  <rguenther@suse.de>
23938         PR middle-end/56294
23939         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23940         (insert_updated_phi_nodes_compare_uids): New function.
23941         (update_ssa): Sort symbols_to_rename after UID before
23942         traversing it to insert PHI nodes.
23944 2013-03-06  Richard Biener  <rguenther@suse.de>
23946         PR middle-end/50494
23947         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23948         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23950         Revert
23951         2013-02-13  Richard Biener  <rguenther@suse.de>
23953         PR lto/50494
23954         * varasm.c (output_constant_def_1): Get the decl representing
23955         the constant as argument.
23956         (output_constant_def): Wrap output_constant_def_1.
23957         (make_decl_rtl): Use output_constant_def_1 with the decl
23958         representing the constant.
23959         (build_constant_desc): Optionally re-use a decl already
23960         representing the constant.
23961         (tree_output_constant_def): Adjust.
23963 2013-03-06  Joey Ye  <joey.ye@arm.com>
23965         PR lto/50293
23966         * gcc.c (convert_white_space): New function.
23967         (main): Handles white space in function name.
23969 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23971         PR target/56529
23972         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23973         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23974         to SH_DIV_CALL_TABLE for TARGET_SH2.
23975         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23976         list.
23977         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23978         call-table options.
23980 2013-03-05  Sterling Augustine  <saugustine@google.com>
23981             Cary Coutant  <ccoutant@google.com>
23983         PR debug/55364
23984         * dwarf2out.c (resolve_addr): Don't call
23985         remove_loc_list_addr_table_entries a second time for the same
23986         expression.
23988 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23990         PR debug/56510
23991         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23992         (avoid_complex_debug_insns): New function.
23993         (expand_debug_locations): Call it.
23995         PR rtl-optimization/56484
23996         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23997         lifetimes of hard registers on small register class machines.
23999 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
24001         * config/microblaze/microblaze-protos.h: Rename
24002         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
24003         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
24004         fast_interrupt.
24005         (microblaze_fast_interrupt_function_p): New function.
24006         (microblaze_is_interrupt_handler): Rename to
24007         microblaze_is_interrupt_variant and add fast_interrupt check.
24008         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
24009         (save_restore_insns): Likewise.
24010         (compute_frame_size): Likewise.
24011         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
24012         (microblaze_globalize_label): Likewise.
24013         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
24014         * config/microblaze/microblaze.md: Use wrapper
24015         microblaze_is_interrupt_variant.
24017 2013-03-05  Kai Tietz  <ktietz@redhat.com>
24019         * sdbout.c (sdbout_one_type): Switch to current function's section
24020         supporting cold/hot.
24022 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
24024         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
24025         -mxl-reorder.
24027 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24029         PR middle-end/56461
24030         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
24031         if VALGRIND_GET_VBITS is defined, temporarily make object
24032         memory all defined, and restore previous valgrind addressability
24033         and definability afterwards.  Free this_object at the end.
24035         PR middle-end/56461
24036         * lra.c (lra): Call lra_clear_live_ranges if live_p,
24037         right before calling lra_create_live_ranges, also call it
24038         when clearing live_p.  Only call lra_clear_live_ranges
24039         at the end if live_p.
24041         PR middle-end/56461
24042         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
24044 2013-03-05  Richard Biener  <rguenther@suse.de>
24046         PR tree-optimization/56521
24047         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
24048         value-id.
24050 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
24052         PR c++/55135
24053         * except.h (remove_unreachable_eh_regions): New prototype.
24054         * except.c (remove_eh_handler_splicer): New function, split out
24055         of remove_eh_handler.
24056         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
24057         warning about running it on many EH regions one at a time.
24058         (remove_unreachable_eh_regions_worker): New function, walk the
24059         EH tree in depth-first order and remove non-marked regions.
24060         (remove_unreachable_eh_regions): New function.
24061         * tree-eh.c (mark_reachable_handlers): New function, split out
24062         from remove_unreachable_handlers.
24063         (remove_unreachable_handlers): Use mark_reachable_handlers and
24064         remove_unreachable_eh_regions.
24065         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
24066         and remove_unreachable_eh_regions.
24068 2013-03-05  Richard Biener  <rguenther@suse.de>
24070         PR middle-end/56525
24071         * loop-init.c (fix_loop_structure): Remove loops in two stages,
24072         not freeing them until the end.
24074 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
24076         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
24078 2013-03-05  Richard Biener  <rguenther@suse.de>
24080         PR tree-optimization/56270
24081         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
24082         of loads after scheduling an SLP instance.
24084 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24086         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
24087         tic6x.exp.
24088         (check_gcc_parallelize): Run guality.exp as a separate job from
24089         vect.exp with unsorted.exp and $(dg_target_exps) separately from
24090         struct-layout-1.exp with stackalign.exp.
24092         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
24094         PR middle-end/56461
24095         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
24096         load_index sbitmap even if some bit in it isn't set.
24098         PR middle-end/56461
24099         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
24100         (discover_iteration_bound_by_body_walk): Change queues to
24101         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
24102         spelling in comment.  Call safe_push on queues[bound_index] directly.
24103         Release queues[queue_index] in every iteration unconditionally.
24104         Release bounds vector.
24106         PR middle-end/56461
24107         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
24108         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
24109         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
24110         inner_phis vector.
24112 2013-03-05  Richard Biener  <rguenther@suse.de>
24114         PR lto/56515
24115         * tree-inline.c (remap_blocks_to_null): New function.
24116         (expand_call_inline): When expanding a call stmt without
24117         an associated block inline remap all callee blocks to NULL.
24119 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24121         PR rtl-optimization/56494
24122         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
24123         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
24124         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
24126         PR middle-end/56461
24127         * sel-sched-ir.c (free_sched_pools): Release
24128         succs_info_pool.stack[succs_info_pool.max_top] vectors too
24129         if succs_info_pool.max_top isn't -1.
24131         PR bootstrap/56509
24132         * opts.c (opts_obstack, opts_concat): Moved to...
24133         * opts-common.c (opts_obstack, opts_concat): ... here.
24135 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
24137         PR middle-end/56461
24138         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
24140 2013-03-04  Martin Jambor  <mjambor@suse.cz>
24142         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
24143         all appropriate places.
24145 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
24147         PR tree-optimization/56424
24148         * ipa-split.c (split_function): Do not set the RSO flag if result is
24149         not by reference and its type is a register type.
24151 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
24153         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
24154         (microblaze_legitimate_pic_operand): Likewise
24155         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
24156         new function microblaze_legitimate_pic_operand
24157         * config/microblaze/microblaze-protos.h
24158         (microblaze_legitimate_pic_operand): Declare.
24160 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24162         * config/microblaze/predicates.md (call_insn_simple_operand):
24163         New predicate for supported rtx code types.
24164         * config/microblaze/microblaze.md (call_internal1): Use
24165         call_insn_simple_operand predicate.
24167 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
24169         PR middle-end/56461
24170         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
24171         partitions.ordered_remove.
24173         PR middle-end/56461
24174         * tree-vect-stmts.c (vectorizable_conversion): Don't call
24175         vec_oprnds0.create (1) for modifier == NONE.
24177         PR middle-end/56461
24178         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
24179         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
24180         vec_oprnds1 right before pushing anything to it for
24181         scalar_shift_arg.
24183         PR middle-end/56461
24184         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
24185         set nbbs to 0 instead of having separate code path.
24186         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
24187         instead of false as last argument if returning NULL.
24189 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
24191         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
24192         the attribute is now called "target" instead of "option".
24193         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
24194         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
24195         attribute/pragma name for TARGET_OPTION_VALID_P and
24196         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
24197         * doc/tm.texi: Regenerated.
24199 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
24201         * config/microblaze/microblaze.c:
24202         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
24203         * config/microblaze/microblaze.h: Add -mxl-reorder to
24204         DRIVER_SELF_SPECS.
24205         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
24206         instructions emitted if TARGET_REORDER.
24207         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
24208         or 0 for -m/-mno case, but initialises as 2 to detect default use case
24209         separately.
24211 2013-03-01  Xinliang David Li  <davidxl@google.com>
24213         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
24214         walk length.
24216 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
24218         PR middle-end/56461
24219         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
24220         vector even when returning true.  Fix up function comment formatting.
24222         PR middle-end/56461
24223         * ira-build.c (ira_loop_nodes_count): New variable.
24224         (create_loop_tree_nodes): Initialize it.
24225         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
24227         PR middle-end/56461
24228         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
24229         method on dr_chain and result_chain.
24230         * tree-vect-stmts.c (vectorizable_store): Only call
24231         result_chain.create if j == 0.
24233         PR middle-end/56461
24234         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
24235         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
24236         before overwriting it.
24238 2013-03-01  Tobias Burnus  <burnus@net-b.de>
24240         * doc/extended.texi (C Extensions): Change order in @menu
24241         to match @node.
24242         (Other MIPS Built-in Functions): Move last MIPS entry before
24243         "picoChip Built-in Functions".
24244         (SH Built-in Functions): Move after RX Built-in Functions.
24245         * doc/gcc.texi (Introduction): Change order in @menu
24246         to match @node.
24247         * doc/md.texi (Constraints): Ditto.
24248         * gty.texi (Type Information): Ditto.
24249         (User-provided marking routines for template types): Make
24250         subsection.
24251         * doc/invoke.texi (AArch64 Options): Move before
24252         "Adapteva Epiphany Options".
24254 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
24255             Jakub Jelinek  <jakub@redhat.com>
24257         PR sanitizer/56454
24258         * asan.c (gate_asan): Lookup no_sanitize_address instead of
24259         no_address_safety_analysis attribute.
24260         * doc/extend.texi (no_address_safety_attribute): Rename to
24261         no_sanitize_address attribute, mention no_address_safety_analysis
24262         attribute as deprecated alias.
24264 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
24266         PR middle-end/56461
24267         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
24268         type to vec<vec<tree> > *.
24269         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
24270         to be vec<tree> instead of vec<tree> *, set vec_defs
24271         to vNULL and call vec_defs.create (number_of_vects), adjust other
24272         uses of vec_defs.
24273         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
24274         vectorizable_condition): Adjust vect_get_slp_defs callers.
24276 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24278         * config/aarch64/aarch64.c
24279         (aarch64_float_const_representable): Remove unused variable.
24281 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24283         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
24285 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24287         * config/aarch64/aarch64-builtins.c
24288         (aarch64_init_simd_builtins): Make static.
24290 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24292         * config/aarch64/aarch64.c
24293         (aarch64_simd_make_constant): Make static.
24295 2013-02-28  Martin Jambor  <mjambor@suse.cz>
24297         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
24298         with no initialization to the RHS of debug statements.
24300 2013-02-28  Martin Jambor  <mjambor@suse.cz>
24302         PR tree-optimization/56294
24303         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
24304         Adjust dumping.
24305         (get_access_replacement): Do not call create_access_replacement.
24306         Assert a replacement exists.
24307         (get_repl_default_def_ssa_name): Create the replacement declaration
24308         itself.
24310 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
24312         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
24313         final_end_function.
24315 2013-02-28  Marek Polacek  <polacek@redhat.com>
24317         PR rtl-optimization/56466
24318         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
24319         if we're changing a loop.
24320         (peel_loops_completely): Likewise.
24322 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
24324         PR c++/55813
24325         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
24327 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24329         PR target/56445
24330         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
24331         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
24332         INTX_FTYPE_FX, FX_FTYPE_INTX.
24333         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
24335 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24337         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
24338         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
24339         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
24340         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
24341         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
24342         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
24343         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
24344         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
24345         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
24346         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
24347         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
24348         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
24349         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
24350         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
24351         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
24352         (avrxmega6): Increase max flash segments from 5 to 6.
24353         * config/avr/t-multilib: Regenerate.
24354         * config/avr/avr-tables.opt: Regenerate.
24355         * doc/avr-mmcu.texi: Regenerate.
24357 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24359         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
24360         (avr_device_to_arch): Rename to avr_device_to_ld.
24361         (avr_device_to_as): New prototype.
24362         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
24363         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
24364         * config/avr/driver-avr.c (avr_device_to_as): New.
24365         (avr_device_to_arch): Rename to avr_device_to_ld.
24367 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24369         PR middle-end/56461
24370         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
24371         method on dr_chain and result_chain.
24373         PR middle-end/56461
24374         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
24375         pointer_set_destroy on not_executed_last_iteration.
24377         PR middle-end/56461
24378         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
24380         PR middle-end/56461
24381         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
24382         FOR_EACH_DEFINED_FUNCTION when freeing state.
24384         PR middle-end/56461
24385         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
24386         pool_free.
24387         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
24388         overwriting it.
24390         PR middle-end/56461
24391         * ipa-cp.c (decide_whether_version_node): Call vec_free on
24392         known_aggs[i].items and release known_aggs vector.
24394         PR middle-end/56461
24395         * ipa-reference.c (propagate): Free node_info even for alias nodes.
24397 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24399         * config/microblaze/microblaze.c (microblaze_emit_compare):
24400         Use xor for EQ/NE comparisions.
24401         * config/microblaze/microblaze.md (cstoresf4): Add constraints
24402         (cbranchsf4): Adjust operator to comparison_operator.
24404 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24406         PR middle-end/56461
24407         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
24408         vector.
24409         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
24410         vec_safe_push, always update *slot.
24411         (redirect_edge_var_map_clear): Use vec_free.
24412         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
24413         (free_var_map_entry): Use vec_free.
24414         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
24415         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
24417 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
24419         PR middle-end/45472
24420         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
24421         when the may_trap_p bit of the exprs being merged differs.
24422         Reorder tests for speculativeness in the logical and operator.
24424 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24426         * incpath.c (add_standard_paths): Use reconcat instead of concat
24427         where appropriate and avoid leaking memory.
24429         * opts.h: Include obstack.h.
24430         (opts_concat): New prototype.
24431         (opts_obstack): New declaration.
24432         * opts.c (opts_concat): New function.
24433         (opts_obstack): New variable.
