* config/i386/constraints.md (B): New.
[official-gcc.git] / gcc / ChangeLog
blob6702604ef551487fd51037ff7d30539719610ea0
1 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
3         * config/i386/constraints.md (B): New.
4         (Ti): New.
5         (Tb): New.
6         * config/i386/i386-c.c (ix86_target_macros_internal): Add __MPX__.
7         * config/i386/i386-modes.def (BND32): New.
8         (BND64): New.
9         * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): New.
10         * config/i386/i386.c (isa_opts): Add mmpx.
11         (regclass_map): Add bound registers.
12         (dbx_register_map): Likewise.
13         (dbx64_register_map): Likewise.
14         (svr4_dbx_register_map): Likewise.
15         (PTA_MPX): New.
16         (ix86_option_override_internal): Support MPX ISA.
17         (ix86_conditional_register_usage): Support bound registers.
18         (print_reg): Likewise.
19         (ix86_code_end): Add MPX bnd prefix.
20         (output_set_got): Likewise.
21         (ix86_output_call_insn): Likewise.
22         (ix86_print_operand): Add '!' (MPX bnd) print prefix support.
23         (ix86_print_operand_punct_valid_p): Likewise.
24         (ix86_print_operand_address): Support UNSPEC_BNDMK_ADDR and
25         UNSPEC_BNDMK_ADDR.
26         (ix86_class_likely_spilled_p): Add bound regs support.
27         (ix86_hard_regno_mode_ok): Likewise.
28         (x86_order_regs_for_local_alloc): Likewise.
29         (ix86_bnd_prefixed_insn_p): New.
30         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Fix to new value.
31         (FIXED_REGISTERS): Add bound registers.
32         (CALL_USED_REGISTERS): Likewise.
33         (REG_ALLOC_ORDER): Likewise.
34         (HARD_REGNO_NREGS): Likewise.
35         (TARGET_MPX): New.
36         (VALID_BND_REG_MODE): New.
37         (FIRST_BND_REG): New.
38         (LAST_BND_REG): New.
39         (reg_class): Add BND_REGS.
40         (REG_CLASS_NAMES): Likewise.
41         (REG_CLASS_CONTENTS): Likewise.
42         (BND_REGNO_P): New.
43         (ANY_BND_REG_P): New.
44         (BNDmode): New.
45         (HI_REGISTER_NAMES): Add bound registers.
46         * config/i386/i386.md (UNSPEC_BNDMK): New.
47         (UNSPEC_BNDMK_ADDR): New.
48         (UNSPEC_BNDSTX): New.
49         (UNSPEC_BNDLDX): New.
50         (UNSPEC_BNDLDX_ADDR): New.
51         (UNSPEC_BNDCL): New.
52         (UNSPEC_BNDCU): New.
53         (UNSPEC_BNDCN): New.
54         (UNSPEC_MPX_FENCE): New.
55         (BND0_REG): New.
56         (BND1_REG): New.
57         (type): Add mpxmov, mpxmk, mpxchk, mpxld, mpxst.
58         (length_immediate): Likewise.
59         (prefix_0f): Likewise.
60         (memory): Likewise.
61         (prefix_rep): Check for bnd prefix.
62         (length_nobnd): New.
63         (length): Use length_nobnd if specified.
64         (BND): New.
65         (bnd_ptr): New.
66         (BNDCHECK): New.
67         (bndcheck): New.
68         (*jcc_1): Add bnd prefix and rename length attr to length_nobnd.
69         (*jcc_2): Likewise.
70         (jump): Likewise.
71         (simple_return_internal): Likewise.
72         (simple_return_pop_internal): Likewise.
73         (*indirect_jump): Add MPX bnd prefix.
74         (*tablejump_1): Likewise.
75         (simple_return_internal_long): Likewise.
76         (simple_return_indirect_internal): Likewise.
77         (<mode>_mk): New.
78         (*<mode>_mk): New.
79         (mov<mode>): New.
80         (*mov<mode>_internal_mpx): New.
81         (<mode>_<bndcheck>): New.
82         (*<mode>_<bndcheck>): New.
83         (<mode>_ldx): New.
84         (*<mode>_ldx): New.
85         (<mode>_stx): New.
86         (*<mode>_stx): New.
87         * config/i386/predicates.md (lea_address_operand): Rename to...
88         (address_no_seg_operand): ... this.
89         (address_mpx_no_base_operand): New.
90         (address_mpx_no_index_operand): New.
91         (bnd_mem_operator): New.
92         * config/i386/i386.opt (mmpx): New.
93         * doc/invoke.texi: Add documentation for the flags -mmpx, -mno-mpx.
94         * doc/rtl.texi Add documentation for BND32mode and BND64mode.
96 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
98         * mode-classes.def (MODE_POINTER_BOUNDS): New.
99         * tree.def (POINTER_BOUNDS_TYPE): New.
100         * genmodes.c (complete_mode): Support MODE_POINTER_BOUNDS.
101         (POINTER_BOUNDS_MODE): New.
102         (make_pointer_bounds_mode): New.
103         * machmode.h (POINTER_BOUNDS_MODE_P): New.
104         * stor-layout.c (int_mode_for_mode): Support MODE_POINTER_BOUNDS.
105         (layout_type): Support POINTER_BOUNDS_TYPE.
106         * tree-pretty-print.c (dump_generic_node): Support POINTER_BOUNDS_TYPE.
107         * tree.c (build_int_cst_wide): Support POINTER_BOUNDS_TYPE.
108         (type_contains_placeholder_1): Likewise.
109         * tree.h (POINTER_BOUNDS_TYPE_P): New.
110         * varasm.c (output_constant): Support POINTER_BOUNDS_TYPE.
111         * doc/rtl.texi (MODE_POINTER_BOUNDS): New.
113 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
115         * expr.c (expand_expr_real_1): Use mode of memory reference rather
116         than mode of address computation when calling memory_address_addr_space.
118 2013-08-24  Richard Henderson  <rth@twiddle.net>
120         PR rtl/58542
121         * optabs.c (maybe_emit_atomic_exchange): Use create_input_operand
122         instead of create_convert_operand_to.
123         (maybe_emit_sync_lock_test_and_set): Likewise.
124         (expand_atomic_compare_and_swap): Likewise.
125         (maybe_emit_compare_and_swap_exchange_loop): Don't convert_modes.
127 2013-08-24  Sriraman Tallam  <tmsriram@google.com>
129         * cgraph.c (cgraph_fnver_htab): Move GTY((...)) to be before htab_t.
130         Change param_is to use the struct and not the pointer to the struct.
132 2013-10-24  Andrew MacLeod  <amacleod@redhat.com>
134         * builtins.c (dummy_object, gimplify_va_arg_expr): Move to gimplify.c.
135         * gimplify.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
136         Move to targhooks.c.
137         (dummy_object, gimplify_va_arg_expr): Relocate from builtins.c.
138         * targhooks.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
139         Relocate from gimplify.c.
140         * targhooks.h: Add 2 prototypes.
141         * tree.h: Delete 2 prototypes.
143 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
145         * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p ): Check both
146         [reg+mult*reg] and [mult*reg] to determine if multiplier is allowed.
148 2013-10-24  Cong Hou  <congh@google.com>
150         * convert.c (convert_to_real): Guard those unsafe math function
151         convertions with flag_unsafe_math_optimizations.  Handle sqrt()
152         specially. 
154 2013-10-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
155         
156         PR ipa/58712
157         * cgraph.c (cgraph_create_edge_1): Add indirect_unknown_callee
158         as argument.
159         (cgraph_create_edge): Use the new argument.
160         (cgraph_create_indirect_edge): Likewise.
162 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
164         * gcc/config/arc/arc.c (arc_ccfsm_post_advance): Also handle
165         TYPE_UNCOND_BRANCH.
166         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
167         changes statep->state.
169 2013-10-24  Tobias Burnus  <burnus@net-b.de>
171         PR other/33426
172         * tree-cfg.c (replace_loop_annotate): New function.
173         (execute_build_cfg): Call it.
174         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
175         * internal-fn.c (expand_ANNOTATE): New function.
176         * internal-fn.def (ANNOTATE): Define as new internal function.
177         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
178         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
179         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
180         * doc/extend.texi (Pragmas): Document #pragma ivdep.
181         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
183 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
184             Marcus Shawcroft  <marcus.shawcroft@arm.com>
186         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
187         Special case reload SP+C into none GENERAL_REGS.
189 2013-10-24  Michael Matz  <matz@suse.de>
191         * gengtype.c (is_file_equal): Check that files will be same
192         length.
194 2013-10-25  Christian Bruel  <christian.bruel@st.com>
196         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
197         * gcc/config/sh/t-sh (sh-mem.o): New rule.
198         * gcc/config/sh/sh-mem.cc (expand_block_move): Moved here.
199         (sh4_expand_cmpstr): New function.
200         * gcc/config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c
201         * gcc/config/sh/sh-protos.h (sh4_expand_cmpstr): Declare.
202         * gcc/config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
203         (rotlhi3_8): Rename.
205 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
207         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
208         * Makefile.in (CXXDEPMODE): Assign and change users.
209         (CCDEPMODE): Delete.
210         * configure: Regenerate.
212 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
214         * gengtype-parse.c (require_without_advance): New.
215         (type): For GTY-marked types that are not GTY((user)), parse any
216         base classes, requiring them to be single-inheritance, and not
217         be templates.  For non-GTY-marked types and GTY((user)),
218         continue to skip over any C++ inheritance specification.
219         * gengtype-state.c (state_writer::write_state_struct_type):
220         Write base class of type (if any).
221         (read_state_struct_type): Read base class of type (if any).
222         * gengtype.c (new_structure): Add a "base_class" parameter.
223         (create_optional_field_): Update for new parameter to
224         new_structure.
225         (adjust_field_rtx_def): Likewise.
226         (adjust_field_tree_exp): Likewise.
227         * gengtype.h (struct type): Add "base_class" field to the s
228         union field.
229         (new_structure): Add "base" parameter.
231 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
233         PR target/57756
234         * config/i386/i386.c (ix86_option_override_internal):
235         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
236         (ix86_valid_target_attribute_tree):
237         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
238         Change TARGET_SSE to TARGET_SSE_P (opts->...)
240 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
242         * tree-ssa-loop.h: Remove include files.
243         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
244         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
245         * graphite-clast-to-gimple.c: Likewise.
246         * graphite-scop-detection.c: Likewise.
247         * graphite-sese-to-poly.c: Likewise.
248         * ipa-inline-analysis.c: Likewise.
249         * ipa-pure-const.c: Likewise.
250         * loop-init.c: Likewise.
251         * passes.c: Likewise.
252         * predict.c: Likewise.
253         * tree-cfg.c: Likewise.
254         * tree-cfgcleanup.c: Likewise.
255         * tree-chrec.c: Likewise.
256         * tree-data-ref.c: Likewise.
257         * tree-loop-distribution.c: Likewise.
258         * tree-parloops.c: Likewise.
259         * tree-predcom.c: Likewise.
260         * tree-scalar-evolution.c: Likewise.
261         * tree-ssa-address.c: Likewise.
262         * tree-ssa.c: Likewise.
263         * tree-ssa-dce.c: Likewise.
264         * tree-ssa-loop.c: Likewise.
265         * tree-ssa-loop-im.c: Likewise.
266         * tree-ssa-loop-ivcanon.c: Likewise.
267         * tree-ssa-loop-ivopts.c: Likewise.
268         * tree-ssa-loop-manip.c: Likewise.
269         * tree-ssa-loop-niter.c: Likewise.
270         * tree-ssa-loop-prefetch.c: Likewise.
271         * tree-ssa-loop-unswitch.c: Likewise.
272         * tree-ssa-reassoc.c: Likewise.
273         * tree-vect-data-refs.c: Likewise.
274         * tree-vect-loop.c: Likewise.
275         * tree-vect-loop-manip.c: Likewise.
276         * tree-vectorizer.c: Likewise.
277         * tree-vect-stmts.c: Likewise.
278         * tree-vrp.c: Likewise.
280 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
282         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
284 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
286         PR tree-optimization/58775
287         PR tree-optimization/58791
288         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
289         (insert_stmt_after): Rewritten, don't move the stmt, but really
290         insert it.
291         (get_stmt_uid_with_default): Remove.
292         (build_and_add_sum): Use insert_stmt_after and
293         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only
294         labels.
295         (update_range_test): Set uid on stmts added by
296         force_gimple_operand_gsi.  Don't immediately modify statements
297         in inter-bb optimization, just update oe->op values.
298         (optimize_range_tests): Return bool whether any changed have
299         been made.
300         (update_ops): New function.
301         (struct inter_bb_range_test_entry): New type.
302         (maybe_optimize_range_tests): Perform statement changes here.
303         (not_dominated_by, appears_later_in_bb, get_def_stmt,
304         ensure_ops_are_available): Remove.
305         (find_insert_point): Rewritten.
306         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
307         return LHS of the (new resp. old) stmt.  Don't call
308         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
309         instead of last, move new stmt at the right place.
310         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
311         (negate_value): Likewise.  Set uids.
312         (break_up_subtract_bb): Initialize uids.
313         (reassociate_bb): Adjust rewrite_expr_tree caller.
314         (do_reassoc): Don't call renumber_gimple_stmt_uids.
316 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
318         PR target/58838
319         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
320         TARGET_32BIT final condition.
321         (mulsi3_internal2 and splitter): Same.
323 2013-10-23  Jeff Law  <law@redhat.com>
325         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
326         through joiner blocks with abnormal outgoing edges.
328         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
329         Add parameter JOINERS, to allow/disallow threading through joiner
330         blocks.
331         (thread_block): New.  Call thread_block_1.
332         (mark_threaded_blocks): Remove code to filter out certain cases
333         of threading through joiner blocks.
334         (thread_through_all_blocks): Document how we can have a dangling
335         edge AUX field and clear it.
337 2013-10-23  Ian Lance Taylor  <iant@google.com>
339         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
340         (C++ Dialect Options): Likewise.
341         (Optimize Options): Likewise.
343 2013-10-23  Tom de Vries  <tom@codesourcery.com>
345         PR tree-optimization/58805
346         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
348 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
350         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
351         sequence based on get_range_info returned range.
353 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
355         * tree-ssa.h: Remove all #include's
356         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
357         * alias.c: Move required includes from tree-ssa.h.
358         * asan.c: Likewise.
359         * builtins.c: Likewise.
360         * calls.c: Likewise.
361         * cfgexpand.c: Likewise.
362         * cfghooks.c: Likewise.
363         * cfgloop.c: Likewise.
364         * cfgloopmanip.c: Likewise.
365         * cgraph.c: Likewise.
366         * cgraphbuild.c: Likewise.
367         * cgraphclones.c: Likewise.
368         * cgraphunit.c: Likewise.
369         * dse.c: Likewise.
370         * except.c: Likewise.
371         * expr.c: Likewise.
372         * final.c: Likewise.
373         * fold-const.c: Likewise.
374         * ggc-page.c: Likewise.
375         * gimple-builder.c: Likewise.
376         * gimple-fold.c: Likewise.
377         * gimple-iterator.c: Likewise.
378         * gimple-low.c: Likewise.
379         * gimple-pretty-print.c: Likewise.
380         * gimple-ssa-strength-reduction.c: Likewise.
381         * gimple-streamer-in.c: Likewise.
382         * gimple-streamer-out.c: Likewise.
383         * gimplify.c: Likewise.
384         * graphite-blocking.c: Likewise.
385         * graphite-clast-to-gimple.c: Likewise.
386         * graphite-dependences.c: Likewise.
387         * graphite-interchange.c: Likewise.
388         * graphite-optimize-isl.c: Likewise.
389         * graphite-poly.c: Likewise.
390         * graphite-scop-detection.c: Likewise.
391         * graphite-sese-to-poly.c: Likewise.
392         * graphite.c: Likewise.
393         * ipa-cp.c: Likewise.
394         * ipa-inline-analysis.c: Likewise.
395         * ipa-inline-transform.c: Likewise.
396         * ipa-inline.c: Likewise.
397         * ipa-prop.c: Likewise.
398         * ipa-pure-const.c: Likewise.
399         * ipa-reference.c: Likewise.
400         * ipa-split.c: Likewise.
401         * ipa-utils.c: Likewise.
402         * loop-init.c: Likewise.
403         * lto-cgraph.c: Likewise.
404         * lto-section-in.c: Likewise.
405         * lto-section-out.c: Likewise.
406         * lto-streamer-in.c: Likewise.
407         * lto-streamer-out.c: Likewise.
408         * lto-streamer.c: Likewise.
409         * omp-low.c: Likewise.
410         * passes.c: Likewise.
411         * predict.c: Likewise.
412         * print-tree.c: Likewise.
413         * profile.c: Likewise.
414         * sese.c: Likewise.
415         * targhooks.c: Likewise.
416         * tracer.c: Likewise.
417         * trans-mem.c: Likewise.
418         * tree-call-cdce.c: Likewise.
419         * tree-cfg.c: Likewise.
420         * tree-cfgcleanup.c: Likewise.
421         * tree-chrec.c: Likewise.
422         * tree-complex.c: Likewise.
423         * tree-data-ref.c: Likewise.
424         * tree-dfa.c: Likewise.
425         * tree-eh.c: Likewise.
426         * tree-emutls.c: Likewise.
427         * tree-if-conv.c: Likewise.
428         * tree-inline.c: Likewise.
429         * tree-into-ssa.c: Likewise.
430         * tree-loop-distribution.c: Likewise.
431         * tree-mudflap.c: Likewise.
432         * tree-nested.c: Likewise.
433         * tree-nrv.c: Likewise.
434         * tree-object-size.c: Likewise.
435         * tree-outof-ssa.c: Likewise.
436         * tree-parloops.c: Likewise.
437         * tree-phinodes.c: Likewise.
438         * tree-predcom.c: Likewise.
439         * tree-pretty-print.c: Likewise.
440         * tree-profile.c: Likewise.
441         * tree-scalar-evolution.c: Likewise.
442         * tree-sra.c: Likewise.
443         * tree-ssa-address.c: Likewise.
444         * tree-ssa-alias.c: Likewise.
445         * tree-ssa-ccp.c: Likewise.
446         * tree-ssa-coalesce.c: Likewise.
447         * tree-ssa-copy.c: Likewise.
448         * tree-ssa-copyrename.c: Likewise.
449         * tree-ssa-dce.c: Likewise.
450         * tree-ssa-dom.c: Likewise.
451         * tree-ssa-dse.c: Likewise.
452         * tree-ssa-forwprop.c: Likewise.
453         * tree-ssa-ifcombine.c: Likewise.
454         * tree-ssa-live.c: Likewise.
455         * tree-ssa-loop-ch.c: Likewise.
456         * tree-ssa-loop-im.c: Likewise.
457         * tree-ssa-loop-ivcanon.c: Likewise.
458         * tree-ssa-loop-ivopts.c: Likewise.
459         * tree-ssa-loop-manip.c: Likewise.
460         * tree-ssa-loop-niter.c: Likewise.
461         * tree-ssa-loop-prefetch.c: Likewise.
462         * tree-ssa-loop-unswitch.c: Likewise.
463         * tree-ssa-loop.c: Likewise.
464         * tree-ssa-math-opts.c: Likewise.
465         * tree-ssa-operands.c: Likewise.
466         * tree-ssa-phiopt.c: Likewise.
467         * tree-ssa-phiprop.c: Likewise.
468         * tree-ssa-pre.c: Likewise.
469         * tree-ssa-propagate.c: Likewise.
470         * tree-ssa-reassoc.c: Likewise.
471         * tree-ssa-sccvn.c: Likewise.
472         * tree-ssa-sink.c: Likewise.
473         * tree-ssa-strlen.c: Likewise.
474         * tree-ssa-structalias.c: Likewise.
475         * tree-ssa-tail-merge.c: Likewise.
476         * tree-ssa-ter.c: Likewise.
477         * tree-ssa-threadedge.c: Likewise.
478         * tree-ssa-threadupdate.c: Likewise.
479         * tree-ssa-uncprop.c: Likewise.
480         * tree-ssa-uninit.c: Likewise.
481         * tree-ssa.c: Likewise.
482         * tree-ssanames.c: Likewise.
483         * tree-stdarg.c: Likewise.
484         * tree-streamer-in.c: Likewise.
485         * tree-switch-conversion.c: Likewise.
486         * tree-tailcall.c: Likewise.
487         * tree-vect-data-refs.c: Likewise.
488         * tree-vect-generic.c: Likewise.
489         * tree-vect-loop-manip.c: Likewise.
490         * tree-vect-loop.c: Likewise.
491         * tree-vect-patterns.c: Likewise.
492         * tree-vect-slp.c: Likewise.
493         * tree-vect-stmts.c: Likewise.
494         * tree-vectorizer.c: Likewise.
495         * tree-vrp.c: Likewise.
496         * tree.c: Likewise.
497         * tsan.c: Likewise.
498         * value-prof.c: Likewise.
499         * var-tracking.c: Likewise.
500         * varpool.c: Likewise.
501         * vtable-verify.c: Likewise.
503 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
505         * config/tilegx/tilegx.c: Include "tree.h".
507 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
509         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
510         the info, not after it.
511         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
512         dump_ssaname_info call.
513         (pp_gimple_stmt_1): Adjust caller.
514         (dump_phi_nodes): Likewise.  Don't print "# " here.
516 2013-10-22  Jan Hubicka  <jh@suse.cz>
518         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New tuning flag.
519         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
520         * i386.c (expand_small_movmem_or_setmem): New function.
521         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New function
522         (alg_usable_p): Add support for value ranges; cleanup.
523         (ix86_expand_set_or_movmem): Add support for misaligned moves.
525 2013-10-22  Sterling Augustine  <saugustine@google.com>
527         * doc/invoke.texi: Document -ggnu-pubnames.
528         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
529         logic.
530         * dwarf2out.c: Include gdb/gdb-index.h.
531         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
532         debug_generate_pub_sections.
533         (is_java, output_pubtables, output_pubname): New functions.
534         (include_pubname_in_output): Handle debug_generate_pub_sections at
535         level 2.
536         (size_of_pubnames): Use new local space_for_flags based on
537         debug_generate_pub_sections.
538         (output_pubnames): Unify pubnames and pubtypes output logic.
539         Genericize comments.  Call output_pubname.
540         (dwarf2out_finish): Move logic to output_pubnames and call it.
542 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
544         PR target/58779
545         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
546         Remove CCCmode handling.
547         <case LTU>: Return 'c' suffix for CCCmode.
548         <case GEU>: Return 'nc' suffix for CCCmode.
549         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
550         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
551         (*sub<mode>3_cc_overflow): Ditto.
552         (*subsi3_zext_cc_overflow): Ditto.
554 2013-10-22  Steve Ellcey  <sellcey@mips.com>
556         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
557         (AND (NOT OP1) (NOT OP2)).
559 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
561         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
562         meaning of merge-high and merge-low masks for little endian; avoid
563         use of vector-pack masks for little endian for mismatched modes.
565 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
567         * config/tilepro/tilepro.c: Include "tree.h".
569 2013-10-22  Andreas Schwab  <schwab@suse.de>
571         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
572         with PRE_DEC.
574 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
576         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
577         [Ed Smith-Rowland]): New entries.
578         ([Stephen M. Webb]): Update.
580 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
582         * tree-ssa-ter.h: Remove duplicate copy of file contents.
584 2013-10-21  Marek Polacek  <polacek@redhat.com>
586         PR middle-end/58809
587         * fold-const.c (fold_range_test): Return 0 if the type is not
588         an integral type.
590 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
592         * system.h: Move hwint.h include further down.
593         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
594         gcc_checking_asserts.
595         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
597 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
599         Fix volatile issues in optimize_bit_field_compare.
600         * fold-const.c (optimize_bit_field_compare): Bail out if
601         lvolatilep or rvolatilep.
603 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
605         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
606         and get_inner_reference returning different pmode for non-volatile
607         bit-field members dependent on flag_strict_volatile_bitfields.
608         * stor-layout.c (layout_decl): Remove special handling of
609         flag_strict_volatile_bitfields.
610         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
611         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
613 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
615         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
616         calculations.
618 2013-10-21  Jeff Law  <law@redhat.com>
620         * tree-ssa-threadedge.c (thread_through_normal_block): New
621         argument VISITED.  Remove VISISTED as a local variable.  When we
622         have a threadable jump, verify the destination of the jump has not
623         been visised.
624         (thread_across_edge): Allocate VISITED bitmap once at function
625         scope and use it throughout.  Make sure to set appropriate bits in
626         VISITED for E (start of jump thread path).
627         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
628         through a joiner if any edge on the path has a recorded jump thread.
630 2013-10-21  Ian Lance Taylor  <iant@google.com>
632         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
633         don't imply that attributes can solve all problems.
634         (Directory Options): Fix typo.
636 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
638         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
639         extend and extend_add.
641 2013-10-21  Richard Biener  <rguenther@suse.de>
643         PR tree-optimization/58794
644         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
645         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
647 2013-10-21  Richard Biener  <rguenther@suse.de>
649         PR middle-end/58742
650         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
651         to (T) X for sign-changing conversions (or no conversion).
653 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
655         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
657 2013-10-20  Jan Hubicka  <jh@suse.cz>
659         * config/i386/i386-tune.def: Add comment; organize into categories
661 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
663         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
664         argument.  Update function comment.
665         (expand_set_or_movmem_via_rep): New function combining
666         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
667         (expand_movmem_via_rep_mov): Remove.
668         expand_setmem_via_rep_stos): Remove.
669         (expand_movmem_epilogue): Update calls correspondingly.
670         (expand_setmem_epilogue_via_loop): Likewise.
671         (emit_memset): New.
672         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
673         (expand_set_or_movmem_prologue): New function combining
674         expand_movmem_prologue and expand_setmem_prologue.
675         (expand_movmem_prologue): Remove.
676         (expand_setmem_prologue): Remove.
677         (expand_set_or_movmem_constant_prologue): New function combining
678         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
679         (expand_constant_movmem_prologue): Remove.
680         (expand_constant_setmem_prologue): Remove.
681         (promote_duplicated_reg): Allow vector-const0 value.
682         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
683         and ix86_expand_setmem.
684         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
685         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
687 2013-10-21  Diego Novillo  <dnovillo@google.com>
689         * asan.c: Include tree.h
690         * bb-reorder.c: Likewise.
691         * cfgcleanup.c: Likewise.
692         * cfgloopmanip.c: Likewise.
693         * data-streamer-in.c: Likewise.
694         * data-streamer-out.c: Likewise.
695         * data-streamer.c: Likewise.
696         * dwarf2cfi.c: Likewise.
697         * graphite-blocking.c: Likewise.
698         * graphite-clast-to-gimple.c: Likewise.
699         * graphite-dependences.c: Likewise.
700         * graphite-interchange.c: Likewise.
701         * graphite-optimize-isl.c: Likewise.
702         * graphite-poly.c: Likewise.
703         * graphite-scop-detection.c: Likewise.
704         * graphite-sese-to-poly.c: Likewise.
705         * graphite.c: Likewise.
706         * ipa-devirt.c: Likewise.
707         * ipa-profile.c: Likewise.
708         * ipa.c: Likewise.
709         * ira.c: Likewise.
710         * loop-init.c: Likewise.
711         * loop-unroll.c: Likewise.
712         * lower-subreg.c: Likewise.
713         * lto/lto-object.c: Likewise.
714         * recog.c: Likewise.
715         * reginfo.c: Likewise.
716         * tree-loop-distribution.c: Likewise.
717         * tree-parloops.c: Likewise.
718         * tree-ssa-strlen.c: Likewise.
719         * tree-streamer.c: Likewise.
720         * value-prof.c: Likewise.
721         * target-globals.c: Likewise.
722         * expr.h: Include tree-core.h instead of tree.h.
723         * gimple.h: Likewise.
724         * ipa-prop.h: Likewise.
725         * ipa-utils.h: Likewise.
726         * lto-streamer.h: Likewise.
727         * streamer-hooks.h: Likewise.
728         * ipa-reference.h: Include cgraph.h instead of tree.h.
729         * cgraph.h: Include basic-block.h instead of tree.h.
730         * tree-streamer.h: Do not include tree.h.
731         * genattrtab.c (write_header): Generate inclusion of
732         tree.h.
733         * genautomata.c (main): Likewise.
734         * genemit.c: Likewise.
735         * genopinit.c: Likewise.
736         * genoutput.c (output_prologue): Likewise.
737         * genpeep.c: Likewise.
739 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
741         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
742         little endian.
743         (vec_unpacku_hi_v8hi): Likewise.
744         (vec_unpacku_lo_v16qi): Likewise.
745         (vec_unpacku_lo_v8hi): Likewise.
747 2013-10-20  Jan Hubicka  <jh@suse.cz>
749         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
750         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
751         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
753 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
755         * config/mips/mips.h (ISA_HAS_WSBH): Define.
756         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
757         constants.
758         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
760 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
762         PR target/58603
763         * system.h: Undef m_slot.
765 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
767         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
768         all elements for both endian flavors.
770 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
772         PR target/58792
773         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
774         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
775         and SI_REG for 64bit SYSV ABI targets.
777 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
779         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
780         to multi_reg_return.  Clarify that we are skipping USEs of multiple
781         return registers.  Use bool type where appropriate.
783 2013-10-18  Jan Hubicka  <jh@suse.cz>
785         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
786         for cold functions.
787         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
788         (X86_TUNE_PUSH_MEMORY): Likewise.
789         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
790         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
791         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
792         New.
793         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
794         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
795         Remove.
796         (ix86_option_override_internal): Update to use tune features instead
797         of variables.
799 2013-10-18  Cong Hou  <congh@google.com>
801         PR tree-optimization/58508
802         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
803         statement that contains data refs with zero-step.
805 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
807         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
808         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
809         * gimple-low.c (gimple_check_call_arg,
810         gimple_check_call_matching_types): Move to cgraph.c.
811         * gimple-low.h: Remove prototype.
812         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
813         Relocate from gimple-low.c.
814         * cgraph.h: Add prototype.  Don't include basic-block.h.
815         * gimplify.c: Add gimple-low to include list.
816         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
817         * tree-eh.c: Add gimple-low to include list.
818         * tree-nested.c: Likewise.
819         * cfgexpand.c: Add tree-ssa-address.h to include list.
820         * expr.c: Likewise.
821         * gimple-fold.c: Likewise.
822         * gimple-ssa-strength-reduction.c: Likewise.
823         * trans-mem.c: Likewise.
824         * tree-mudflap.c: Likewise.
825         * tree-ssa-loop-ivopts.c: Likewise.
826         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
827         (degenerate_phi_result): Move to tree-phinodes.c.
828         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
829         * tree-ssa-threadedge.c: Likewise.
830         * tree-vrp.c: Likewise.
831         * tree-phinodes.c (degenerate_phi_result): Relocate here.
832         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
833         * tree-phinodes.h (degenerate_phi_result): Add prototype.
834         * tree-ssa-copy.c: Include tree-ssa-dom.h.
835         * tree-ssa-forwprop.c: Likewise.
836         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
837         pass_data_cleanup_cfg_post_optimizing,
838         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
839         * tree-optimize.c: Delete File.
840         * graphite.c: Include tree-cfgcleanup.h.
841         * passes.c: Likewise.
842         * tree-cfg.c: Likewise.
843         * tree-profile.c: Likewise.
844         * tree-ssa-dse.c: Likewise.
845         * tree-ssa-loop-ivcanon.c: Likewise.
846         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
847         * tree-outof-ssa.c: Include sbitmap.h.
848         * tree-ssa-live.c: Likewise.
849         * tree-ssa-propagate.c: Likewise.
850         * tree-ssa-structalias.c: Likewise.
851         * tree-stdarg.c: Likewise.
852         * Makefile.in (OBJS): Delete tree-optimize.o.
853         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
854         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
855         * varasm.c: Include basic-block.h.
856         * cfgloop.h: Include function.h instead of basic-block.h
857         (bb_loop_depth): Move to cfgloop.c.
858         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
860 2013-10-18  Teresa Johnson  <tejohnson@google.com>
862         * predict.c (probably_never_executed): Compare frequency-based
863         count to number of training runs.
864         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
866 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
868         * config/arm/arm.c (cortexa9_extra_costs): New table.
869         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
871 2013-10-18  Jeff Law  <law@redhat.com>
873         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
875         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
876         (register_jump_thread): Add "registered_jump_thread" debug
877         counter support.
878         * dbgcnt.def (registered_jump_thread): New debug counter.
880 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
882         * config/rs6000/rs6000.c: Include cgraph.h.
884 2013-10-18  Teresa Johnson  <tejohnson@google.com>
886         * tree-ssa-tail-merge.c (replace_block_by): Update edge
887         weights during merging.
889 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
891         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
892         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
893         * tree-ssa-operands.h: Remove prototype.
894         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
895         * gimple.c (virtual_operand_p): Relocate from gimple.c.
896         * gimple.h: Add prototype.
897         * gimple-ssa.h: Include tree-ssa-operands.h.
898         * tree-dump.c: Add tree-cfg.h to include list.
899         * tree-ssa-alias.c: Add ipa-reference.h to include list.
900         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
901         * config/i386/i386.c: Don't include tree-flow.h.
902         * config/rs6000/rs6000.c: Likewise.
904 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
906         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
908 2013-10-18  Richard Biener  <rguenther@suse.de>
910         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
911         or TYPE_UNSIGNED of integral types.
912         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
913         NULL_TREE for zero-precision integral types.
915 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
917         * config/aarch64/arm_neon.h
918         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
920 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
922         PR target/57756
923         * opth-gen.awk: Define target_flags_explicit.
925 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
927         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
928         fields to the reg_addr array that describes the valid addressing
929         mode for any register, general purpose registers, floating point
930         registers, and Altivec registers.
931         (FIRST_RELOAD_REG_CLASS): Likewise.
932         (LAST_RELOAD_REG_CLASS): Likewise.
933         (struct reload_reg_map_type): Likewise.
934         (reload_reg_map_type): Likewise.
935         (RELOAD_REG_VALID): Likewise.
936         (RELOAD_REG_MULTIPLE): Likewise.
937         (RELOAD_REG_INDEXED): Likewise.
938         (RELOAD_REG_OFFSET): Likewise.
939         (RELOAD_REG_PRE_INCDEC): Likewise.
940         (RELOAD_REG_PRE_MODIFY): Likewise.
941         (reg_addr): Likewise.
942         (mode_supports_pre_incdec_p): New helper functions to say whether
943         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
944         (mode_supports_pre_modify_p): Likewise.
945         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
946         print the valid address mode bits for each mode.
947         (rs6000_debug_print_mode): Likewise.
948         (rs6000_debug_reg_global): Likewise.
949         (rs6000_setup_reg_addr_masks): New function to set up the address
950         mask bits for each type.
951         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
952         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
953         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
954         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
955         PRE_MODIFY are supported.
956         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
957         registers, instead of {src,dest}_av_p.
958         (rs6000_print_options_internal): Tweak the debug output slightly.
960 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
962         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
963         isa attribute.
965 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
967         * tree-flow.h (struct omp_region): Move to omp-low.c.
968         Remove omp_ prototypes and variables.
969         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
970         (copy_var_decl): Relocate prototype from tree-flow.h.
971         * gimple.c (copy_var_decl): Relocate from omp-low.c.
972         * tree.h: Move prototype to omp-low.h.
973         * omp-low.h: New File.  Relocate prototypes here.
974         * omp-low.c (struct omp_region): Make local here.
975         (root_omp_region): Make static.
976         (copy_var_decl) Move to gimple.c.
977         (new_omp_region): Make static.
978         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
979         * tree-cfg.c: Include omp-low.h.
980         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
981         * gimplify.c: Include omp-low.h.
982         * tree-parloops.c: Likewise.
984 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
986         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
987         operands are in memory, prefer to force non-matched operand 1 to
988         the register.
990 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
992         PR target/58673
993         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
994         restrict TImode addresses to single indirect registers if both
995         -mquad-memory and -mvsx-timode are used.
996         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
997         we should emit load/store quad.  Remove using %y for quad memory
998         addresses.
1000         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
1001         constraints to allow load/store quad on machines where TImode is
1002         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
1003         for TImode to load integer constants.
1005 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1007         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
1009 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1011         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
1012         handling of STACK_REG.
1014 2013-10-17  Richard Biener  <rguenther@suse.de>
1016         PR tree-optimization/58143
1017         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
1018         New function.
1019         (rewrite_to_defined_overflow): Likewise.
1020         (move_computations_dom_walker::before_dom): Rewrite stmts
1021         with undefined signed overflow that are not always executed
1022         into unsigned arithmetic.
1024 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1026         PR target/57756
1027         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
1028         explicit isa flag to be an options variable, instead of using
1029         global_options_set.  Remove define from rs6000.h.
1030         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
1032         * config/rs6000/rs6000.c (rs6000_option_override_internal):
1033         Initialize rs6000_isa_flags_explicit.
1034         (rs6000_function_specific_save): Add gcc_options* parameter, so
1035         that the powerpc builds after the 2013-10-15 changes.
1036         (rs6000_function_specific_restore): Likewise.
1038 2013-10-16  DJ Delorie  <dj@redhat.com>
1040         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
1041         op is a REG before checking REGNO.
1042         (rl78_alloc_physical_registers): Verify pattern is a SET before
1043         checking SET_SRC.
1045 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1047         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
1048         endianness.
1049         (vec_unpacks_lo_v4sf): Likewise.
1050         (vec_unpacks_float_hi_v4si): Likewise.
1051         (vec_unpacks_float_lo_v4si): Likewise.
1052         (vec_unpacku_float_hi_v4si): Likewise.
1053         (vec_unpacku_float_lo_v4si): Likewise.
1055 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1057         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
1058         (vsx_concat_v2sf): Likewise.
1060 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
1062         * config/aarch64/aarch64.md
1063         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
1065 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
1067         PR tree-optimization/58697
1068         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
1069         estimated_loop_iterations_int.
1070         (max_stmt_executions_int): Call get_max_loop_iterations_int.
1071         (get_max_loop_iterations_int): New.  HWINT version of
1072         get_max_loop_iterations.
1073         * cfgloop.h: Add prototypes.
1074         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
1075         * loop-unroll.c (decide_peel_once_rolling): Call
1076         get_estimated_loop_iterations_int.
1077         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
1078         * tree-ssa-loop-niter.h: Tweak prototypes.
1080 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
1082         * gengtype-parse.c (struct_field_seq): Ignore access-control
1083         keywords ("public:" etc).
1085 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1087         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
1088         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
1090 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
1092         * config/arm/arm.opt (mlra): New option.
1093         * config/arm/arm.c (arm_lra_p): New function.
1094         (TARGET_LRA_P): Define.
1096 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
1098         * tree-core.h (tree_code_name): Remove.
1099         * tree.h (get_tree_code_name): New prototype.
1100         * tree.c (tree_code_name): Make static.
1101         (get_tree_code_name): New function.
1102         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
1103         tree_class_check_failed, tree_range_check_failed,
1104         tree_not_class_check_failed, omp_clause_check_failed,
1105         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
1106         wrapper get_tree_code_name instead of calling tree_code_name directly.
1107         * tree-vrp.c (dump_asserts_for): Likewise.
1108         * tree-dump.c (dequeue_and_dump): Likewise.
1109         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
1110         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
1111         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
1112         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
1113         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
1114         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
1115         dump_gimple_omp_for): Likewise.
1116         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
1117         * tree-ssa-pre.c (print_pre_expr): Likewise.
1118         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1119         * print-tree.c (print_node_brief, print_node): Likewise.
1120         * gimple.c (gimple_check_failed): Likewise.
1121         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
1122         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
1123         * config/mep/mep.c (mep_validate_vliw): Likewise.
1124         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
1125         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
1127 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
1129         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
1130         for AMD bdver3.
1132 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
1134         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
1135         (MULTILIB_MATCHES): Add multilib for -march=v8.
1137 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
1139         PR target/57756
1140         * optc-save-gen.awk: Add extra parameter to the save and restore
1141         target calls.
1142         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
1143         * tree.c (build_optimization_node): New parameter.  Add extra parameter
1144         to call to cl_optimization_save.
1145         (build_target_option_node): New parameter. Add extra parameter
1146         to call to cl_target_option_save.
1147         * tree.h (build_optimization_node): New parameter.
1148         (build_target_option_node): New parameter.
1149         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
1150         build_optimization_node and build_target_option_node.
1151         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
1152         (handle_pragma_push_options): Ditto.
1153         * toplev.c (process_options): Ditto.
1154         * opts.c (init_options_struct): Check for opts_set non-null.
1155         * target.def (target_option.save): New parameter.
1156         (target_option.restore): New parameter.
1157         * tm.texi: Generate.
1158         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1159         (ix86_pragma_target_parse): Ditto.
1160         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
1161         parameters.
1162         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
1163         to  build_optimization_node and build_target_option_node.
1164         (rs6000_valid_attribute_p): Ditto.
1165         (rs6000_pragma_target_parse): Ditto.
1166         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
1167         data.
1168         * config/i386/i386.h:
1169         TARGET_64BIT_P: New Macro
1170         TARGET_MMX_P: New Macro.
1171         TARGET_3DNOW_P: New Macro.
1172         TARGET_3DNOW_A_P: New Macro.
1173         TARGET_SSE_P: New Macro.
1174         TARGET_SSE2_P: New Macro.
1175         TARGET_SSE3_P: New Macro.
1176         TARGET_SSSE3_P: New Macro.
1177         TARGET_SSE4_1_P: New Macro.
1178         TARGET_SSE4_2_P: New Macro.
1179         TARGET_AVX_P: New Macro.
1180         TARGET_AVX2_P: New Macro.
1181         TARGET_AVX512F_P: New Macro.
1182         TARGET_AVX512PF_P: New Macro.
1183         TARGET_AVX512ER_P: New Macro.
1184         TARGET_AVX512CD_P: New Macro.
1185         TARGET_FMA_P: New Macro.
1186         TARGET_SSE4A_P: New Macro.
1187         TARGET_FMA4_P: New Macro.
1188         TARGET_XOP_P: New Macro.
1189         TARGET_LWP_P: New Macro.
1190         TARGET_ABM_P: New Macro.
1191         TARGET_BMI_P: New Macro.
1192         TARGET_BMI2_P: New Macro.
1193         TARGET_LZCNT_P: New Macro.
1194         TARGET_TBM_P: New Macro.
1195         TARGET_POPCNT_P: New Macro.
1196         TARGET_SAHF_P: New Macro.
1197         TARGET_MOVBE_P: New Macro.
1198         TARGET_CRC32_P: New Macro.
1199         TARGET_AES_P: New Macro.
1200         TARGET_PCLMUL_P: New Macro.
1201         TARGET_CMPXCHG16B_P: New Macro.
1202         TARGET_FSGSBASE_P: New Macro.
1203         TARGET_RDRND_P: New Macro.
1204         TARGET_F16C_P: New Macro.
1205         TARGET_RTM_P: New Macro.
1206         TARGET_HLE_P: New Macro.
1207         TARGET_RDSEED_P: New Macro.
1208         TARGET_PRFCHW_P: New Macro.
1209         TARGET_ADX_P: New Macro.
1210         TARGET_FXSR_P: New Macro.
1211         TARGET_XSAVE_P: New Macro.
1212         TARGET_XSAVEOPT_P: New Macro.
1213         TARGET_LP64_P: New Macro.
1214         TARGET_X32_P: New Macro.
1215         TARGET_FPMATH_DEFAULT_P: New Macro.
1216         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
1217         * config/i386/i386.c (ix86_option_override_internal): New parameters.
1218         opts and opts_set.
1219         Change ix86_tune_string to access opts->x_ix86_tune_string.
1220         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
1221         Change ix86_arch_string to access opts->x_ix86_arch_string.
1222         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
1223         Change ix86_pmode to access opts->x_ix86_pmode.
1224         Change ix86_abi to access opts->x_ix86_abi.
1225         Change ix86_cmodel to access opts->x_ix86_cmodel.
1226         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
1227         Change ix86_isa_flags_explicit to access
1228         opts->x_ix86_isa_flags_explicit.
1229         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
1230         Change ix86_regparm to access opts->x_ix86_regparm.
1231         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
1232         Change ix86_preferred_stack_boundary_arg to access
1233         opts->x_ix86_preferred_stack_boundary_arg.
1234         Change ix86_force_align_arg_pointer to access
1235         opts->x_ix86_force_align_arg_pointer.
1236         Change ix86_incoming_stack_boundar_arg to access
1237         opts->x_ix86_incoming_stack_boundar_arg.
1238         Change ix86_fpmath to access opts->x_ix86_fpmath.
1239         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
1240         Change ix86_recip_name to access opts->x_ix86_recip_name.
1241         Change ix86_stack_protector_guard to access
1242         opts->x_ix86_stack_protector_guard.
1243         Change ix86_tune_memcpy_strategy to access
1244         opts->x_ix86_tune_memcpy_strategy.
1245         Change ix86_tune_memset_strategy to access
1246         opts->x_ix86_tune_memset_strategy.
1247         Change global_options to access opts.
1248         Change global_options_set to access opts_set.
1249         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
1250         Change TARGET_MMX to TARGET_MMX_P (opts->...).
1251         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
1252         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
1253         Change TARGET_SSE to TARGET_SSE_P (opts->...).
1254         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
1255         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
1256         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
1257         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
1258         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
1259         Change TARGET_AVX to TARGET_AVX_P (opts->...).
1260         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
1261         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
1262         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
1263         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
1264         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
1265         Change TARGET_FMA to TARGET_FMA_P (opts->...).
1266         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
1267         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
1268         Change TARGET_XOP to TARGET_XOP_P (opts->...).
1269         Change TARGET_LWP to TARGET_LWP_P (opts->...).
1270         Change TARGET_ABM to TARGET_ABM_P (opts->...).
1271         Change TARGET_BMI to TARGET_BMI_P (opts->...).
1272         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
1273         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
1274         Change TARGET_TBM to TARGET_TBM_P (opts->...).
1275         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
1276         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
1277         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
1278         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
1279         Change TARGET_AES to TARGET_AES_P (opts->...).
1280         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
1281         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
1282         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
1283         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
1284         Change TARGET_F16C to TARGET_F16C_P (opts->...).
1285         Change TARGET_RTM to TARGET_RTM_P (opts->...).
1286         Change TARGET_HLE to TARGET_HLE_P (opts->...).
1287         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
1288         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
1289         Change TARGET_ADX to TARGET_ADX_P (opts->...).
1290         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
1291         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
1292         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
1293         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
1294         Change TARGET_X32 to TARGET_X32_P (opts->...).
1295         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
1296         Change TARGET_FLOAT_RETURNS_IN_80387 to
1297         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
1298         (ix86_function_specific_save): New parameter. Use opts-> fields
1299         to replace global fields.
1300         (ix86_function_specific_restore): Ditto.
1301         (ix86_valid_target_attribute_inner_p): New parameters.
1302         Fix recursive call.
1303         Fix call to ix86_handle_option and set_option.
1304         (ix86_valid_target_attribute_tree): New parameters.
1305         Change global_options to access opts.
1306         Change global_options_set to access opts_set.
1307         Fix call to ix86_valid_target_attribute_inner_p.
1308         Change ix86_tune_string to access opts->x_ix86_tune_string.
1309         Change ix86_arch_string to access opts->x_ix86_arch_string.
1310         Change ix86_fpmath to access opts->x_ix86_fpmath
1311         Fix call to ix86_option_override_internal.
1312         Fix call to ix86_add_new_builtins.
1313         Fix calls to build_optimization_node and build_target_option_node.
1314         (ix86_valid_target_attribute_p): Remove access to global_options.
1315         Use new gcc_options structure func_options.
1316         Fix call to ix86_valid_target_attribute_tree.
1317         Fix call to  build_optimization_node.
1318         (get_builtin_code_for_version): Fix call to
1319         ix86_valid_target_attribute_tree.
1321 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
1323         * Makefile.in (PICFLAG): New.
1324         (enable_host_shared): New.
1325         (INTERNAL_CFLAGS): Use PICFLAG.
1326         (LIBIBERTY): Use pic build of libiberty.a if configured with
1327         --enable-host-shared.
1328         * configure.ac: Add --enable-host-shared, setting up new
1329         PICFLAG variable.
1330         * configure: Regenerate.
1331         * doc/install.texi (--enable-shared): Add note contrasting it with ...
1332         (--enable-host-shared): New option.
1334 2013-10-15  Richard Biener  <rguenther@suse.de>
1336         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
1337         for built-in functions.
1339 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
1341         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
1342         (optimize_range_tests_1): New function,
1343         extracted from optimize_range_tests.
1344         (optimize_range_tests_xor): Similarly.
1345         (optimize_range_tests_diff): New function.
1346         (optimize_range_tests): Use optimize_range_tests_1.
1348 2013-10-15  Cong Hou  <congh@google.com>
1350         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
1351         requirement of the reduction pattern so that one operand of the
1352         reduction operation can come from outside of the loop.
1354 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1356         * config/arm/neon-schedgen.ml: Remove.
1357         * config/arm/cortex-a9-neon.md: Remove comment regarding
1358         neon-schedgen.ml.
1360 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1362         * config/arm/types: Remove old neon types.
1364 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1366         * config/arm/cortex-a7.md
1367         (cortex_a7_neon_type): New.
1368         (cortex_a7_neon_mul): Update for new types.
1369         (cortex_a7_neon_mla): Likewise.
1370         (cortex_a7_neon): Likewise.
1372 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1374         * config/arm/cortex-a15-neon.md
1375         (cortex_a15_neon_type): New,
1377         (cortex_a15_neon_int_1): Remove.
1378         (cortex_a15_neon_int_2): Likewise.
1379         (cortex_a15_neon_int_3): Likewise.
1380         (cortex_a15_neon_int_4): Likewise.
1381         (cortex_a15_neon_int_5): Likewise.
1382         (cortex_a15_neon_vqneg_vqabs): Likewise.
1383         (cortex_a15_neon_vmov): Likewise.
1384         (cortex_a15_neon_vaba): Likewise.
1385         (cortex_a15_neon_vaba_qqq): Likewise.
1386         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1387         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1388         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1389         Likewise.
1390         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1391         (cortex_a15_neon_mla_qqq_8_16): Likewise.
1392         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
1393         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1394         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1395         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
1396         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1397         (cortex_a15_neon_shift_1): Likewise.
1398         (cortex_a15_neon_shift_2): Likewise.
1399         (cortex_a15_neon_shift_3): Likewise.
1400         (cortex_a15_neon_vshl_ddd): Likewise.
1401         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1402         (cortex_a15_neon_vsra_vrsra): Likewise.
1403         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1404         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1405         (cortex_a15_neon_bp_3cycle): Likewise.
1406         (cortex_a15_neon_ldm_2): Likewise.
1407         (cortex_a15_neon_stm_2): Likewise.
1408         (cortex_a15_neon_mcr): Likewise.
1409         (cortex_a15_neon_mrc): Likewise.
1410         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
1411         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
1412         (cortex_a15_neon_fp_vmul_ddd): Likewise.
1413         (cortex_a15_neon_fp_vmul_qqd): Likewise.
1414         (cortex_a15_neon_fp_vmla_ddd): Likewise.
1415         (cortex_a15_neon_fp_vmla_qqq): Likewise.
1416         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1417         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1418         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1419         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1420         (cortex_a15_neon_bp_simple): Likewise.
1421         (cortex_a15_neon_bp_2cycle): Likewise.
1422         (cortex_a15_neon_bp_3cycle): Likewise.
1423         (cortex_a15_neon_vld1_1_2_regs): Likewise.
1424         (cortex_a15_neon_vld1_3_4_regs): Likewise.
1425         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1426         (cortex_a15_neon_vld2_4_regs): Likewise.
1427         (cortex_a15_neon_vld3_vld4): Likewise.
1428         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1429         (cortex_a15_neon_vst1_3_4_regs): Likewise.
1430         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
1431         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
1432         (cortex_a15_neon_vst3_vst4): Rename to...
1433         (cortex_a15_neon_vst4): This, update for new attributes.
1434         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
1435         (cortex_a15_neon_vld3_vld4_lane): Likewise.
1436         (cortex_a15_neon_vst1_vst2_lane): Likewise.
1437         (cortex_a15_neon_vst3_vst4_lane): Likewise.
1438         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
1439         (cortex_a15_neon_ldm_2): Likewise.
1440         (cortex_a15_neon_stm_2): Likewise.
1441         (cortex_a15_neon_mcr): Likewise.
1442         (cortex_a15_neon_mcr_2_mcrr): Likewise.
1443         (cortex_a15_neon_mrc): Likewise.
1444         (cortex_a15_neon_mrrc): Likewise.
1446         (cortex_a15_neon_abd): New.
1447         (cortex_a15_neon_abd_q): Likewise.
1448         (cortex_a15_neon_aba): Likewise.
1449         (cortex_a15_neon_aba_q): Likewise.
1450         (cortex_a15_neon_acc): Likewise.
1451         (cortex_a15_neon_acc_q): Likewise.
1452         (cortex_a15_neon_arith_basic): Likewise.
1453         (cortex_a15_neon_arith_complex): Likewise.
1454         (cortex_a15_neon_multiply): Likewise.
1455         (cortex_a15_neon_multiply_q): Likewise.
1456         (cortex_a15_neon_mla): Likewise.
1457         (cortex_a15_neon_mla_q): Likewise.
1458         (cortex_a15_neon_sat_mla_long): Likewise.
1459         (cortex_a15_neon_shift_acc): Likewise.
1460         (cortex_a15_neon_shift_imm_basic): Likewise.
1461         (cortex_a15_neon_shift_imm_complex): Likewise.
1462         (cortex_a15_neon_shift_reg_basic): Likewise.
1463         (cortex_a15_neon_shift_reg_basic_q): Likewise.
1464         (cortex_a15_neon_shift_reg_complex): Likewise.
1465         (cortex_a15_neon_shift_reg_complex_q): Likewise.
1466         (cortex_a15_neon_fp_negabs): Likewise
1467         (cortex_a15_neon_fp_arith): Likewise
1468         (cortex_a15_neon_fp_arith_q): Likewise
1469         (cortex_a15_neon_fp_cvt_int): Likewise
1470         (cortex_a15_neon_fp_cvt_int_q): Likewise
1471         (cortex_a15_neon_fp_cvt_16): Likewise
1472         (cortex_a15_neon_fp_mul): Likewise
1473         (cortex_a15_neon_fp_mul_q): Likewise
1474         (cortex_a15_neon_fp_mla): Likewise
1475         (cortex_a15_neon_fp_mla_q): Likewise
1476         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
1477         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
1478         (cortex_a15_neon_bitops): Likewise.
1479         (cortex_a15_neon_bitops_q): Likewise.
1480         (cortex_a15_neon_from_gp): Likewise.
1481         (cortex_a15_neon_from_gp_q): Likewise.
1482         (cortex_a15_neon_tbl3_tbl4): Likewise.
1483         (cortex_a15_neon_zip_q): Likewise.
1484         (cortex_a15_neon_to_gp): Likewise.
1485         (cortex_a15_neon_load_a): Likewise.
1486         (cortex_a15_neon_load_b): Likewise.
1487         (cortex_a15_neon_load_c): Likewise.
1488         (cortex_a15_neon_load_d): Likewise.
1489         (cortex_a15_neon_load_e): Likewise.
1490         (cortex_a15_neon_load_f): Likewise.
1491         (cortex_a15_neon_store_a): Likewise.
1492         (cortex_a15_neon_store_b): Likewise.
1493         (cortex_a15_neon_store_c): Likewise.
1494         (cortex_a15_neon_store_d): Likewise.
1495         (cortex_a15_neon_store_e): Likewise.
1496         (cortex_a15_neon_store_f): Likewise.
1497         (cortex_a15_neon_store_g): Likewise.
1498         (cortex_a15_neon_store_h): Likewise.
1499         (cortex_a15_vfp_to_from_gp): Likewise.
1501 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1503         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
1505         (cortex_a9_neon_vshl_ddd): Remove.
1506         (cortex_a9_neon_vst3_vst4): Likewise.
1507         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1509         (cortex_a9_neon_bit_ops_q): New.
1511         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
1512         (cortex_a9_neon_int_2): Likewise.
1513         (cortex_a9_neon_int_3): Likewise.
1514         (cortex_a9_neon_int_4): Likewise.
1515         (cortex_a9_neon_int_5): Likewise.
1516         (cortex_a9_neon_vqneg_vqabs): Likewise.
1517         (cortex_a9_neon_vmov): Likewise.
1518         (cortex_a9_neon_vaba): Likewise.
1519         (cortex_a9_neon_vaba_qqq): Likewise.
1520         (cortex_a9_neon_shift_1): Likewise.
1521         (cortex_a9_neon_shift_2): Likewise.
1522         (cortex_a9_neon_shift_3): Likewise.
1523         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1524         (cortex_a9_neon_vsra_vrsra): Likewise.
1525         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1526         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1527         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1528         Likewise.
1529         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1530         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1531         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1532         Likewise.
1533         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1534         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1535         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1536         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1537         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1538         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1539         (cortex_a9_neon_fp_vsum): Likewise.
1540         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1541         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1542         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1543         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1544         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1545         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1546         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1547         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1548         (cortex_a9_neon_bp_simple): Likewise.
1549         (cortex_a9_neon_bp_2cycle): Likewise.
1550         (cortex_a9_neon_bp_3cycle): Likewise.
1551         (cortex_a9_neon_ldr): Likewise.
1552         (cortex_a9_neon_str): Likewise.
1553         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1554         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1555         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1556         (cortex_a9_neon_vld2_4_regs): Likewise.
1557         (cortex_a9_neon_vld3_vld4): Likewise.
1558         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1559         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1560         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1561         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1562         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1563         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1564         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1565         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1566         (cortex_a9_neon_mcr): Likewise.
1567         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1568         (cortex_a9_neon_mrc): Likewise.
1569         (cortex_a9_neon_mrrc): Likewise.
1571 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1573         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
1575         (cortex_a8_neon_vshl_ddd): Remove.
1576         (cortex_a8_neon_vst3_vst4): Likewise.
1577         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1579         (cortex_a8_neon_bit_ops_q): New.
1581         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
1582         (cortex_a8_neon_int_2): Likewise..
1583         (cortex_a8_neon_int_3): Likewise.
1584         (cortex_a8_neon_int_5): Likewise.
1585         (cortex_a8_neon_vqneg_vqabs): Likewise.
1586         (cortex_a8_neon_int_4): Likewise.
1587         (cortex_a8_neon_vaba): Likewise.
1588         (cortex_a8_neon_vaba_qqq): Likewise.
1589         (cortex_a8_neon_shift_1): Likewise.
1590         (cortex_a8_neon_shift_2): Likewise.
1591         (cortex_a8_neon_shift_3): Likewise.
1592         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1593         (cortex_a8_neon_vsra_vrsra): Likewise.
1594         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1595         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1596         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
1597         Likewise.
1598         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1599         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1600         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
1601         Likewise.
1602         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1603         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1604         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1605         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1606         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1607         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1608         (cortex_a8_neon_fp_vsum): Likewise.
1609         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1610         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1611         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1612         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1613         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1614         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1615         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1616         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1617         (cortex_a8_neon_bp_simple): Likewise.