24434         (init_options_struct): Call gcc_init_obstack on opts_obstack.
24435         (finish_options): Use opts_concat instead of concat
24436         and XOBNEWVEC instead of XNEWVEC.
24437         * opts-common.c (generate_canonical_option, decode_cmdline_option,
24438         generate_option): Likewise.
24439         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
24440         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
24442         PR target/56455
24443         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
24444         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
24446 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24448         PR middle-end/56461
24449         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
24451 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24453         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
24454         (arm_block_move_unaligned_straight): Likewise.
24455         (arm_adjust_block_mem): Likewise.
24457 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24459         PR target/48901
24460         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
24461         temp, cond and label.
24462         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
24464         PR target/52500
24465         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
24466         * config/c6x/c6x.h (dbx_register_map): Update declaration.
24468         PR target/52501
24469         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
24470         of prologue/epilogue functions.
24472         PR target/52550
24473         * config/tilegx/tilegx.c (tilegx_expand_prologue):
24474         Remove unused variable cfa_offset.
24475         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24477         PR target/54639
24478         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24479         type promotion to unsigned.
24481         PR target/54640
24482         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24483         for HOST_WIDE_INT of 32 bit / same size as int.
24484         (arm_block_move_unaligned_straight): Likewise.
24485         (arm_adjust_block_mem): Likewise.
24487         PR target/54662
24488         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24489         ALL_CFLAGS.
24491 2013-02-26  Marek Polacek  <polacek@redhat.com>
24493         PR tree-optimization/56426
24494         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24496 2013-02-26  Richard Biener  <rguenther@suse.de>
24498         PR target/56444
24499         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24500         unused variable loops.
24502 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24504         PR tree-optimization/56448
24505         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24506         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24507         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24508         later operands of the references, or even first operand for
24509         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24511         PR tree-optimization/56443
24512         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24513         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24514         to type_for_mode langhook.
24516 2013-02-25  Matt Turner  <mattst88@gmail.com>
24518         * doc/invoke.texi: Document r4700.
24520 2013-02-25  Richard Biener  <rguenther@suse.de>
24522         PR tree-optimization/56175
24523         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24524         split out from ...
24525         (simplify_bitwise_binary): ... here.  Also guard the conversion
24526         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24528 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24530         Revert:
24531         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24532             Maciej W. Rozycki  <macro@codesourcery.com>
24533             Tom de Vries  <tom@codesourcery.com>
24534             Nathan Sidwell  <nathan@codesourcery.com>
24535             Iain Sandoe  <iain@codesourcery.com>
24536             Nathan Froyd  <froydnj@codesourcery.com>
24537             Chao-ying Fu  <fu@mips.com>
24539         * doc/extend.texi: (micromips, nomicromips, nocompression):
24540         Document new function attributes.
24541         * doc/invoke.texi (minterlink-compressed, mmicromips,
24542         m14k, m14ke, m14kec): Document new options.
24543         (minterlink-mips16): Update documentation.
24544         * doc/md.texi (ZC, ZD): Document new constraints.
24545         * configure.ac (gcc_cv_as_micromips): Check if linker
24546         supports the .set micromips directive.
24547         * configure: Regenerate.
24548         * config.in: Regenerate.
24549         * config/mips/mips-tables.opt: Regenerate.
24550         * config/mips/micromips.md: New file.
24551         * constraints.md (ZC, AD): New constraints.
24552         * config/mips/predicates.md (movep_src_register): New predicate.
24553         (movep_src_operand): New predicate.
24554         (non_volatile_mem_operand): New predicate.
24555         * config/mips/mips.md (multimem): New type.
24556         (length): Differentiate between 17-bit and 18-bit branch offsets.
24557         (MOVEP1, MOVEP2): New mode iterator.
24558         (mov_<load>l): Use ZC constraint.
24559         (mov_<load>r): Likewise.
24560         (mov_<store>l): Likewise.
24561         (mov_<store>r): Likewise.
24562         (*branch_equality<mode>_inverted): Add microMIPS support.
24563         (*branch_equality<mode>): Likewise.
24564         (*jump_absolute): Likewise.
24565         (indirect_jump_<mode>): Likewise.
24566         (tablejump_<mode>): Likewise.
24567         (<optab>_internal): Likewise.
24568         (sibcall_internal): Likewise.
24569         (sibcall_value_internal): Likewise.
24570         (prefetch): Use constraint ZD.
24571         * config/mips/mips.opt (minterlink-compressed): New option.
24572         (minterlink-mips16): Now an alias for minterlink-compressed.
24573         (mmicromips): New option.
24574         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24575         (compare_and_swap_12): Likewise.
24576         (sync_add<mode>): Likewise.
24577         (sync_<optab>_12): Likewise.
24578         (sync_old_<optab>_12): Likewise.
24579         (sync_new_<optab>_12): Likewise.
24580         (sync_nand_12): Likewise.
24581         (sync_old_nand_12): Likewise.
24582         (sync_new_nand_12): Likewise.
24583         (sync_sub<mode>): Likewise.
24584         (sync_old_add<mode>): Likewise.
24585         (sync_old_sub<mode>): Likewise.
24586         (sync_new_add<mode>): Likewise.
24587         (sync_new_sub<mode>): Likewise.
24588         (sync_<optab><mode>): Likewise.
24589         (sync_old_<optab><mode>): Likewise.
24590         (sync_new_<optab><mode>): Likewise.
24591         (sync_nand<mode>): Likewise.
24592         (sync_old_nand<mode>): Likewise.
24593         (sync_new_nand<mode>): Likewise.
24594         (sync_lock_test_and_set<mode>): Likewise.
24595         (test_and_set_12): Likewise.
24596         (atomic_compare_and_swap<mode>): Likewise.
24597         (atomic_exchange<mode>_llsc): Likewise.
24598         (atomic_fetch_add<mode>_llsc): Likewise.
24599         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24600         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24601         (umips_save_restore_pattern_p): Likewise.
24602         (umips_load_store_pair_p): Likewise.
24603         (umips_output_load_store_pair): Likewise.
24604         (umips_movep_target_p): Likewise.
24605         (umips_12bit_offset_address_p): Likewise.
24606         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24607         (mips_base_mips16): Rename this...
24608         (mips_base_compression_flags): ...to this. Update all uses.
24609         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24610         (mips_mips16_decl_p): Delete.
24611         (mips_nomips16_decl_p): Delete.
24612         (mips_get_compress_on_flags): New function.
24613         (mips_get_compress_off_flags): New function.
24614         (mips_get_compress_mode): New function.
24615         (mips_get_compress_on_name): New function.
24616         (mips_get_compress_off_name): New function.
24617         (mips_insert_attributes): Support multiple compression types.
24618         (mips_merge_decl_attributes): Likewise.
24619         (umips_12bit_offset_address_p): New function.
24620         (mips_start_function_definition): Emit .set micromips directive.
24621         (mips_call_may_need_jalx_p): New function.
24622         (mips_function_ok_for_sibcall): Add microMIPS support.
24623         (mips_print_operand_punctuation): Support short delay slots and
24624         compact jumps.
24625         (umips_swm_mask, umips_swm_encoding): New.
24626         (umips_build_save_restore): New function.
24627         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24628         (was_mips16_p): Remove.
24629         (old_compression_mode): New.
24630         (mips_set_compression_mode): New function.
24631         (mips_set_current_function): Add microMIPS support.
24632         (mips_option_override): Likewise.
24633         (umips_save_restore_pattern_p): New function.
24634         (umips_output_save_restore): New function.
24635         (umips_load_store_pair_p_1): New function.
24636         (umips_load_store_pair_p): New function.
24637         (umips_output_load_store_pair_1): New function.
24638         (umips_output_load_store_pair): New function.
24639         (umips_movep_target_p) New function.
24640         (mips_prepare_pch_save): Add microMIPS support.
24641         * config/mips/mips.h (TARGET_COMPRESSION): New.
24642         (TARGET_CPU_CPP_BUILTINS): Update macro
24643         to use new compression flags and to support microMIPS.
24644         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24645         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24646         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24647         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24648         (ASM_SPEC): Support mmicromips and mno-micromips.
24649         (M16STORE_REG_P): New macro.
24650         (MIPS_CALL): Support TARGET_MICROMIPS.
24651         (MICROMIPS_J): New macro.
24652         (mips_base_mips16): Rename this...
24653         (mips_base_compression_flags): ...to this.
24654         (UMIPS_12BIT_OFFSET_P): New macro.
24655         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24656         (MULTILIB_DIRNAMES): Likewise.
24658 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24660         PR rtl-optimization/56131
24661         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24662         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24663         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24665 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24667         * doc/invoke.texi (-fsanitize=): Move from optimization
24668         to debugging options.
24670 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24672         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24674 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24675             Alexander Monakov  <amonakov@ispras.ru>
24677         PR middle-end/56077
24678         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24679         flush pending lists also on non-jumps.  Adjust comment.
24681 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24682             Maciej W. Rozycki  <macro@codesourcery.com>
24683             Tom de Vries  <tom@codesourcery.com>
24684             Nathan Sidwell  <nathan@codesourcery.com>
24685             Iain Sandoe  <iain@codesourcery.com>
24686             Nathan Froyd  <froydnj@codesourcery.com>
24687             Chao-ying Fu  <fu@mips.com>
24689         * doc/extend.texi: (micromips, nomicromips, nocompression):
24690         Document new function attributes.
24691         * doc/invoke.texi (minterlink-compressed, mmicromips,
24692         m14k, m14ke, m14kec): Document new options.
24693         (minterlink-mips16): Update documentation.
24694         * doc/md.texi (ZC, ZD): Document new constraints.
24695         * configure.ac (gcc_cv_as_micromips): Check if linker
24696         supports the .set micromips directive.
24697         * configure: Regenerate.
24698         * config.in: Regenerate.
24699         * config/mips/mips-tables.opt: Regenerate.
24700         * config/mips/micromips.md: New file.
24701         * constraints.md (ZC, AD): New constraints.
24702         * config/mips/predicates.md (movep_src_register): New predicate.
24703         (movep_src_operand): New predicate.
24704         (non_volatile_mem_operand): New predicate.
24705         * config/mips/mips.md (multimem): New type.
24706         (length): Differentiate between 17-bit and 18-bit branch offsets.
24707         (MOVEP1, MOVEP2): New mode iterator.
24708         (mov_<load>l): Use ZC constraint.
24709         (mov_<load>r): Likewise.
24710         (mov_<store>l): Likewise.
24711         (mov_<store>r): Likewise.
24712         (*branch_equality<mode>_inverted): Add microMIPS support.
24713         (*branch_equality<mode>): Likewise.
24714         (*jump_absolute): Likewise.
24715         (indirect_jump_<mode>): Likewise.
24716         (tablejump_<mode>): Likewise.
24717         (<optab>_internal): Likewise.
24718         (sibcall_internal): Likewise.
24719         (sibcall_value_internal): Likewise.
24720         (prefetch): Use constraint ZD.
24721         * config/mips/mips.opt (minterlink-compressed): New option.
24722         (minterlink-mips16): Now an alias for minterlink-compressed.
24723         (mmicromips): New option.
24724         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24725         (compare_and_swap_12): Likewise.
24726         (sync_add<mode>): Likewise.
24727         (sync_<optab>_12): Likewise.
24728         (sync_old_<optab>_12): Likewise.
24729         (sync_new_<optab>_12): Likewise.
24730         (sync_nand_12): Likewise.
24731         (sync_old_nand_12): Likewise.
24732         (sync_new_nand_12): Likewise.
24733         (sync_sub<mode>): Likewise.
24734         (sync_old_add<mode>): Likewise.
24735         (sync_old_sub<mode>): Likewise.
24736         (sync_new_add<mode>): Likewise.
24737         (sync_new_sub<mode>): Likewise.
24738         (sync_<optab><mode>): Likewise.
24739         (sync_old_<optab><mode>): Likewise.
24740         (sync_new_<optab><mode>): Likewise.
24741         (sync_nand<mode>): Likewise.
24742         (sync_old_nand<mode>): Likewise.
24743         (sync_new_nand<mode>): Likewise.
24744         (sync_lock_test_and_set<mode>): Likewise.
24745         (test_and_set_12): Likewise.
24746         (atomic_compare_and_swap<mode>): Likewise.
24747         (atomic_exchange<mode>_llsc): Likewise.
24748         (atomic_fetch_add<mode>_llsc): Likewise.
24749         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24750         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24751         (umips_save_restore_pattern_p): Likewise.
24752         (umips_load_store_pair_p): Likewise.
24753         (umips_output_load_store_pair): Likewise.
24754         (umips_movep_target_p): Likewise.
24755         (umips_12bit_offset_address_p): Likewise.
24756         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24757         (mips_base_mips16): Rename this...
24758         (mips_base_compression_flags): ...to this. Update all uses.
24759         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24760         (mips_mips16_decl_p): Delete.
24761         (mips_nomips16_decl_p): Delete.
24762         (mips_get_compress_on_flags): New function.
24763         (mips_get_compress_off_flags): New function.
24764         (mips_get_compress_mode): New function.
24765         (mips_get_compress_on_name): New function.
24766         (mips_get_compress_off_name): New function.
24767         (mips_insert_attributes): Support multiple compression types.
24768         (mips_merge_decl_attributes): Likewise.
24769         (umips_12bit_offset_address_p): New function.
24770         (mips_start_function_definition): Emit .set micromips directive.
24771         (mips_call_may_need_jalx_p): New function.
24772         (mips_function_ok_for_sibcall): Add microMIPS support.
24773         (mips_print_operand_punctuation): Support short delay slots and
24774         compact jumps.
24775         (umips_swm_mask, umips_swm_encoding): New.
24776         (umips_build_save_restore): New function.
24777         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24778         (was_mips16_p): Remove.
24779         (old_compression_mode): New.
24780         (mips_set_compression_mode): New function.
24781         (mips_set_current_function): Add microMIPS support.