1618         (cortex_a8_neon_bp_2cycle): Likewise.
1619         (cortex_a8_neon_bp_3cycle): Likewise.
1620         (cortex_a8_neon_ldr): Likewise.
1621         (cortex_a8_neon_str): Likewise.
1622         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1623         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1624         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1625         (cortex_a8_neon_vld2_4_regs): Likewise.
1626         (cortex_a8_neon_vld3_vld4): Likewise.
1627         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1628         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1629         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1630         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1631         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1632         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1633         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1634         (cortex_a8_neon_mcr): Likewise.
1635         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1636         (cortex_a8_neon_mrc): Likewise.
1637         (cortex_a8_neon_mrrc): Likewise.
1639 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1641         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1642         (fp): New.
1643         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1644         (aarch64_simd_dup<mode>): Add "type" attribute.
1645         (aarch64_dup_lane<mode>): Likewise.
1646         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1647         (*aarch64_simd_mov<mode>): Likewise.
1648         (aarch64_simd_mov_from_<mode>low): Likewise.
1649         (aarch64_simd_mov_from_<mode>high): Likewise.
1650         (orn<mode>3): Likewise.
1651         (bic<mode>3): Likewise.
1652         (add<mode>3): Likewise.
1653         (sub<mode>3): Likewise.
1654         (mul<mode>3): Likewise.
1655         (*aarch64_mul3_elt<mode>): Likewise.
1656         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1657         (*aarch64_mul3_elt_to_128df): Likewise.
1658         (*aarch64_mul3_elt_to_64v2df): Likewise.
1659         (neg<mode>2): Likewise.
1660         (abs<mode>2): Likewise.
1661         (abd<mode>_3): Likewise.
1662         (aba<mode>_3): Likewise.
1663         (fabd<mode>_3): Likewise.
1664         (*fabd_scalar<mode>3): Likewise.
1665         (and<mode>3): Likewise.
1666         (ior<mode>3): Likewise.
1667         (xor<mode>3): Likewise.
1668         (one_cmpl<mode>2): Likewise.
1669         (aarch64_simd_vec_set<mode>): Likewise.
1670         (aarch64_simd_lshr<mode>): Likewise.
1671         (aarch64_simd_ashr<mode>): Likewise.
1672         (aarch64_simd_imm_shl<mode>): Likewise.
1673         (aarch64_simd_reg_sshl<mode): Likewise.
1674         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1675         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1676         (aarch64_simd_vec_setv2di): Likewise.
1677         (aarch64_simd_vec_set<mode>): Likewise.
1678         (aarch64_mla<mode>): Likewise.
1679         (*aarch64_mla_elt<mode>): Likewise.
1680         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1681         (aarch64_mls<mode>): Likewise.
1682         (*aarch64_mls_elt<mode>): Likewise.
1683         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1684         (<su><maxmin><mode>3): Likewise.
1685         (move_lo_quad_<mode>): Likewise.
1686         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1687         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1688         (vec_pack_trunc_<mode>): Likewise.
1689         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1690         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1691         (*aarch64_<su>mlal_lo<mode>): Likewise.
1692         (*aarch64_<su>mlal_hi<mode>): Likewise.
1693         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1694         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1695         (*aarch64_<su>mlal<mode>): Likewise.
1696         (*aarch64_<su>mlsl<mode>): Likewise.
1697         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1698         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1699         (add<mode>3): Likewise.
1700         (sub<mode>3): Likewise.
1701         (mul<mode>3): Likewise.
1702         (div<mode>3): Likewise.
1703         (neg<mode>2): Likewise.
1704         (abs<mode>2): Likewise.
1705         (fma<mode>4): Likewise.
1706         (*aarch64_fma4_elt<mode>): Likewise.
1707         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1708         (*aarch64_fma4_elt_to_128df): Likewise.
1709         (*aarch64_fma4_elt_to_64v2df): Likewise.
1710         (fnma<mode>4): Likewise.
1711         (*aarch64_fnma4_elt<mode>): Likewise.
1712         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1713         (*aarch64_fnma4_elt_to_128df): Likewise.
1714         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1715         (<frint_pattern><mode>2): Likewise.
1716         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1717         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1718         (vec_unpacks_lo_v4sf): Likewise.
1719         (aarch64_float_extend_lo_v2df): Likewise.
1720         (vec_unpacks_hi_v4sf): Likewise.
1721         (aarch64_float_truncate_lo_v2sf): Likewise.
1722         (aarch64_float_truncate_hi_v4sf): Likewise.
1723         (aarch64_vmls<mode>): Likewise.
1724         (<su><maxmin><mode>3): Likewise.
1725         (<maxmin_uns><mode>3): Likewise.
1726         (reduc_<sur>plus_<mode>): Likewise.
1727         (reduc_<sur>plus_v2di): Likewise.
1728         (reduc_<sur>plus_v2si): Likewise.
1729         (reduc_<sur>plus_<mode>): Likewise.
1730         (aarch64_addpv4sf): Likewise.
1731         (clz<mode>2): Likewise.
1732         (reduc_<maxmin_uns>_<mode>): Likewise.
1733         (reduc_<maxmin_uns>_v2di): Likewise.
1734         (reduc_<maxmin_uns>_v2si): Likewise.
1735         (reduc_<maxmin_uns>_<mode>): Likewise.
1736         (reduc_<maxmin_uns>_v4sf): Likewise.
1737         (aarch64_simd_bsl<mode>_internal): Likewise.
1738         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1739         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1740         (aarch64_get_lane<mode>): Likewise.
1741         (*aarch64_combinez<mode>): Likewise.
1742         (aarch64_combine<mode>): Likewise.
1743         (aarch64_simd_combine<mode>): Likewise.
1744         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1745         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1746         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1747         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1748         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1749         (aarch64_<sur>h<addsub><mode>): Likewise.
1750         (aarch64_<sur><addsub>hn<mode>): Likewise.
1751         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1752         (aarch64_pmul<mode>): Likewise.
1753         (aarch64_<su_optab><optab><mode>): Likewise.
1754         (aarch64_<sur>qadd<mode>): Likewise.
1755         (aarch64_sqmovun<mode>): Likewise.
1756         (aarch64_<sur>qmovn<mode>): Likewise.
1757         (aarch64_s<optab><mode>): Likewise.
1758         (aarch64_sq<r>dmulh<mode>): Likewise.
1759         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1760         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1761         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1762         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1763         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1764         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1765         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1766         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1767         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1768         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1769         (aarch64_sqdmull<mode>): Likewise.
1770         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1771         (aarch64_sqdmull_n<mode>): Likewise.
1772         (aarch64_sqdmull2<mode>_internal): Likewise.
1773         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1774         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1775         (aarch64_<sur>shl<mode>): Likewise.
1776         (aarch64_<sur>q<r>shl<mode>
1777         (aarch64_<sur>shll_n<mode>): Likewise.
1778         (aarch64_<sur>shll2_n<mode>): Likewise.
1779         (aarch64_<sur>shr_n<mode>): Likewise.
1780         (aarch64_<sur>sra_n<mode>): Likewise.
1781         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1782         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1783         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1784         (aarch64_cm<optab><mode>): Likewise.
1785         (aarch64_cm<optab>di): Likewise.
1786         (aarch64_cm<optab><mode>): Likewise.
1787         (aarch64_cm<optab>di): Likewise.
1788         (aarch64_cmtst<mode>): Likewise.
1789         (aarch64_cmtstdi): Likewise.
1790         (aarch64_cm<optab><mode>): Likewise.
1791         (*aarch64_fac<optab><mode>): Likewise.
1792         (aarch64_addp<mode>): Likewise.
1793         (aarch64_addpdi): Likewise.
1794         (sqrt<mode>2): Likewise.
1795         (vec_load_lanesoi<mode>): Likewise.
1796         (vec_store_lanesoi<mode>): Likewise.
1797         (vec_load_lanesci<mode>): Likewise.
1798         (vec_store_lanesci<mode>): Likewise.
1799         (vec_load_lanesxi<mode>): Likewise.
1800         (vec_store_lanesxi<mode>): Likewise.
1801         (*aarch64_mov<mode>): Likewise.
1802         (aarch64_ld2<mode>_dreg): Likewise.
1803         (aarch64_ld2<mode>_dreg): Likewise.
1804         (aarch64_ld3<mode>_dreg): Likewise.
1805         (aarch64_ld3<mode>_dreg): Likewise.
1806         (aarch64_ld4<mode>_dreg): Likewise.
1807         (aarch64_ld4<mode>_dreg): Likewise.
1808         (aarch64_tbl1<mode>): Likewise.
1809         (aarch64_tbl2v16qi): Likewise.
1810         (aarch64_combinev16qi): Likewise.
1811         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1812         (aarch64_st2<mode>_dreg): Likewise.
1813         (aarch64_st2<mode>_dreg): Likewise.
1814         (aarch64_st3<mode>_dreg): Likewise.
1815         (aarch64_st3<mode>_dreg): Likewise.
1816         (aarch64_st4<mode>_dreg): Likewise.
1817         (aarch64_st4<mode>_dreg): Likewise.
1818         (*aarch64_simd_ld1r<mode>): Likewise.
1819         (aarch64_frecpe<mode>): Likewise.
1820         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1821         (aarch64_frecps<mode>): Likewise.
1823 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1825         * config/arm/iterators.md (V_elem_ch): New.
1826         (q): Likewise.
1827         (VQH_type): Likewise.
1828         * config/arm/arm.md (is_neon_type): New.
1829         (conds): Use is_neon_type.
1830         (anddi3_insn): Update type attribute.
1831         (xordi3_insn): Likewise.
1832         (one_cmpldi2): Likewise.
1833         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1834         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1835         (*movmisalign<mode>_neon_store): Likewise.
1836         (*movmisalign<mode>_neon_load): Likewise.
1837         (vec_set<mode>_internal): Likewise.
1838         (vec_set<mode>_internal): Likewise.
1839         (vec_setv2di_internal): Likewise.
1840         (vec_extract<mode>): Likewise.
1841         (vec_extract<mode>): Likewise.
1842         (vec_extractv2di): Likewise.
1843         (*add<mode>3_neon): Likewise.
1844         (adddi3_neon): Likewise.
1845         (*sub<mode>3_neon): Likewise.
1846         (subdi3_neon): Likewise.
1847         (fma<VCVTF:mode>4): Likewise.
1848         (fma<VCVTF:mode>4_intrinsic): Likewise.
1849         (*fmsub<VCVTF:mode>4): Likewise.
1850         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1851         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1852         (ior<mode>3): Likewise.
1853         (and<mode>3): Likewise.
1854         (orn<mode>3_neon): Likewise.
1855         (orndi3_neon): Likewise.
1856         (bic<mode>3_neon): Likewise.
1857         (bicdi3_neon): Likewise.
1858         (xor<mode>3): Likewise.
1859         (one_cmpl<mode>2): Likewise.
1860         (abs<mode>2): Likewise.
1861         (neg<mode>2): Likewise.
1862         (negdi2_neon): Likewise.
1863         (*umin<mode>3_neon): Likewise.
1864         (*umax<mode>3_neon): Likewise.
1865         (*smin<mode>3_neon): Likewise.
1866         (*smax<mode>3_neon): Likewise.
1867         (vashl<mode>3): Likewise.
1868         (vashr<mode>3_imm): Likewise.
1869         (vlshr<mode>3_imm): Likewise.
1870         (ashl<mode>3_signed): Likewise.
1871         (ashl<mode>3_unsigned): Likewise.
1872         (neon_load_count): Likewise.
1873         (ashldi3_neon_noclobber): Likewise.
1874         (ashldi3_neon): Likewise.
1875         (signed_shift_di3_neon): Likewise.
1876         (unsigned_shift_di3_neon): Likewise.
1877         (ashrdi3_neon_imm_noclobber): Likewise.
1878         (lshrdi3_neon_imm_noclobber): Likewise.
1879         (<shift>di3_neon): Likewise.
1880         (widen_ssum<mode>3): Likewise.
1881         (widen_usum<mode>3): Likewise.
1882         (quad_halves_<code>v4si): Likewise.
1883         (quad_halves_<code>v4sf): Likewise.
1884         (quad_halves_<code>v8hi): Likewise.
1885         (quad_halves_<code>v16qi): Likewise.
1886         (reduc_splus_v2di): Likewise.
1887         (neon_vpadd_internal<mode>): Likewise.
1888         (neon_vpsmin<mode>): Likewise.
1889         (neon_vpsmax<mode>): Likewise.
1890         (neon_vpumin<mode>): Likewise.
1891         (neon_vpumax<mode>): Likewise.
1892         (*ss_add<mode>_neon): Likewise.
1893         (*us_add<mode>_neon): Likewise.
1894         (*ss_sub<mode>_neon): Likewise.
1895         (*us_sub<mode>_neon): Likewise.
1896         (neon_vadd<mode>_unspec): Likewise.
1897         (neon_vaddl<mode>): Likewise.
1898         (neon_vaddw<mode>): Likewise.
1899         (neon_vhadd<mode>): Likewise.
1900         (neon_vqadd<mode>): Likewise.
1901         (neon_vaddhn<mode>): Likewise.
1902         (neon_vmul<mode>): Likewise.
1903         (neon_vfms<VCVTF:mode>): Likewise.
1904         (neon_vmlal<mode>): Likewise.
1905         (neon_vmls<mode>): Likewise.
1906         (neon_vmlsl<mode>): Likewise.
1907         (neon_vqdmulh<mode>): Likewise.
1908         (neon_vqdmlal<mode>): Likewise.
1909         (neon_vqdmlsl<mode>): Likewise.
1910         (neon_vmull<mode>): Likewise.
1911         (neon_vqdmull<mode>): Likewise.
1912         (neon_vsub<mode>_unspec): Likewise.
1913         (neon_vsubl<mode>): Likewise.
1914         (neon_vsubw<mode>): Likewise.
1915         (neon_vqsub<mode>): Likewise.
1916         (neon_vhsub<mode>): Likewise.
1917         (neon_vsubhn<mode>): Likewise.
1918         (neon_vceq<mode>): Likewise.
1919         (neon_vcge<mode>): Likewise.
1920         (neon_vcgeu<mode>): Likewise.
1921         (neon_vcgt<mode>): Likewise.
1922         (neon_vcgtu<mode>): Likewise.
1923         (neon_vcle<mode>): Likewise.
1924         (neon_vclt<mode>): Likewise.
1925         (neon_vcage<mode>): Likewise.
1926         (neon_vcagt<mode>): Likewise.
1927         (neon_vtst<mode>): Likewise.
1928         (neon_vabd<mode>): Likewise.
1929         (neon_vabdl<mode>): Likewise.
1930         (neon_vaba<mode>): Likewise.
1931         (neon_vabal<mode>): Likewise.
1932         (neon_vmax<mode>): Likewise.
1933         (neon_vmin<mode>): Likewise.
1934         (neon_vpaddl<mode>): Likewise.
1935         (neon_vpadal<mode>): Likewise.
1936         (neon_vpmax<mode>): Likewise.
1937         (neon_vpmin<mode>): Likewise.
1938         (neon_vrecps<mode>): Likewise.
1939         (neon_vrsqrts<mode>): Likewise.
1940         (neon_vqabs<mode>): Likewise.
1941         (neon_vqneg<mode>): Likewise.
1942         (neon_vcls<mode>): Likewise.
1943         (clz<mode>2): Likewise.
1944         (popcount<mode>2): Likewise.
1945         (neon_vrecpe<mode>): Likewise.
1946         (neon_vrsqrte<mode>): Likewise.
1947         (neon_vget_lane<mode>_sext_internal): Likewise.
1948         (neon_vget_lane<mode>_zext_internal): Likewise.
1949         (neon_vdup_n<mode>): Likewise.
1950         (neon_vdup_n<mode>): Likewise.
1951         (neon_vdup_nv2di): Likewise.
1952         (neon_vdup_lane<mode>_interal): Likewise.
1953         (*neon_vswp<mode>): Likewise.
1954         (neon_vcombine<mode>): Likewise.
1955         (float<mode><V_cvtto>2): Likewise.
1956         (floatuns<mode><V_cvtto>2): Likewise.
1957         (fix_trunc<mode><V_cvtto>2): Likewise.
1958         (fixuns_trunc<mode><V_cvtto>2
1959         (neon_vcvt<mode>): Likewise.
1960         (neon_vcvt<mode>): Likewise.
1961         (neon_vcvtv4sfv4hf): Likewise.
1962         (neon_vcvtv4hfv4sf): Likewise.
1963         (neon_vcvt_n<mode>): Likewise.
1964         (neon_vcvt_n<mode>): Likewise.
1965         (neon_vmovn<mode>): Likewise.
1966         (neon_vqmovn<mode>): Likewise.
1967         (neon_vqmovun<mode>): Likewise.
1968         (neon_vmovl<mode>): Likewise.
1969         (neon_vmul_lane<mode>): Likewise.
1970         (neon_vmul_lane<mode>): Likewise.
1971         (neon_vmull_lane<mode>): Likewise.
1972         (neon_vqdmull_lane<mode>): Likewise.
1973         (neon_vqdmulh_lane<mode>): Likewise.
1974         (neon_vqdmulh_lane<mode>): Likewise.
1975         (neon_vmla_lane<mode>): Likewise.
1976         (neon_vmla_lane<mode>): Likewise.
1977         (neon_vmlal_lane<mode>): Likewise.
1978         (neon_vqdmlal_lane<mode>): Likewise.
1979         (neon_vmls_lane<mode>): Likewise.
1980         (neon_vmls_lane<mode>): Likewise.
1981         (neon_vmlsl_lane<mode>): Likewise.
1982         (neon_vqdmlsl_lane<mode>): Likewise.
1983         (neon_vext<mode>): Likewise.
1984         (neon_vrev64<mode>): Likewise.
1985         (neon_vrev32<mode>): Likewise.
1986         (neon_vrev16<mode>): Likewise.
1987         (neon_vbsl<mode>_internal): Likewise.
1988         (neon_vshl<mode>): Likewise.
1989         (neon_vqshl<mode>): Likewise.
1990         (neon_vshr_n<mode>): Likewise.
1991         (neon_vshrn_n<mode>): Likewise.
1992         (neon_vqshrn_n<mode>): Likewise.
1993         (neon_vqshrun_n<mode>): Likewise.
1994         (neon_vshl_n<mode>): Likewise.
1995         (neon_vqshl_n<mode>): Likewise.
1996         (neon_vqshlu_n<mode>): Likewise.
1997         (neon_vshll_n<mode>): Likewise.
1998         (neon_vsra_n<mode>): Likewise.
1999         (neon_vsri_n<mode>): Likewise.
2000         (neon_vsli_n<mode>): Likewise.
2001         (neon_vtbl1v8qi): Likewise.
2002         (neon_vtbl2v8qi): Likewise.
2003         (neon_vtbl3v8qi): Likewise.
2004         (neon_vtbl4v8qi): Likewise.
2005         (neon_vtbl1v16qi): Likewise.
2006         (neon_vtbl2v16qi): Likewise.
2007         (neon_vcombinev16qi): Likewise.
2008         (neon_vtbx1v8qi): Likewise.
2009         (neon_vtbx2v8qi): Likewise.
2010         (neon_vtbx3v8qi): Likewise.
2011         (neon_vtbx4v8qi): Likewise.
2012         (*neon_vtrn<mode>_insn): Likewise.
2013         (*neon_vzip<mode>_insn): Likewise.
2014         (*neon_vuzp<mode>_insn): Likewise.
2015         (neon_vld1<mode>): Likewise.
2016         (neon_vld1_lane<mode>): Likewise.
2017         (neon_vld1_lane<mode>): Likewise.
2018         (neon_vld1_dup<mode>): Likewise.
2019         (neon_vld1_dup<mode>): Likewise.
2020         (neon_vld1_dupv2di): Likewise.
2021         (neon_vst1<mode>): Likewise.
2022         (neon_vst1_lane<mode>): Likewise.
2023         (neon_vst1_lane<mode>): Likewise.
2024         (neon_vld2<mode>): Likewise.
2025         (neon_vld2<mode>): Likewise.
2026         (neon_vld2_lane<mode>): Likewise.
2027         (neon_vld2_lane<mode>): Likewise.
2028         (neon_vld2_dup<mode>): Likewise.
2029         (neon_vst2<mode>): Likewise.
2030         (neon_vst2<mode>): Likewise.
2031         (neon_vst2_lane<mode>): Likewise.
2032         (neon_vst2_lane<mode>): Likewise.
2033         (neon_vld3<mode>): Likewise.
2034         (neon_vld3qa<mode>): Likewise.
2035         (neon_vld3qb<mode>): Likewise.
2036         (neon_vld3_lane<mode>): Likewise.
2037         (neon_vld3_lane<mode>): Likewise.
2038         (neon_vld3_dup<mode>): Likewise.
2039         (neon_vst3<mode>): Likewise.
2040         (neon_vst3qa<mode>): Likewise.
2041         (neon_vst3qb<mode>): Likewise.
2042         (neon_vst3_lane<mode>): Likewise.
2043         (neon_vst3_lane<mode>): Likewise.
2044         (neon_vld4<mode>): Likewise.
2045         (neon_vld4qa<mode>): Likewise.
2046         (neon_vld4qb<mode>): Likewise.
2047         (neon_vld4_lane<mode>): Likewise.
2048         (neon_vld4_lane<mode>): Likewise.
2049         (neon_vld4_dup<mode>): Likewise.
2050         (neon_vst4<mode>): Likewise.
2051         (neon_vst4qa<mode>): Likewise.
2052         (neon_vst4qb<mode>): Likewise.
2053         (neon_vst4_lane<mode>): Likewise.
2054         (neon_vst4_lane<mode>): Likewise.
2055         (neon_vec_unpack<US>_lo_<mode>): Likewise.
2056         (neon_vec_unpack<US>_hi_<mode>): Likewise.
2057         (neon_vec_<US>mult_lo_<mode>): Likewise.
2058         (neon_vec_<US>mult_hi_<mode>): Likewise.
2059         (neon_vec_<US>shiftl_<mode>): Likewise.
2060         (neon_unpack<US>_<mode>): Likewise.
2061         (neon_vec_<US>mult_<mode>): Likewise.
2062         (vec_pack_trunc_<mode>): Likewise.
2063         (neon_vec_pack_trunc_<mode>): Likewise.
2064         (neon_vabd<mode>_2): Likewise.
2065         (neon_vabd<mode>_3): Likewise.
2067 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2069         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
2070         (load_pair): Update type attribute.
2071         (store_pair): Update type attribute.
2072         * config/aarch64/iterators.md (q): New.
2074 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
2076         * config/arm/types.md: Add new types for Neon insns.
2078 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2079             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2080             Sergey Lega  <sergey.s.lega@intel.com>
2081             Anna Tikhonova  <anna.tikhonova@intel.com>
2082             Ilya Tocar  <ilya.tocar@intel.com>
2083             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2084             Ilya Verbin  <ilya.verbin@intel.com>
2085             Kirill Yukhin  <kirill.yukhin@intel.com>
2086             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2088         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
2089         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
2090         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
2091         (rcp14<mode>): New.
2092         (srcp14<mode>): Ditto.
2093         (rsqrt14<mode>): Ditto.
2094         (rsqrt14<mode>): Ditto.
2095         (avx512f_vmscalef<mode>): Ditto.
2096         (avx512f_scalef<mode>): Ditto.
2097         (avx512f_getexp<mode>): Ditto.
2098         (avx512f_sgetexp<mode>): Ditto.
2099         (avx512f_fixupimm<mode>): Ditto.
2100         (avx512f_sfixupimm<mode>): Ditto.
2101         (avx512f_rndscale<mode>): Ditto.
2102         (*avx512er_exp2<mode>): Ditto.
2103         (*avx512er_rcp28<mode>): Ditto.
2104         (avx512er_rsqrt28<mode>): Ditto.
2105         (avx512f_getmant<mode>): Ditto.
2106         (avx512f_getmant<mode>): Ditto.
2107         (avx512f_rndscale<mode>): Fix formatting.
2109 2013-10-15  Martin Jambor  <mjambor@suse.cz>
2111         * ipa-utils.h (ipa_edge_within_scc): Declare.
2112         * ipa-cp.c (edge_within_scc): Moved...
2113         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
2115 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2116             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2117             Sergey Lega  <sergey.s.lega@intel.com>
2118             Anna Tikhonova  <anna.tikhonova@intel.com>
2119             Ilya Tocar  <ilya.tocar@intel.com>
2120             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2121             Ilya Verbin  <ilya.verbin@intel.com>
2122             Kirill Yukhin  <kirill.yukhin@intel.com>
2123             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2125         * config/i386/predicates.md (const_8_to_15_operand): New.
2126         (const_16_to_31_operand): Ditto.
2127         * config/i386/sse.md (V8FI): New.
2128         (V16FI): Ditto.
2129         (reduc_splus_v8df): Ditto.
2130         (reduc_splus_v16sf): Ditto.
2131         (avx512f_vextract<shuffletype>32x4_1): Ditto.
2132         (vec_extract_hi_<mode>): Ditto.
2133         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
2134         (vec_set_lo_<mode>): Ditto.
2135         (vec_set_hi_<mode>): Ditto.
2136         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
2137         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
2138         (avx512f_pshufd_1): Ditto.
2139         (avx512f_broadcast<mode>): Ditto.
2140         (avx512f_broadcast<mode>): Ditto.
2141         (define_split): Split vec_extract_lo into move.
2142         (ssequartermode): Ditto.
2143         (ssedoublemode): Extened with wider modes.
2144         (vec_extract_lo_<mode>): Ditto.
2146 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2147             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2148             Sergey Lega  <sergey.s.lega@intel.com>
2149             Anna Tikhonova  <anna.tikhonova@intel.com>
2150             Ilya Tocar  <ilya.tocar@intel.com>
2151             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2152             Ilya Verbin  <ilya.verbin@intel.com>
2153             Kirill Yukhin  <kirill.yukhin@intel.com>
2154             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2156         * config/i386/predicates.md (register_or_constm1_operand): New.
2157         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
2158         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
2159         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
2160         UNSPEC_SCATTER_PREFETCH
2161         (VI48_512): New.
2162         (avx512f_ucmp<mode>3): Ditto.
2163         (avx512f_vternlog<mode>): Ditto.
2164         (avx512f_align<mode>): Ditto.
2165         (<shift_insn><mode>3): Ditto.
2166         (avx512f_<rotate>v<mode>): Ditto.
2167         (avx512f_<rotate><mode>): Ditto.
2168         (avx512f_eq<mode>3): Ditto.
2169         (avx512f_eq<mode>3_1): Ditto.
2170         (avx512f_gt<mode>3): Ditto.
2171         (avx512f_testm<mode>3): Ditto.
2172         (avx512f_testnm<mode>3): Ditto.
2173         (avx512pf_gatherpf<mode>): Ditto.
2174         (*avx512pf_gatherpf<mode>_mask): Ditto.
2175         (*avx512pf_gatherpf<mode>): Ditto.
2176         (avx512pf_scatterpf<mode>): Ditto.
2177         (*avx512pf_scatterpf<mode>_mask): Ditto.
2178         (*avx512pf_scatterpf<mode>): Ditto.
2179         (avx512f_vec_dup_gpr<mode>): Ditto.
2180         (clz<mode>2): Ditto.
2181         (conflict<mode>): Ditto.
2182         (REDUC_SMINMAX_MODE): Extened with wider modes.
2183         (reduc_<code>_<mode>): Ditto.
2184         (vlshr<mode>3): Ditto.
2185         (vashl<mode>3): Ditto.
2187 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2188             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2189             Sergey Lega  <sergey.s.lega@intel.com>
2190             Anna Tikhonova  <anna.tikhonova@intel.com>
2191             Ilya Tocar  <ilya.tocar@intel.com>
2192             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2193             Ilya Verbin  <ilya.verbin@intel.com>
2194             Kirill Yukhin  <kirill.yukhin@intel.com>
2195             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2197         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
2198         UNSPEC_SCATTER.
2199         (VI48F_512): New.
2200         (avx512fmaskmode): Ditto.
2201         (bcstscalarsuff): Ditto.
2202         (avx512f_blendm<mode>): Ditto.
2203         (cmp_imm_predicate): Ditto.
2204         (avx512f_cmp<mode>3): Ditto.
2205         (avx512f_vec_dup<mode>): Ditto.
2206         (avx512f_vec_dup_mem<mode>): Ditto.
2207         (avx512f_vpermi2var<mode>3): Ditto.
2208         (avx512f_vpermt2var<mode>3): Ditto.
2209         (vec_init<mode>): Ditto.
2210         (avx512f_gathersi<mode>): Ditto.
2211         (*avx512f_gathersi<mode>): Ditto.
2212         (*avx512f_gathersi<mode>_2): Ditto.
2213         (avx512f_gatherdi<mode>): Ditto.
2214         (*avx512f_gatherdi<mode>): Ditto.
2215         (*avx512f_gatherdi<mode>_2): Ditto.
2216         (avx512f_scattersi<mode>): Ditto.
2217         (*avx512f_scattersi<mode>): Ditto.
2218         (avx512f_scatterdi<mode>): Ditto.
2219         (*avx512f_scatterdi<mode>): Ditto.
2220         (sseintprefix): Extened with wider modes.
2221         (VEC_GATHER_IDXSI): Ditto.
2222         (VEC_GATHER_IDXDI): Ditto.
2223         (VEC_GATHER_SRCDI): Ditto.
2225 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
2226             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2228         * config/arm/t-aprofile: New file.
2229         * config.gcc: Handle --with-multilib-list option.
2231 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
2233         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
2234         the input for r2.
2236 2013-10-15  Richard Biener  <rguenther@suse.de>
2238         * tree-loop-distribution.c (build_empty_rdg): Inline into
2239         single user.
2240         (rdg_flag_vertex): Inline into single user.
2241         (rdg_flag_vertex_and_dependent): Likewise.
2242         (build_rdg_partition_for_vertex): Remove processed bitmap.
2243         (rdg_build_partitions): Simplify.
2245 2013-10-15  Richard Biener  <rguenther@suse.de>
2247         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
2248         Restructure forwarding through conversions and copies to
2249         avoid performing copy-propagation the wrong way.  Adjust
2250         recursion invocations.
2251         (forward_propagate_addr_expr): Add argument stating if we
2252         are recursing from a single-use.
2253         (ssa_forward_propagate_and_combine): Adjust.
2255 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
2257         * dumpfile.h (gcc::dump_manager): New class, to hold state
2258         relating to dumpfile management.
2259         (get_dump_file_name): Remove in favor of method of dump_manager.
2260         (dump_initialized_p): Likewise.
2261         (dump_start): Likewise.
2262         (dump_finish): Likewise.
2263         (dump_switch_p): Likewise.
2264         (dump_register): Likewise.
2265         (get_dump_file_info): Likewise.
2266         * context.c (gcc::context::context): Construct the dump_manager
2267         instance.
2268         * context.h (gcc::context::get_dumps): New.
2269         (gcc::context::m_dumps): New.
2270         * coverage.c (coverage_init): Port to dump_manager API.
2271         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
2272         (extra_dump_files_in_use): Likewise.
2273         (extra_dump_files_alloced): Likewise.
2274         (gcc::dump_manager::dump_manager): New.
2275         (dump_register): Convert to...
2276         (gcc::dump_manager::dump_register): ...method, replacing
2277         function-static next_dump with m_next_dump field.
2278         (get_dump_file_info): Convert to...
2279         (gcc::dump_manager::get_dump_file_info): ...method.
2280         (get_dump_file_name): Convert to...
2281         (gcc::dump_manager::get_dump_file_name): ...method.
2282         (dump_start): Convert to...
2283         (gcc::dump_manager::dump_start): ...method.
2284         (dump_finish): Convert to...
2285         (gcc::dump_manager::dump_finish): ...method.
2286         (dump_begin): Replace body with...
2287         (gcc::dump_manager::dump_begin): ...new method.
2288         (dump_phase_enabled_p): Convert to...
2289         (gcc::dump_manager::dump_phase_enabled_p): ...method.
2290         (dump_phase_enabled_p): Convert to...
2291         (gcc::dump_manager::dump_phase_enabled_p): ...method.
2292         (dump_initialized_p):  Convert to...
2293         (gcc::dump_manager::dump_initialized_p): ...method.
2294         (dump_flag_name): Replace body with...
2295         (gcc::dump_manager::dump_flag_name): ...new method.
2296         (dump_enable_all): Convert to...
2297         (gcc::dump_manager::dump_enable_all): ...new method.
2298         (opt_info_enable_passes): Convert to...
2299         (gcc::dump_manager::opt_info_enable_passes): ...new method.
2300         (dump_switch_p_1): Convert to...
2301         (gcc::dump_manager::dump_switch_p_1): ...new method.
2302         (dump_switch_p):  Convert to...
2303         (gcc::dump_manager::dump_switch_p): ...new method.
2304         (opt_info_switch_p): Port to dump_manager API.
2305         (enable_rtl_dump_file): Likewise.
2306         * opts-global.c (handle_common_deferred_options): Port to new
2307         dump_manager API.
2308         * passes.c (pass_manager::finish_optimization_passes): Likewise.
2309         (pass_manager::register_one_dump_file): Likewise.
2310         (pass_manager::register_pass): Likewise.
2311         (pass_init_dump_file): Likewise.
2312         (pass_fini_dump_file): Likewise.
2313         * statistics.c (statistics_early_init): Likewise.
2315 2013-10-14  Richard Biener  <rguenther@suse.de>
2317         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
2318         iterative_hash_canonical_type, gimple_canonical_type_hash,
2319         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
2320         gimple_register_canonical_type, print_gimple_types_stats,
2321         free_gimple_type_tables): Move to lto/lto.c
2322         (gt-gimple.h): Do not include.
2323         * gimple.h (gimple_register_canonical_type,
2324         print_gimple_types_stats, free_gimple_type_tables): Remove.
2325         * Makefile.in (GTFILES): Remove gimple.c.
2327 2013-10-14  Travis Snoozy  <quandary@remstate.com>
2329         PR target/58716
2330         * config/msp430/msp430.c (msp430_option_override): Correct thinko
2331         scanning for msp430x targets.
2333 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
2335         PR bootstrap/58509
2336         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
2337         (registers_ok_for_ldd_peep): Move around.
2338         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
2339         * config/sparc/sparc.md (widening peepholes): Use it.
2341 2013-10-14  Richard Biener  <rguenther@suse.de>
2343         PR middle-end/58712
2344         PR middle-end/55358
2345         * gimple.c (iterative_hash_canonical_type): Make sure to
2346         record the hash into the correct hashtable slot.
2348 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
2350         PR rtl-optimization/58662
2351         * combine.c (try_combine): Take into account death nodes on I2 when
2352         splitting a PARALLEL of two independent SETs.  Fix dump message.
2354 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
2356         PR target/51244
2357         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
2358         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
2359         * config/sh/t-sh (sh_treg_combine.o): New entry.
2360         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
2361         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
2362         (register_sh_passes): New function.  Register sh_treg_combine pass.
2363         (sh_option_override): Invoke it.
2364         (sh_canonicalize_comparison): Handle op0_preserve_value.
2365         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
2366         opportunities.  Canonicalize branch condition.
2367         (nott): Allow only if pseudos can be created for non-SH2A.
2369 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
2371         PR target/58690
2372         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
2373         (ix86_expand_movmem): Replace copy_addr_to_reg with
2374         ix86_copy_addr_to_reg.
2375         (ix86_expand_setmem): Likewise.
2377 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
2379         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
2380         provided by ix86_fp_compare_mode instead of CCFPUmode.
2382 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
2384         * config/aarch64/arm_neon.h
2385         (vtbx<1,3>_<psu>8): Fix register constriants.
2387 2013-10-11  Jeff Law  <law@redhat.com>
2389         PR tree-optimization/58640
2390         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
2391         threading paths that cross over two loop entry points.
2393 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2395         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
2396         handle vector float as well.
2397         (*vsx_le_perm_load_v4si): Likewise.
2398         (*vsx_le_perm_store_v2di): Likewise.
2399         (*vsx_le_perm_store_v4si): Likewise.
2401 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2403         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
2404         directly to circumvent subtract from splat{31} workaround.
2405         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
2406         prototype.
2407         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
2408         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
2409         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
2410         (altivec_vperm_<mode>): Convert to define_insn_and_split to
2411         separate big and little endian logic.
2412         (*altivec_vperm_<mode>_internal): New define_insn.
2413         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
2414         separate big and little endian logic.
2415         (*altivec_vperm_<mode>_uns_internal): New define_insn.
2416         (vec_permv16qi): Add little endian logic.
2418 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
2420         * doc/extend.texi (returns_nonnull): Remove arguments.
2422 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2423             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2424             Sergey Lega  <sergey.s.lega@intel.com>
2425             Anna Tikhonova  <anna.tikhonova@intel.com>
2426             Ilya Tocar  <ilya.tocar@intel.com>
2427             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2428             Ilya Verbin  <ilya.verbin@intel.com>
2429             Kirill Yukhin  <kirill.yukhin@intel.com>
2430             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2432         * config/i386/sse.md (VI48F_256_512): New.
2433         (avx2_permvar<mode>): Change to ...
2434         (<avx2_avx512f>_permvar<mode>): This.
2436 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2437             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2438             Sergey Lega  <sergey.s.lega@intel.com>
2439             Anna Tikhonova  <anna.tikhonova@intel.com>
2440             Ilya Tocar  <ilya.tocar@intel.com>
2441             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2442             Ilya Verbin  <ilya.verbin@intel.com>
2443             Kirill Yukhin  <kirill.yukhin@intel.com>
2444             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2446         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
2447         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
2448         * config/i386/sse.md (VI4_AVX): New.
2449         (sf2simodelower): Ditto.
2450         (sse2_cvtps2dq): Change to ...
2451         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
2453 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2454             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2455             Sergey Lega  <sergey.s.lega@intel.com>
2456             Anna Tikhonova  <anna.tikhonova@intel.com>
2457             Ilya Tocar  <ilya.tocar@intel.com>
2458             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2459             Ilya Verbin  <ilya.verbin@intel.com>
2460             Kirill Yukhin  <kirill.yukhin@intel.com>
2461             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2463         * config/i386/sse.md (V_512): New.
2464         (VI_512): Ditto.
2465         (vcond<V_512:mode><VF_512:mode>): Ditto.
2466         (vcond<V_512:mode><VI_512:mode>): Ditto.
2467         (vcondu<V_512:mode><VI_512:mode>): Ditto.
2469 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2470             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2471             Sergey Lega  <sergey.s.lega@intel.com>
2472             Anna Tikhonova  <anna.tikhonova@intel.com>
2473             Ilya Tocar  <ilya.tocar@intel.com>
2474             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2475             Ilya Verbin  <ilya.verbin@intel.com>
2476             Kirill Yukhin  <kirill.yukhin@intel.com>
2477             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2479         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
2480         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
2481         (FMAMODE): Ditto.
2482         (fma<mode>4): Removed condition.
2483         (fms<mode>4): Ditto.
2484         (fnma<mode>4): Ditto.
2485         (fnms<mode>4): Ditto.
2486         (fma4i_fmadd_<mode>): Ditto.
2487         (*fma_fmadd_<mode>): Ditto.
2488         (*fma_fmsub_<mode>): Ditto.
2489         (*fma_fnmadd_<mode>): Ditto.
2490         (*fma_fnmsub_<mode>): Ditto.
2491         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
2492         (*fma_fmaddsub_<mode>): Ditto.
2493         (*fma_fmsubadd_<mode>): Ditto.
2494         (*fmai_fmadd_<mode>): Ditto.
2495         (*fmai_fmsub_<mode>): Ditto.
2496         (*fmai_fnmadd_<mode>): Ditto.
2497         (*fmai_fnmsub_<mode>): Ditto.
2499 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2500             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2501             Sergey Lega  <sergey.s.lega@intel.com>
2502             Anna Tikhonova  <anna.tikhonova@intel.com>
2503             Ilya Tocar  <ilya.tocar@intel.com>
2504             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2505             Ilya Verbin  <ilya.verbin@intel.com>
2506             Kirill Yukhin  <kirill.yukhin@intel.com>
2507             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2509         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
2510         (VI124_256): Changed to ...
2511         (VI124_256_48_512): This.
2512         (ssepackmode): Extended with wider modes.
2513         (<code><mode>3): Changed iterator.
2514         (*avx2_<code><mode>3): Ditto.
2515         (vec_pack_trunc_<mode>): Ditto.
2517 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2518             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2519             Sergey Lega  <sergey.s.lega@intel.com>
2520             Anna Tikhonova  <anna.tikhonova@intel.com>
2521             Ilya Tocar  <ilya.tocar@intel.com>
2522             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2523             Ilya Verbin  <ilya.verbin@intel.com>
2524             Kirill Yukhin  <kirill.yukhin@intel.com>
2525             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2527         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
2528         (VI8F_256_512): Ditto.
2529         (abs<mode>2): Changed iterator.
2530         (avx2_perm<mode>): Changed to ...
2531         (<avx2_avx512f>_perm<mode>): This.
2532         (avx2_perm<mode>_1): Changed to ...
2533         (<avx2_avx512f>_perm<mode>_1): This.
2535 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2536             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2537             Sergey Lega  <sergey.s.lega@intel.com>
2538             Anna Tikhonova  <anna.tikhonova@intel.com>
2539             Ilya Tocar  <ilya.tocar@intel.com>
2540             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2541             Ilya Verbin  <ilya.verbin@intel.com>
2542             Kirill Yukhin  <kirill.yukhin@intel.com>
2543             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2545         * config/i386/sse.md (VI48_AVX512F): New.
2546         (VI48_AVX2): Changed to ...
2547         (VI48_AVX2_48_AVX512F): This.
2548         (avx2_ashrv<mode>): Changed to ...
2549         (<avx2_avx512f>_ashrv<mode>): This.
2550         (avx2_<shift_insn>v<mode>): Changed to ...
2551         (<avx2_avx512f>_<shift_insn>v<mode>): This.
2553 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2554             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2555             Sergey Lega  <sergey.s.lega@intel.com>
2556             Anna Tikhonova  <anna.tikhonova@intel.com>
2557             Ilya Tocar  <ilya.tocar@intel.com>
2558             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2559             Ilya Verbin  <ilya.verbin@intel.com>
2560             Kirill Yukhin  <kirill.yukhin@intel.com>
2561             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2563         * config/i386/sse.md (VI4_AVX512F): New.
2564         (VI8_AVX2_AVX512F): Ditto.
2565         (mul<mode>3): Extended with wider modes.
2566         (*<sse4_1_avx2>_mul<mode>3): Ditto.
2567         (mul<mode>3): Ditto.
2568         (vec_widen_<s>mult_odd_<mode>): Ditto.
2570 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2571             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2572             Sergey Lega  <sergey.s.lega@intel.com>
2573             Anna Tikhonova  <anna.tikhonova@intel.com>
2574             Ilya Tocar  <ilya.tocar@intel.com>
2575             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2576             Ilya Verbin  <ilya.verbin@intel.com>
2577             Kirill Yukhin  <kirill.yukhin@intel.com>
2578             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2580         * config/i386/sse.md (VI2_AVX512F): New.
2581         (VI124_AVX512F): Ditto.
2582         (sseunpackmode): Extended with wider modes.
2583         (sseunpackfltmode): Ditto.
2584         (vec_unpacks_float_hi_<mode>): Ditto.
2585         (vec_unpacks_float_lo_<mode>): Ditto.
2586         (vec_unpacku_float_hi_<mode>): Ditto.
2587         (vec_unpacku_float_lo_<mode>): Ditto.
2588         (vec_unpacks_lo_<mode>): Ditto.
2589         (vec_unpacks_hi_<mode>): Ditto.
2590         (vec_unpacku_lo_<mode>): Ditto.
2591         (vec_unpacku_hi_<mode>): Ditto.
2593 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2594             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2595             Sergey Lega  <sergey.s.lega@intel.com>
2596             Anna Tikhonova  <anna.tikhonova@intel.com>
2597             Ilya Tocar  <ilya.tocar@intel.com>
2598             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2599             Ilya Verbin  <ilya.verbin@intel.com>
2600             Kirill Yukhin  <kirill.yukhin@intel.com>
2601             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2603         * config/i386/i386.md (multdiv): New.
2604         (multdiv_mnemonic): Ditto.
2605         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
2606         (<sse>_vm<multdiv_mnemonic><mode>3): This.
2607         (<sse>_vmdiv<mode>3): Removed.
2609 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2610             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2611             Sergey Lega  <sergey.s.lega@intel.com>
2612             Anna Tikhonova  <anna.tikhonova@intel.com>
2613             Ilya Tocar  <ilya.tocar@intel.com>
2614             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2615             Ilya Verbin  <ilya.verbin@intel.com>
2616             Kirill Yukhin  <kirill.yukhin@intel.com>
2617             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2619         * config/i386/sse.md (V): Extended with wider modes.
2620         (VF2): Ditto.
2621         (ssehalfvecmode): Ditto.
2622         (i128): Ditto.
2623         (ssepackfltmode): Ditto.
2624         (avx_vec_concat<mode>): Ditto.
2625         (V_256_512): New iterator.
2626         (VF2_512_256): Ditto.
2627         (si2dfmode): New attribute.
2628         (si2dfmodelower): Ditto.
2629         (sf2dfmode): Ditto.
2630         (concat_tg_mode): Ditto.
2631         (floatv4siv4df2): Changed to ...
2632         (float<si2dfmodelower><mode>2): This.
2633         (avx_cvtps2pd256): Changed to ...
2634         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
2635         (vec_pack_trunc_v4df): Changed to ...
2636         (vec_pack_trunc_<mode>): This.
2637         (avx_vpermil<mode>): Changed to ...
2638         (<sse2_avx_avx512f>_vpermil<mode>): This.
2639         (<fixsuffix>fix_truncv8dfv8si2): New.
2640         (vec_pack_sfix_trunc_v8df): Ditto.
2641         (avx512f_rndscale<mode>): Ditto.
2642         (avx512f_roundpd512): Ditto.
2643         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2645 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2646             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2647             Sergey Lega  <sergey.s.lega@intel.com>
2648             Anna Tikhonova  <anna.tikhonova@intel.com>
2649             Ilya Tocar  <ilya.tocar@intel.com>
2650             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2651             Ilya Verbin  <ilya.verbin@intel.com>
2652             Kirill Yukhin  <kirill.yukhin@intel.com>
2653             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2655         * config/i386/i386.md (any_fix): New iterator.
2656         (fixsuffix): New attribute.
2657         * config/i386/sse.md (VF1): Extened with wider modes.
2658         (VI): Ditto.
2659         (VI_AVX2): Ditto.
2660         (VI8): Ditto.
2661         (sseintvecmodelower): Ditto.
2662         (ssescalarmode): Ditto.
2663         (ssescalarnum): Ditto.
2664         (VF1_128_256): New.
2665         (ssexmmmode): Ditto.
2666         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2667         (<sse>_rcp<mode>2): Change iterator.
2668         (rsqrt<mode>2): Ditto.
2669         (<sse>_rsqrt<mode>2): Ditto.
2670         (avx2_vec_dup<mode>): Ditto.
2671         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2672         (round<mode>2_sfix): Ditto.
2673         (avx2_pbroadcast<mode>): Ditto.
2674         (*andnot<mode>3): Handle XI mode.
2675         (*<code><mode>3): Ditto.
2676         (AVXTOSSEMODE): Removed.
2677         (avx_vpermil<mode>): Changed to ...
2678         (<sse2_avx_avx512f>_vpermil<mode>): This.
2680 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2681             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2682             Sergey Lega  <sergey.s.lega@intel.com>
2683             Anna Tikhonova  <anna.tikhonova@intel.com>
2684             Ilya Tocar  <ilya.tocar@intel.com>
2685             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2686             Ilya Verbin  <ilya.verbin@intel.com>
2687             Kirill Yukhin  <kirill.yukhin@intel.com>
2688             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2690         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2691         (<sse>_comi): Ditto.
2692         (<sse>_ucomi): Ditto.
2693         (sse_cvtss2siq_2): Ditto.
2694         (sse2_cvtsd2si): Ditto.
2695         (sse2_cvtsd2siq): Ditto.
2696         (sse2_cvttsd2si): Ditto.
2697         (sse2_cvttsd2siq): Ditto.
2698         (<shift_insn><mode>3): Ditto.
2699         (sse2_cvtsi2sdq): Update constraint and prefix.
2700         (sse_cvtsi2ss): Update prefix.
2701         (sse_cvtsi2ssq): Ditto.
2703 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2705         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2706         ignore internal calls.
2708 2013-10-11  Richard Biener  <rguenther@suse.de>
2710         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2711         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2712         TREE_OVERFLOW is set.
2714 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2716         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2718         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2719         * gimple.h: Likewise.
2721         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2723         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2724         macros.
2726         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2728         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2729         * configure: Regenerate.
2731 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2733         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2734         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2735         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2736         similarly pp_right_paren.
2737         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2738         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2739         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2740         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2741         clauses.
2742         (omp_declare_simd_clauses_equal,
2743         omp_remove_redundant_declare_simd_attrs): New functions.
2744         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2745         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2746         * tree.h (OMP_LOOP_CHECK): Define.
2747         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2748         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2749         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2750         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2751         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2752         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2753         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2754         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2755         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2756         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2757         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2758         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2759         (omp_remove_redundant_declare_simd_attrs): New prototype.
2760         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2761         GIMPLE_OMP_TEAMS): New codes.
2762         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2763         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2764         fields.
2765         (target_nesting_level): New variable.
2766         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2767         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2768         for collapse > 1 static schedule unless ordered.
2769         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2770         (determine_parallel_type): Adjust get_ws_args_for caller.
2771         (install_var_field): Handle mask & 4 for double indirection.
2772         (scan_sharing_clauses): Ignore shared clause on teams construct.
2773         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2774         (create_omp_child_function): If inside target or declare target
2775         constructs, set "omp declare target" attribute on the child function.
2776         (find_combined_for): New function.
2777         (scan_omp_parallel): Handle combined loops.
2778         (scan_omp_target, scan_omp_teams): New functions.
2779         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2780         restrictions and set ctx->cancellable for cancellable constructs.
2781         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2782         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2783         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2784         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2785         (omp_clause_aligned_alignment): New function.
2786         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2787         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2788         on teams constructs.  Handle user defined reductions and new
2789         OpenMP 4.0 clauses.
2790         (lower_reduction_clauses): Don't set placeholder to address of ref
2791         if it has already the right type.
2792         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2793         (expand_parallel_call): Use the new non-_start suffixed builtins,
2794         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2795         and GOMP_parallel_end after the call.
2796         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2797         (expand_omp_for_init_counts): Handle combined loops.
2798         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2799         loops.
2800         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2801         end of cancellable loops.
2802         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2803         Likewise.  Handle collapse > 1 loops.
2804         (expand_omp_simd): Handle combined loops.
2805         (expand_omp_for): Add inner_stmt argument, adjust callers of
2806         expand_omp_for* functions, use expand_omp_for_static*chunk even
2807         for collapse > 1 unless ordered.
2808         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2809         of cancellable sections.
2810         (expand_omp_single): Remove need_barrier variable, just rely on
2811         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2812         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2813         (expand_omp_atomic_load, expand_omp_atomic_store,
2814         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2815         (expand_omp_target): New function.
2816         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2817         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2818         (build_omp_regions_1): Immediately close region for
2819         GF_OMP_TARGET_KIND_UPDATE.
2820         (maybe_add_implicit_barrier_cancel): New function.
2821         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2822         cancellation.
2823         (lower_omp_single): Likewise.  Add clobber after the barrier.
2824         (lower_omp_taskgroup): New function.
2825         (lower_omp_for): Handle combined loops.  Adjust
2826         lower_rec_input_clauses caller.  Handle cancellation.
2827         (lower_depend_clauses): New function.
2828         (lower_omp_taskreg): Lower depend clauses.  Adjust
2829         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2830         cancellation.
2831         (lower_omp_target, lower_omp_teams): New functions.
2832         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2833         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2834         and GOMP_cancellation_point calls.
2835         (lower_omp): Fold stmts inside of target region.
2836         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2837         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2838         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2839         (BT_FN_VOID_OMPFN_PTR_UINT,
2840         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2841         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2842         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2843         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2844         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2845         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2846         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2847         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2848         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2849         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2850         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2851         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2852         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2853         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2854         BUILT_IN_GOMP_PARALLEL_END.
2855         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2856         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2857         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2858         GF_OMP_FOR_KIND_DISTRIBUTE.
2859         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2860         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2861         (dump_gimple_omp_return): Print lhs if it has any.
2862         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2863         gimple_omp_atomic_seq_cst_p.
2864         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2865         and GIMPLE_OMP_TEAMS.
2866         * langhooks.c (lhd_omp_mappable_type): New function.
2867         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2868         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2869         hook.
2870         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2871         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2872         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2873         (struct gimplify_omp_ctx): Add combined_loop field.
2874         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2875         on stmts inside of target region.
2876         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2877         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2878         ORT_TARGET and ORT_TARGET_DATA.
2879         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2880         Handle GOVD_MAP.
2881         (omp_notice_threadprivate_variable): Complain about threadprivate
2882         variables in target region.
2883         (omp_notice_variable): Complain about vars with non-mappable type
2884         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2885         (omp_check_private): Ignore ORT_TARGET* regions.
2886         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2887         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2888         (find_combined_omp_for): New function.
2889         (gimplify_omp_for): Handle gimplification of combined loops.
2890         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2891         OMP_TEAMS.
2892         (gimplify_omp_target_update): New function.
2893         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2894         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2895         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2896         (gimplify_body): If fndecl has "omp declare target" attribute, add
2897         implicit ORT_TARGET context around it.
2898         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2899         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2900         * tree-nested.c (convert_nonlocal_reference_stmt,
2901         convert_local_reference_stmt, convert_gimple_call): Handle
2902         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2903         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2904         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2905         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2906         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2907         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2908         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2909         BUILT_IN_GOMP_LOOP_END_CANCEL,
2910         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2911         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2912         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2913         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2914         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2915         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2916         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2917         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2918         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2919         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2920         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2921         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2922         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2923         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2924         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2925         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2926         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2927         gimple_build_omp_teams): New functions.
2928         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2929         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2930         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2931         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2932         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2933         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2934         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2935         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2936         GF_OMP_ATOMIC_SEQ_CST): New.
2937         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2938         gimple_build_omp_teams): New prototypes.
2939         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2940         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2941         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2942         GIMPLE_OMP_SINGLE as end of range.
2943         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2944         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2945         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2946         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2947         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2948         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2949         gimple_omp_target_kind, gimple_omp_target_set_kind,
2950         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2951         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2952         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2953         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2954         gimple_omp_teams_set_clauses): New inlines.
2955         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2956         and GIMPLE_OMP_TASKGROUP.
2957         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2958         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2959         enum omp_clause_proc_bind_kind): New.
2960         (union omp_clause_subcode): Add depend_kind, map_kind and
2961         proc_bind_kind fields.
2962         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2963         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2964         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2965         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2966         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2968 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2970         * predict.c (tree_estimate_probability): Add new parameter
2971         for estimate_bb_frequencies.
2972         (estimate_bb_frequencies): Add new parameter to force estimation.
2973         (rebuild_frequencies): When max frequency in function is small,
2974         recompute counts from frequencies.
2975         * predict.h (estimate_bb_frequencies): New parameter.
2977 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2979         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2980         optimizations are disabled.
2982 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2984         * coverage.c (coverage_finish): Fix leak of da_file_name.
2986 2013-10-10  Jan Hubicka  <jh@suse.cz>
2988         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2989         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2990         Bobcat and generic.
2992 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2994         PR middle-end/58670
2995         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2996         if any labels are in FALLTHRU_BB, use a special label emitted
2997         immediately after the asm goto insn rather than label_rtx
2998         of the LABEL_DECL.
2999         (expand_asm_stmt): Adjust caller.
3000         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
3001         edge if the last insn in predecessor is a jump with single successor,
3002         but it isn't simplejump_p.
3004 2013-10-10  Richard Biener  <rguenther@suse.de>
3006         PR tree-optimization/58656
3007         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
3009 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
3011         * gimplify.c: Include expr.h and tm_p.h for targets with special
3012         va-arg padding requirements.
3014 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
3016         * tree-flow.h: Move some prototypes to gimple.h.
3017         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
3018         * gimple.h: Relocate some prototypes from tree-flow.h
3019         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
3020         Move to gimplify.c.
3021         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
3022         (build_va_arg_indirect_ref): Relocate and make static.
3023         (std_gimplify_va_arg_expr): Relocate here.
3024         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
3025         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
3027 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3029         * doc/md.texi: Document the mnemonic attribute.
3031 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3033         PR target/57377
3034         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
3035         (set_attr_alternative x ...) when searching for user defined
3036         mnemonic attribute.
3038 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
3040         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
3042 2013-10-09  Easwaran Raman <eraman@google.com>
3044         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
3045         * cfgexpand.c (defer_stack_allocation): ...use here
3046         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
3048 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
3050         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
3051         (operand_equal_for_value_replacement): New function, extracted from
3052         value_replacement and enhanced to catch more cases.
3053         (value_replacement): Use operand_equal_for_value_replacement.
3055 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
3057         * loop-doloop.c (doloop_modify, doloop_optimize): Use
3058         get_max_loop_iterations.
3060 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3062         * config/arm/aarch-common.c (arm_early_load_addr_dep):
3063         Place comment above function.
3065 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
3067         * tree-flow.h: Remove all remaining prototypes, enums and structs that
3068         are not related to tree-cfg.c.
3069         * tree-ssa-address.h: New file.  Relocate prototypes.
3070         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
3071         (addr_for_mem_ref): New.  Combine call to get_address_description and
3072         return addr_for_mem_ref.
3073         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
3074         * tree-ssa-live.h: Adjust prototypes.
3075         * passes.c: Include tree-ssa-live.h.
3076         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
3077         * graphite.c (graphite_transform_loops): Make static.
3078         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
3079         make_pass_graphite, pass_data_graphite_transforms,
3080         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
3081         * ipa-pure-const.c (warn_function_noreturn): Make static.
3082         (execute_warn_function_noreturn, gate_warn_function_noreturn,
3083         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
3084         Relocate from tree-cfg.c
3085         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
3086         static.
3087         (execute_warn_function_noreturn, gate_warn_function_noreturn,
3088         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
3089         Move to ipa-pure-const.c.
3090         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
3091         Relocate from tree-optimize.c.
3092         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
3093         make_pass_fixup_cfg): Move to tree-cfg.c.
3094         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
3095         Relocate some prototypes.
3096         * tree-data-ref.h (tree_check_data_deps) Add prototype.
3097         * tree-dump.c (dump_function_to_file): Remove prototype.
3098         Add tree-flow.h to the include file.
3099         * tree-dump.h: Remove prototype.
3100         * tree-parloops.h: New File.  Add prototypes.
3101         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
3102         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
3103         from tree-ssa-loop.c.
3104         * tree-predcom.c (run_tree_predictive_commoning,
3105         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
3106         Relocate here from tree-ssa-loop.c.
3107         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
3108         ssa_name_values.release ().