24782         (mips_option_override): Likewise.
24783         (umips_save_restore_pattern_p): New function.
24784         (umips_output_save_restore): New function.
24785         (umips_load_store_pair_p_1): New function.
24786         (umips_load_store_pair_p): New function.
24787         (umips_output_load_store_pair_1): New function.
24788         (umips_output_load_store_pair): New function.
24789         (umips_movep_target_p) New function.
24790         (mips_prepare_pch_save): Add microMIPS support.
24791         * config/mips/mips.h (TARGET_COMPRESSION): New.
24792         (TARGET_CPU_CPP_BUILTINS): Update macro
24793         to use new compression flags and to support microMIPS.
24794         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24795         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24796         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24797         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24798         (ASM_SPEC): Support mmicromips and mno-micromips.
24799         (M16STORE_REG_P): New macro.
24800         (MIPS_CALL): Support TARGET_MICROMIPS.
24801         (MICROMIPS_J): New macro.
24802         (mips_base_mips16): Rename this...
24803         (mips_base_compression_flags): ...to this.
24804         (UMIPS_12BIT_OFFSET_P): New macro.
24805         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24806         (MULTILIB_DIRNAMES): Likewise.
24808 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24810         PR target/52555
24811         * target-globals.c (save_target_globals): For init_reg_sets and
24812         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24814 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
24816         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24817         * config/aarch64/t-aarch64
24818         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24820 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24822         PR inline-asm/56148
24823         * lra-constraints.c (process_alt_operands): Reload operand
24824         conflicting with earlier clobber only if no more other conflicting
24825         operands.
24827 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24829         PR sanitizer/56393
24830         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24831         if not linking a shared library.
24833 2013-02-22  Seth LaForge  <sethml@google.com>
24835         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24837 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24839         * config/arm/arm.md (split for extendsidi): Update condition.
24840         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24841         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24842         (qhs_zextenddi_cstr): Likewise.
24844 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24846         PR middle-end/56420
24847         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24848         avoid signed wrapping.
24849         (expand_mult): Handle properly multiplication by
24850         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24851         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24852         in the compiler if coeff is HOST_WIDE_INT_MIN.
24853         (expand_divmod): Don't make ext_op1 static, change it's type to
24854         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24856         PR rtl-optimization/50339
24857         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24858         field.
24859         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24860         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24861         into splitting_ashiftrt field.
24862         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24863         ASHIFTRT.
24864         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24865         choices.
24867 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24869         PR middle-end/56108
24870         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24871         are sure to go irrevocable.
24873 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24875         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24876         scalars are valid operands.
24878 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24880         PR tree-optimization/56310
24881         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24882         only matching indices and non-negative final offsets.
24883         (intersect_aggregates_with_edge): Pass src_idx to
24884         agg_replacements_to_vector.  Pass src_idx insstead of index to
24885         intersect_with_agg_replacements.
24887 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24889         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24890         instead of hard-wired defaults.
24892 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24894         * doc/invoke.texi (MIPS Options): Update documentation of the
24895         floating-point multiply-accumulate instruction restrictions.
24897 2013-02-21  Kostya Serebryany  <kcc@google.com>
24899         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24900         asan_shadow_offset on x86_64 linux.
24902 2013-02-21  Richard Biener  <rguenther@suse.de>
24904         PR tree-optimization/56415
24905         Revert
24906         2013-02-11  Richard Biener  <rguenther@suse.de>
24908         PR tree-optimization/56273
24909         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24910         first VRP run.
24912 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24914         PR bootstrap/56258
24915         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24916         instead of @itemx.
24918         PR inline-asm/56405
24919         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24920         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24922 2013-02-20  Jan Hubicka  <jh@suse.cz>
24924         PR tree-optimization/56265
24925         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24926         when target is referenced for first time.
24928 2013-02-20  Richard Biener  <rguenther@suse.de>
24930         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24931         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24932         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24933         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24934         not return anything.
24935         (rename_ssa_copies): Do not remove unused locals.
24936         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24937         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24938         * passes.c (execute_function_todo): Do not schedule unused locals
24939         removal if cleanup_tree_cfg did something.
24940         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24941         about the number of removed locals.
24943 2013-02-20  Richard Biener  <rguenther@suse.de>
24945         PR tree-optimization/56398
24946         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24948 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24950         PR tree-optimization/55334
24951         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24952         restricted pointers to arrays.
24954 2013-02-20  Richard Biener  <rguenther@suse.de>
24955             Jakub Jelinek  <jakub@redhat.com>
24957         PR tree-optimization/56396
24958         * tree-ssa-ccp.c (n_const_val): New static variable.
24959         (get_value): Return NULL for SSA names we don't have a lattice
24960         entry for.
24961         (ccp_initialize): Initialize n_const_val.
24962         * tree-ssa-copy.c (n_copy_of): New static variable.
24963         (init_copy_prop): Initialize n_copy_of.
24964         (get_value): Return NULL_TREE for SSA names we don't have a
24965         lattice entry for.
24967 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24969         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24971 2013-02-20  Richard Biener  <rguenther@suse.de>
24973         * genpreds.c (write_lookup_constraint): Do not compare first
24974         letter of the constraint again.
24976 2013-02-20  Richard Biener  <rguenther@suse.de>
24978         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24979         and ceil_log2.
24980         (get_use_iv_cost): Terminate hashtable walk when coming across
24981         an empty entry.
24983 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24985         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24986         reassociation for avx2 targets.
24988 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24990         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24991         Add version check for v8.10.a to enable microblaze_has_clz
24992         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24993         version and TARGET_PATTERN_COMPARE check
24994         * config/microblaze/microblaze.md: New clzsi2 instruction
24996 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24998         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24999         function before branching.
25001 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
25003         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
25004         DUMP_INSN_RTX_UID.
25005         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
25007 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
25009         PR middle-end/55889
25010         * sel-sched.c: Include ira.h.
25011         (implicit_clobber_conflict_p): New function.
25012         (moveup_expr): Use it.
25013         * Makefile.in (sel-sched.o): Depend on ira.h.
25015 2013-02-19  Richard Biener  <rguenther@suse.de>
25017         PR tree-optimization/56384
25018         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
25019         (vn_hash_type): Split out from ...
25020         (vn_hash_constant_with_type): ... here.
25021         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
25022         (vn_phi_eq): Compare types from vn_phi_s structure.
25023         (vn_phi_lookup): Populate vn_phi_s type.
25024         (vn_phi_insert): Likewise.
25026 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
25028         PR tree-optimization/56350
25029         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
25030         if haven't found reduction or nested cycle operand, rather than
25031         asserting we must find it.
25033         PR tree-optimization/56381
25034         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
25035         to fold_build3.
25037 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
25038             Jakub Jelinek  <jakub@redhat.com>
25040         PR target/52555
25041         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
25042         (swap_optab_enable): Same.
25043         (init_all_optabs): Use argument instead of global.
25044         * tree.h (struct tree_optimization_option): New field target_optabs.
25045         * expr.h (init_all_optabs): Add argument to prototype.
25046         (TREE_OPTIMIZATION_OPTABS): New.
25047         (save_optabs_if_changed): Protoize.
25048         * optabs.h: Declare this_fn_optabs.
25049         * optabs.c (save_optabs_if_changed): New.
25050         Declare this_fn_optabs.
25051         (init_optabs): Add argument to init_all_optabs() call.
25052         * function.c (invoke_set_current_function_hook): Handle per
25053         function optabs.
25054         * function.h (struct function): New field optabs.
25055         * config/mips/mips.c (mips_set_mips16_mode): Handle when
25056         optimization_current_node has changed.
25057         * target-globals.h (save_target_globals_default_opts): Protoize.
25058         * target-globals.c (save_target_globals_default_opts): New.
25060 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25062         PR target/56347
25063         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
25064         registers %fr12 and %fr12R as call used.
25066         PR target/56214
25067         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
25068         and HImode, require all displacements to be an integer multiple of
25069         their mode size.
25070         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
25071         only allow QImode and HImode when reload is in progress and strict is
25072         true.  Likewise for symbolic addresses.  Use base14_operand to check
25073         displacements in REG+BASE addresses.
25075 2013-02-18  Richard Biener  <rguenther@suse.de>
25077         PR tree-optimization/56366
25078         * tree-vect-loop.c (get_initial_def_for_induction): Properly
25079         handle sign-conversion of outer-loop initial induction value.
25081 2013-02-18  Richard Biener  <rguenther@suse.de>
25083         PR middle-end/56349
25084         * cfghooks.c (merge_blocks): If we merge a latch into another
25085         block adjust references to it.
25086         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
25087         (verify_loop_structure): Verify that a recorded latch is in fact
25088         a latch.
25090 2013-02-18  Richard Biener  <rguenther@suse.de>
25092         PR tree-optimization/56321
25093         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
25094         order SSA name release and virtual operand unlinking.
25096 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25098         * config/microblaze/microblaze.md (save_stack_block): Define.
25099         (restore_stack_block): Likewise.
25101 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25103         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
25104         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
25105         * config/microblaze/microblaze.c (microblaze_option_override):
25106         Bail out early for PIC modes when target does not support PIC.
25108 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25110         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
25111         Replace with a microblaze version.
25112         (microblaze_trampoline_init): Adapt for microblaze.
25113         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
25114         microblaze.
25116 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
25117             Dodji Seketeli  <dodji@redhat.com>
25119         PR asan/56330
25120         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
25121         (instrument_mem_region_access): Do not forget to always put
25122         instrumentation of the of 'base' and 'base + len' in a "if (len !=
25123         0) statement, even for cases where either 'base' or 'base + len'
25124         are not instrumented -- because they have been previously
25125         instrumented.  Simplify the logic by putting all the statements
25126         instrument 'base + len' inside a sequence, and then insert that
25127         sequence right before the current insertion point.  Then, to
25128         instrument 'base + len', just get an iterator on that statement.
25129         And do not forget to update the pointer to iterator the function
25130         received as argument.
25132 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
25134         PR rtl-optimization/56348
25135         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
25137 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
25139         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
25140         (clean_graph_dump_file): Pass base to start_graph_dump.
25142 2013-02-14  Richard Henderson  <rth@redhat.com>
25144         PR target/55941
25145         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
25147 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
25149         * collect2-aix.h: Define F_LOADONLY.
25151 2013-02-14  Richard Biener  <rguenther@suse.de>
25153         PR lto/50494
25154         * varasm.c (output_constant_def_1): Get the decl representing
25155         the constant as argument.
25156         (output_constant_def): Wrap output_constant_def_1.
25157         (make_decl_rtl): Use output_constant_def_1 with the decl
25158         representing the constant.
25159         (build_constant_desc): Optionally re-use a decl already
25160         representing the constant.
25161         (tree_output_constant_def): Adjust.
25163 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
25165         Fix an asan crash
25166         * asan.c (instrument_builtin_call):  Really put the length of the
25167         second source argument into src1_len.
25169 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
25171         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
25172         argument.  If it is false, don't create edge from then_bb to
25173         fallthru_bb.
25174         (insert_if_then_before_iter): Pass true to it.
25175         (build_check_stmt): Pass false to it.
25176         (transform_statements): Flush hash table only on extended basic
25177         block boundaries, rather than at the beginning of every bb.
25178         Don't flush hash table on nonfreeing_call_p calls.
25179         * tree-flow.h (nonfreeing_call_p): New prototype.
25180         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
25182 2013-02-13  David S. Miller  <davem@davemloft.net>
25184         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
25186 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
25188         PR target/56184
25189         * ira.c (max_regno_before_ira): Move from ...
25190         (ira): ... here.
25191         (fix_reg_equiv_init): Use max_regno_before_ira instead of
25192         vec_safe_length.
25194 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
25196         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
25198 2013-02-13  Richard Biener  <rguenther@suse.de>
25200         PR lto/56295
25201         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
25202         globals in MEM_REFs.
25204 2013-02-13  Richard Biener  <rguenther@suse.de>
25206         * loop-init.c (loop_optimizer_init): Clear loop state when
25207         re-initializing preserved loops.
25208         * loop-unswitch.c (unswitch_single_loop): Return whether
25209         we unswitched the loop.  Do not verify loop state here.
25210         (unswitch_loops): When we unswitched a loop discover new loops.
25212 2013-02-13  Kostya Serebryany  <kcc@google.com>
25214         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
25215         on x86_64 linux.
25216         * sanitizer.def: Rename __asan_init to __asan_init_v1.
25218 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
25220         Avoid instrumenting duplicated memory access in the same basic block
25221         * Makefile.in (asan.o): Add new dependency on hash-table.h
25222         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
25223         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
25224         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
25225         (free_mem_ref_resources, has_mem_ref_been_instrumented)
25226         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
25227         (get_mem_ref_of_assignment): New functions.
25228         (get_mem_refs_of_builtin_call): Extract from
25229         instrument_builtin_call and tweak a little bit to make it fit with
25230         the new signature.
25231         (instrument_builtin_call): Use the new
25232         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
25233         of is_gimple_builtin_call.
25234         (instrument_derefs, instrument_mem_region_access): Insert the
25235         instrumented memory reference into the hash table.
25236         (maybe_instrument_assignment): Renamed instrument_assignment into
25237         this, and change it to advance the iterator when instrumentation
25238         actually happened and return true in that case.  This makes it
25239         homogeneous with maybe_instrument_assignment, and thus give a
25240         chance to callers to be more 'regular'.
25241         (transform_statements): Clear the memory reference hash table
25242         whenever we enter a new BB, when we cross a function call, or when
25243         we are done transforming statements.  Use
25244         maybe_instrument_assignment instead of instrumentation.  No more
25245         need to special case maybe_instrument_assignment and advance the
25246         iterator after calling it; it's now handled just like
25247         maybe_instrument_call.  Update comment.
25249 2013-02-13  Richard Biener  <rguenther@suse.de>
25251         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
25252         Fix loop discovery code.
25254 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
25256         PR inline-asm/56148
25257         * lra-constraints.c (process_alt_operands): Match early clobber
25258         operand with itself.  Check conflicts with earlyclobber only if
25259         the operand is not reloaded.  Prefer to reload conflicting operand
25260         if earlyclobber and matching operands are the same.