3109         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
3110         (ssa_name_values): Relocate from tree-flow.h.
3111         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
3112         * tree-ssa-loop.c (run_tree_predictive_commoning,
3113         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
3114         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
3115         make_pass_graphite, pass_data_graphite_transforms,
3116         make_pass_graphite_transforms, gate_tree_parallelize_loops,
3117         tree_parallelize_loops, pass_data_parallelize_loops,
3118         make_pass_parallelize_loops): Move to other files.
3119         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
3120         moved here.
3121         * tree.h: Remove prototypes from tree-address.c.
3123 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
3125         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
3126         (struct int_tree_map): Move to tree-hasher.h
3127         (SCALE, LABEL, PERCENT): Move to gimple.h
3128         * tree-flow-inline.h: Delete.  Move functions to other files.
3129         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
3130         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
3131         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
3132         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
3133         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
3134         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
3135         * tree-hasher.h: Don't include tree-flow.h.
3136         (struct int_tree_map): Relocate from tree-flow.h.
3137         * tree-sra.c (contains_view_convert_expr_p): Relocate from
3138         tree-flow-inline.h and make static.
3139         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
3140         tree-flow-inline.h.
3141         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
3142         tree-flow-inline.h and make static.
3143         * tree.h (is_global_var, may_be_aliased): Relocate from
3144         tree-flow-inline.h.
3145         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
3146         * value-prof.c: No longer include tree-flow-inline.h.
3147         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
3149 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
3151         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
3152         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
3153         (enum move_pos): Move to tree-ssa-loop-im.h
3154         * cfgloop.h: Move some prototypes.
3155         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
3156         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
3157         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
3158         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
3159         (loop_containing_stmt): Relocate from tree-flow-inline.h.
3160         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
3161         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
3162         (enum move_pos): Relocate here.
3163         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
3164         tree-ssa-loop.c.
3165         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
3166         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
3167         make_pass_lim): Relocate here from tree-ssa-loop.c.
3168         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
3169         tree-ssa-loop.c.
3170         (loop_edge_to_cancel, unloop_loops): Make static.
3171         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
3172         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
3173         (tree_complete_unroll, gate_tree_complete_unroll,
3174         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
3175         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
3176         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
3177         * tree-ssa-loop-ivopts.c: Remove local prototypes.
3178         (stmt_invariant_in_loop_p): Remove unused function.
3179         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
3180         * tree-ssa-loop-manip.h: New file.  Add prototypes.
3181         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
3182         (gcov_type_to_double_int): Move to cfgloop.h.
3183         (double_int_cmp, bound_index,
3184         estimate_numbers_of_iterations_loop): Make static.
3185         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
3186         (max_loop_iterations): Factor out get_max_loop_iterations.
3187         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
3188         cfgloop.c.
3189         * tree-ssa-loop-niter.h: New file.  Add prototypes.
3190         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
3191         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
3192         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
3193         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
3194         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
3195         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
3196         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
3197         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
3198         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
3199         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
3200         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
3201         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
3202         pass_data_complete_unroll, make_pass_complete_unroll,
3203         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
3204         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
3205         tree-ssa-loop-ivcanon.c.
3206         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
3207         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
3208         tree-ssa-loop-prefetch.c.
3209         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
3210         tree-ssa-loop-im.c.
3211         (get_lsm_tmp_name): Relocate and add suffix parameter.
3212         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
3213         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
3214         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
3215         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
3216         (get_estimated_loop_iterations): Factor out accessor from
3217         estimated_loop_iterations in tree-ssa-loop-niter.c.
3218         (get_max_loop_iterations): Factor out accessor from
3219         _max_loop_iterations in tree-ssa-niter.c.
3220         * loop-unroll.c (decide_unroll_constant_iterations,
3221         decide_unroll_runtime_iterations, decide_peel_simple,
3222         decide_unroll_stupid): Use new get_* accessors.
3224 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
3226         PR tree-optimization/20318
3227         * doc/extend.texi (returns_nonnull): New function attribute.
3228         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
3229         attribute.
3230         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
3231         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
3233 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
3235         PR middle-end/58570
3236         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
3237         false if both components are bitfields.
3239 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3241         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
3242         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
3243         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
3244         * config/aarch64/aarch64.h
3245         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
3246         * config/aarch64/aarch64-simd-builtins.def
3247         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
3249 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3251         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
3253 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3255         * config/aarch64/arm_neon.h (vdiv_f64): Added.
3257 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
3259         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
3260         (vneg_f64): New intrinsic.
3261         (vneg_s8): Asm replaced with C.
3262         (vneg_s16): Likewise.
3263         (vneg_s32): Likewise.
3264         (vneg_s64): New intrinsic.
3265         (vnegq_f32): Asm replaced with C.
3266         (vnegq_f64): Likewise.
3267         (vnegq_s8): Likewise.
3268         (vnegq_s16): Likewise.
3269         (vnegq_s32): Likewise.
3270         (vnegq_s64): Likewise.
3272 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
3274         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
3276 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3278         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
3279         packed stack special handling.
3280         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
3281         back to fixed stack slots for FPRs saved due to stdarg.
3283 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3285         * config/s390/s390.c (s390_frame_info): Restructure function.
3287 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3289         * config/s390/s390.c (struct s390_frame_layout): New field
3290         gpr_save_slots.
3291         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
3292         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
3293         regs_ever_clobbered to char*.
3294         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
3295         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
3296         (s390_register_info): Enable FPR save slots.  Move/Copy some
3297         functionality into ...
3298         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
3299         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
3300         function.
3301         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
3302         not imply a stack frame.
3303         (s390_init_frame_layout): Remove variable clobbered_regs.
3304         (s390_update_register_info): Remove function.
3305         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
3306         cannot be used for register renaming.
3307         (s390_hard_regno_scratch_ok): New function.
3308         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
3309         (s390_initial_elimination_offset): Change offset calculation of
3310         the return address pointer.
3311         (save_gprs): Deal with only r6 being saved from the call-saved regs.
3312         (restore_gprs): Set frame related flag.
3313         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
3314         (s390_emit_prologue): Call s390_register_info instead of
3315         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
3316         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
3317         (s390_optimize_prologue): Call s390_optimize_register_info.
3318         Try to remove also FPR slot save/restore INSNs.  Remove frame
3319         related flags from restore INSNs.
3321 2013-10-08  DJ Delorie  <dj@redhat.com>
3323         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
3324         (movhi): Likewise.
3326         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
3327         avoid conflict with the MI use of %c.
3328         * config/rl78/rl78-real.md: change %c to %C throughout.
3329         * config/rl78/rl78-virt.md: Likewise.
3331 2013-10-08  Jan Hubicka  <jh@suse.cz>
3333         * config/i386/i386.c (ix86_option_override_internal): Switch
3334         to SSE math for -ffast-math when target ISA supports SSE2.
3336 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
3338         * tree-flow.h: Remove some prototypes.
3339         * tree.h: Remove some protypes, add a couple.
3340         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
3341         using_eh_for_cleanups_p): Add interface routines for front ends.
3342         * tree-eh.h: New file.  Add protoptyes.
3343         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
3344         (add_stmt_to_eh_lp_fn): Make static.
3345         (lower_try_finally): Use new using_eh_for_cleanups_p.
3346         * emit-rtl.c: Include tree-eh.h.
3347         * gimple.h: Include tree-eh.h.
3349 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
3351         PR tree-optimization/58480
3352         * tree-vrp.c (infer_nonnull_range): New function.
3353         (infer_value_range): Call infer_nonnull_range.
3355 2013-10-08  Dehao Chen  <dehao@google.com>
3357         PR tree-optimization/58619
3358         * tree-inline.c (copy_phis_for_bb): Combine location data
3359         only if non-null.
3361 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3363         PR target/58423
3364         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
3365         RTX_FRAME_RELATED_P on INSN.
3367 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3369         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
3370         (altivec_expand_vec_perm_const): Call it.
3372 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3374         * config/rs6000/vector.md (mov<mode>): Emit permuted move
3375         sequences for LE VSX loads and stores at expand time.
3376         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
3377         prototype.
3378         * config/rs6000/rs6000.c (rs6000_const_vec): New.
3379         (rs6000_gen_le_vsx_permute): New.
3380         (rs6000_gen_le_vsx_load): New.
3381         (rs6000_gen_le_vsx_store): New.
3382         (rs6000_gen_le_vsx_move): New.
3383         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
3384         (*vsx_le_perm_load_v4si): New.
3385         (*vsx_le_perm_load_v8hi): New.
3386         (*vsx_le_perm_load_v16qi): New.
3387         (*vsx_le_perm_store_v2di): New.
3388         (*vsx_le_perm_store_v4si): New.
3389         (*vsx_le_perm_store_v8hi): New.
3390         (*vsx_le_perm_store_v16qi): New.
3391         (*vsx_xxpermdi2_le_<mode>): New.
3392         (*vsx_xxpermdi4_le_<mode>): New.
3393         (*vsx_xxpermdi8_le_V8HI): New.
3394         (*vsx_xxpermdi16_le_V16QI): New.
3395         (*vsx_lxvd2x2_le_<mode>): New.
3396         (*vsx_lxvd2x4_le_<mode>): New.
3397         (*vsx_lxvd2x8_le_V8HI): New.
3398         (*vsx_lxvd2x16_le_V16QI): New.
3399         (*vsx_stxvd2x2_le_<mode>): New.
3400         (*vsx_stxvd2x4_le_<mode>): New.
3401         (*vsx_stxvd2x8_le_V8HI): New.
3402         (*vsx_stxvd2x16_le_V16QI): New.
3404 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
3406         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
3408 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3410         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
3411         loop to work also for 31bit ABI.
3412         Save the stack pointer for frame_size > 0.
3414 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3416         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
3417         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
3418         constraint letters from expanders.
3419         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
3420         retry count to general_operand.
3421         ("tabort"): Give operand 0 a mode.
3422         ("tabort_1"): Add mode and constraint letter for operand 0.
3423         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
3425 2013-10-04  Jeff Law  <law@redhat.com>
3427         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
3429         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
3430         out of ...
3431         (thread_across_edge): Here.  Call it.
3433 2013-10-04  Cary Coutant  <ccoutant@google.com>
3435         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
3436         discarding a location list expression (or a piece of one).
3438 2013-10-03  Jan Hubicka  <jh@suse.cz>
3440         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
3441         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
3442         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
3444 2013-10-03  Jan Hubicka  <jh@suse.cz>
3446         * config/i386/i386.c (ix86_option_override_internal): Do not enable
3447         accumulate-outgoing-args when producing unwind info.
3449 2013-10-03  Wei Mi  <wmi@google.com>
3451         * lra-constraints.c (insert_move_for_subreg): New function
3452         extracted from simplify_operand_subreg.
3453         (simplify_operand_subreg): Add reload for paradoxical subreg.
3455 2013-10-03  Rong Xu  <xur@google.com>
3457         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
3458         the candidate of builtin_expect such that we should fix the
3459         size/time estimation.
3460         (estimate_function_body_sizes): Do the acutally size/time fix-up
3461         for builtin_expect.
3463 2013-10-03  Rong Xu  <xur@google.com>
3465         * predict.c (tree_predict_by_opcode): Get the probability
3466         for builtin_expect from param builtin_expect_probability.
3467         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
3468         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
3469         * doc/invoke.texi: Add documentation for builtin-expect-probability.
3471 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
3473         PR c++/19476
3474         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
3475         * calls.c (alloca_call_p): Use get_callee_fndecl.
3476         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
3477         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
3478         Likewise.
3479         (vrp_visit_stmt): Remove duplicated code.
3481 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
3483         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
3484         ceildf2, btruncdf2, instead of vsx_* name.
3486         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
3487         iterators to only do V2DF and V4SF here.  Move the DF code to
3488         rs6000.md where it is combined with SF mode.  Replace <VSv> with
3489         just 'v' since only vector operations are handled with these insns
3490         after moving the DF support to rs6000.md.
3491         (vsx_sub<mode>3): Likewise.
3492         (vsx_mul<mode>3): Likewise.
3493         (vsx_div<mode>3): Likewise.
3494         (vsx_fre<mode>2): Likewise.
3495         (vsx_neg<mode>2): Likewise.
3496         (vsx_abs<mode>2): Likewise.
3497         (vsx_nabs<mode>2): Likewise.
3498         (vsx_smax<mode>3): Likewise.
3499         (vsx_smin<mode>3): Likewise.
3500         (vsx_sqrt<mode>2): Likewise.
3501         (vsx_rsqrte<mode>2): Likewise.
3502         (vsx_fms<mode>4): Likewise.
3503         (vsx_nfma<mode>4): Likewise.
3504         (vsx_copysign<mode>3): Likewise.
3505         (vsx_btrunc<mode>2): Likewise.
3506         (vsx_floor<mode>2): Likewise.
3507         (vsx_ceil<mode>2): Likewise.
3508         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
3509         (vsx_sminsf3): Likewise.
3510         (vsx_fmadf4): Likewise.
3511         (vsx_fmsdf4): Likewise.
3512         (vsx_nfmadf4): Likewise.
3513         (vsx_nfmsdf4): Likewise.
3514         (vsx_cmpdf_internal1): Likewise.
3516         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
3517         simpler to select whether a target has SPE or traditional floating
3518         point support in iterators.
3519         (TARGET_DF_SPE): Likewise.
3520         (TARGET_SF_FPR): Likewise.
3521         (TARGET_DF_FPR): Likewise.
3522         (TARGET_SF_INSN): Macros to say whether floating point support
3523         exists for a given operation for expanders.
3524         (TARGET_DF_INSN): Likewise.
3526         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
3527         combining of SF/DF mode operations, using both traditional and VSX
3528         registers.
3529         (Fvsx): Likewise.
3530         (Ff): Likewise.
3531         (Fv): Likewise.
3532         (Fs): Likewise.
3533         (Ffre): Likewise.
3534         (FFRE): Likewise.
3535         (abs<mode>2): Combine SF/DF modes using traditional floating point
3536         instructions.  Add support for using the upper DF registers with
3537         VSX support, and SF registers with power8-vector support.  Update
3538         expanders for operations supported by both the SPE and traditional
3539         floating point units.
3540         (abs<mode>2_fpr): Likewise.
3541         (nabs<mode>2): Likewise.
3542         (nabs<mode>2_fpr): Likewise.
3543         (neg<mode>2): Likewise.
3544         (neg<mode>2_fpr): Likewise.
3545         (add<mode>3): Likewise.
3546         (add<mode>3_fpr): Likewise.
3547         (sub<mode>3): Likewise.
3548         (sub<mode>3_fpr): Likewise.
3549         (mul<mode>3): Likewise.
3550         (mul<mode>3_fpr): Likewise.
3551         (div<mode>3): Likewise.
3552         (div<mode>3_fpr): Likewise.
3553         (sqrt<mode>3): Likewise.
3554         (sqrt<mode>3_fpr): Likewise.
3555         (fre<Fs>): Likewise.
3556         (rsqrt<mode>2): Likewise.
3557         (cmp<mode>_fpr): Likewise.
3558         (smax<mode>3): Likewise.
3559         (smin<mode>3): Likewise.
3560         (smax<mode>3_vsx): Likewise.
3561         (smin<mode>3_vsx): Likewise.
3562         (negsf2): Delete SF operations that are merged with DF.
3563         (abssf2): Likewise.
3564         (addsf3): Likewise.
3565         (subsf3): Likewise.
3566         (mulsf3): Likewise.
3567         (divsf3): Likewise.
3568         (fres): Likewise.
3569         (fmasf4_fpr): Likewise.
3570         (fmssf4_fpr): Likewise.
3571         (nfmasf4_fpr): Likewise.
3572         (nfmssf4_fpr): Likewise.
3573         (sqrtsf2): Likewise.
3574         (rsqrtsf_internal1): Likewise.
3575         (smaxsf3): Likewise.
3576         (sminsf3): Likewise.
3577         (cmpsf_internal1): Likewise.
3578         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
3579         (negdf2): Delete DF operations that are merged with SF.
3580         (absdf2): Likewise.
3581         (nabsdf2): Likewise.
3582         (adddf3): Likewise.
3583         (subdf3): Likewise.
3584         (muldf3): Likewise.
3585         (divdf3): Likewise.
3586         (fred): Likewise.
3587         (rsqrtdf_internal1): Likewise.
3588         (fmadf4_fpr): Likewise.
3589         (fmsdf4_fpr): Likewise.
3590         (nfmadf4_fpr): Likewise.
3591         (nfmsdf4_fpr): Likewise.
3592         (sqrtdf2): Likewise.
3593         (smaxdf3): Likewise.
3594         (smindf3): Likewise.
3595         (cmpdf_internal1): Likewise.
3596         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
3597         (btrunc<mode>2): Delete separate expander, and combine with the
3598         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
3599         (btrunc<mode>2_fpr): Likewise.
3600         (ceil<mode>2): Likewise.
3601         (ceil<mode>2_fpr): Likewise.
3602         (floor<mode>2): Likewise.
3603         (floor<mode>2_fpr): Likewise.
3604         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
3605         Add support for using the upper registers with VSX and
3606         power8-vector.  Move insns to be closer to the define_expands. On
3607         VSX systems, prefer the traditional form of FMA over the VSX
3608         version, since the traditional form allows the target not to
3609         overlap with the inputs.
3610         (fms<mode>4_fpr): Likewise.
3611         (nfma<mode>4_fpr): Likewise.
3612         (nfms<mode>4_fpr): Likewise.
3614 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3615             Richard Earnshaw  <richard.earnshaw@arm.com>
3617         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
3618         (struct mult_cost_table): Likewise.
3619         (struct mem_cost_table): Likewise.
3620         (struct fp_cost_table): Likewise.
3621         (struct vector_cost_table): Likewise.
3622         (cpu_cost_table): Likewise.
3623         * config/arm/arm.opt (mold-rts-costs): New option.
3624         (mnew-generic-costs): Likewise.
3625         * config/arm/arm.c (generic_extra_costs): New table.
3626         (cortexa15_extra_costs): Likewise.
3627         (arm_slowmul_tune): Use NULL as new costs.
3628         (arm_fastmul_tune): Likewise.
3629         (arm_strongarm_tune): Likewise.
3630         (arm_xscale_tune): Likewise.
3631         (arm_9e_tune): Likewise.
3632         (arm_v6t2_tune): Likewise.
3633         (arm_cortex_a5_tune): Likewise.
3634         (arm_cortex_a9_tune): Likewise.
3635         (arm_v6m_tune): Likewise.
3636         (arm_fa726te_tune): Likewise.
3637         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3638         (arm_cortex_tune): Use generict_extra_costs.
3639         (shifter_op_p): New function.
3640         (arm_unspec_cost): Likewise.
3641         (LIBCALL_COST): Define.
3642         (arm_new_rtx_costs): New function.
3643         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3644         table is available. Use old costs otherwise unless mnew-generic-costs
3645         is specified.
3646         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3647         (cpu_cost_table): Declare.
3649 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3651         PR target/58460
3652         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3653         (*subs_mul_imm_<mode>)
3654         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3655         (*sub_<shift>_<mode>)
3656         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3657         Remove k constraint.
3659 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3661         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3662         code.
3663         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3665 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3667         * predict.c (probably_never_executed): New function.
3668         (probably_never_executed_bb_p): Invoke probably_never_executed.
3669         (probably_never_executed_edge_p): Ditto.
3670         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3671         Treat profile insanities conservatively.
3673 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3675         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3677 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3679         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3680         use it.  Use smaller increase for scratch.  Don't increase reject
3681         for early clobber scratch.
3682         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3683         setting eliminated regs except setting fp from hfp.
3684         (lra_eliminate): Check lra_insn_recog_data on NULL.
3686 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3688         PR target/58587
3689         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3690         setting -mvsx-timode by default until the underlying problem is fixed.
3691         (RS6000_CPU, power7 defaults): Likewise.
3693 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3695         * config/x-linux (host-linux.o): Remove header dependencies.
3696         Use $(COMPILE) and $(POSTCOMPILE).
3697         * config/t-linux-android (linux-android.o): Ditto.
3699 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3701         * Makefile.in (expmed.o-warn): Remove.
3703 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3705         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3706         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3708 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3710         * dojump.c (do_jump_1): Divide probability between
3711         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3713 2013-10-02  Tom Tromey  <tromey@redhat.com>
3715         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3717 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3719         * tree-flow.h: Remove some prototypes.
3720         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3721         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3722         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3723         mark_virtual_phi_result_for_renaming): Relocate here.
3724         * tree-into-ssa.h: Add prototypes.
3725         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3726         single_pred_before_succ_order.
3727         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3728         (nonfreeing_call_p) Move to gimple.c.
3729         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3730         tree-ssa-phiopt.c.
3731         * basic-block.h (single_pred_before_succ_order): Add prototype.
3732         * gimple.c (nonfreeing_call_p): Relocate here.
3733         * gimple.h: Add prototype.
3734         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3735         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3736         * tree-ssa.h: Include tree-ssa-dom.h.
3738 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3740         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3741         Use $(COMPILE) and $(POSTCOMPILE).
3743         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3745 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3747         * tree-flow.h: Remove some prototypes.
3748         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3749         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3750         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3751         propagate_tree_value*): Move from here to...
3752         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3753         propagate_tree_value*): Relocate here.
3754         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3755         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3756         * gimple-fold.c: Remove gimple-fold.h from include list.
3757         * tree-vrp.c: Remove gimple-fold.h from include list.
3758         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3759         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3760         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3761         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3762         * sese.c: Add tree-ssa-propagate.h to include list.
3764 2013-10-02  Richard Biener  <rguenther@suse.de>
3766         * tree-loop-distribution.c: Include tree-vectorizer.h for
3767         find_loop_location.
3768         (enum partition_kind): Remove PKIND_REDUCTION.
3769         (struct partition_s): Remove has_writes member, add reduction_p member.
3770         (partition_alloc): Adjust.
3771         (partition_builtin_p): Likewise.
3772         (partition_has_writes): Remove.
3773         (partition_reduction_p): New function.
3774         (partition_merge_into): Likewise.
3775         (generate_code_for_partition): Commonize builtin partition
3776         handling tail.
3777         (rdg_cannot_recompute_vertex_p): Remove.
3778         (already_processed_vertex_p): Likewise.
3779         (rdg_flag_vertex): Do not set has_writes.
3780         (classify_partition): Adjust.
3781         (rdg_build_partitions): Do not set has_writes, treat all
3782         partitions as useful.
3783         (distribute_loop): Record number of library calls generated.  Adjust.
3784         (tree_loop_distribution): Report number of loops and library
3785         calls generated as opt-info.
3787 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3789         * tree-flow.h: Include new .h files.  Move prototypes.
3790         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3791         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3792         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3793         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3794         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3795         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3796         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3797         * tree.h (get_ref_base_and_extent): Move prototype out.
3798         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3799         tree-dfa.h.
3800         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3801         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3802         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3803         * tree-scalar-evolution.c: Include tree.h.
3804         * sese.c: Include tree.h.
3805         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3806         * dwarf2out.c: Include tree-dfa.h.
3807         * tree-chrec.c: Include tree.h.
3808         * tree-data-ref.c: Include tree.h.
3810 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3812         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3813         Fix whitespace.
3815 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3817         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3818         Use $(COMPILE) and $(POSTCOMPILE).
3819         (sol2-cxx.o): Likewise.
3820         (sol2-stubs.o): Likewise.
3821         (sol2.o): Likewise.
3822         * config/x-solaris (host-solaris.o): Likewise.
3824         * config/sparc/t-sparc (sparc.o): Remove.
3825         (sparc-c.o): Remove header dependencies.
3826         Use $(COMPILE) and $(POSTCOMPILE).
3827         * config/sparc/x-sparc: Likewise.
3829 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3831         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3832         * config/arc/arc700.md: Likewise.
3833         * config/arc/arc-modes.def: Likewise.
3834         * config/arc/arc-simd.h: Likewise.
3835         * config/arc/t-arc-uClibc: Likewise.
3836         * config/arc/t-arc-newlib: Likewise.
3838 2013-10-02  Renlin Li  <renlin.li@arm.com>
3840         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3841         plus_constant.
3842         (aarch64_expand_epilogue): Likewise.
3844 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3845             Yufeng Zhang  <yufeng.zhang@arm.com>
3847         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3848         declaration.
3849         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3850         'base_in' represent a conversion and legal_cast_p_1 holds; set
3851         'base_in' with the returned value from get_unwidened.
3853 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3855         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3856         plus_constant is not used.
3858 2013-10-01  Wei Mi  <wmi@google.com>
3860         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3861         * config/i386/i386.md: Add define_peephole2 to
3862         break partial reg stall for cvtss2sd/cvtsd2ss.
3864 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3866         * config/arc/arc.c (pass_arc_ifcvt::clone):
3867         Update for ctxt_ -> m_ctxt change.
3869 2013-10-01  Jeff Law  <law@redhat.com>
3871         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3872         outgoing_edge and intermediate_edge fields.  Instead store the path.
3873         (redirection_data::hash): Hash on the last edge's destination index.
3874         (redirection_data::equal): Check the entire thread path.
3875         (lookup_redirectio_data): Corresponding changes.
3876         (create_edge_and_update_destination_phis): Likewise.
3877         (thread_single_edge): Likewise.
3879 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3880             Diego Novillo <dnovillo@google.com>
3882         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3883         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3884         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3885         versions of these patterns.
3887         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3888         (__builtin_arc_aligned): Likewise.
3890         * config/arc/arc.md: Expand adc_0 comment stating the intended
3891         purpose and why it isn't ready.
3892         Replace commented out call_value_via_label_mixed with a
3893         plain comment about bl_s.
3895         * config/arc/arc.c (stdio.h): Don't include directly.
3896         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3897         Remove if (1) condition.
3898         (arc_encode_section_info): Fix comment.
3900 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3902         * config/arc/arc.c (arc_conditional_register_usage):
3903         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3904         Also set reg_alloc_order for DMA config regs.
3906 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3907             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3909         * doc/install.texi (--with-cpu): Mention ARC.
3910         (arc-*-elf32): New paragraph.
3911         (arc-linux-uclibc): Likewise.
3912         * doc/md.texi (Machine Constraints): Add ARC part.
3913         * doc/invoke.texi: (menu): Add ARC Options.
3914         (Machine Dependent Options) <ARC Options>: Add synopsis.
3915         (node ARC Options): Add.
3916         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3917         (ARC Built-in Functions): New section defining
3918         generic ARC built-in functions.
3919         (ARC SIMD Built-in Functions): New section defining SIMD specific
3920         built-in functions.
3921         (Declaring Attributes of Functions): Extended
3922         description of short_call and long_call attributes for ARC and
3923         added index entries.
3925 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3926             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3927             Joern Rennecke  <joern.rennecke@embecosm.com>
3928             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3929             Brendan Kehoe  <brendan@zen.org>
3930             Michael Eager  <eager@eagercon.com>
3931             Simon Cook  <simon.cook@embecosm.com>
3932             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3934         * config/arc, common/config/arc: New directories.
3936 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3937             Brendan Kehoe  <brendan@zen.org>
3938             Simon Cook  <simon.cook@embecosm.com>
3940         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3942 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3944         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3945         * tree-ssa-coalesce.h: New. Move prototype to here.
3946         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3947         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3948         (gimple_can_coalesce_p): Move to...
3949         * gimple.c (gimple_can_coalesce_p): Here.
3951 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3953         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3954         (dump_decl_set): Move to gimple.c.
3955         * gimple.h: Don't include tree-ssa-operands.h.
3956         (dump_decl_set): Add prototype.
3957         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3958         Move to gimple-ssa.h.
3959         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3960         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3961         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3962         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3963         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3964         than PHI_ARG_DEF.
3965         (dump_decl_set): Relocate here.
3966         * gimple-ssa.h: New file.
3967         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3968         Relocate from gimple.h.
3969         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3970         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3971         swap_tree_operands and remove non-ssa path.
3972         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3973         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3974         swap_ssa_operands.
3975         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3976         vect_is_simple_reduction_1): Use swap_ssa_operands.
3977         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3978         (enum need_phi_state): Move to tree-into-ssa.c.
3979         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3980         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3981         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3982         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3983         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3984         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3985         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3986         num_imm_uses): Move to ssa-iterators.h.
3987         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3988         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3989         tree-phinodes.h.
3990         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3991         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3992         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3993         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3994         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3995         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3996         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3997         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3998         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3999         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
4000         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
4001         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
4002         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
4003         (set_phi_nodes): Move to tree-phinodes.h.
4004         * tree-ssa-operands.h (enum ssa_op_iter_type,
4005         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
4006         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
4007         (dump_decl_set): Remove prototype.
4008         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
4009         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
4010         (set_phi_nodes): Relocate from tree-flow-inline.h.
4011         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
4012         tree-flow-inline.h
4013         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
4014         include list.  Temporarily add gimple.h to include list.
4015         * ssa-iterators.h: New file.
4016         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
4017         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
4018         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
4019         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
4020         Relocate from tree-ssa-operands.h.
4021         (delink_imm_use, link_imm_use_to_list, link_imm_use,
4022         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
4023         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
4024         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
4025         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
4026         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
4027         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
4028         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
4029         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
4030         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
4031         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
4032         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
4033         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
4034         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
4035         Relocate from tree-flow-inline.h.
4036         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
4038 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
4040         * aarch64-simd.md
4041         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
4042         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
4043         Insert '\t' to output template.
4044         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
4045         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
4046         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
4047         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
4049 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
4051         * doc/install.texi (Host/target specific installation notes for GCC):
4052         Put @anchor before @heading.
4053         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
4054         Use @email for email addresses.
4056 2013-10-01  Jeff Law  <law@redhat.com>
4058         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
4059         a vec.  Only delete the path if we create one without successfully
4060         registering a jump thread.