25262 2013-02-12  Richard Biener  <rguenther@suse.de>
25264         PR lto/56297
25265         * lto-streamer-out.c (write_symbol): Do not output symbols
25266         for hard register variables.
25268 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
25270         PR target/54222
25271         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
25272         (umulsidi3_insn, mulsidi3_insn): New insns.
25274 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
25276         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
25277         (struct tune_params): Add vec_costs field.
25278         * config/arm/arm.c (arm_builtin_vectorization_cost)
25279         (arm_add_stmt_cost): New functions.
25280         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
25281         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
25282         (arm_default_vec_cost): New struct of type cpu_vec_costs.
25283         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
25284         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
25285         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
25286         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
25288 2013-02-12  Richard Biener  <rguenther@suse.de>
25290         PR lto/56295
25291         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
25292         decls again if possible.
25294 2013-02-12  Richard Biener  <rguenther@suse.de>
25296         PR middle-end/56288
25297         * tree-ssa.c (verify_ssa_name): Fix check, move
25298         SSA_NAME_IN_FREE_LIST check up.
25300 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
25301             Steven Bosscher   <steven@gcc.gnu.org>
25303         PR rtl-optimization/56151
25304         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
25305         equal to op0 or op1, and last_insn pattern is CODE operation
25306         with MEM dest and one of the operands matches that MEM.
25308 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
25310         * doc/extend.texi: Document Function Multiversioning and "default"
25311         parameter string to target attribute.
25312         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
25313         target attribute parameter is "default".
25314         (ix86_compare_version_priority): Remove checks for target attribute.
25315         (ix86_mangle_function_version_assembler_name): Change error to sorry.
25316         Remove check for target attribute equal to NULL. Add assert.
25317         (ix86_generate_version_dispatcher_body): Change error to sorry.
25319 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
25320             Jack Howarth  <howarth@bromo.med.uc.edu>
25321             Patrick Marlier  <patrick.marlier@gmail.com>
25323         PR libitm/55693
25324         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
25325         define ENDFILE_SPEC as TM_DESTRUCTOR.
25326         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
25328 2013-02-11  Alexander Potapenko  <glider@google.com>
25329             Jack Howarth  <howarth@bromo.med.uc.edu>
25330             Jakub Jelinek  <jakub@redhat.com>
25332         PR sanitizer/55617
25333         * config/darwin.c (cdtor_record): Rename ctor_record.
25334         (sort_cdtor_records): Rename sort_ctor_records.
25335         (finalize_dtors): New routine to sort destructors by
25336         priority before use in assemble_integer.
25337         (machopic_asm_out_destructor): Use finalize_dtors if needed.
25339 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
25341         PR rtl-optimization/56275
25342         * simplify-rtx.c (avoid_constant_pool_reference): Check that
25343         offset is non-negative and less than cmode size before
25344         calling simplify_subreg.
25346 2013-02-11  Richard Biener  <rguenther@suse.de>
25348         PR tree-optimization/56264
25349         * cfgloop.h (fix_loop_structure): Adjust prototype.
25350         * loop-init.c (fix_loop_structure): Return the number of
25351         newly discovered loops.
25352         * tree-cfgcleanup.c (repair_loop_structures): When new loops
25353         are discovered, do a full loop-closed SSA rewrite.
25355 2013-02-11  Richard Biener  <rguenther@suse.de>
25357         PR tree-optimization/56273
25358         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
25359         first VRP run.
25360         (check_array_ref): Fix missing newline in dumps.
25361         (search_for_addr_array): Likewise.
25363 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
25365         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
25367 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
25369         PR target/56256
25370         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
25372 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
25374         PR rtl-optimization/56246
25375         * lra-constraints.c (simplify_operand_subreg): Try to reuse
25376         reload pseudo.
25377         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
25378         constraints are satisfied.
25380 2013-02-08  Jeff Law  <law@redhat.com>
25382         PR debug/53948
25383         * emit-rtl.c (reg_is_parm_p): New function.
25384         * regs.h (reg_is_parm_p): New prototype.
25385         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
25386         callee-clobbered registers.
25388 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
25390         PR target/56043
25391         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
25392         If there is no implicit builtin declaration, just return NULL.
25394 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25396         * config/i386/sse.md (FMAMODEM): New mode iterator.
25397         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
25398         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
25400 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25402         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
25403         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
25404         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
25406 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25408         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
25409         (microblaze*-*-elf): Likewise.
25410         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
25411         LINK_SPEC.
25412         * config/microblaze/microblaze-c.c: Add builtin defines for
25413         _LITTLE_ENDIAN and _BIG_ENDIAN.
25414         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
25415         add to TARGET_DEFAULT flags.
25416         Expand ASM_SPEC and LINK_SPEC.
25417         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
25418         * config/microblaze/microblaze.md: Update extendsidi2 and
25419         movdi_internal instructions to use low-order / high-order reg
25420         print_operands.
25421         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
25422         options and inversemask / mask of LITTLE_ENDIAN.
25423         * config/microblaze/t-microblaze: Expand multilib options to
25424         include mlittle-endian (le) and update exceptions patterns.
25426 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
25428         PR rtl-optimization/56195
25429         * lra-constraints.c (get_reload_reg): Don't reuse regs
25430         if they have smaller mode than requested, if they have
25431         wider mode than requested, try to return a SUBREG.
25433         PR tree-optimization/56250
25434         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
25435         if type is unsigned and code isn't MULT_EXPR.
25437 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25439         PR tree-optimization/56064
25440         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
25441         bits according to mode.
25442         * fixed-value.h (fixed_from_double_int)
25443         (const_fixed_from_double_int): Adjust comments.
25445 2013-02-08  Richard Biener  <rguenther@suse.de>
25447         PR lto/56231
25448         * lto-streamer.h (struct data_in): Remove current_file, current_line
25449         and current_col members.
25450         * lto-streamer-out.c (lto_output_location): Stream changed bits
25451         en-block for efficiency.
25452         * lto-streamer-in.c (clear_line_info): Remove.
25453         (lto_input_location): Cache current file, line and column
25454         globally via local statics.  Read changed bits en-block.
25455         (input_function): Do not call clear_line_info.
25456         (lto_read_body): Likewise.
25457         (lto_input_toplevel_asms): Likewise.
25459 2013-02-08  Michael Matz  <matz@suse.de>
25461         PR tree-optimization/52448
25462         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
25463         (nt_call_phase): New static.
25464         (add_or_mark_expr): Only mark accesses with newer phase than any
25465         call seen.
25466         (nonfreeing_call_p): New.
25467         (nt_init_block): Update nt_call_phase, mark blocks as visited.
25468         (nt_fini_block): Keep blocks marked as visited.
25469         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
25471 2013-02-08  Richard Biener  <rguenther@suse.de>
25473         * ira.c (ira): Free broken dominator information.
25475 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25477         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25479 2013-02-08  Marek Polacek  <polacek@redhat.com>
25481         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25483 2013-02-08  Richard Biener  <rguenther@suse.de>
25485         PR middle-end/56181
25486         * cfgloop.h (flow_loops_find): Adjust.
25487         (bb_loop_header_p): Declare.
25488         * cfgloop.c (bb_loop_header_p): New function split out from ...
25489         (flow_loops_find): ... here.  Adjust function signature,
25490         support incremental loop structure update.
25491         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25492         * cfgloopmanip.c (fix_loop_structure): Move ...
25493         * loop-init.c (fix_loop_structure): ... here.
25494         (apply_loop_flags): Split out from ...
25495         (loop_optimizer_init): ... here.
25496         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25497         in incremental mode, only remove dead loops here.
25499 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25501         PR target/54222
25502         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25503         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25504         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25505         (*round<mode>3.libgcc): New insns for fixed-modes.
25506         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25507         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25508         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25509         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25510         implementations.  Define to __builtin_avr_absFX,
25511         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25512         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25513         __builtin_avr_countlsFX, respectively.
25514         * config/avr/avr-c.c (target.h): Include it.
25515         (enum avr_builtin_id): New enum.
25516         (avr_resolve_overloaded_builtin): New static function.
25517         (avr_register_target_pragmas): Use it to set
25518         targetm.resolve_overloaded_builtin.
25519         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25520         tree nodes used by DEF_BUILTIN.
25521         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25522         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25523         <AVR_BUILTIN_xxBITS>: Same.
25525 2013-02-08  Richard Biener  <rguenther@suse.de>
25527         * cfgloop.c (verify_loop_structure): Properly handle
25528         a loop exiting to another loop header.
25529         * ira-int.h (ira_loops): Remove.
25530         * ira.c (ira_loops): Remove.
25531         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25532         (do_reload): Use loop_optimizer_finalize.
25533         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25534         number_of_loops to access the loop tree.
25535         (more_one_region_p): Likewise.
25536         (finish_loop_tree_nodes): Likewise.
25537         (rebuild_regno_allocno_maps): Likewise.
25538         (mark_loops_for_removal): Likewise.
25539         (mark_all_loops_for_removal): Likewise.
25540         (remove_unnecessary_regions): Likewise.
25541         (ira_build): Likewise.
25542         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25544 2013-02-08  Richard Biener  <rguenther@suse.de>
25546         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25547         * ipa-pure-const.c (analyze_function): Avoid calling
25548         mark_irreducible_loops twice.
25549         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25551 2013-02-07  David S. Miller  <davem@davemloft.net>
25553         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25554         on 'reg'.
25555         * var-tracking.c (vt_add_function_parameter): Test the presence of
25556         HAVE_window_save properly and do not remap argument registers when
25557         we have a leaf function.
25559 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25561         PR bootstrap/56227
25562         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25563         instead of "ll".
25564         * config/i386/i386.c (ix86_print_operand): Ditto.
25566 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25568         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25570 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25572         PR rtl-optimization/56225
25573         * lra-constraints.c (process_alt_operands): Check that reload hard
25574         reg can hold value for strict_low_part.
25576 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25578         PR debug/56154
25579         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25580         dwarf2out_end_function.
25581         (in_first_function_p, maybe_at_text_label_p,
25582         first_loclabel_num_not_at_text_label): New variables.
25583         (dwarf2out_var_location): In the first function find out
25584         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25585         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25586         functions.
25588 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25590         PR rtl-optimization/56178
25591         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25592         SUBREG of a register.  Tidy up related block of code.
25593         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25594         note if the source is a register or a SUBREG of a register.
25596 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25598         PR target/56228
25599         * config/rs6000/rs6000.md (ptrm): New mode attr.
25600         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25601         call_value_indirect_aix<pttrsize>,
25602         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25603         m in constraints.
25605 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25607         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25608         if -bnortl. Convert to strcmp and strncmp.
25610 2013-02-07  Alan Modra  <amodra@gmail.com>
25612         PR target/54009
25613         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25614         addresses won't wrap when offsetting.
25615         (rs6000_secondary_reload): Provide secondary reloads needed for
25616         wrapping LO_SUM addresses.
25618 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25620         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25621         MACH, just __MACH__.
25623 2013-02-06  Richard Biener  <rguenther@suse.de>
25625         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25626         instead of calling fix_loop_structure.
25628 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25630         PR middle-end/56217
25631         * omp-low.c (use_pointer_for_field): Return false if
25632         lower_send_shared_vars doesn't generate any copy-out code.
25634 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25636         PR rtl-optimization/56131
25637         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25638         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25639         of the label is NULL.  Add comment.
25641 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25643         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25645         PR sanitizer/55374
25646         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25647         (STATIC_LIBTSAN_LIBS): Likewise.
25648         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25649         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25650         is defined, don't add anything else beyond that.
25651         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25652         (LINK_COMMAND_SPEC): Use them.
25654         PR tree-optimization/56205
25655         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25656         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25657         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25659 2013-02-05  Richard Biener  <rguenther@suse.de>
25661         PR tree-optimization/53342
25662         PR tree-optimization/53185
25663         * tree-vectorizer.h (vect_check_strided_load): Remove.
25664         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25665         not disallow peeling for vectorized strided loads.
25666         (vect_check_strided_load): Make static and simplify.
25667         (vect_analyze_data_refs): Adjust.
25668         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25669         correctly when vectorizing strided loads.
25671 2013-02-05  Richard Biener  <rguenther@suse.de>
25673         * doc/install.texi: Refer to ISL, not PPL.
25675 2013-02-05  Jan Hubicka  <jh@suse.cz>
25677         PR tree-optimization/55789
25678         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25680 2013-02-05  Jan Hubicka  <jh@suse.cz>
25682         PR tree-optimization/55789
25683         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25684         the dead call anyway.
25686 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25688         PR sanitizer/55374
25689         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25691 2013-02-04  Alexander Potapenko  <glider@google.com>
25692             Jack Howarth  <howarth@bromo.med.uc.edu>
25693             Jakub Jelinek  <jakub@redhat.com>
25695         PR sanitizer/55617
25696         * config/darwin.c (sort_ctor_records): Stabilized qsort
25697         on constructor priority by using original position.
25698         (finalize_ctors): New routine to sort constructors by
25699         priority before use in assemble_integer.
25700         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25702 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25704         PR libstdc++/54314
25705         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25706         about visibility on artificial decls.
25707         * config/sol2.c (solaris_assemble_visibility): Likewise.
25709 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25711         PR target/56186
25712         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25713         argument and improve checking of return-argument types for 16-byte
25714         modes.
25715         (ix86_function_value_1): Add additional valtype argument on call
25716         of function_value_64.
25717         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25718         handling infunction_value_64 function.
25720 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25722         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25724 2013-02-04  Richard Biener  <rguenther@suse.de>
25726         PR tree-optimization/56188
25727         * tree-ssa-structalias.c (label_visit): Consider case with
25728         initially non-empty points-to set.
25729         (perform_var_substitution): Dump node mapping and clean up.
25731 2013-02-04  Richard Guenther  <rguenther@suse.de>
25733         PR lto/56168
25734         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25735         node prevail as last resort.
25736         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25737         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25739 2013-02-04  Richard Biener  <rguenther@suse.de>
25741         PR tree-optimization/56113
25742         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25743         Merge into ...
25744         (equiv_class_lookup_or_add): ... this.
25745         (label_visit): Adjust and fix error in previous patch.
25746         (perform_var_substitution): Adjust.
25748 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25750         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25751         * config/sh/sh4-300.md: Likewise.
25752         * config/sh/sh4a.md: Likewise.
25753         * config/sh/constraints.md: Likewise.
25754         * config/sh/sh.md: Likewise.
25755         * config/sh/netbsd-elf.h: Likewise.
25756         * config/sh/predicates.md: Likewise.
25757         * config/sh/sh-protos.h: Likewise.
25758         * config/sh/ushmedia.h: Likewise.
25759         * config/sh/linux.h: Likewise.
25760         * config/sh/sh.c: Likewise.
25761         * config/sh/superh.h: Likewise.
25762         * config/sh/elf.h: Likewise.
25763         * config/sh/sh4.md: Likewise.
25764         * config/sh/sh.h: Likewise.
25766 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25768         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25769         constraint to match_test floating_point_store_memory_operand().
25770         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25771         (base14_operand): New.
25772         (floating_point_store_memory_operand): New.
25773         (integer_store_memory_operand): Revise to use base14_operand and
25774         reg_plus_base_memory_operand.
25775         (move_dest_operand): Allow symbolic_memory_operands.
25776         (symbolic_memory_operand): Check for LO_SOM.
25777         (symbolic_operand): Change default case to break.
25778         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25779         CONST_DOUBLE values to be reloaded by putting them into memory when
25780         the destination is a floating point register.
25781         (movdf): Remove code to handle CONST_DOUBLE.
25782         (movsf): Likewise.
25783         (reload_indf_r1): New.
25784         (reload_insf_r1): New.
25785         Consistently use "Q" and "T" constraints with integer and floating
25786         point move instructions, respectively.
25787         (movdi): Remove FAIL.
25788         Change predicate for source operand unamed DImode move from
25789         general_operand to move_src_operand.
25790         (umulsidi3): Change predicate for destination operand to
25791         register_operand.
25792         Likewise for similar unamed patterns.
25793         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25794         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25795         (hppa_legitimize_address): Simplify mask calculation.
25796         (pa_emit_move_sequence): Revised handling of secondary reloads from
25797         REG+D addresses for floating point loads and stores.  Directly handle
25798         loading CONST0_RTX (mode) to a floating point register.
25799         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25800         to floating point registers.  Don't restrict secondary reloads to
25801         floating point registers to integer modes.  Revise some comments and
25802         cleanup some code.
25803         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25804         (pa_legitimate_address_p): New.
25805         (pa_legitimize_reload_address): New.
25806         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25807         (STRICT_REG_OK_FOR_BASE_P): New.
25808         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25809         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25811 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25812             Andrew Dixie  <andrewd@gentrack.com>
25814         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25815         flag set.
25817 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25819         * expmed.c (extract_bit_field_1): Pass the full width of the
25820         structure to get_best_reg_extraction_insn.
25822 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25824         PR target/54601
25825         * configure.ac (use_cxa_atexit): Add AIX.
25826         * configure: Regenerate.
25828         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25830 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25832         PR debug/54793
25833         * final.c (need_profile_function): New variable.
25834         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25835         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25836         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25837         notes, targetm.asm_out.function_prologue doesn't emit anything,
25838         HAVE_prologue and profiler should be emitted before prologue,
25839         set need_profile_function instead of emitting it.
25840         (final_scan_insn): If need_profile_function, emit
25841         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25842         NOTE_INSN_FUNCTION_BEG note.
25844 2013-02-01  Richard Henderson  <rth@redhat.com>
25846         * config/rs6000/rs6000.md (smulditi3): New.
25847         (umulditi3): New.
25849         * config/alpha/alpha.md (umulditi3): New.
25851 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25853         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25854         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25856 2013-02-01  Richard Biener  <rguenther@suse.de>
25858         PR tree-optimization/56113
25859         * tree-ssa-structalias.c (label_visit): Reduce work for
25860         single-predecessor nodes.
25862 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25864         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25865         range isn't testing for zero.
25867 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25869         PR middle-end/56113
25870         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25872 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25873             Nick Clifton  <nickc@redhat.com>
25875         * config/v850/constraints.md (Q): Define as a memory constraint.
25876         * config/v850/predicates.md (label_ref_operand): New predicate.
25877         (e3v5_shift_operand): New predicate.
25878         (ior_operator): New predicate.
25879         * config/v850/t-v850: Add e3v5 multilib.
25880         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25881         (v850_gen_movdi): Prototype.
25882         * config/v850/v850.c: Add support for e3v5 architecture.
25883         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25884         TARGET_V850E_UP.
25885         (construct_save_jarl): Add e3v5 long JARL support.
25886         (v850_adjust_insn_length): New function.  Adjust length of call
25887         insns when using e3v5 instructions.
25888         (v850_gen_movdi): New function: Generate instructions to move a
25889         DImode value.
25890         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25891         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25892         (TARGET_USE_FPU): Enable for e3v5.
25893         (CONST_OK_FOR_W): New macro.
25894         (ADJUST_INSN_LENGTH): Define.
25895         * config/v850/v850.md (UNSPEC_LOOP): Define.
25896         (attr cpu): Add v850e3v5.
25897         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25898         (movdi): New pattern.
25899         (movdi_internal): New pattern.
25900         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25901         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25902         (cstoresf4): Likewise.
25903         (cstoredf4): Likewise.
25904         (insv): New pattern.
25905         (rotlso3_a): New pattern.
25906         (rotlsi3_b): New pattern
25907         (rotlsi3_v850e3v5): New pattern.
25908         (doloop_begin): New pattern.
25909         (fix_loop_counter): New pattern.
25910         (doloop_end): New pattern.
25911         (branch_normal): Add e3v5 long branch support.
25912         (branch_invert): Likewise.
25913         (branch_z_normal): Likewise.
25914         (branch_z_invert): Likewise.
25915         (branch_nz_normal): Likewise.
25916         (branch_nz_invert): Likewise.
25917         (call_internal_short): Add e3v5 register-indirect JARL support.
25918         (call_internal_long): Likewise.
25919         (call_value_internal_short): Likewise.
25920         (call_value_internal_long): Likewise.
25921         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25922         (mloop): New option.
25923         * config.gcc: Add support for configuring v840e3v5 target.
25924         * doc/invoke.texi: Document new v850 specific command line options.
25926 2013-01-31  Paul Koning  <ni1d@arrl.net>
25928         PR debug/55059
25929         PR debug/54508
25930         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25931         children if parent is a class.
25932         (prune_unused_types_prune): Don't add DW_AT_declaration.
25934 2013-01-31  Richard Biener  <rguenther@suse.de>
25936         PR tree-optimization/56157
25937         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25938         match up operand with SLP child.
25940 2013-01-31  Jason Merrill  <jason@redhat.com>
25942         PR debug/54410
25943         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25944         parameters the first time.
25945         (gen_scheduled_generic_parms_dies): Check completeness here.
25947 2013-01-31  Richard Biener  <rguenther@suse.de>
25949         PR middle-end/53073
25950         * common.opt (faggressive-loop-optimizations): New flag,
25951         enabled by default.
25952         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25953         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25954         infer_loop_bounds_from_undefined by it.
25956 2013-01-31  Richard Biener  <rguenther@suse.de>
25958         PR tree-optimization/56150
25959         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25960         visit virtual operands.
25961         (find_uses_to_rename_bb): Likewise.
25963 2013-01-31  Richard Biener  <rguenther@suse.de>
25965         PR tree-optimization/56150
25966         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25967         mixed store non-store stmts.
25969 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25971         PR sanitizer/55374
25972         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25973         LIBASAN_EARLY_SPEC is defined.
25974         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25975         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25976         before %o.
25977         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25979         PR c++/55742
25980         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25981         invalid args instead of ICEing on it.
25982         (ix86_valid_target_attribute_tree): Return error_mark_node if
25983         ix86_valid_target_attribute_inner_p failed.
25984         (ix86_valid_target_attribute_p): Return false only if
25985         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25986         target("default") attribute.
25987         (sorted_attr_string): Change argument from const char * to tree,
25988         merge in all target attribute arguments rather than just one.
25989         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25990         instead of free.  Avoid using strcat.
25991         (ix86_mangle_function_version_assembler_name): Mangle
25992         target("default") as if no target attribute is present.  Adjust
25993         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25994         instead of xmalloc and XDELETEVEC instead of free.
25995         (ix86_function_versions): Don't return true if one of the decls
25996         doesn't have target attribute.  If they don't and one of the decls
25997         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25998         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25999         (ix86_supports_function_versions): Remove.
26000         (make_name): Fix up formatting.
26001         (make_dispatcher_decl): Remove resolver_name and its initialization.
26002         Avoid leaking memory.
26003         (is_function_default_version): Return true if there is
26004         target("default") attribute rather than no target attribute at all.
26005         (make_resolver_func): Avoid leaking memory.
26006         (ix86_generate_version_dispatcher_body): Likewise.
26007         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
26008         * target.def (supports_function_versions): Remove.
26009         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
26010         * doc/tm.texi: Regenerated.
26012 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
26014         PR rtl-optimization/56144
26015         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
26016         for values with side effects.
26018 2013-01-30  Richard Biener  <rguenther@suse.de>
26020         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
26021         (sparseset_pop): Likewise.
26022         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
26023         to be able to use quick_push in the worker loop.
26025 2013-01-30  Marek Polacek  <polacek@redhat.com>
26027         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
26029 2013-01-30  Richard Biener  <rguenther@suse.de>
26031         PR lto/56147
26032         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
26034 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
26036         PR tree-optimization/56064
26037         * fixed-value.c (fixed_from_double_int): New function.
26038         * fixed-value.h (fixed_from_double_int): New prototype.
26039         (const_fixed_from_double_int): New static inline function.
26040         * fold-const.c (native_interpret_fixed): New static function.
26041         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
26042         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
26043         (native_encode_fixed): New static function.
26044         (native_encode_expr) <FIXED_CST>: Use it.
26045         (native_interpret_int): Move double_int worker code to...
26046         * double-int.c (double_int::from_buffer): ...this new static method.
26047         * double-int.h (double_int::from_buffer): Prototype it.
26049 2013-01-30  Richard Biener  <rguenther@suse.de>
26051         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
26052         New pointer-map and obstack.
26053         (init_alias_vars): Allocate pointer-map and obstack.
26054         (delete_points_to_sets): Free them.
26055         (find_what_var_points_to): Cache result.
26056         (find_what_p_points_to): Adjust for changed interface of
26057         find_what_var_points_to.
26058         (compute_points_to_sets): Likewise.
26059         (ipa_pta_execute): Likewise.
26061 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
26063         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
26064         * configure: Regenerate.
26065         * config.in: Regenerate.
26066         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
26067         #nobits/#progbits if supported.
26069 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
26071         PR target/56121
26072         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
26073         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
26074         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
26076 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26078         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
26079         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
26081 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26083         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
26084         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
26086 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26088         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
26089         declaration.
26090         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
26091         * config/arm/cortex-a7.md: New bypasses using
26092         arm_mac_accumulator_is_result.
26094 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26096         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
26097         (cortex_a7_neon_mla): Likewise.
26098         (cortex_a7_fpfmad): New reservation.
26099         (cortex_a7_fpmacs): Use ffmas and update required units.
26100         (cortex_a7_fpmuld): Update required units and latency.
26101         (cortex_a7_fpmacd): Likewise.
26102         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
26103         (cortex_a7_neon). Likewise.
26104         (bypass) Update participating units.
26106 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26108         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
26109         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
26110         from fmac to ffma.
26111         * config/arm/vfp11.md (vfp_farith): Use ffmas.
26112         (vfp_fmul): Use ffmad.
26113         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
26114         (cortex_r4_fmacd): Use ffmad.
26115         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
26116         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
26117         (cortex_a9_fmacd): Use ffmad.
26118         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
26119         (cortex_a8_vfp_macd): Use ffmad.
26120         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
26121         (cortex_a5_fpmacd): Use ffmad.
26122         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
26123         (cortex_a15_vfp_macd): Use ffmad.
26124         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
26126 2013-01-29  Jason Merrill  <jason@redhat.com>
26128         PR libstdc++/54314
26129         * varasm.c (default_assemble_visibility): Don't warn about
26130         visibility on artificial decls.
26132 2013-01-29  Richard Biener  <rguenther@suse.de>
26134         PR tree-optimization/56113
26135         * tree-ssa-structalias.c (equiv_class_lookup): Also return
26136         the bitmap leader.
26137         (label_visit): Free duplicate bitmaps and record the leader instead.
26138         (perform_var_substitution): Adjust.
26140 2013-01-29  Richard Biener  <rguenther@suse.de>
26142         PR tree-optimization/55270
26143         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
26144         the CFG, schedule loops for fixup.
26146 2013-01-29  Nick Clifton  <nickc@redhat.com>
26148         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
26149         SP_REG.
26151 2013-01-28  Leif Ekblad  <leif@rdos.net>
26153         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
26154         * config/i386/i386.h (TARGET_RDOS): New macro.
26155         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
26156         * config/i386/i386.c (ix86_option_override_internal): For 64bit
26157         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
26158         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
26159         DEFAULT_LARGE_SECTION_THRESHOLD.
26160         * config/i386/i386.md (R14_REG, R15_REG): New constants.
26161         * config/i386/rdos.h: New file.
26162         * config/i386/rdos64.h: New file.
26164 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
26166         PR other/54814
26167         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
26168         TEST_HARD_REG_BIT.
26170 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
26172         PR rtl-optimization/56117
26173         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
26174         call cselib_lookup_from_insn on the MEM before calling
26175         add_insn_mem_dependence.