4061         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
4062         as a pointer.
4063         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
4064         (paths): New vector of jump threading paths.
4065         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
4066         (THREAD_PATH): New accessor macro for the entire thread path.
4067         (lookup_redirection_data): Get intermediate and final outgoing edge
4068         from the thread path.
4069         (create_edge_and_update_destination_phis): Copy the threading path.
4070         (ssa_fix_duplicate_block_edges): Get edges and block types from the
4071         jump threading path.
4072         (ssa_redirect_edges): Get edges and block types from the jump threading
4073         path.  Free the path vector.
4074         (thread_block): Get edges from the jump threading path.  Look at the
4075         entire path to see if we thread to a loop exit.  If we cancel a jump
4076         thread request, then free the path vector.
4077         (thread_single_edge): Get edges and block types from the jump threading
4078         path.  Free the path vector.
4079         (thread_through_loop_header): Get edges and block types from the jump
4080         threading path.  Free the path vector.
4081         (mark_threaded_blocks): Iterate over the vector of paths and store
4082         the path on the appropriate edge.  Get edges and block types from the
4083         jump threading path.
4084         (mark_threaded_blocks): Get edges and block types from the jump
4085         threading path.  Free the path vector.
4086         (thread_through_all_blocks): Use the vector of paths rather than
4087         a vector of 3-edge sets.
4088         (register_jump_thread): Accept pointer to a path vector rather
4089         than the path vector itself.  Store the path vector for later use.
4090         Simplify.
4092 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
4093             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4095         PR target/58574
4096         * config/s390/s390.c (s390_split_branches): Modify check for table
4097         jump insns.
4098         (s390_chunkify_start): Rearrange table jump insn check in order to
4099         deal with compare and branch insns correctly.
4101 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
4103         PR target/58578
4104         Revert
4105         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
4106         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
4107         define_insn_and_split.
4108         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
4109         (shiftsi3_compare): New pattern.
4110         (rrx): New pattern.
4111         * config/arm/unspecs.md (UNSPEC_RRX): New.
4113 2013-10-01  Alan Modra  <amodra@gmail.com>
4115         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
4116         casing inout operands.
4118 2013-10-01  Richard Biener  <rguenther@suse.de>
4120         PR tree-optimization/58553
4121         * tree-loop-distribution.c (struct partition_s): Add niter member.
4122         (classify_partition): Populate niter member for the partition
4123         and properly identify whether the relevant store happens before
4124         or after the loop exit.
4125         (generate_memset_builtin): Use niter member from the partition.
4126         (generate_memcpy_builtin): Likewise.
4128 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
4130         * vec.h (vec_prefix, vec): Prefix member names with "m_".
4131         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
4133 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
4135         * basic-block.h (edge_list): Prefix member names with "m_".
4136         * context.h (context): Likewise.
4137         * domwalk.h (dom_walker): Likewise.
4138         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
4139         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
4140         * hash-table.h (hash_table): Likewise.
4141         * machmode.h (bit_field_mode_iterator): Likewise.
4142         * pass_manager.h (pass_list): Likewise.
4143         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
4144         * tree-pass.h (pass_data): Likewise.
4145         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
4146         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
4147         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
4148         * asan.c (pass_data_asan): Update accordingly.
4149         * cfganal.c (control_dependences::find_control_dependence): Likewise.
4150         (control_dependences::control_dependences): Likewise.
4151         (control_dependences::~control_dependences): Likewise.
4152         (control_dependences::~control_dependences): Likewise.
4153         (control_dependences::get_edges_dependent_on): Likewise.
4154         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
4155         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
4156         * context.c (gcc::context::context): Likewise.
4157         * cprop.c (pass_rtl_cprop::clone): Likewise.
4158         * domwalk.c (dom_walker::walk): Likewise.
4159         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
4160         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
4161         * mode-switching.c (pass_mode_switching::clone): Likewise.
4162         * passes.c (opt_pass::opt_pass): Likewise.
4163         (pass_manager::pass_manager): Likewise.
4164         * predict.c (pass_strip_predict_hints::clone): Likewise.
4165         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
4166         (pass_split_all_insns::clone): Likewise.
4167         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4168         Likewise.
4169         (bit_field_mode_iterator::next_mode): Likewise.
4170         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
4171         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
4172         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
4173         * tree-complex.c (pass_lower_complex::clone): Likewise.
4174         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
4175         * tree-object-size.c (pass_object_sizes::clone): Likewise.
4176         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
4177         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
4178         (pass_fold_builtins::clone): Likewise.
4179         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
4180         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
4181         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
4182         (pass_cd_dce::clone): Likewise.
4183         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
4184         (pass_phi_only_cprop::clone): Likewise.
4185         * tree-ssa-dse.c (pass_dse::clone): Likewise.
4186         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
4187         * tree-ssa-loop.c (pass_lim::clone): Likewise.
4188         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
4189         * tree-ssa-pre.c (pass_fre::clone): Likewise.
4190         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
4191         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
4192         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
4193         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
4194         * tree-vrp.c (pass_vrp::clone): Likewise.
4195         * tsan.c (pass_tsan::clone): Likewise.
4197 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
4199         PR middle-end/58564
4200         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
4201         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
4203         PR middle-end/58564
4204         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
4205         optimization, punt if sign_bit_p looked through any zero extension.
4207 2013-09-30  Teresa Johnson  <tejohnson@google.com>
4209         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
4210         Update redirected out edge count in joiner case.
4211         (ssa_redirect_edges): Common the joiner and non-joiner cases
4212         so that joiner case gets profile updates.
4214 2013-09-30  Richard Biener  <rguenther@suse.de>
4216         PR tree-optimization/58554
4217         * tree-loop-distribution.c (classify_partition): Require
4218         unconditionally executed stores for memcpy and memset recognition.
4219         (tree_loop_distribution): Calculate dominance info.
4221 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
4223         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
4224         (NO_PROFILE_COUNTERS): Likewise.
4225         (PROFILE_HOOK): Likewise.
4226         (FUNCTION_PROFILER): Likewise.
4227         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
4229 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
4231         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
4232         calls and defines in TARGET_MACHO conditional.
4233         (load_macho_picbase_di): Likewise.
4234         (reload_macho_picbase): Likewise.
4235         (reload_macho_picbase_si): Likewise.
4236         (reload_macho_picbase_di): Likewise.
4237         (nonlocal_goto_receiver): Likewise.
4239 2013-09-30  Nick Clifton  <nickc@redhat.com>
4241         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
4242         that use the MSP430X ISA.
4243         (msp430_option_override): Scan -mmcu command line option for any
4244         MCU name that supports the MSP430X ISA.
4245         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
4246         -mmcu options which enable the MSP430X ISA.
4248 2013-09-30  Richard Biener  <rguenther@suse.de>
4250         PR middle-end/58532
4251         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
4252         before looking for setjmp-like calls.
4254 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
4256         PR target/10901
4257         * config/darwin-protos.h (machopic_get_function_picbase): New.
4258         * config/darwin.c (machopic_get_function_picbase): New.
4259         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
4260         label for a new func.  (load_macho_picbase_di): Likewise.
4261         (reload_macho_picbase): New expand.
4262         (reload_macho_picbase_si): New insn.
4263         (reload_macho_picbase_di): New insn.
4264         (nonlocal_goto_receiver): New define and split.
4265         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
4266         (unspecv enum): Add UNSPECV_NLGR.
4268 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
4270         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
4271         that altivec registers are correctly sized on Darwin.
4273 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
4275         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
4276         darwin-driver.o): Use COMPILE and POSTCOMPILE.
4277         * config/x-darwin (host-darwin.o): Likewise.
4278         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
4279         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
4280         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
4282 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
4284         * doc/invoke.texi: Fix usage of @tie{} command.
4286 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
4288         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
4290 2013-09-28  Jan Hubicka  <jh@suse.cz>
4292         * config/i386/x86-tune.def: Add documentation for each of the options;
4293         add whitespace.
4295 2013-09-28  Jan Hubicka  <jh@suse.cz>
4297         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
4298         generic.
4299         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
4300         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
4301         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
4302         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
4303         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
4305 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
4307         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
4308         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
4309         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
4310         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
4311         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
4312         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
4313         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
4314         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
4315         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
4316         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
4317         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
4318         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
4319         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
4320         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
4321         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
4322         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
4323         gimple.h, godump.c, graphite-clast-to-gimple.c,
4324         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
4325         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
4326         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
4327         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
4328         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
4329         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
4330         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
4331         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
4332         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
4333         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
4334         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
4335         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
4336         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
4337         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
4338         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
4339         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
4340         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
4341         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
4342         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
4343         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
4344         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
4345         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
4346         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
4347         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
4348         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
4349         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
4350         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
4351         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
4352         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
4353         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
4354         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
4355         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
4356         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
4357         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
4358         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
4359         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
4360         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
4361         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
4362         whitespace before "(".
4364 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
4366         * expr.h (extract_bit_field): Remove packedp parameter.
4367         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
4368         from forward declaration.
4369         (store_split_bit_field): Remove packedp arg from calls to
4370         extract_fixed_bit_field.
4371         (extract_bit_field_1): Remove packedp parameter and packedp
4372         argument from recursive calls and calls to extract_fixed_bit_field.
4373         (extract_bit_field): Remove packedp parameter and corresponding
4374         arg to extract_bit_field_1.
4375         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
4376         to issue warnings.
4377         (extract_split_bit_field): Remove packedp arg from call to
4378         extract_fixed_bit_field.
4379         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
4380         (copy_blkmode_from_reg): Likewise.
4381         (copy_blkmode_to_reg): Likewise.
4382         (read_complex_part): Likewise.
4383         (store_field): Likewise.
4384         (expand_expr_real_1): Likewise.
4385         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
4386         to extract_bit_field.
4387         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
4388         call to extract_bit_field.
4389         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
4390         call to extract_bit_field.
4391         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
4392         and special packedp behavior from -fstrict-volatile-bitfields
4393         documentation.
4395 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4397         * lra-eliminations.c (init_elim_table): Guard value_p.
4399 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
4401         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
4402         DFmode, DImode, and SFmode in the upper VSX registers based on the
4403         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
4404         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
4405         of the VSX register handling.
4407         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
4408         (f32_sv): Likewise.
4409         (zero_extendsidi2_lfiwzx): Add support for loading into the
4410         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
4411         only do VSX memory options on Altivec registers.
4412         (extendsidi2_lfiwax): Likewise.
4413         (extendsfdf2_fpr): Likewise.
4414         (mov<mode>_hardfloat, SF/SD modes): Likewise.
4415         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
4416         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
4417         (movdi_internal64): Likewise.
4419 2013-09-27  Xinliang David Li  <davidxl@google.com>
4421         * opts.c (finish_options): Adjust parameters
4422         according to vect cost model.
4423         (common_handle_option): Set dynamic vect cost
4424         model for FDO.
4425         targhooks.c (default_add_stmt_cost): Compute stmt cost
4426         unconditionally.
4427         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
4428         Use helper function.
4429         * tree-vectorizer.h (unlimited_cost_model): New function.
4430         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
4431         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
4432         function.
4433         (vect_enhance_data_refs_alignment): Ditto.
4434         * flag-types.h: New enum.
4435         * common/config/i386/i386-common.c (ix86_option_init_struct):
4436         No need to initialize vect_cost_model flag.
4437         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
4438         unconditionally.
4440 2013-09-27  Diego Novillo  <dnovillo@google.com>
4442         * gimple.h (enum ssa_mode): Remove.
4444 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
4446         * cfgloop.h (number_of_loops): Fix typo in check for null.
4448 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
4450         PR middle-end/58551
4451         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
4452         are children of outermost saved_cfun's loop, and set it up to
4453         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
4454         if loop != loop0 and SESE region contains bbs that belong to loop0.
4456 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
4458         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
4459         (binary_scale_code_p, get_base_term, get_index_term): New functions.
4460         (set_address_segment, set_address_base, set_address_index)
4461         (set_address_disp): Accept the argument unconditionally.
4462         (baseness): Remove must_be_base_p and must_be_index_p checks.
4463         (decompose_normal_address): Classify as much as possible in the
4464         main loop.
4466 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
4468         * cse.c (count_reg_usage): Handle INT_LIST.
4469         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
4470         * reginfo.c (reg_scan_mark_refs): Likewise.
4471         * reload1.c (eliminate_regs_1): Likewise.
4473 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
4475         PR middle-end/58547
4476         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
4477         signed.
4479 2013-09-27  Richard Biener  <rguenther@suse.de>
4481         PR tree-optimization/58459
4482         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
4483         restriction not propagating into loops.
4485 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
4487         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
4488         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
4489         * doc/tree-ssa.texi (Walking use-def chains): Delete.
4491 2013-09-26  Richard Biener  <rguenther@suse.de>
4493         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
4495 2013-09-26  Richard Biener  <rguenther@suse.de>
4497         * alias.h (component_uses_parent_alias_set): Rename to ...
4498         (component_uses_parent_alias_set_from): ... this.
4499         * alias.c (component_uses_parent_alias_set): Rename to ...
4500         (component_uses_parent_alias_set_from): ... this and return
4501         the desired parent.
4502         (reference_alias_ptr_type_1): Use the result from
4503         component_uses_parent_alias_set_from instead of stripping
4504         components one at a time.
4505         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
4507 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
4509         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
4510         Move prototypes to...
4511         * tree-ssa-ter.h: New File.  Move prototypes here.
4512         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
4513         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
4514         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
4515         from is_replaceable_p.
4516         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
4517         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
4518         (process_replaceable): Use ter_is_replaceable_p.
4519         (find_replaceable_in_bb): Use ter_is_replaceable_p.
4520         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
4521         newly refactored ssa_is_replaceable_p.
4522         * cfgexpand.c: Include tree-outof-ssa.h.
4523         * ssaexpand.h: Delete.
4525 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
4527         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
4528         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
4529         tree-ssa.c
4530         (create_gimple_tmp): Delete.
4531         (get_expr_type, build_assign, build_type_cast): Move to...
4532         * gimple-builder.c: New File.
4533         (get_expr_type): Relocate from gimple.c.
4534         (build_assign, build_type_cast): Change to only create ssanames.
4535         * gimple.h: Move prototypes to...
4536         * gimple-builder.h: New File. Here.
4537         * tree-ssa.h: And here.
4538         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
4539         count_uses_and_derefs): Relocate from gimple.c.
4540         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
4541         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
4542         * tree-ssa-math-opts (execute_cse_reciprocals): Use
4543         gimple_replace_ssa_lhs.
4544         * asan.c: Include gimple-builder.h.
4545         * Makefile.in: Add gimple-builder.o.
4547 2013-09-26  Richard Biener  <rguenther@suse.de>
4549         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
4550         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
4551         (loe_visit_block): Use gcc_checking_assert.
4552         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
4553         gimple_assign_ssa_name_copy_p.
4554         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
4555         change.
4557 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
4559         * config/rs6000/t-rs6000 (rs6000.o): Remove.
4560         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
4562 2013-09-26  Richard Biener  <rguenther@suse.de>
4564         PR tree-optimization/58539
4565         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
4566         the fact that debug statements are not taking part in loop-closed
4567         SSA construction.
4569 2013-09-26  Nick Clifton  <nickc@redhat.com>
4571         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
4572         time warning message.
4573         (msp430_print_operand_raw): Delete unused letter parameter.
4574         (TARGET_PRINT_OPERAND_ADDRESS): Define.
4575         (msp430_print_operand_address): New function.
4576         (msp430_print_operand): Move address printing code from here to
4577         new function.
4578         * config/msp430/msp430.md (movsipsi2): Add comment in generated
4579         assembler.
4580         (zero_extendpsisi2): Likewise.
4581         (extendpsisi2): New pattern.
4582         (andneghi3): New pattern.
4584 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
4586         * config/aarch64/aarch64.opt (mlra): New option.
4587         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
4588         (TARGET_LRA_P): Define.
4590 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
4592         * expr.c (expand_assignment): Remove obsolete comment.
4594 2013-09-25  Jeff Law  <law@redhat.com>
4596         * tree-flow.h (thread_through_all_blocks): Prototype moved into
4597         tree-ssa-threadupdate.h.
4598         (register_jump_thread): Similarly.
4599         * tree-ssa-threadupdate.h: New header file.
4600         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
4601         * tree-vrp.c: Likewise.
4602         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
4603         (thread_around_empty_blocks): Change type of path vector argument to
4604         an edge,type pair from just an edge.  Initialize both elements when
4605         appending to a jump threading path.  Tweak references to elements
4606         appropriately.
4607         (thread_across_edge): Similarly.  Release memory for the elements
4608         as needed.
4609         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
4610         (dump_jump_thread_path): New function broken out from
4611         register_jump_thread.
4612         (register_jump_thread): Use dump_jump_thread_path.  Change type of
4613         path vector entries.  Search the path for NULL edges and dump
4614         the path if one is found.  Tweak the conversion of path to 3-edge
4615         form to use the block copy type information embedded in the path.
4617 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4619         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
4621 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
4622             Vladimir Makarov  <vmakarov@redhat.com>
4624         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
4625         from the least significant bit.
4626         (strip_address_mutations): Add bitfield operations handling.
4627         (must_be_index_p): Add shifting and rotate operations handling.
4628         (set_address_base): Use must_be_base_p predicate.
4629         (set_address_index): Use must_be_index_p predicate.
4631 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4632             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4633             Sergey Lega  <sergey.s.lega@intel.com>
4634             Anna Tikhonova  <anna.tikhonova@intel.com>
4635             Ilya Tocar  <ilya.tocar@intel.com>
4636             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4637             Ilya Verbin  <ilya.verbin@intel.com>
4638             Kirill Yukhin  <kirill.yukhin@intel.com>
4639             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4641         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4642         Use new names.
4643         (ix86_expand_vector_move_misalign): Support new unaligned load and
4644         stores and use new names.
4645         (CODE_FOR_sse2_storedqu): Rename to ...
4646         (CODE_FOR_sse2_storedquv16qi): ... this.
4647         (CODE_FOR_sse2_loaddqu): Rename to ...
4648         (CODE_FOR_sse2_loaddquv16qi): ... this.
4649         (CODE_FOR_avx_loaddqu256): Rename to ...
4650         (CODE_FOR_avx_loaddquv32qi): ... this.
4651         (CODE_FOR_avx_storedqu256): Rename to ...
4652         (CODE_FOR_avx_storedquv32qi): ... this.
4653         * config/i386/i386.md (fpint_logic): New.
4654         * config/i386/sse.md (VMOVE): Extend for AVX512.
4655         (VF): Ditto.
4656         (VF_128_256): New.
4657         (VF_512): Ditto.
4658         (VI_UNALIGNED_LOADSTORE): Ditto.
4659         (sse2_avx_avx512f): Ditto.
4660         (sse2_avx2): Extend for AVX512.
4661         (sse4_1_avx2): Ditto.
4662         (avx2_avx512f): New.
4663         (sse): Extend for AVX512.
4664         (sse2): Ditto.
4665         (sse4_1): Ditto.
4666         (avxsizesuffix): Ditto.
4667         (sseintvecmode): Ditto.
4668         (ssePSmode): Ditto.
4669         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4670         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4671         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4672         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4673         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4674         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4675         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4676         (STORENT_MODE): Extend for AVX512.
4677         (*absneg<mode>2): Replace constraint "x" with "v".
4678         (*mul<mode>3): Ditto.
4679         (*ieee_smin<mode>3): Ditto.
4680         (*ieee_smax<mode>3): Ditto.
4681         (avx_cmp<mode>3): Replace VF with VF_128_256.
4682         (*<sse>_maskcmp<mode>3_comm): Ditto.
4683         (<sse>_maskcmp<mode>3): Ditto.
4684         (<sse>_andnot<mode>3): Extend for AVX512.
4685         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4686         (<code><mode>3, fpint_logic): New.
4687         (*<code><mode>3): Extend for AVX512.
4688         (avx512flogicsuff): New.
4689         (avx512f_<logic><mode>): Ditto.
4690         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4691         VF_128_256.
4692         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4693         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4694         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4695         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4696         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4697         (xop_vpermil2<mode>3): Ditto.
4698         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4699         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4700         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4701         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4703 2013-09-25  Tom Tromey  <tromey@redhat.com>
4705         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4706         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4707         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4708         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4709         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4710         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4711         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4712         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4713         (GRAPHITE_HTAB_H): Remove.
4715 2013-09-25  Tom Tromey  <tromey@redhat.com>
4717         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4719 2013-09-25  Tom Tromey  <tromey@redhat.com>
4721         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4723 2013-09-25  Tom Tromey  <tromey@redhat.com>
4725         * config/i386/t-i386 (i386.o): Remove.
4726         (i386-c.o): Use COMPILE and POSTCOMPILE.
4728 2013-09-25  Tom Tromey  <tromey@redhat.com>
4730         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4732 2013-09-25  Tom Tromey  <tromey@redhat.com>
4734         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4735         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4736         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4737         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4738         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4739         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4740         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4741         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4742         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4743         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4744         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4745         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4746         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4747         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4748         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4749         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4750         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4751         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4752         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4753         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4754         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4755         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4756         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4757         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4758         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4759         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4760         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4761         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4762         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4763         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4764         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4765         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4766         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4767         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4768         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4769         (graphite-clast-to-gimple.o, graphite-dependences.o)
4770         (graphite-interchange.o, graphite-poly.o)
4771         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4772         (graphite-optimize-isl.o, tree-vect-loop.o)
4773         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4774         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4775         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4776         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4777         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4778         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4779         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4780         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4781         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4782         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4783         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4784         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4785         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4786         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4787         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4788         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4789         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4790         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4791         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4792         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4793         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4794         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4795         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4796         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4797         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4798         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4799         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4800         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4801         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4802         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4803         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4804         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4805         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4806         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4807         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4808         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4809         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4810         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4811         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4812         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4813         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4814         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4815         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4816         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4817         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4818         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4819         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4820         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4821         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4822         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4823         (gcov-dump.o): Remove.
4824         (default-c.o): Use COMPILE and POSTCOMPILE.
4825         (CFLAGS-gcc.o): New variable.
4826         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4828 2013-09-25  Tom Tromey  <tromey@redhat.com>
4830         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4831         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4832         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4833         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4834         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4835         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4836         (c-family/array-notation-common.o, c-family/stub-objc.o)
4837         (c-family/c-ubsan.o): Remove.
4839 2013-09-25  Tom Tromey  <tromey@redhat.com>
4841         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4843 2013-09-25  Tom Tromey  <tromey@redhat.com>
4845         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4846         to add -DENABLE_SHARED_LIBGCC.
4847         (gcc.o): Don't use subshell.
4849 2013-09-25  Tom Tromey  <tromey@redhat.com>
4851         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4852         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4853         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4854         * configure, config.in: Rebuild.
4856 2013-09-25  Tom Tromey  <tromey@redhat.com>
4858         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4859         (COMPILE, POSTCOMPILE): New variables.
4860         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4861         (DEPFILES): New variable.
4862         Include ".Po" files.
4863         * configure.ac: Add checks for dependency checking.
4864         * configure, aclocal.m4: Regenerate.
4866 2013-09-25  Tom Tromey  <tromey@redhat.com>
4868         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4869         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4871 2013-09-25  Tom Tromey  <tromey@redhat.com>
4873         * Makefile.in (generated_files): Add options.h,
4874         target-hooks-def.h, insn-opinit.h,
4875         common/common-target-hooks-def.h, pass-instances.def,
4876         c-family/c-target-hooks-def.h.
4878 2013-09-25  Jeff Law  <law@redhat.com>
4880         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4881         than foo[foo.length () - 1] to access last member in a vec.
4882         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4884 2013-09-25  Richard Biener  <rguenther@suse.de>
4886         PR middle-end/58521
4887         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4889 2013-09-25  Jan Hubicka  <jh@suse.cz>
4891         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4892         test.
4894 2013-09-25  Marek Polacek  <polacek@redhat.com>
4896         PR sanitizer/58420
4897         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4898         when determining the type name.
4900 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4902         * config/sh/sh.md: Fix formatting.
4904 2013-09-24  Xinliang David Li  <davidxl@google.com>
4906         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4907         max peel iterations parameter.
4908         * param.def: New parameter.
4909         * doc/invoke.texi: Document New parameter.
4911 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4913         * gimple-pretty-print.c: Various whitespace tweaks.
4914         * tree-core.h: Likewise.
4915         * tree-pretty-print.c: Likewise.
4916         * tree-ssa-alias.c: Likewise.
4917         * tree-ssa-copy.c: Likewise.
4918         * tree-ssanames.c: Likewise.
4919         * tree-ssanames.h: Likewise.
4920         * tree-vrp.c: Likewise.
4922 2013-09-24  Alan Modra  <amodra@gmail.com>
4924         PR middle-end/57134
4925         PR middle-end/57586
4926         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4927         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4928         inout operands.
4930 2013-09-24  Richard Biener  <rguenther@suse.de>
4932         PR middle-end/58513
4933         * tree.c (reference_alias_ptr_type): Move ...
4934         * alias.c (reference_alias_ptr_type): ... here and implement
4935         in terms of the new reference_alias_ptr_type_1.
4936         (ref_all_alias_ptr_type_p): New helper.
4937         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4938         use ref_all_alias_ptr_type_p.
4939         (get_deref_alias_set): Add flag_strict_aliasing check here.
4940         (reference_alias_ptr_type_1): New function, split out from ...
4941         (get_alias_set): ... here.
4942         (alias_ptr_types_compatible_p): New function.
4943         * alias.h (reference_alias_ptr_type): Declare.
4944         (alias_ptr_types_compatible_p): Likewise.
4945         * tree.h (reference_alias_ptr_type): Remove.
4946         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4947         to compare MEM_REF alias types.
4949 2013-09-24  Richard Biener  <rguenther@suse.de>
4951         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4953 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4955         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4956         reload helper function arrays into a single array reg_addr.
4957         (reload_fpr_gpr): Likewise.
4958         (reload_gpr_vsx): Likewise.
4959         (reload_vsx_gpr): Likewise.
4960         (struct rs6000_reg_addr): Likewise.
4961         (reg_addr): Likewise.
4962         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4963         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4964         (rs6000_init_hard_regno_mode_ok): Likewise.
4965         (rs6000_secondary_reload_direct_move): Likewise.
4966         (rs6000_secondary_reload): Likewise.
4968         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4969         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4970         power8 changes.  Put wg constraint in alphabetical order.
4972         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4973         for future work to add ISA 2.07 VSX single precision support.
4974         (-mvsx-scalar-double): Change default from -1 to 1, update
4975         documentation comment.
4976         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4977         (-mupper-regs-df): New debug switch to control whether DF values
4978         can go in the traditional Altivec registers.
4979         (-mupper-regs-sf): New debug switch to control whether SF values
4980         can go in the traditional Altivec registers.
4982         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4983         and wy constraints.
4984         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4985         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4986         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4987         (rs6000_debug_legitimate_address_p): Print if we are running
4988         before, during, or after reload.
4989         (rs6000_secondary_reload): Add a comment.
4990         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4992         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4993         constraints.  Update documentation string.
4994         (wd constraint): Likewise.
4995         (wf constraint): Likewise.
4996         (wg constraint): Likewise.
4997         (wn constraint): Likewise.
4998         (ws constraint): Likewise.
4999         (wt constraint): Likewise.
5000         (wx constraint): Likewise.
5001         (wz constraint): Likewise.
5002         (wu constraint): New constraint for ISA 2.07 SFmode scalar
5003         instructions.
5004         (ww constraint): Likewise.
5005         (wy constraint): Likewise.
5006         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
5007         in the previous submissions.
5008         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
5010 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
5012         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
5013         (REG_BR_PROB): Say that the probability is stored in an int_list.
5014         * reg-notes.def: Update commentary to mention INT_LIST.
5015         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
5016         (INT_LIST): New rtx.
5017         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
5018         * rtlanal.c (int_reg_note_p): New function.
5019         (alloc_reg_note): Assert that the note does not have an int argument.
5020         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
5021         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
5022         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
5023         rather than an INSN_LIST.  Handle INT_LIST.
5024         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
5025         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
5026         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
5027         Manipulate them as ints rather than rtxes.
5028         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
5029         * regmove.c (copy_src_to_dest): Likewise.
5030         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
5032         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
5033         into the cases that need it.
5034         * config/arm/arm.c (arm_unwind_emit): Likewise.
5036         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
5037         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
5038         * loop-doloop.c (add_test, doloop_modify): Likewise.
5039         * loop-unswitch.c (compare_and_jump_seq): Likewise.
5040         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
5041         * predict.c (combine_predictions_for_insn): Likewise.
5042         * print-rtl.c (print_rtx): Handle INT_LIST.
5043         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
5044         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
5045         * config/arm/arm.c (emit_unlikely_jump): Likewise.
5046         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
5047         (ix86_split_fp_branch, predict_jump): Likewise.
5048         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
5049         * config/sh/sh.c (expand_cbranchsi4): Likewise.
5050         * config/spu/spu.c (ea_load_store_inline): Likewise.
5052         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
5053         value of a REG_BR_PROB note.
5054         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
5055         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
5056         * emit-rtl.c (try_split): Likewise.
5057         * predict.c (br_prob_note_reliable_p): Likewise.
5058         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
5059         * reorg.c (mostly_true_jump): Likewise.
5060         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
5061         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
5062         * config/i386/i386.c (ix86_print_operand): Likewise.
5063         * config/ia64/ia64.c (ia64_print_operand): Likewise.
5064         * config/mmix/mmix.c (mmix_print_operand): Likewise.
5065         * config/rs6000/rs6000.c (output_cbranch): Likewise.
5066         * config/s390/s390.c (s390_expand_tbegin): Likewise.