26177 2013-01-28  Richard Biener  <rguenther@suse.de>
26179         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
26180         to a stmt that didn't have one.
26181         (copy_phis_for_bb): Likewise for PHI arguments.
26182         (copy_debug_stmt): Likewise for debug stmts.
26184 2013-01-28  Richard Biener  <rguenther@suse.de>
26186         PR tree-optimization/56034
26187         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
26188         (partition_builtin_p): Adjust.
26189         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
26190         it is the last partition.
26191         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
26192         up the vertex for the definition.
26193         (classify_partition): Classify whether a partition is a
26194         PKIND_REDUCTION, thus has uses outside of the loop.
26195         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
26196         Merge all PKIND_REDUCTION partitions into the last partition.
26197         (tree_loop_distribution): Seed partitions from reductions as well.
26199 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
26201         PR tree-optimization/56125
26202         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
26203         pow(x,c) into sqrt(x) * powi(x, n/2) or
26204         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
26205         optimizing for size.
26206         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
26207         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
26208         integer.
26210         PR tree-optimization/56094
26211         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
26212         to UNKNOWN_LOCATION while gimplifying expr.
26214 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
26216         PR target/56114
26217         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
26218         operand 0 in movabs insn template for -masm=intel asm alternative.
26219         (*movabs<mode>_2): Ditto for operand 1.
26221 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
26223         PR target/54663
26224         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
26225         of microblaze-c.o
26227 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26229         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
26230         tm_file.
26232 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26234         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
26235         Undef to avoid warning.
26237 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
26239         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
26240         * configure: Regenerate.
26242 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
26244         PR tree-optimization/56098
26245         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
26246         for stmts with volatile ops.
26247         (cond_store_replacement): Don't optimize if assign has volatile ops.
26248         (cond_if_else_store_replacement_1): Don't optimize if either
26249         then_assign or else_assign have volatile ops.
26250         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
26251         volatile ops.
26253 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
26255         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
26257 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
26259         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
26260         missing ':' in asm example.
26262 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
26264         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
26265         entries into lane and laneq entries.
26266         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
26267         Remove AdvSIMD scalar modes.
26268         (aarch64_sq<r>dmulh_laneq<mode>): New.
26269         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
26270         modes.
26271         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
26272         builtin implementations to relfect changes in RTL in aarch64-simd.md.
26273         * config/aarch64/iterators.md (VCOND): New.
26274         (VCONQ): New.
26276 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
26278         PR target/54222
26279         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
26280         Add NULL LIBNAME argument to existing definitions.
26281         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
26282         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
26283         * config/avr/avr.c (DEF_BUILTIN): Same.
26284         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
26285         (avr_expand_builtin): Expand to a vanilla call if a libgcc
26286         implementation is available (DECL_ASSEMBLER_NAME is set).
26287         (avr_fold_absfx): New static function.
26288         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
26289         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
26290         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
26291         AVR_BUILTIN_ABSLLK.
26292         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
26293         (abshk, absk, abslk, absllk): Provide as static inline functions.
26295 2013-01-25  Marek Polacek  <polacek@redhat.com>
26297         PR tree-optimization/56035
26298         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
26300 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
26302         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
26303         (*movtf_internal_rex64): Add (!o,C) alternative
26304         (*movxf_internal_rex64): Ditto.
26305         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
26307 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
26309         * doc/invoke.texi: fix typo.
26310         * doc/objc.texi: fix typo.
26312 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
26314         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
26315         for the first two alternatives.
26317 2013-01-24  Diego Novillo  <dnovillo@google.com>
26319         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
26320         (ggc-zone.o): Remove.
26321         * configure.ac: Remove option --with-gc.
26322         * configure: Re-generate.
26323         * doc/install.texi: Remove documentation for --with-gc.
26324         * gengtype.c (write_enum_defn): Remove.  Update all users.
26325         (write_Types_process_field): Remove generation of gt_e_* argument.
26326         (output_type_enum): Remove.  Update all users.
26327         (write_enum_defn): Remove.  Update all users.
26328         (enum alloc_zone): Remove.  Update all users.
26329         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
26330         * ggc-common.c (ggc_splay_alloc): Remove first argument.
26331         Update all callers.
26332         (struct ptr_data): Remove field TYPE.  Update all users.
26333         (gt_pch_note_object): Remove argument TYPE.  Update all users.
26334         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
26335         Update all users.
26336         * ggc-none.c (ggc_alloc_typed_stat): Remove.
26337         (struct alloc_zone): Remove.
26338         (ggc_internal_alloc_zone_stat): Remove.
26339         (ggc_internal_cleared_alloc_zone_stat): Remove.
26340         * ggc-page.c (ggc_alloc_typed_stat): Remove.
26341         (ggc_pch_count_object): Remove last argument.  Update all users.
26342         (ggc_pch_alloc_object): Remove last argument.  Update all users.
26343         (struct alloc_zone): Remove.
26344         * ggc-zone.c: Remove.
26345         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
26346         (struct alloc_zone): Remove.
26347         (ggc_alloc_typed_stat): Remove.
26348         (ggc_alloc_typed): Remove.
26349         (ggc_splay_alloc): Remove first argument.
26350         (rtl_zone): Remove.  Update all users.
26351         (tree_zone): Remove.  Update all users.
26352         (tree_id_zone): Remove.  Update all users.
26353         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
26354         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
26355         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
26356         * tree-ssanames.c: Remove references to zone allocator in comments.
26358 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
26360         * config/avr/avr.c (avr_out_fract): Make register numbers that
26361         might be outside of source operand signed.
26363 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
26365         * config/i386/constraints.md (Yf): New constraint.
26366         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
26367         of f constraint to conditionaly disable x87 register preferences.
26368         (*movdf_internal): Ditto.
26369         (*movsf_internal): Ditto.
26371 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
26373         PR inline-asm/55934
26374         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
26375         that have operands with impossible constraints.
26376         Add a FIXME for a speed-up opportunity.
26377         * lra-constraints.c (process_alt_operands): Verify that a class
26378         selected from constraints on asms is valid for the operand mode.
26379         (curr_insn_transform): Remove incorrect comment.
26381 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
26383         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
26384         TOC operand is a valid symbol ref in the constant pool.
26386 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26388         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
26390 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
26392         PR target/54222
26393         * config/avr/stdfix.h: New file.
26394         * t-avr (stdfix-gcc.h): New rule to build it.
26395         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
26397 2013-01-23  Kostya Serebryany  <kcc@google.com>
26399         * config/darwin.h: remove dependency on
26400         CoreFoundation (asan on Mac OS).
26402 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
26404         PR target/49069
26405         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
26406         instead of cmpdi_operand for first comparison operand.
26407         Don't assert that comparison operands aren't both constants.
26409 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26411         * doc/install.texi (Downloading the Source): Update references to
26412         downloading separate components.
26414 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26416         * doc/extend.texi (__int128): Improve grammar.
26418 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
26420         PR target/56028
26421         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
26422         alternative to (o,r).
26423         (*movdi_internal_rex64): Remove (!o,n) alternative.
26424         (DImode immediate->memory splitter): Remove.
26425         (DImode immediate->memory peephole2): Remove.
26426         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
26427         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
26428         alternative to (!o,*r).
26429         (*movtf_internal_sse): New pattern.
26430         (*movxf_internal_rex64): New pattern.
26431         (*movxf_internal): Disable for TARGET_64BIT.
26432         (*movdf_internal_rex64): Remove (!o,F) alternative.
26434 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
26436         PR middle-end/56074
26437         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
26438         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
26439         * tree-vect-loop-manip.c (find_loop_location): Also ignore
26440         stmt locations where LOCATION_LOCUS of the stmt location is
26441         UNKNOWN_LOCATION or BUILTINS_LOCATION.
26443         PR target/55686
26444         * config/i386/i386.md (UNSPEC_STOS): New.
26445         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
26446         *strsetqi_1): Add UNSPEC_STOS.
26448 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
26450         PR c++/56067
26451         * doc/invoke.texi: Remove left over -Wsynth example.
26453 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26455         PR tree-optimization/56051
26456         * fold-const.c (fold_binary_loc): Don't fold
26457         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
26458         a narrowing conversion, or widening conversion from signed
26459         to unsigned.
26461 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26463         PR rtl-optimization/56023
26464         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
26465         dependent on debug instruction.
26467 2013-01-21  Martin Jambor  <mjambor@suse.cz>
26469         PR middle-end/56022
26470         * function.c (allocate_struct_function): Call
26471         invoke_set_current_function_hook earlier.
26473 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26475         * reload1.c (init_reload): Only initialize reload_obstack
26476         during the first call.
26478 2013-01-21  Marek Polacek  <polacek@redhat.com>
26480         * cfgloop.c (verify_loop_structure): Fix up grammar.
26482 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26484         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26485         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26487 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26489         PR target/56058
26490         * config/arm/marvell-pj4.md: Update copyright year.
26491         Fix up use of alu to alu_reg and simple_alu_imm.
26493 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26495         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26497 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26499         PR target/55433
26500         * lra-constraints.c (curr_insn_transform): Don't reuse original
26501         insn for secondary memory move when memory mode should be different.
26503 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26505         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26506         atomic_storedi_1): New patterns.
26508 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26510         btver2 pipeline descriptions.
26511         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26512         descriptions.
26513         * config/i386/i386.md (btver2_decode): New type attributes.
26514         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26515         type attributes.
26516         * config/i386/btver2.md: New file describing btver2 pipelines.
26518 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26520         PR tree-optimization/52631
26521         * tree-ssa-sccvn (visit_use): Before looking up the original
26522         statement, try looking up the simplified expression.
26524 2013-01-19  Anthony Green  <green@moxielogic.com>
26526         * config/moxie/moxie.c (moxie_expand_prologue): Set
26527         current_function_static_stack_size.
26529 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26531         PR tree-optimization/56029
26532         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26533         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26535 2013-01-18  Sharad Singhai  <singhai@google.com>
26537         PR tree-optimization/55995
26538         * dumpfile.c (dump_loc): Print location only if available.
26539         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26541 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26543         PR target/55433
26544         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26545         secondary memory move.
26546         (inherit_reload_reg): Use rclass instead of cl for
26547         check_secondary_memory_needed_p.
26549 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26551         PR middle-end/56015
26552         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26553         the case where writing real complex part of target modifies op1.
26555 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26557         * config/aarch64/aarch64-simd.md
26558         (aarch64_vcond_internal<mode>): Handle unordered cases.
26559         * config/aarch64/iterators.md (v_cmp_result): New.
26561 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26562             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26564         * config/arm/marvell-pj4.md: New file.
26565         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26566         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26567         (generic_vfp): Likewise.
26568         * config/arm/arm-cores.def: Add marvell-pj4.
26569         * config/arm/arm-tune.md: Regenerate.
26570         * config/arm/arm-tables.opt: Regenerate.
26571         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26572         * doc/invoke.texi: Document marvell-pj4.
26574 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26576         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26578 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26580         PR debug/54114
26581         PR debug/54402
26582         PR debug/49888
26583         * var-tracking.c (negative_power_of_two_p): New.
26584         (global_get_addr_cache, local_get_addr_cache): New.
26585         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26586         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26587         heading comment.
26588         (vt_stack_offset_p): Remove.
26589         (vt_canon_true_dep): Always canonicalize loc's address.
26590         (clobber_overlapping_mems): Make sure we have a MEM.
26591         (local_get_addr_clear_given_value): New.
26592         (val_reset): Clear local cached entries.
26593         (compute_bb_dataflow): Create and release the local cache.
26594         Disable duplicate MEMs clobbering.
26595         (emit_notes_in_bb): Clobber MEMs likewise.
26596         (vt_emit_notes): Create and release the local cache.
26597         (vt_initialize, vt_finalize): Create and release the global
26598         cache, respectively.
26599         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26601 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26603         PR libmudflap/53359
26604         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26605         not found in the symtab.
26607 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26609         PR debug/56006
26610         PR rtl-optimization/55547
26611         PR rtl-optimization/53827
26612         PR debug/53671
26613         PR debug/49888
26614         * alias.c (offset_overlap_p): New, factored out of...
26615         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26616         the conservative special case for symbolic constants.  Don't
26617         adjust zero sizes on alignment.
26619 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26621         PR rtl-optimization/52573
26622         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26623         REG_UNUSED for the same register.
26625 2013-01-17  Richard Biener  <rguenther@suse.de>
26626             Marek Polacek  <polacek@redhat.com>
26628         PR rtl-optimization/55833
26629         * loop-unswitch.c (unswitch_loops): Move loop verification...
26630         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26631         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26632         Set it to true when we're removing a loop from hierarchy tree in
26633         an irreducible region.
26634         (fix_bb_placements): Adjust caller.
26635         (fix_loop_placements): Likewise.
26637 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26639         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26640         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26641         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26642         Remove ID.  Adjust comments.
26643         * config/avr/avr-c.c (avr_builtin_name): Remove.
26644         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26645         * config/avr/avr.c (avr_tolower): New static function.
26646         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26647         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26648         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26649         default expansion.
26651 2013-01-17  Jan Hubicka  <jh@suse.cz>
26653         PR tree-optimization/55273
26654         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26656 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26658         PR target/55981
26659         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26660         store through atomic_store<mode>_1.
26661         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26663 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26665         PR tree-optimizations/55264
26666         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26667         for virtual methods.
26668         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26669         virtual methods before inlining is over.
26670         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26671         virtual functions.
26672         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26673         non-virtual.
26675 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26677         PR rtl-optimization/56005
26678         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26679         pending reads for prefetch.
26681 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26683         * config/aarch64/aarch64.md
26684         (*cstoresi_neg_uxtw): New pattern.
26685         (*cmovsi_insn_uxtw): New pattern.
26686         (*<optab>si3_uxtw): New pattern.
26687         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26688         (*<optab>si3_insn_uxtw): New pattern.
26689         (*bswapsi2_uxtw): New pattern.
26691 2013-01-16  Richard Biener  <rguenther@suse.de>
26693         * tree-inline.c (tree_function_versioning): Remove set but
26694         never used variable.
26696 2013-01-16  Richard Biener  <rguenther@suse.de>
26698         PR tree-optimization/55964
26699         * tree-flow.h (rename_variables_in_loop): Remove.
26700         (rename_variables_in_bb): Likewise.
26701         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26702         (copy_loop_before): Adjust and delete update-ssa status.
26703         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26704         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26705         (rename_variables_in_loop): Remove.
26706         (slpeel_update_phis_for_duplicate_loop): Likewise.
26707         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26708         use available cfg machinery instead of duplicating it.
26709         Update PHI nodes and perform poor-mans SSA update here.
26710         (slpeel_tree_peel_loop_to_edge): Adjust.
26712 2013-01-16  Richard Biener  <rguenther@suse.de>
26714         PR tree-optimization/54767
26715         PR tree-optimization/53465
26716         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26717         (vrp_visit_phi_node): For PHI arguments coming via backedges
26718         drop all symbolical range information.
26719         (execute_vrp): Compute backedges.
26721 2013-01-16  Richard Biener  <rguenther@suse.de>
26723         * doc/install.texi: Update CLooG and ISL requirements to
26724         0.18.0 and 0.11.1.
26726 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26728         PR target/55301
26729         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26730         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26731         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26733 2013-01-16  DJ Delorie  <dj@redhat.com>
26735         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26736         (UNSPECV_SP_SWITCH_E): New.
26737         (sp_switch_1): Change to an unspec.
26738         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26739         replace $r15.
26741 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26743         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26744         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26745         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26746         (expand_mem_thread_fence): Ditto.
26747         (expand_mem_signal_fence): Ditto.
26748         (expand_atomic_load): Ditto.
26749         (expand_atomic_store): Ditto.
26751 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26753         PR rtl-optimization/55547
26754         PR rtl-optimization/53827
26755         PR debug/53671
26756         PR debug/49888
26757         * alias.c (memrefs_conflict_p): Set sizes to negative after
26758         AND adjustments.
26760 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26762         PR target/55940
26763         * function.c (thread_prologue_and_epilogue_insns): Always
26764         add crtl->drap_reg to set_up_by_prologue.set, even if
26765         stack_realign_drap is false.
26767 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26769         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26770         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26771         *call): Fix indention.
26773 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26775         PR target/55876
26776         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26777         Update comment.
26779 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26781         PR rtl-optimization/55153
26782         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26784 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26786         PR tree-optimization/55920
26787         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26788         accesses as grp_to_be_debug_replaced.
26790 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26792         PR tree-optimization/55920
26793         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26794         there is non-useless type conversion needed from debug rhs to lhs,
26795         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26797 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26798             Mikael Pettersson  <mikpe@it.uu.se>
26800         PR target/43961
26801         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26802         Thumb.
26803         (ASM_OUTPUT_CASE_LABEL): Remove.
26804         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26805         * final.c (shorten_branches): Update alignment of labels before
26806         jump tables if CASE_VECTOR_SHORTEN_MODE.
26808 2013-01-15  Richard Biener  <rguenther@suse.de>
26810         PR bootstrap/55961
26811         * system.h: Do not include gmp.h for building host tools.
26813 2013-01-15  Richard Biener  <rguenther@suse.de>
26815         PR middle-end/55882
26816         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26817         account for bitpos when computing alignment.
26819 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26821         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26822         (ix86_target_macros_internal): Likewise.
26824         * config/i386/i386.c (m_CORE2I7): Removed.
26825         (m_CORE_HASWELL): New macro.
26826         (m_CORE_ALL): Likewise.
26827         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26828         (initial_ix86_arch_features): Likewise.
26829         (processor_target_table): Initializations for Core avx2.
26830         (cpu_names): New names "core-avx2".
26831         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26832         PROCESSOR_CORE_HASWELL.
26833         (ix86_issue_rate): New case.
26834         (ia32_multipass_dfa_lookahead): Likewise.
26835         (ix86_sched_init_global): Likewise.
26837         * config/i386/i386.h (TARGET_HASWELL): New macro.
26838         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26839         (processor_type): New PROCESSOR_HASWELL.
26841 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26843         PR tree-optimization/55955
26844         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26845         *SHIFT_EXPR and *ROTATE_EXPR codes.
26847         PR tree-optimization/48766
26848         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26849         -ftrapv disable -fwrapv.
26851 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26853         PR target/55974
26854         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26855         etc. to 1 and not to __flash.
26856         Use LL suffix for __INT24_MAX__ with -mint8.
26857         Use ULL suffix for __UINT24_MAX__ with -mint8.
26859 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26861         * config/avr/avr-arch.h
26862         (struct base_arch_s): Use typedef avr_arch_t instead.
26863         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26864         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26865         * config/avr/avr.c: Same.
26866         * config/avr/avr-devices.c: Same.
26867         * config/avr/driver-avr.c: Same.
26868         * config/avr/gen-avr-mmcu-texi.c: Same.
26869         * config/avr/avr-mcus.def: Adjust comment.
26871 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26873         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26874         * config/aarch64/iterators.md (VALLDI): New.
26876 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26877             Andi Kleen  <ak@linux.intel.com>
26879         PR target/55948
26880         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26881         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26882         memmodel flag.
26884 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26886         * config/avr/avr-stdint.h: Remove trailing blanks.
26887         * config/avr/avr-log.h: Same.
26888         * config/avr/avr-arch.h: Same.
26889         * config/avr/avr-devices.c: Same.
26890         * config/avr/avr-dimode.md: Same.
26891         * config/avr/predicates.md: Same.
26892         * config/avr/avr-c.c: Same.  And fix typo.
26894         * config/avr/avr-protos.h: Same.  And:
26895         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26896         (init_cumulative_args): Rename to avr_init_cumulative_args.
26897         (expand_prologue): Rename to avr_expand_prologue.
26898         (expand_epilogue): Rename to avr_expand_epilogue.
26899         (adjust_insn_length): Rename to avr_adjust_insn_length.
26900         (notice_update_cc): Rename to avr_notice_update_cc.
26901         (final_prescan_insn): Rename to avr_final_prescan_insn.
26902         * config/avr/avr.c: Same.
26903         * config/avr/avr.h: Same.
26904         * config/avr/avr.md: Remove trailing blanks.
26905         (prologue): Use avr_expand_prologue.
26906         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26908 2013-01-14  Richard Biener  <rguenther@suse.de>
26910         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26911         verify_location, collect_subblocks): New functions.
26912         (verify_gimple_in_cfg): Verify that locations only reference
26913         BLOCKs in the functions BLOCK tree.
26915 2013-01-14  Richard Biener  <rguenther@suse.de>
26917         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26918         PHI argument.
26919         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26920         unshare reference.
26921         (insert_out_of_ssa_copy_on_edge): Likewise.
26922         (rewrite_close_phi_out_of_ssa): Likewise.
26923         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26924         debug expressions.
26925         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26926         propagated constants.
26927         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26928         can not be shared.
26930 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26932         * config/avr/avr-modes.def: Add GPL copyright notice.
26934 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26936         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26937         MEMMODEL_MASK to determine memory model.
26938         (atomic_store<mode>): Ditto from operands[2].
26939         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26941 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26943         PR fortran/55935
26944         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26945         (fold_gimple_assign): Don't call unshare_expr here.
26946         (fold_ctor_reference): Call unshare_expr.
26948 2013-01-13  Terry Guo  <terry.guo@arm.com>
26950         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26951         * doc/fragments.texi: Document MULTILIB_REUSE.
26952         * gcc.c (multilib_reuse): New internal spec.
26953         (set_multilib_dir): Also search multilib from multilib_reuse.
26954         * genmultilib (tmpmultilib3): Refactor code.
26955         (tmpmultilib4): Ditto.
26956         (multilib_reuse): New multilib argument.
26958 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26960         * Makefile.in: Update copyright.
26962 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26964         PR middle-end/55890
26965         * calls.c (expand_call): Check if arg_nr is valid.
26967 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26969         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26970         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26971         documentation.  Add missing '__' in front of
26972         __builtin_ia32_packssdw256.
26974 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26976         PR target/55719
26977         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26978         NO_REGS for larl operands.
26979         (s390_reload_larl_operand): Use s390_load_address instead of
26980         emit_move_insn.
26982 2013-01-11  Richard Biener  <rguenther@suse.de>
26984         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26985         (verify_node_sharing): ... here.
26986         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26988 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26990         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26991         Substitute TREECHECKING.
26992         * configure: Regenerate.
26993         * Makefile.in (TREECHECKING): New.
26995 2013-01-11  Richard Guenther  <rguenther@suse.de>
26997         PR tree-optimization/44061
26998         * tree-vrp.c (extract_range_basic): Compute zero as
26999         value-range for __builtin_constant_p of function parameters.
27001 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
27003         Update copyright years.
27005 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
27007         PR rtl-optimization/55672
27008         * lra-eliminations.c (mark_not_eliminable): Permit addition with
27009         const to be eliminable.
27011 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
27013         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
27014         * configure: Regenerate.
27016 2013-01-10  Richard Biener  <rguenther@suse.de>
27018         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
27020 2013-01-10  Richard Biener  <rguenther@suse.de>
27022         PR bootstrap/55792
27023         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
27024         locations for virtual PHI arguments.
27025         (rewrite_update_phi_arguments): Likewise.
27027 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
27029         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
27030         on to assembler.
27032 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
27034         PR tree-optimization/55921
27035         * tree-complex.c (expand_complex_asm): New function.
27036         (expand_complex_operations_1): Call it for GIMPLE_ASM.
27038 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
27040         PR target/55718
27041         * config/s390/s390.c (s390_symref_operand_p)
27042         (s390_loadrelative_operand_p): Merge the two functions.
27043         (s390_check_qrst_address, print_operand_address): Add parameters
27044         to s390_loadrelative_operand_p invokation.
27045         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
27046         (s390_reload_larl_operand, s390_secondary_reload): Use
27047         s390_loadrelative_operand_p instead of s390_symref_operand_p.
27048         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
27050 2013-01-09  Mike Stump  <mikestump@comcast.net>
27052         * dse.c (record_store): Remove unnecessary assert.
27054 2013-01-09  Jan Hubicka  <jh@suse.cz>
27056         PR tree-optimization/55569
27057         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
27058         * cfgloop.h (scale_loop_profile): Likewise.
27060 2013-01-09  Jan Hubicka  <jh@suse.cz>
27062         PR lto/45375
27063         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
27064         functions.
27065         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
27067 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
27069         PR middle-end/55114
27070         * expr.h (maybe_emit_group_store): Declare.
27071         * expr.c (maybe_emit_group_store): New function.
27072         * builtins.c (expand_builtin_int_roundingfn): Call it.
27073         (expand_builtin_int_roundingfn_2): Likewise.
27075 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
27077         PR rtl-optimization/55829
27078         * lra-constraints.c (match_reload): Add code for absent output.
27079         (curr_insn_transform): Add code for reloads of matched inputs
27080         without output.
27082 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
27084         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
27085         attribute of movddup insn to DF.
27086         (*vec_interleave_lowv2df): Ditto.
27087         (vec_dupv2df): Ditto.
27089 2013-01-09  Jan Hubicka  <jh@suse.cz>
27091         PR tree-optimiation/55875
27092         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
27093         EVERY_ITERATION parameter.
27094         (number_of_iterations_exit): Check if exit is executed every iteration.
27095         (idx_infer_loop_bounds): Similarly here.
27096         (n_of_executions_at_most): Simplify
27097         to only test for cases where statement is dominated by the
27098         particular bound; handle correctly the "postdominance" test.
27099         (scev_probably_wraps_p): Use max loop iterations info
27100         as a global bound first.
27102 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
27103             Nick Clifton  <nickc@redhat.com>
27105         * config/v850/v850.md (cbranchsf4): New pattern.
27106         (cstoresf4): New pattern.
27107         (cbranchdf4): New pattern.
27108         (cstoredf4): New pattern.
27109         (movsicc): Disallow floating point comparisons.
27110         (cmpsf_le_insn): Fix order of operators.
27111         (cmpsf_lt_insn): Likewise.
27112         (cmpsf_eq_insn): Likewise.
27113         (cmpdf_le_insn): Likewise.
27114         (cmpdf_lt_insn): Likewise.
27115         (cmpdf_eq_insn): Likewise.
27116         (cmpsf_ge_insn): Use LE comparison.
27117         (cmpdf_ge_insn): Likewise.
27118         (cmpsf_gt_insn): Use LT comparison.
27119         (cmpdf_gt_insn): Likewise.
27120         (cmpsf_ne_insn): Delete pattern.
27121         (cmpdf_ne_insn): Delete pattern.
27122         * config/v850/v850.c (v850_gen_float_compare): Use
27123         gen_cmpdf_eq_insn for NE comparison.
27124         (v850_float_z_comparison_operator)
27125         (v850_float_nz_comparison_operator): Move from here ...
27126         * config/v850/predicates.md: ... to here.  Move GT and GE
27127         comparisons into v850_float_z_comparison_operator.
27128         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
27129         Delete prototype.
27130         (v850_float_nz_comparison_operator): Likewise.
27132 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27134         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
27135         with calls to gen_insvsi/gen_insvdi.
27137 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
27139         * config/i386/i386.c (initial_ix86_tune_features): Set up
27140         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
27142 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
27143             Jakub Jelinek  <jakub@redhat.com>
27145         PR tree-optimization/48189
27146         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
27147         If nitercst is 0, don't predict the exit edge.
27149 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
27151         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
27152         in asm_fprintf with reg_names.
27153         (aarch64_print_operand_address): Likewise.
27154         (aarch64_return_addr): Likewise.
27155         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
27157 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27159         * config/pa/pa.h (VAL_U6_BITS_P): Define.
27160         (INT_U6_BITS): Likewise.
27161         * config/pa/predicates.md (uint6_operand): New predicate.