5067         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
5068         * config/sparc/sparc.c (output_cbranch): Likewise.
5069         * config/spu/spu.c (get_branch_target): Likewise.
5070         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
5071         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
5073 2013-09-23  Jan Hubicka  <jh@suse.cz>
5075         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
5076         for ipa-devirt.
5077         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
5078         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
5079         of external calls
5080         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
5081         (gimple_fold_call): Dump inconsistent devirtualizations; add
5082         sanity check for type based devirtualizations.
5083         * ipa-prop.c: Include ipa-utils.h
5084         (ipa_intraprocedural_devirtualization): Add sanity check.
5085         (try_make_edge_direct_virtual_call): Likewise.
5087 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
5089         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
5090         assignment statements.
5092 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
5094         * gimple-pretty-print.c (dump_ssaname_info): New function.
5095         (dump_gimple_phi): Call it.
5096         (pp_gimple_stmt_1): Likewise.
5097         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
5098         (range_info_def): Declare.
5099         * tree-pretty-print.c (pp_double_int): New function.
5100         (dump_generic_node): Call it.
5101         * tree-pretty-print.h (pp_double_int): Declare.
5102         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
5103         * tree-ssanames.h (range_info_def): New structure.
5104         (value_range_type): Move definition here.
5105         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
5106         Declare.
5107         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
5108         initialization.
5109         (set_range_info): New function.
5110         (get_range_info): Likewise.
5111         (duplicate_ssa_name_range_info): Likewise.
5112         (duplicate_ssa_name_fn): Check pointer type and call
5113         duplicate_ssa_name_range_info.
5114         * tree-ssa-copy.c (fini_copy_prop): Likewise.
5115         * tree-vrp.c (value_range_type): Remove definition, now in
5116         tree-ssanames.h.
5117         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
5118         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
5119         (SSA_NAME_RANGE_INFO): New macro.
5121 2013-09-23  Richard Biener  <rguenther@suse.de>
5123         PR tree-optimization/58464
5124         * tree-ssa-pre.c (phi_trans_lookup): Remove.
5125         (phi_trans_add): Change to add conditionally on being not present.
5126         (phi_translate_1): Remove recursion detection here.
5127         (phi_translate): Pre-seed the cache with NULL to catch
5128         recursion here in a more generic way.
5129         (bitmap_find_leader): Adjust comment.
5130         (get_representative_for): Dump value-numbers.
5131         (create_expression_by_pieces): Likewise.
5132         (insert_into_preds_of_block): Likewise.
5134 2013-09-23  Christian Bruel  <christian.bruel@st.com>
5136         PR target/58475
5137         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
5138         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
5140 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
5142         Revert r202780:
5143         2013-09-20  Renlin Li  <renlin.li@arm.com>
5145         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5146         plus_constant.
5147         (aarch64_expand_epilogue): Likewise.
5148         (aarch64_legitimize_reload_address): Likewise.
5150 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
5152         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
5153         NULL_TREE before pushing them onto the vector.  Likewise for labels.
5155 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
5157         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
5158         comparison operators when -fno-trapping-math is in effect.
5159         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
5160         comparison operators in TFmode and assert that unsupported operators
5161         cannot reach here.
5162         (ia64_print_operand): Likewise.
5164 2013-09-21  Jan Hubicka  <jh@suse.cz>
5166         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
5167         (sse_typeless_stores): Enable for core
5168         (sse_load0_by_pxor): Likewise.
5169         (four_jump_limit): Disable for core.
5170         (pad_returns): Likewise.
5171         (avoid_vector_decode): Likewise.
5172         (fuse_cmp_and_branch): Enable for cores.
5173         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
5175 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
5177         PR middle-end/56791
5178         * config/pa/pa.c (pa_option_override): Disable auto increment and
5179         decrement instructions until reload is completed.
5181         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
5182         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
5183         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
5185 2013-09-20  DJ Delorie  <dj@redhat.com>
5186             Nick Clifton  <nickc@redhat.com>
5188         * config/rl78/rl78.c: Various whitespace and comment tweaks.
5189         (need_to_save): Save bank 0 on interrupts.
5190         (characterize_address): Strip far address wrappers.
5191         (rl78_as_legitimate_address): Likewise.
5192         (transcode_memory_rtx): Likewise.
5193         (rl78_peep_movhi_p): Disable this peephole after devirt.
5194         (rl78_propogate_register_origins): Forget all origins when a
5195         CLOBBER is seen.
5196         * config/rl78/rl78-virt.md: Various whitespace tweaks.
5197         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
5198         peephole2's.
5199         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
5200         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
5201         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
5202         relocating.
5203         * config/rl78/constraints.md: Various whitespace and paren tweaks.
5205 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
5207         * config/pa/pa.md: In "scc" insn patterns, change output template to
5208         handle const0_rtx in reg_or_0_operand operands.
5210 2013-09-20  Martin Husemann  <martin@NetBSD.org>
5212         PR target/56875
5213         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
5214         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
5216 2013-09-20  Richard Biener  <rguenther@suse.de>
5218         PR middle-end/58484
5219         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
5220         remembering SSA name version and block index.
5221         (new_scev_info_str): Adjust.
5222         (hash_scev_info): Likewise.  Also hash the block index.
5223         (eq_scev_info): Adjust.
5224         (find_var_scev_info): Likewise.
5225         (struct instantiate_cache_entry): Remove.
5226         (struct instantiate_cache_type): Use a htab to map name, block
5227         to chrec.
5228         (instantiate_cache_type::~instantiate_cache_type): Adjust.
5229         (get_instantiated_value_entry): Likewise.
5230         (hash_idx_scev_info, eq_idx_scev_info): New functions.
5231         (instantiate_scev_name): Adjust.
5233 2013-09-20  Jeff Law  <law@redhat.com>
5235         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
5237 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
5239         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5240         Call aarch64_simd_expand_args to update op[argc].
5242 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
5244         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
5245         plugin argument.
5247 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
5249         * gengtype.c (file_rules): Added rule for *.cc files.
5250         (get_output_file_with_visibility): Give fatal message when no
5251         rules found.
5253 2013-09-20  Renlin Li  <renlin.li@arm.com>
5255         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
5256         (aarch64_expand_epilogue): Likewise.
5257         (aarch64_legitimize_reload_address): Likewise.
5259 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5261         PR middle-end/57748
5262         * expr.c (expand_assignment): Remove misalignp code path.
5264 2013-09-20  Marek Polacek  <polacek@redhat.com>
5266         PR sanitizer/58413
5267         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
5268         TYPE_PRECISION.  Add asserts.
5270 2013-09-20  Richard Biener  <rguenther@suse.de>
5272         PR tree-optimization/58453
5273         * tree-loop-distribution.c (distribute_loop): Apply the cost
5274         model for -ftree-loop-distribute-patterns, too.
5276 2013-09-20  Richard Biener  <rguenther@suse.de>
5278         PR middle-end/58473
5279         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
5280         make type comparison less strict.
5282 2013-09-20  Alan Modra  <amodra@gmail.com>
5284         * configure: Regenerate.
5285         * aclocal.m4: Regenerate.
5287 2013-09-20  Marek Polacek  <polacek@redhat.com>
5289         PR other/58467
5290         * doc/extend.texi: Document that attribute used is meant to be used
5291         on variables with static storage duration.
5293 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
5295         PR tree-optimization/58472
5296         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
5297         simd_lane_access set inv_p = false.
5298         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
5299         the simduid magic VAR_DECL.
5301 2013-09-19  Jan Hubicka  <jh@suse.cz>
5303         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
5305 2013-09-17  Jeff Law  <law@redhat.com>
5307         * tree-ssa-dom.c (record_temporary_equivalences): New function
5308         split out of dom_opt_dom_walker::after_dom_children.
5309         (dom_opt_dom_walker::thread_across_edge): Move common code
5310         in here from dom_opt_dom_walker::after_dom_children.
5311         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
5313 2013-09-19  Jan Hubicka  <jh@suse.cz>
5315         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
5316         (TARGET_GENERIC): Use PROCESOR_GENERIC
5317         (enum processor_type): Unify generic32 and 64.
5318         * i386.md (cpu): Likewise.
5319         * x86-tune.def (use_leave): Enable for generic32.
5320         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
5321         * athlon.md: Change generic64 to generic in all occurences.
5322         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
5323         (ix86_target_macros_internal): Likewise.
5324         * driver-i386.c (host_detect_local_cpu): Likewise.
5325         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
5326         to ..
5327         (generic_memcpy, generic_memset, generic_cost): This one.
5328         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
5329         (m_GENERIC32, m_GENERIC64): Remove.
5330         (m_GENERIC): Turn into one flag.
5331         (processor_target): Unify generic tunnings.
5332         (ix86_option_override_internal): Replace generic32/64 by generic.
5333         (ix86_issue_rate): Likewise.
5334         (ix86_adjust_cost): Likewise.
5336 2013-09-19  Jan Hubicka  <jh@suse.cz>
5338         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
5339         of speculative flag.
5341 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
5343         * omp-low.c (expand_omp_sections): Always pass len - 1 to
5344         GOMP_sections_start, even if !exit_reachable.
5346 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
5348         * lra-constraints.c (need_for_all_save_p): Use macro
5349         HARD_REGNO_CALL_PART_CLOBBERED.
5350         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
5351         set up pseudo conflict hard regs.
5353 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
5355         PR target/58452
5356         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
5357         operands.
5359 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
5361         PR rtl-optimization/58438
5362         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
5363         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
5364         from previous subpasses.
5366 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
5368         * arm.c (arm_get_frame_offsets): Validate architecture supports
5369         LDRD/STRD before accepting the tuning preference.
5370         (arm_expand_prologue): Likewise.
5371         (arm_expand_epilogue): Likewise.
5373 2013-09-18  Richard Biener  <rguenther@suse.de>
5375         PR tree-optimization/58417
5376         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
5377         have chrecs with symbols defined in the loop as operands.
5378         (chrec_fold_multiply): Likewise.
5379         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
5380         parameters before folding binary operations.
5381         (struct instantiate_cache_entry_hasher): Remove.
5382         (struct instantiate_cache_type): Use a pointer-map.
5383         (instantiate_cache_type::instantiate_cache_type): New function.
5384         (instantiate_cache_type::get): Likewise.
5385         (instantiate_cache_type::set): Likewise.
5386         (instantiate_cache_type::~instantiate_cache_type): Adjust.
5387         (get_instantiated_value_entry): Likewise.
5388         (global_cache): New global.
5389         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
5390         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
5391         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
5392         (instantiate_scev_name): Adjust.
5393         (instantiate_scev): Construct global instead of local cache.
5394         (resolve_mixers): Likewise.
5396 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
5397             Paolo Carlini  <paolo.carlini@oracle.com>
5399         PR c++/58458
5400         * doc/implement-cxx.texi: Fix references to the C++ standards.
5402 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
5404         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
5405         * tree-vect-data-refs.c (vect_analyze_data_refs): For
5406         simd_lane_access drs, update also DR_ALIGNED_TO.
5408 2013-09-18  Marek Polacek  <polacek@redhat.com>
5410         PR sanitizer/58411
5411         * doc/extend.texi: Document no_sanitize_undefined attribute.
5412         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
5413         no_sanitize_undefined attribute.
5415 2013-09-18  Nick Clifton  <nickc@redhat.com>
5417         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
5418         (ASM_DECLARE_FUNCTION_NAME): Define.
5420 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
5422         * compare-elim.c (find_comparison_dom_walker): New class
5423         (find_comparisons_in_bb): Rename to
5424         find_comparison_dom_walker::before_dom_children
5425         (find_comparisons): Adjust
5426         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
5427         adjust.
5428         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
5429         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
5430         (init_walk_dominator_tree): Remove declaration.
5431         (fini_walk_dominator_tree): Remove declaration.
5432         * fwprop.c (single_def_use_dom_walker): New class
5433         (single_def_use_enter_block): Convert to
5434         single_def_use_dom_walker::before_dom_children.
5435         (single_def_use_leave_block): Convert to
5436         single_def_use_dom_walker::after_dom_children.
5437         (build_single_def_use_links): Adjust.
5438         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
5439         class.
5440         (find_candidates_in_block): Convert to
5441         find_candidates_dom_walker::before_dom_children.
5442         (execute_strength_reduction): Adjust.
5443         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
5444         (sese_dom_walker): New class.
5445         (sese_dom_walker::sese_dom_walker): New constructor.
5446         (sese_dom_walker::~sese_dom_walker): New destructor.
5447         (build_sese_conditions_before): Convert to
5448         sese_dom_walker::before_dom_children.
5449         (build_sese_conditions_after): Convert to
5450         sese_dom_walker::after_dom_children.
5451         (build_poly_scop): Adjust
5452         * tree-into-ssa.c (rewrite_dom_walker): New class
5453         (rewrite_enter_block): Convert to
5454         rewrite_dom_walker::before_dom_children.
5455         (rewrite_leave_block): Convert to
5456         rewrite_dom_walker::after_dom_children.
5457         (rewrite_update_dom_walker): New class.
5458         (rewrite_update_enter_block): Convert to
5459         rewrite_update_dom_walker::before_dom_children.
5460         (rewrite_update_leave_block): Convert to
5461         rewrite_update_dom_walker::after_dom_children.
5462         (rewrite_blocks, rewrite_into_ssa): Adjust.
5463         (mark_def_dom_walker): New class.
5464         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
5465         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
5466         (mark_def_sites_blocks): Convert to
5467         mark_def_dom_walker::before_dom_children.
5468         (mark_def_site_blocks): Remove.
5469         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
5470         (tree_ssa_dominator_optimize): Adjust.
5471         (dom_thread_across_edge): Convert to method
5472         dom_opt_dom_walker::thread_across_edge.
5473         (dom_opt_enter_block): Convert to member function
5474         dom_opt_dom_walker::before_dom_children.
5475         (dom_opt_leave_block): Convert to member function
5476         dom_opt_dom_walker::after_dom_children.
5477         * tree-ssa-dse.c (dse_dom_walker): New class.
5478         (dse_enter_block): Convert to member function
5479         dse_dom_walker::before_dom_children.
5480         (tree_ssa_dse): Adjust.
5481         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
5482         (determine_invariantness_stmt): Convert to method
5483         invariantness_dom_walker::before_dom_children.
5484         (determine_invariantness): Remove
5485         (move_computations_dom_walker): New class.
5486         (move_computations_stmt): Convert to method
5487         move_computations_dom_walker::before_dom_children.
5488         (move_computations, tree_ssa_lim): Adjust.
5489         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
5490         (nt_init_block): Convert to method
5491         notrappping_dom_walker::before_dom_children.
5492         (nt_fini_block): Convert to method
5493         method nontrapping_dom_walker::after_dom_children.
5494         (get_non_trapping): Adjust.
5495         * tree-ssa-pre.c (eliminate_dom_walker): New class.
5496         (eliminate_bb): Convert to method
5497         eliminate_dom_walker::before_dom_children.
5498         (eliminate_leave_block): Convert to method
5499         eliminate_dom_walker::after_dom_children.
5500         (eliminate): Adjust.
5501         * tree-ssa-strlen.c (strlen_dom_walker): New class.
5502         (strlen_enter_block): Convert to method
5503         strlen_dom_walker::before_dom_children.
5504         (strlen_leave_block): Convert to method
5505         method strlen_dom_walker::after_dom_children.
5506         (tree_ssa_strlen): Adjust.
5507         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
5508         (tree_ssa_uncprop): Adjust.
5509         (uncprop_leave_block): Convert to method
5510         uncprop_dom_walker::after_dom_children.
5511         (uncprop_leave_block): Convert to method
5512         uncprop_dom_walker::before_dom_children.
5514 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
5516         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
5517         branch in same basic block.  Check both src and dest of the move insn.
5519 2013-09-17  Nick Clifton  <nickc@redhat.com>
5521         * config/rl78/rl78-real.md (bf): New pattern.
5522         (bt): New pattern.
5523         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
5524         (rl78_print_operand): Do not put a # before a %B.
5525         * config/rl78/rl78.opt: Tweak doc strings.
5527 2013-09-17  DJ Delorie  <dj@redhat.com>
5529         * config/rl78/constraints.md (Wcv): Allow up to $r31.
5530         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
5531         (rl78_option_override): Likewise, if -mallregs.
5532         (is_virtual_register): Likewise.
5533         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
5534         (REGNO_OK_FOR_BASE_P): Likewise.
5535         * config/rl78/rl78.opt (-mallregs): New.
5537 2013-09-17  Nick Clifton  <nickc@redhat.com>
5539         * config/rl78/rl78.c (need_to_save): Change return type to bool.
5540         For interrupt functions: save all call clobbered registers if the
5541         interrupt handler is not a leaf function.
5542         (rl78_expand_prologue): Always recompute the frame information.
5543         For interrupt functions: only select bank 0 if one of the bank 0
5544         registers is going to be psuhed.
5546 2013-09-17  DJ Delorie  <dj@redhat.com>
5548         * config/rl78/constraints.md: For each W* constraint, rename to C*
5549         and create a W* constraint that checks for an optional ES: prefix
5550         pattern also.
5551         * config/rl78/rl78.md (UNS_ES_ADDR): New.
5552         (es_addr): New.  Used to wrap far addresses.
5553         * config/rl78/rl78-protos.h (rl78_es_addr): New.
5554         (rl78_es_base): New.
5555         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
5556         wrapped far addresses.
5557         (rl78_print_operand_1): Unwrap far addresses before processing.
5558         (rl78_lo16): Wrap far addresses in unspecs.
5559         (rl78_es_addr): New.
5560         (rl78_es_base): New.
5561         (insn_ok_now): Check for not-yet-wrapped far addresses.
5562         (transcode_memory_rtx): Properly re-wrap far addresses.
5564 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
5566         * config/sparc/t-rtems: Add leon3 multilibs.
5568 2013-09-17  Cong Hou  <congh@google.com>
5570         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
5571         when checking the dot production pattern. The type of rhs operand
5572         of multiply is now checked correctly.
5574 2013-09-17  Jeff Law  <law@redhat.com>
5576         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5577         edge implied equivalences into successor phis.
5578         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
5579         here from tree-ssa-threadedge.c.
5580         (mark_threaded_blocks): When threading through a joiner, if both
5581         successors of the joiner's clone reach the same block, verify the
5582         PHI arguments are equal.  If not, cancel the jump threading request.
5583         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
5584         tree-ssa-threadupdate.c
5585         (thread_across_edge): Don't check PHI argument equality when
5586         threading through joiner block here.
5588 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
5590         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
5591         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
5592         (warn_uninit, warn_uninitialized_vars,
5593         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5594         Move to tree-ssa-uninit.c.
5595         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
5596         (has_undefined_value_p): New.  Pass dependant parts of
5597         ssa_undefined_value_p.
5598         (uninit_undefined_value_p): Use has_undefined_value_p.
5599         (warn_uninit, warn_uninitialized_vars,
5600         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
5601         Move from tree-ssa.c.
5602         * tree-ssa.h: Adjust prototypes.
5604 2013-09-17  Jan Hubicka  <jh@suse.cz>
5606         PR middle-end/58332
5607         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
5608         * ipa-inline.c (can_inline_edge_p): Do not downgrade
5609         FUNCTION_NOT_OPTIMIZED.
5610         * ipa-inline-analysis.c (compute_inline_parameters): Function
5611         not optimized is not inlinable unless it is alwaysinline.
5612         (inline_analyze_function): Force calls in not optimized
5613         function not inlinable.
5615 2013-09-17  Jan Hubicka  <jh@suse.cz>
5617         PR middle-end/58329
5618         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
5619         to return NULL.
5620         * ipa.c (function_and_variable_visibility): Likewise.
5621         * ipa-profile.c (ipa_profile): Likewise.
5623 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5625         PR ipa/58398
5626         * cgraph.c (cgraph_function_body_availability): Check for ifunc
5627         attribute, and don't inline the resolver in this case.
5629 2013-09-17  Teresa Johnson  <tejohnson@google.com>
5631         * coverage.c (get_coverage_counts): Add missing newline.
5633 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5635         PR tree-optimization/58088
5636         * fold-const.c (mask_with_trailing_zeros): New function.
5637         (fold_binary_loc): Make sure we don't recurse infinitely
5638         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5639         Use mask_with_trailing_zeros where appropriate.
5641 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5643         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5644         of 'prev' var to get better distance estimation.
5646 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5648         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5649         * tree-inline.c (is_parameter_of): New predicate.
5650         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5651         a parameter has been remapped.
5652         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5653         (optimize_inline_calls): Initialize transform_parameter.
5654         (copy_gimple_seq_and_replace_locals): Likewise.
5655         (tree_function_versioning): Likewise.
5656         (maybe_inline_call_in_expr): Likewise.
5658 2013-09-17  Nick Clifton  <nickc@redhat.com>
5660         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5661         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5662         interrupt handlers.
5663         (is_attr_func): New function.
5664         (msp430_is_interrupt_func): New function.
5665         (is_naked_func): New function.
5666         (is_reentrant_func): New function.
5667         (is_critical_func): New function.
5668         (msp430_start_function): Add annotations for function attributes.
5669         (msp430_attr): New function.
5670         (msp430_attribute_table): New.
5671         (msp430_function_section): New function.
5672         (TARGET_ASM_FUNCTION_SECTION): Define.
5673         (msp430_builtin): New enum.
5674         (msp430_init_builtins): New function.
5675         (msp430_builtin_devl): New function.
5676         (msp430_expand_builtin): New function.
5677         (TARGET_INIT_BUILTINS): Define.
5678         (TARGET_EXPAND_BUILTINS): Define.
5679         (TARGET_BUILTIN_DECL): Define.
5680         (msp430_expand_prologue): Add support for naked, interrupt,
5681         critical and reentrant functions.
5682         (msp430_expand_epilogue): Likewise.
5683         (msp430_print_operand): Handle 'O' character.
5684         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5685         NO_TRAMPOLINES.
5686         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5687         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5688         (pushm): Use a 'n' rather than an 'i' constraint.
5689         (msp_return): Add generation of the interrupt return instruction.
5690         (disable_interrupts): New pattern.
5691         (enable_interrupts): New pattern.
5692         (push_intr_state): New pattern.
5693         (pop_intr_state): New pattern.
5694         (bic_SR): New pattern.
5695         (bis_SR): New pattern.
5696         * doc/extend.texi: Document MSP430 function attributes and builtin
5697         functions.
5699 2013-09-17  Richard Biener  <rguenther@suse.de>
5701         PR tree-optimization/58432
5702         * tree-loop-distribution.c (tree_loop_distribution): Also
5703         scan PHIs for outside loop uses and seed a partition from them.
5705 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5707         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5708         (restructure_reference): Call backtrace_base_for_ref.
5710 2013-09-17  Alan Modra  <amodra@gmail.com>
5712         PR target/57589
5713         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5714         patch.
5716 2013-09-16  DJ Delorie  <dj@redhat.com>
5718         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5719         vregs location for RL78/G10.
5720         (rl78_expand_prologue): Avoid SEL on G10.
5721         (rl78_expand_epilogue): Likewise.
5722         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5723         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5724         __RL78_G10__ when appropriate.
5725         (ASM_SPEC): Pass -mg10 along to the assembler.
5726         * config/rl78/rl78.md (sel_rb): Disable for G10.
5727         * config/rl78/rl78.opt: Add -mg10 option.
5728         * config/rl78/t-rl78: Add -mg10 multilib.
5730 2013-09-16  Xinliang David Li  <davidxl@google.com>
5732         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5733         * omp-low.c (omp_max_vf): Ditto.
5734         (expand_omp_simd): Ditto.
5735         * tree-vectorizer.c (vectorize_loops): Ditto.
5736         (gate_vect_slp): Ditto.
5737         (gate_increase_alignment): Ditto.
5738         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5739         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5740         (gate_tree_vectorize): Name change.
5741         (tree_vectorize): Ditto.
5742         (pass_vectorize::gate): Call new function.
5743         (pass_vectorize::execute): Ditto.
5744         * opts.c: O3 default setting change.
5745         (finish_options): Check new flag.
5746         * doc/invoke.texi: Document new flags.
5747         * common.opt: New flags.
5749 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5751         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5752         (Condition Code): Don't reference it.
5753         * doc/tm.texi: Regenerate.
5755 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5757         PR middle-end/58418
5758         * lra-constraints.c (undo_optional_reloads): Consider all optional
5759         reload even if it did not get a hard reg.
5761 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5763         * dumpfile.c (dump_loc): Remove newline emission.
5764         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5765         emission to dump_printf_loc calls where missing.
5766         (vect_mark_for_runtime_alias_test): Ditto.
5767         (vect_analyze_data_ref_dependence): Ditto.
5768         (vect_analyze_data_ref_dependences): Ditto.
5769         (vect_slp_analyze_data_ref_dependence): Ditto.
5770         (vect_slp_analyze_data_ref_dependences): Ditto.
5771         (vect_compute_data_ref_alignment): Ditto.
5772         (vect_update_misalignment_for_peel): Ditto.
5773         (vect_verify_datarefs_alignment): Ditto.
5774         (vector_alignment_reachable_p): Ditto.
5775         (vect_get_data_access_cost): Ditto.
5776         (vect_enhance_data_refs_alignment): Ditto.
5777         (vect_find_same_alignment_drs): Ditto.
5778         (vect_analyze_data_refs_alignment): Ditto.
5779         (vect_analyze_group_access): Ditto.
5780         (vect_analyze_data_ref_access): Ditto.
5781         (vect_analyze_data_ref_accesses): Ditto.
5782         (vect_prune_runtime_alias_test_list): Ditto.
5783         (vect_analyze_data_refs): Ditto.
5784         (vect_create_addr_base_for_vector_ref): Ditto.
5785         (vect_create_data_ref_ptr): Ditto.
5786         (vect_grouped_store_supported): Ditto.
5787         (vect_grouped_load_supported): Ditto.
5788         * value-prof.c (check_counter): Ditto.
5789         (check_ic_target): Ditto.
5790         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5791         (vect_recog_widen_mult_pattern): Ditto.
5792         (vect_recog_widen_sum_pattern): Ditto.
5793         (vect_recog_over_widening_pattern): Ditto.
5794         (vect_recog_widen_shift_pattern): Ditto.
5795         (vect_recog_rotate_pattern): Ditto.
5796         (vect_recog_vector_vector_shift_pattern): Ditto.
5797         (vect_recog_divmod_pattern): Ditto.
5798         (vect_recog_mixed_size_cond_pattern): Ditto.
5799         (vect_recog_bool_pattern): Ditto.
5800         (vect_pattern_recog_1): Ditto.
5801         (vect_pattern_recog): Ditto.
5802         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5803         (vect_is_simple_iv_evolution): Ditto.
5804         (vect_analyze_scalar_cycles_1): Ditto.
5805         (vect_get_loop_niters): Ditto.
5806         (vect_analyze_loop_1): Ditto.
5807         (vect_analyze_loop_form): Ditto.
5808         (vect_analyze_loop_operations): Ditto.
5809         (vect_analyze_loop_2): Ditto.
5810         (vect_analyze_loop): Ditto.
5811         (report_vect_op): Ditto.
5812         (vect_is_slp_reduction): Ditto.
5813         (vect_is_simple_reduction_1): Ditto.
5814         (vect_get_known_peeling_cost): Ditto.
5815         (vect_estimate_min_profitable_iters): Ditto.
5816         (vect_model_reduction_cost): Ditto.
5817         (vect_model_induction_cost): Ditto.
5818         (get_initial_def_for_induction): Ditto.
5819         (vect_create_epilog_for_reduction): Ditto.
5820         (vectorizable_reduction): Ditto.
5821         (vectorizable_induction): Ditto.
5822         (vectorizable_live_operation): Ditto.
5823         (vect_loop_kill_debug_uses): Ditto.
5824         (vect_transform_loop): Ditto.
5825         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5826         (vect_stmt_relevant_p): Ditto.
5827         (process_use): Ditto.
5828         (vect_mark_stmts_to_be_vectorized): Ditto.
5829         (vect_model_simple_cost): Ditto.
5830         (vect_model_promotion_demotion_cost): Ditto.
5831         (vect_model_store_cost): Ditto.
5832         (vect_get_store_cost): Ditto.
5833         (vect_model_load_cost): Ditto.
5834         (vect_get_load_cost): Ditto.
5835         (vect_init_vector_1): Ditto.
5836         (vect_get_vec_def_for_operand): Ditto.
5837         (vect_finish_stmt_generation): Ditto.
5838         (vectorizable_call): Ditto.
5839         (vectorizable_conversion): Ditto.
5840         (vectorizable_assignment): Ditto.
5841         (vectorizable_shift): Ditto.
5842         (vectorizable_operation): Ditto.
5843         (vectorizable_store): Ditto.
5844         (vectorizable_load): Ditto.
5845         (vectorizable_condition): Ditto.
5846         (vect_analyze_stmt): Ditto.
5847         (vect_transform_stmt): Ditto.
5848         (vect_is_simple_use): Ditto.
5849         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5850         (vect_can_advance_ivs_p): Ditto.
5851         (vect_update_ivs_after_vectorizer): Ditto.
5852         (vect_do_peeling_for_loop_bound): Ditto.
5853         (vect_gen_niters_for_prolog_loop): Ditto.
5854         (vect_update_inits_of_drs): Ditto.
5855         (vect_create_cond_for_alias_checks): Ditto.
5856         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5857         (vect_build_slp_tree_1): Ditto.
5858         (vect_supported_load_permutation_p): Ditto.
5859         (vect_analyze_slp_instance): Ditto.
5860         (vect_analyze_slp): Ditto.
5861         (vect_make_slp_decision): Ditto.
5862         (vect_detect_hybrid_slp): Ditto.
5863         (vect_bb_vectorization_profitable_p): Ditto.