27162         (shift5_operand, shift6_operand): Likewise.
27163         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
27164         arith32_operand.
27165         (lshrdi3): Use shift6_operand.
27166         (shrpsi4, shrpdi4): New insn patterns.
27167         (extzv): Delete expander.
27168         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
27169         predicates in unamed zero extract patterns.  Tighten common constraint.
27170         (extv): Delete expander.
27171         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
27172         predicates in unamed sign extract patterns.  Tighten common constraint.
27173         (insv): Delete expander.
27174         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
27175         predicates in unamed insert patterns.  Tighten common constraint.
27176         Change uint32_operand predicate to uint6_operand predicate in unamed
27177         DImode pattern to insert constant values of type 1...1xxxx.
27179 2013-01-04  Jan Hubicka  <jh@suse.cz>
27181         PR tree-optimization/55823
27182         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
27183         issue.
27185 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27186             Uros Bizjak  <ubizjak@gmail.com>
27188         PR rtl-optimization/55845
27189         * df-problems.c (can_move_insns_across): Stop scanning at
27190         volatile_insn_p source instruction or give up if
27191         across_from .. across_to range contains any volatile_insn_p
27192         instructions.
27194 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
27196         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
27197         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
27198         Declare.
27199         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
27200         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
27202 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27204         PR fortran/55341
27205         * asan.c (asan_clear_shadow): New function.
27206         (asan_emit_stack_protection): Use it.
27208 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
27210         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
27211         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
27212         with tab instead of space.
27214 2013-01-08  Nick Clifton  <nickc@redhat.com>
27216         * config/rl78/rl78.c (rl78_expand_prologue): Always select
27217         register bank 0 at the start of an interrupt handler.
27218         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
27219         MDBH registers.
27221 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
27223         * config/aarch64/aarch64-simd.md
27224         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
27225         (aarch64_simd_bsl): Likewise.
27226         (aarch64_vcond_internal<mode>): Likewise.
27227         (vcond<mode><mode>): Likewise.
27228         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
27229         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
27231 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
27233         * config/aarch64/aarch64-builtins.c
27234         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
27236 2013-01-08  Martin Jambor  <mjambor@suse.cz>
27238         PR debug/55579
27239         * tree-sra.c (analyze_access_subtree): Return true also after
27240         potentially creating a debug-only replacement.
27242 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27244         PR middle-end/55890
27245         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
27247         PR tree-optimization/54120
27248         * tree-vrp.c (range_fits_type_p): Don't allow
27249         src_precision < precision from signed vr to unsigned_p
27250         if vr->min or vr->max is negative.
27251         (simplify_float_conversion_using_ranges): Test can_float_p
27252         against CODE_FOR_nothing.
27254 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27255             Richard Biener  <rguenther@suse.de>
27257         PR middle-end/55851
27258         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
27259         types instead of just INTEGER_TYPE types.
27261 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
27263         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
27264         TF_SIZE): Define.
27266 2013-01-07  Steve Ellcey  <sellcey@mips.com>
27268         PR target/42661
27269         * config/mips/mips.opt: Change mad to mmad to match documentation.
27271 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27273         PR target/55897
27274         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
27275         .progmemx.data now.
27277 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27279         PR target/55897
27280         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
27281         (avr_addrspace_t): Add .section_name field.
27282         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
27283         array size.
27284         (avr_addrspace): Same.  Initialize .section_name.  Remove last
27285         NULL entry.  Put __memx into .progmemx.data.
27286         (progmem_section_prefix): Remove.
27287         (avr_asm_init_sections): No need to initialize progmem_section.
27288         (avr_asm_named_section): Use avr_addrspace[].section_name to get
27289         section name prefix.
27290         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
27291         retrieve the progmem section.
27292         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
27293         boundary to run over avr_addrspace[].
27294         (avr_register_target_pragmas): Ditto.
27296 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
27298         * varasm.c (output_constant_def_contents): For asan_protect_global
27299         protected strings, adjust DECL_ALIGN if needed, before testing for
27300         anchored symbols.
27301         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
27302         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
27303         normal decls.
27304         (output_object_block): For asan protected decls, emit asan padding
27305         after their contents.
27306         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
27307         (asan_finish_file): Test it here instead.
27309 2013-01-07  Nick Clifton  <nickc@redhat.com>
27310             Matthias Klose  <doko@debian.org>
27311             Doug Kwan  <dougkwan@google.com>
27312             H.J. Lu  <hongjiu.lu@intel.com>
27314         PR driver/55470
27315         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
27317         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
27319         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
27321         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
27323         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
27325 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27327         PR target/54461
27328         * doc/install.texi (Cross-Compiler-Specific Options): Document
27329         --with-avrlibc.
27331 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
27333         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
27334         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
27335         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
27336         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
27337         vqmovun_high_s64): Fix source operand number and update copyright.
27339 2013-01-07  Richard Biener  <rguenther@suse.de>
27341         PR middle-end/55890
27342         * gimple.h (gimple_call_builtin_p): New overload.
27343         * gimple.c (validate_call): New function.
27344         (gimple_call_builtin_p): Likewise.
27345         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27346         Use gimple_call_builtin_p.
27347         (find_func_clobbers): Likewise.
27348         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
27349         (strlen_optimize_stmt): Likewise.
27351 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27353         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
27354         (vld1q_dup_*): Likewise.
27355         (vld1_*): Likewise.
27356         (vld1q_*): Likewise.
27357         (vld1_lane_*): Likewise.
27358         (vld1q_lane_*): Likewise.
27360 2013-01-07  Richard Biener  <rguenther@suse.de>
27362         * lto-streamer.h (LTO_minor_version): Bump to 2.
27364 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27366         * config/aarch64/aarch64-protos.h
27367         (aarch64_const_double_zero_rtx_p): Rename to...
27368         (aarch64_float_const_zero_rtx_p): ...this.
27369         (aarch64_float_const_representable_p): New.
27370         (aarch64_output_simd_mov_immediate): Likewise.
27371         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
27372         move immediate case.
27373         * config/aarch64/aarch64.c
27374         (aarch64_const_double_zero_rtx_p): Rename to...
27375         (aarch64_float_const_zero_rtx_p): ...this.
27376         (aarch64_print_operand): Allow printing of new constants.
27377         (aarch64_valid_floating_const): New.
27378         (aarch64_legitimate_constant_p): Check for valid floating-point
27379         constants.
27380         (aarch64_simd_valid_immediate): Likewise.
27381         (aarch64_vect_float_const_representable_p): New.
27382         (aarch64_float_const_representable_p): Likewise.
27383         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
27384         (aarch64_output_simd_mov_immediate): New.
27385         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
27386         (*movdf_aarch64): Likewise.
27387         * config/aarch64/constraints.md (Ufc): New.
27388         (Y): call aarch64_float_const_zero_rtx.
27389         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
27391 2013-01-07  Richard Biener  <rguenther@suse.de>
27393         PR tree-optimization/55888
27394         PR tree-optimization/55862
27395         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
27396         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
27397         not if it is contained therein.
27399 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27401         * config/avr/t-avr: Typo.
27403 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27405         PR55243
27406         * config/avr/t-avr: Don't automatically rebuild
27407         $(srcdir)/config/avr/t-multilib
27408         $(srcdir)/config/avr/avr-tables.opt
27409         $(srcdir)/doc/avr-mmcu.texi
27410         (avr-mcus): New phony target to build them on request.
27411         (s-avr-mlib, s-avr-mmcu-texi): Remove.
27412         * avr/avr-mcus.def: Adjust comments.
27414 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
27416         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
27418 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27420         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
27422 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27424         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
27426 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
27428         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
27429         to generate profiling.
27430         * config/rs6000/aix64.h (LIB_SPEC): Same.
27432 2013-01-04  Andrew Pinski  <apinski@cavium.com>
27434         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
27435         New function.
27436         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
27438 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
27440         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
27441         unconditionally.
27442         (ix86_expand_move): Ditto.
27443         (ix86_zero_extend_to_Pmode): Ditto.
27444         (ix86_expand_call): Ditto.
27445         (ix86_expand_special_args_builtin): Ditto.
27446         (ix86_expand_builtin): Ditto.
27448 2013-01-04  Richard Biener  <rguenther@suse.de>
27450         PR tree-optimization/55862
27451         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
27452         translating them through PHI nodes.
27454 2013-01-04  Martin Jambor  <mjambor@suse.cz>
27456         PR tree-optimization/55755
27457         * tree-sra.c (sra_modify_assign): Do not check that an access has no
27458         children when trying to avoid producing a VIEW_CONVERT_EXPR.
27460 2013-01-04  Marek Polacek  <polacek@redhat.com>
27462         PR middle-end/55859
27463         * opts.c (default_options_optimization): Clarify error message.
27465 2013-01-04  Richard Biener  <rguenther@suse.de>
27467         PR middle-end/55863
27468         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
27469         reassociation.
27471 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27473         PR target/53789
27474         * config/pa/pa.md (movsi): Revert previous change.
27475         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27476         references.
27478 2013-01-03  Richard Henderson  <rth@redhat.com>
27480         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27481         after eliminating TLS symbols.
27483 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27485         PR bootstrap/50167
27486         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27487         * graphite-poly.c (debug_gmp_value): Likewise.
27489 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27491         PR target/55712
27492         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27493         selected code model, define __code_mode_small__, __code_model_medium__,
27494         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27495         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27496         xchg temporary register with %k.  Declare temporary register as
27497         early clobbered.
27498         [__x86_64__]: For medium and large code models, preserve %rbx register.
27500 2013-01-03  Richard Biener  <rguenther@suse.de>
27502         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27503         (dump_subscript): Adjust.
27504         (finalize_ddr_dependent): Do not dump redundant info.
27505         (analyze_siv_subscript): Adjust.
27506         (subscript_dependence_tester): Likewise.
27507         (compute_affine_dependence): Likewise.
27509 2013-01-03  Richard Biener  <rguenther@suse.de>
27511         Revert
27512         2013-01-03  Richard Biener  <rguenther@suse.de>
27514         PR tree-optimization/55857
27515         * tree-vect-stmts.c (vectorizable_load): Do not setup
27516         re-alignment for invariant loads.
27518         2013-01-02  Richard Biener  <rguenther@suse.de>
27520         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27521         invariant load do not generate a vector load from the scalar location.
27523 2013-01-03  Richard Biener  <rguenther@suse.de>
27525         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27526         for not vectorizing.
27527         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27528         not build INDIRECT_REFs, call get_name once only.
27529         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27530         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27532 2013-01-03  Richard Biener  <rguenther@suse.de>
27534         PR tree-optimization/55857
27535         * tree-vect-stmts.c (vectorizable_load): Do not setup
27536         re-alignment for invariant loads.
27538 2013-01-03  Richard Biener  <rguenther@suse.de>
27540         PR lto/55848
27541         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27542         prefer a built-in decl.
27544 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27546         * gcc.c (process_command): Update copyright notice dates.
27547         * gcov.c (print_version): Likewise.
27548         * gcov-dump.c (print_version): Likewise.
27550         PR rtl-optimization/55838
27551         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27552         iv0.step, iv1.step and step.
27554 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27555             Marc Glisse  <marc.glisse@inria.fr>
27557         PR tree-optimization/55832
27558         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27559         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27560         integer_{one,zero}_node.
27562 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27564         PR debug/54402
27565         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27566         * var-tracking.c (reverse_op): Don't add reverse ops to
27567         VALUEs that have already
27568         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27570 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27572         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27574 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27576         * dumpfile.c (dump_loc): Print filename with location.
27577         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27578         new location_t parameter to emit complete unroll message with
27579         new dump framework.
27580         (canonicalize_loop_induction_variables): Compute loops location
27581         and pass to try_unroll_loop_completely.
27582         * loop-unroll.c (report_unroll_peel): New function.
27583         (peel_loops_completely): Use new dump format with location
27584         for main dumpfile message, and invoke report_unroll_peel on success.
27585         (decide_unrolling_and_peeling): Ditto.
27586         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27587         by report_unroll_peel.
27588         (decide_peel_completely): Ditto.
27589         (decide_unroll_constant_iterations): Ditto.
27590         (decide_unroll_runtime_iterations): Ditto.
27591         (decide_peel_simple): Ditto.
27592         (decide_unroll_stupid): Ditto.
27593         * cfgloop.c (get_loop_location): New function.
27594         * cfgloop.h (get_loop_location): Declare.
27596 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27598         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27599         NULL.
27601 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27603         PR middle-end/55198
27604         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27605         BLKmode objects when EXPAND_MEMORY is specified.
27607 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27609         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27610         in loop predicate.
27611         (fold_builtin_cpu): Do not share cpu model decls across statements.
27613 2013-01-02  Jason Merrill  <jason@redhat.com>
27615         PR c++/55804
27616         * tree.c (build_array_type_1): Revert earlier change.
27618 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27620         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27621         "cortex-a57".
27622         * config/aarch64/aarch64-tune.md: Re-generate.
27624 2013-01-02  Richard Biener  <rguenther@suse.de>
27626         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27627         invariant load do not generate a vector load from the scalar location.
27629 2013-01-02  Richard Biener  <rguenther@suse.de>
27631         PR bootstrap/55784
27632         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27633         * configure: Regenerate.
27635 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27637         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27638         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27639         (expand_builtin_int_roundingfn_2): Keep the original target around
27640         for the fallback case.
27642 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27644         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27645         to be clear for sign changes.
27647 2013-01-01  Jan Hubicka  <jh@suse.cz>
27649         * ipa-inline-analysis.c: Fix formatting.
27651 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27653         PR tree-optimization/55831
27654         * tree-vect-loop.c (get_initial_def_for_induction): Use
27655         gsi_after_labels instead of gsi_start_bb.
27657 Copyright (C) 2013 Free Software Foundation, Inc.
27659 Copying and distribution of this file, with or without modification,
27660 are permitted in any medium without royalty provided the copyright
27661 notice and this notice are preserved.