5864         (vect_slp_analyze_bb_1): Ditto.
5865         (vect_update_slp_costs_according_to_vf): Ditto.
5866         (vect_get_mask_element): Ditto.
5867         (vect_transform_slp_perm_load): Ditto.
5868         (vect_schedule_slp_instance): Ditto.
5869         (vect_schedule_slp): Ditto.
5870         (vect_slp_transform_bb): Ditto.
5871         * profile.c (read_profile_edge_counts): Ditto.
5872         (compute_branch_probabilities): Ditto.
5873         * coverage.c (get_coverage_counts): Ditto.
5875 2013-09-16  Diego Novillo  <dnovillo@google.com>
5877         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5879 2013-09-16  Jan Hubicka  <jh@suse.cz>
5881         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5882         abstract functions; for static functions check the presence of body.
5884 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5886         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5887         * config/aarch64/aarch64-simd.md
5888         (aarch64_mla_elt<mode>): New.
5889         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5890         (aarch64_mls_elt<mode>): Likewise.
5891         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5892         (aarch64_fma4_elt<mode>): Likewise.
5893         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5894         (aarch64_fma4_elt_to_128v2df): Likewise.
5895         (aarch64_fma4_elt_to_64df): Likewise.
5896         (fnma<mode>4): Likewise.
5897         (aarch64_fnma4_elt<mode>): Likewise.
5898         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5899         (aarch64_fnma4_elt_to_128v2df): Likewise.
5900         (aarch64_fnma4_elt_to_64df): Likewise.
5901         * config/aarch64/iterators.md (VDQSF): New.
5902         * config/aarch64/arm_neon.h
5903         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5904         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5906 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5908         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5909         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5910         (aarch64_mul3_elt_to_128df): Likewise.
5911         (aarch64_mul3_elt_to_64v2df): Likewise.
5912         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5913         (VMUL): New.
5914         (VMUL_CHANGE_NLANES) Likewise.
5915         (h_con): Likewise.
5916         (f): Likewise.
5917         * config/aarch64/arm_neon.h
5918         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5920 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5922         * config/aarch64/arm_neon.h
5923         (vcvtx_high_f32_f64): Fix parameters.
5925 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5926             Uros Bizjak  <ubizjak@gmail.com>
5928         * config/alpha.c: Include tree-ssanames.h.
5930 2013-09-16  Richard Biener  <rguenther@suse.de>
5932         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5933         (dot_rdg_1): Handle control_dd.
5934         (create_edge_for_control_dependence): New function.
5935         (create_rdg_edges): Add control dependences if asked for.
5936         (build_rdg): Likewise.
5937         (generate_loops_for_partition): If there are not necessary
5938         control stmts remove all their dependencies.
5939         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5940         (distribute_loop): Pass on control dependences.
5941         (tree_loop_distribution): Compute control dependences and remove
5942         restriction on number of loop nodes.
5944 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5946         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5947         for internal calls.
5949 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5951         * cse.c (try_const_anchors): Punt on CC modes.
5953 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5955         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5957 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5959         PR target/58382
5960         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5961         calls to word_mode.
5963 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5965         PR target/48094
5966         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5967         seen.
5968         (darwin_objc1_section): Likewise.
5969         (darwin_file_end): Emit Image Info section when required.
5971 2013-09-14  Jan Hubicka  <jh@suse.cz>
5973         * tree-into-ssa.c (gate_into_ssa): New.
5974         (pass_data_build_ssa): Use it.
5975         * cgraph.h (expand_thunk): Update prototype.
5976         * cgraphunit.c (analyze_function): Expand thunks early.
5977         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5978         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5979         set lowered flag; do not add new function.
5980         (assemble_thunks_and_aliases): Update.
5981         * tree-ssa.c (gate_init_datastructures): New gate.
5982         (pass_data_init_datastructures): Use it.
5984 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5986         PR target/58269
5987         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5988         xmm register set described in the psABI.
5990 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5992         * dwarf2out.c (should_emit_struct_debug): Add check
5993         for type_decl variable is not NULL.
5995 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5997         * config.gcc: Use new winnt-c.c target hooks
5998         * config/t-winnt: New file
5999         * config/winnt-c.c: New file
6000         * doc/tm.texi.in: Document new hook
6001         * doc/tm.texi: Regenerated
6003 2013-09-13  Jan Hubicka  <jh@suse.cz>
6005         PR middle-end/58094
6006         * ipa-inline.c (check_callers): New function.
6007         (check_caller_edge): Remove.
6008         (want_inline_function_to_all_callers_p): Also permit alises that are
6009         called dirrectly.
6010         (inline_to_all_callers): Terminate the walk when devirtualization
6011         introduce new calls.
6013 2013-09-13  Jan Hubicka  <jh@suse.cz>
6015         * ipa-inline-analysis.c (struct growth_data): Add node.
6016         (do_estimate_growth_1): Fix detection of recursion.
6018 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
6020         PR tree-optimization/58392
6021         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
6022         to avoid shadowing of outer loop variable.  If
6023         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
6024         replace_by_duplicate_decl simduid of loops that have it set and
6025         set dest_cfun->has_simduid_loops and/or
6026         dest_cfun->has_force_vect_loops.
6027         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
6028         instead of maybe_lookup_decl.
6029         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
6030         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
6031         call.  Copy over force_vect and copy and remap simduid.  Set
6032         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
6033         (copy_cfg_body): Remove blocks_to_copy argument.  Use
6034         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
6035         caller.  Don't set cfun->has_simduid_loops and/or
6036         cfun->has_force_vect_loops here.
6037         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
6038         caller.
6039         (expand_call_inline, tree_function_versioning): Adjust copy_body
6040         callers.
6042 2013-09-13  Martin Jambor  <mjambor@suse.cz>
6044         PR bootstrap/58388
6045         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
6046         the assert if the edge was a speculative one.
6048 2013-09-13  Richard Biener  <rguenther@suse.de>
6050         * tree-data-ref.h (known_dependences_p): Move here ...
6051         * tree-loop-distribution.c (known_dependences_p): ... from here.
6052         (dump_rdg_component, debug_rdg_component): Remove.
6053         (dump_rdg): Adjust.
6054         (generate_loops_for_partition): Use gimple_uid instead of
6055         relying on matching stmt visit order.
6056         (rdg_build_partitions): Take starting stmt vector.
6057         (ldist_gen): Merge into ...
6058         (distribute_loop): ... this function.  Do not compute starting
6059         vertices vector.
6060         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
6062 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6064         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
6065         Set type attribute correctly. Set predicable_short_it attribute.
6066         (cmpsi_shiftsi): Remove %? from output template.
6068 2013-09-13  Richard Biener  <rguenther@suse.de>
6070         * tree-loop-distribution.c (struct rdg_component,
6071         rdg_defs_used_in_other_loops_p, free_rdg_components,
6072         rdg_build_components): Remove.
6073         (stmts_from_loop): Do not record virtual PHIs.
6074         (generate_loops_for_partition): Skip virtual PHIs.
6075         (build_rdg_partition_for_component): Rename to ...
6076         (build_rdg_partition_for_vertex): ... this and adjust.
6077         (rdg_build_partitions): Take a vector of starting vertices
6078         instead of components.  Remove unnecessary leftover handling.
6079         (ldist_gen): Do not build components or record other stores.
6080         (distribute_loop): Do not distribute loops containing stmts
6081         with side-effects.
6083 2013-09-13  Christian Bruel  <christian.bruel@st.com>
6085         PR target/58314
6086         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
6088 2013-09-13  Kai Tietz  <ktietz@redhat.com>
6090         * config.gcc: Separate cases for mingw and cygwin targets,
6091         and add 64-bit cygwin target case.
6093         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
6094         dll-export inline-functions.
6095         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
6098 2013-09-13  Jeff Law  <law@redhat.com>
6100         PR middle-end/58387
6101         Revert:
6102         2013-09-06  Jeff Law  <law@redhat.com>
6104         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6105         edge implied equivalences into successor phis.
6107 2013-09-12  DJ Delorie  <dj@redhat.com>
6109         * config/rl78/rl78-virt.md: Change from | to \; for asm line
6110         separators.
6112 2013-09-12  Brooks Moses  <bmoses@google.com>
6114         PR driver/42955
6115         * Makefile.in: Do not install driver binaries in $(target)/bin.
6117 2013-09-12  DJ Delorie  <dj@redhat.com>
6119         * config/rl78/rl78.opt (mrelax): New.
6120         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
6121         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
6123         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
6124         between SP and FP.
6125         (rl78_expand_epilogue): Likewise.
6127 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
6129         PR middle-end/58335
6130         * lra-eliminations.c (remove_reg_equal_offset_note): New.
6131         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
6132         pointer elimination with using remove_reg_equal_offset_note.
6134 2013-09-12  DJ Delorie  <dj@redhat.com>
6136         * config/msp430/: New port.
6137         * config.gcc (msp430): Added.
6138         * doc/invoke.texi: Document MSP430 options.
6139         * doc/install.texi: Document msp430-elf
6140         * doc/md.texi: Document msp430-elf
6141         * doc/contrib.texi: Document msp430-elf
6143         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
6144         PSImode.
6146 2013-09-12  Martin Jambor  <mjambor@suse.cz>
6148         PR ipa/58389
6149         * ipa-prop.c (remove_described_reference): Give up if the edge in the
6150         reference descriptor is NULL.
6151         (ipa_edge_removal_hook): If owning a reference descriptor, set its
6152         edge to NULL.
6154 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
6156         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
6157         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
6158         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
6159         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
6160         make_temp_ssa_name): move to tree-ssanames.h
6161         * tree-ssa-alias.h: Move prototype.
6162         * tree-ssa.h: Include tree-ssanames.h.
6163         * tree-ssanames.c (FREE_SSANAMES): Move to here.
6164         * tree-ssanames.h: New.  Move items from tree-flow*.h
6165         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
6167 2013-09-12  Richard Biener  <rguenther@suse.de>
6169         PR tree-optimization/58404
6170         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
6171         propagate non-invariant addresses into dereferences wrapped
6172         in component references.
6174 2013-09-12  Richard Biener  <rguenther@suse.de>
6176         PR tree-optimization/58402
6177         * passes.def: Move pass_late_warn_uninitialized later.
6179 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
6181         * tree-ssa.h: New.  Move content from tree-flow.h and
6182         tree-flow-inline.h.
6183         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
6184         Move prototypes belonging to tree-ssa.c.
6185         * tree-flow-inline.h (redirect_edge_var_map_def,
6186         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
6187         tree-ssa.h.
6188         * gimple.h: Adjust prototypes.
6189         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
6190         to...
6191         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
6192         * tree.h: Move prototype to tree-ssa.h.
6193         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
6194         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
6195         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
6196         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
6197         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
6198         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
6199         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
6200         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
6201         graphite-blocking.c, graphite-clast-to-gimple.c,
6202         graphite-dependences.c, graphite-interchange.c,
6203         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
6204         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
6205         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
6206         ipa-reference.c, ipa-split.c, ipa-utils.c,
6207         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
6208         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
6209         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
6210         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
6211         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
6212         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
6213         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
6214         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
6215         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
6216         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
6217         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
6218         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
6219         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
6220         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
6221         value-prof.c, var-tracking.c,
6222         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
6224 2013-09-12  Richard Biener  <rguenther@suse.de>
6226         PR tree-optimization/58396
6227         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
6228         (build_rdg): Take a loop-nest parameter, fix memleaks.
6229         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
6231 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
6233         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
6234         for SLM.
6236 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
6238         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
6239         The FMA instruction names should have a 'v' prefix.
6241 2013-09-12  Richard Biener  <rguenther@suse.de>
6243         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
6244         (dot_rdg): Use popen instead of system in optional code.
6245         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
6246         (already_processed_vertex_p): Adjust.
6247         (has_anti_or_output_dependence, predecessor_has_mem_write,
6248         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
6249         rdg_flag_uses): Remove.
6250         (rdg_flag_vertex): Simplify.
6251         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
6252         remove recursion.
6253         (build_rdg_partition_for_component): Process the first vertex
6254         of a component only.
6255         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
6257 2013-09-12  Alan Modra  <amodra@gmail.com>
6259         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
6261 2013-09-11  DJ Delorie  <dj@redhat.com>
6262             Nick Clifton  <nickc@redhat.com>
6264         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
6265         (rl78_stack_based_mem): New.
6266         * config/rl78/constraints.md (Iv08): New.
6267         (Iv16): New.
6268         (Iv24): New.
6269         (Is09): New.
6270         (Is17): New.
6271         (Is25): New.
6272         (ISsi): New.
6273         (IShi): New.
6274         (ISqi): New.
6275         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
6276         (movhi): Likewise.
6277         (movsi): Change from expand to insn-and-split.
6278         (ashrsi3): Clobber AX.
6279         (lshrsi3): New.
6280         (ashlsi3): New.
6281         (cbranchsi4): New.
6282         * config/rl78/rl78.md (CC_REG): Fix.
6283         (addsi3): Allow memory and immediate operands.
6284         (addsi3_internal): Split into...
6285         (addsi3_internal_virt): ...new, and ...
6286         (addsi3_internal_real): ...new.
6287         (subsi): New.
6288         (subsi3_internal_virt): New.
6289         (subsi3_internal_real): New.
6290         (mulsi3): Add memory operand.
6291         (mulsi3_rl78): Likewise.
6292         (mulsi3_g13): Likewise.
6293         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
6294         (cbranchqi4_real): Add more constraint options.
6295         (cbranchhi4_real): Expand pattern.
6296         (cbranchhi4_real_signed): New.
6297         (cbranchhi4_real_inverted): New.
6298         (cbranchsi4_real_lt): New.
6299         (cbranchsi4_real_ge): New.
6300         (cbranchsi4_real_signed): New.
6301         (cbranchsi4_real): New.
6302         (peephole2): New.
6303         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
6304         constant shifts.
6305         (lshrsi3_virt): Likewise.
6306         (ashlsi3_virt): Likewise.
6307         (cbranchqi4_virt_signed): New.
6308         (cbranchhi4_virt_signed): New.
6309         (cbranchsi4_virt): New.
6310         * config/rl78/rl78.c: Whitespace fixes throughout.
6311         (move_elim_pass): New.
6312         (pass_data_rl78_move_elim): New.
6313         (pass_rl78_move_elim): New.
6314         (make_pass_rl78_move_elim): New.
6315         (rl78_devirt_info): Run devirt earlier.
6316         (rl78_move_elim_info): New.
6317         (rl78_asm_file_start): Register it.
6318         (rl78_split_movsi): New.
6319         (rl78_as_legitimate_address): Allow virtual base registers when
6320         appropriate.
6321         (rl78_addr_space_convert): Remove spurious debug stuff.
6322         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
6323         (rl78_print_operand): More cases for not printing '#'.
6324         (rl78_expand_compare): Remove most of the logic.
6325         (content_memory): New.
6326         (clear_content_memory): New.
6327         (get_content_index): New.
6328         (get_content_name): New.
6329         (display_content_memory): New.
6330         (update_content): New.
6331         (record_content): New.
6332         (already_contains): New.
6333         (insn_ok_now): Re-recog insns with virtual registers.
6334         (add_postponed_content_update): New.
6335         (process_postponed_content_update): New.
6336         (gen_and_emit_move): New.
6337         (transcode_memory_rtx): Record new location content.
6338         Use gen_and_emit_move.
6339         (force_into_acc): New.
6340         (move_to_acc): Use gen_and_emit_move.
6341         (move_from_acc): Likewise.
6342         (move_acc_to_reg): Likewise.
6343         (move_to_x): Likewise.
6344         (move_to_hl): Likewise.
6345         (move_to_de): Likewise.
6346         (rl78_alloc_physical_registers_op1): Record location content.
6347         (has_constraint): New.
6348         (rl78_alloc_physical_registers_op2): Record location content.
6349         Optimize use of HL.
6350         (rl78_alloc_physical_registers_ro1): Likewise.
6351         (rl78_alloc_physical_registers_cmp): Likewise.
6352         (rl78_alloc_physical_registers_umul): Likewise.
6353         (rl78_alloc_address_registers_macax): New.
6354         (rl78_alloc_physical_registers): Initialize and set location
6355         content memory as needed.
6356         (rl78_reorg): Make sure split2 is called.
6357         (rl78_rtx_costs): New.
6359 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
6361         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
6362         for (not (neg ...)) and (neg (not ...)) cases.
6364 2013-09-11  Richard Biener  <rguenther@suse.de>
6366         PR middle-end/58377
6367         * passes.def: Split critical edges before late uninit warning passes.
6368         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
6370 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
6372         PR tree-optimization/58385
6373         * fold-const.c (build_range_check): If both low and high are NULL,
6374         use omit_one_operand_loc to preserve exp side-effects.
6376 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6378         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
6380 2013-09-11  Richard Biener  <rguenther@suse.de>
6382         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
6383         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
6384         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
6385         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
6386         stmts_from_loop, known_dependences_p, build_empty_rdg,
6387         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
6388         * tree-loop-distribution.c: ... here.
6389         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
6390         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
6391         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
6392         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
6393         * tree-loop-distribution.c: ... here.
6394         * tree-loop-distribution.c: Include gimple-pretty-print.h.
6395         (struct partition_s): Add loops member.
6396         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
6397         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
6398         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
6400 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6401             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
6402             Sergey Lega  <sergey.s.lega@intel.com>
6403             Anna Tikhonova  <anna.tikhonova@intel.com>
6404             Ilya Tocar  <ilya.tocar@intel.com>
6405             Andrey Turetskiy  <andrey.turetskiy@intel.com>
6406             Ilya Verbin  <ilya.verbin@intel.com>
6407             Kirill Yukhin  <kirill.yukhin@intel.com>
6408             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
6410         * config/i386/constraints.md (k): New.
6411         (Yk): Ditto.
6412         * config/i386/i386.c (const regclass_map): Add new mask registers.
6413         (dbx_register_map): Ditto.
6414         (dbx64_register_map): Ditto.
6415         (svr4_dbx_register_map): Ditto.
6416         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
6417         disabled.
6418         (ix86_preferred_reload_class): Disable constants for mask registers.
6419         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
6420         (ix86_hard_regno_mode_ok): Support new mask registers.
6421         (x86_order_regs_for_local_alloc): Ditto.
6422         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
6423         (FIXED_REGISTERS): Add new mask registers.
6424         (CALL_USED_REGISTERS): Ditto.
6425         (REG_ALLOC_ORDER): Ditto.
6426         (VALID_MASK_REG_MODE): New.
6427         (FIRST_MASK_REG): Ditto.
6428         (LAST_MASK_REG): Ditto.
6429         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
6430         (MAYBE_MASK_CLASS_P): New.
6431         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
6432         (REG_CLASS_CONTENTS): Ditto.
6433         (MASK_REGNO_P): New.
6434         (ANY_MASK_REG_P): Ditto.
6435         (HI_REGISTER_NAMES): Add new mask registers.
6436         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
6437         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
6438         mask registers.
6439         (attribute "type"): Add mskmov, msklog.
6440         (attribute "length_immediate"): Support them.
6441         (attribute "memory"): Ditto.
6442         (attribute "prefix_0f"): Ditto.
6443         (*movhi_internal): Support new mask registers.
6444         (*movqi_internal): Ditto.
6445         (define_split): Split out clobber pattern is a logic
6446         insn on mask registers.
6447         (*k<logic><mode>): New.
6448         (*andhi_1): Extend to support mask regs.
6449         (*andqi_1): Extend to support mask regs.
6450         (kandn<mode>): New.
6451         (define_split): Split and-not to and and not if operands
6452         are not mask regs.
6453         (*<code><mode>_1): Separate HI mode to new pattern...
6454         (*<code>hi_1): This.
6455         (*<code>qi_1): Extend to support mask regs.
6456         (kxnor<mode>): New.
6457         (kortestzhi): Ditto.
6458         (kortestchi): Ditto.
6459         (kunpckhi): Ditto.
6460         (*one_cmpl<mode>2_1): Remove HImode and handle it...
6461         (*one_cmplhi2_1): ...Here, now with mask registers support.
6462         (*one_cmplqi2_1): Support new mask registers.
6463         (HI/QImode arithmetics splitter): Don't split if mask registers
6464         are used.
6465         (HI/QImode not splitter): Ditto.
6466         * config/i386/predicated.md (mask_reg_operand): New.
6467         (general_reg_operand): Ditto.
6469 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6471         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
6472         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
6474 2013-09-10  Jeff Law  <law@redhat.com>
6476         PR tree-optimization/58380
6477         * tree-ssa-threadupdate.c (thread_block): Recognize another case
6478         of threading through a buried loop header.
6480         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
6481         return value for single successor case.
6483 2013-09-10  Jan Hubicka  <jh@suse.cz>
6485         * ipa-devirt.c (ipa_devirt): Enable with LTO.
6487 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
6489         PR target/58361
6490         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
6491         support conditional execution.
6492         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
6494 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
6496         * lra.c (lra): Clear lra_optional_reload_pseudos before every
6497         constraint pass.
6498         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
6499         Check destination too to check move insn.
6500         (undo_optional_reloads): Add check that the original peudo did not
6501         changed its allocation and the optional reload was inherited on last
6502         inheritance pass.  Break loop after deciding to keep optional reload.
6503         (lra_undo_inheritance): Add check that inherited pseudo still in
6504         memory.
6506 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
6508         * config/aarch64/aarch64.md (generic_sched): New.
6509         * config/aarch64/aarch64-generic.md (load): Make conditional
6510         on generic_sched attribute.
6511         (nonload): Likewise.
6513 2013-09-10  Jan Hubicka  <jh@suse.cz>
6515         * lto-cgraph.c: Include ipa-utils.h.
6516         (compute_ltrans_boundary): Also add possible targets into the boundary.
6518 2013-09-10  Jan Hubicka  <jh@suse.cz>
6520         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
6521         VAR_DECL of vtable rather than full expression.
6523 2013-09-10  Jan Hubicka  <jh@suse.cz>
6524             Paolo Carlini  <paolo.carlini@oracle.com>
6526         * cgraphunit.c (analyze_functions): Save input_location, set it
6527         to UNKNOWN_LOCATION and restore it at the end.
6529 2013-09-10  Martin Jambor  <mjambor@suse.cz>
6531         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
6532         represented by a thunk.
6534 2013-09-10  Jeff Law  <law@redhat.com>
6536         PR tree-optimization/58343
6537         * tree-ssa-threadupdate.c (thread_block): Identify and disable
6538         jump threading requests through loop headers buried in the middle
6539         of a jump threading path.
6541         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
6542         in return value/type.
6544 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
6546         PR rtl-optimization/58365
6547         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
6548         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
6549         it differs.
6551 2013-09-10  Richard Biener  <rguenther@suse.de>
6553         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
6554         * tree-data-ref.c (create_rdg_vertices): Collect all data
6555         references, signal failure to the caller, use data-ref API.
6556         (build_rdg): Compute data references only once.  Maintain lifetime
6557         of data references and data dependences from within RDG.
6558         (free_rdg): Free dependence relations.
6559         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
6560         inventing extra dependences.
6561         (distribute_loop): Update for RDG API changes.
6563 2013-09-10  Kai Tietz  <ktietz@redhat.com>
6565         * doc/invoke.texi (fms-extensions): Document changed
6566         behavior for ms-abi targets.
6567         * config/i386/i386.c (ix86_option_override_internal):
6568         Set default value of option -fms-extension for ms-abi targets.
6570 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
6572         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
6574 2013-09-10  Alan Modra  <amodra@gmail.com>
6576         PR target/58330
6577         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
6579 2013-09-10  Alan Modra  <amodra@gmail.com>
6581         * config/rs6000/predicates.md (add_cint_operand): New.
6582         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
6583         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
6584         using add_cint_operand.
6585         (largetoc_high_plus_aix): Likewise.
6587 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
6589         PR tree-optimization/58364
6590         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
6591         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
6592         the current range can't be an unconditional true or false.
6594 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
6596         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
6598 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
6600         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
6602 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
6604         PR c++/43452
6605         * doc/invoke.texi (-Wdelete-incomplete): Document it.
6607 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
6609         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
6610         NO_REGS for immediate that can't be moved directly into FP_REGS.
6612 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6614         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6615         comparison with negated operand.
6616         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
6617         RTL form.
6619 2013-09-09  Richard Biener  <rguenther@suse.de>
6621         PR middle-end/58326
6622         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
6623         of a subloop record all its block as affecting loop-closed SSA form.
6625 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6627         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
6628         of an rtx/bitpos pair.
6629         (store_fixed_bit_field): Update accordingly.
6631 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6633         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
6634         GEN_INT.
6635         * builtins.c (expand_errno_check): Likewise.
6636         * dwarf2cfi.c (init_return_column_size): Likewise.
6637         * except.c (sjlj_mark_call_sites): Likewise.
6638         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6639         * lra-constraints.c (emit_inc): Likewise.
6640         * ree.c (combine_set_extension): Likewise.
6641         * regmove.c (fixup_match_2): Likewise.
6642         * reload1.c (inc_for_reload): Likewise.
6644 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6646         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6647         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6648         Use gen_int_mode with the mode of the associated simplify_* call.
6649         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6650         * expmed.c (expand_shift_1): Likewise.
6651         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6652         * loop-iv.c (iv_number_of_iterations): Likewise.
6653         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6654         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6656 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6658         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6659         of the associated expand_* call.
6660         (asan_emit_stack_protection): Likewise.
6661         * builtins.c (round_trampoline_addr): Likewise.
6662         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6663         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6664         (emit_store_flag): Likewise.
6665         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6666         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6667         Likewise.
6668         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6669         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6670         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6671         * modulo-sched.c (generate_prolog_epilog): Likewise.
6672         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6673         (expand_ctz, expand_ffs, expand_unop): Likewise.
6675 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6677         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6678         of the associated gen_rtx_* call.
6679         * caller-save.c (init_caller_save): Likewise.
6680         * combine.c (find_split_point, make_extraction): Likewise.
6681         (make_compound_operation): Likewise.
6682         * dwarf2out.c (mem_loc_descriptor): Likewise.
6683         * explow.c (plus_constant, probe_stack_range): Likewise.
6684         * expmed.c (expand_mult_const): Likewise.
6685         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6686         * reload1.c (init_reload): Likewise.
6687         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6688         * var-tracking.c (adjust_mems): Likewise.
6689         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6690         rather than gen_rtx_fmt_ee.
6692 2013-09-09  Jan Hubicka  <jh@suse.cz>
6694         PR middle-end/58294
6695         * value-prof.c (gimple_ic): Copy also abnormal edges.
6697 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6699         * asan.c (asan_shadow_cst): Use gen_int_mode.
6701 2013-09-08  Jan Hubicka  <jh@suse.cz>
6703         * ipa-profile.c: Add toplevel comment.
6704         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6705         (contains_hot_call_p): New function.
6706         (ipa_propagate_frequency): Set frequencies based on counts when
6707         profile is read.
6708         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6709         profile; do not tamper with profile after inlining if it is read.
6711 2013-09-08  Jan Hubicka  <jh@suse.cz>
6713         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6714         speculative edges.
6716 2013-09-08  Jan Hubicka  <jh@suse.cz>
6718         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6719         summary generation.
6721 2013-09-08  Jeff Law  <law@redhat.com>
6723         PR bootstrap/58340
6724         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6725         of 'found'.
6727 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6729         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6731 2013-09-08  Jan Hubicka  <jh@suse.cz>
6733         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6735 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6737         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6738         for non-debug insns.
6739         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6740         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6741         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6742         accordingly.
6744 2013-09-08  Jan Hubicka  <jh@suse.cz>
6746         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6747         targets and devirtualize to BUILT_IN_UNREACHABLE.
6748         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6749         * ipa.c (walk_polymorphic_call_targets): New function.
6750         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6751         functions; use the new timevar.
6752         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6753         was removed from the program.
6754         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6755         not consider it in the walk when its vtable is dead.
6756         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6757         record_binfo.
6758         (devirt_variable_node_removal_hook): New function.
6759         (possible_polymorphic_call_targets): Also register
6760         devirt_variable_node_removal_hook.
6761         (ipa_devirt): Do not do non-speculative devirtualization.
6762         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6764 2013-09-08  Jan Hubicka  <jh@suse.cz>
6766         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6767         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6768         varpool_remove_variable_insertion_hook): Declare.
6769         * varpool.c (varpool_node_hook_list): New structure.
6770         (first_varpool_node_removal_hook,
6771         first_varpool_variable_insertion_hook): New variables.
6772         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6773         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6774         varpool_remove_variable_insertion_hook,
6775         varpool_call_variable_insertion_hooks): New functions.
6776         (varpool_remove_node): Use it.
6778 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6780         PR c++/54941
6781         * diagnostic.c (diagnostic_build_prefix): When s.file is
6782         "<built-in>" don't output line and column numbers.
6784 2013-09-06  Jan Hubicka  <jh@suse.cz>
6786         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6787         * lto-streamer-out.c: Stream thunks, too.
6788         * lto-streamer-in.c (input_function): Pop cfun here
6789         (lto_read_body): Instead of here.
6791 2013-09-06  Caroline Tice  <cmtice@google.com>
6793         * doc/install.texi: Add documentation for the --enable-vtable-verify
6794         and the --disable-libvtv configure options.
6796 2013-09-06  Jeff Law  <law@redhat.com>
6798         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6799         edge implied equivalences into successor phis.
6801 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6803         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6805 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6807         * resource.c (mark_target_live_regs): Compute resources taking
6808         into account if a call is predicated or not.
6810 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6812         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6813         the compiler in the function names.
6815 2013-09-06  Jan Hubicka  <jh@suse.cz>
6817         PR middle-end/58094
6818         * ipa-inline.c (has_caller_p): New function.
6819         (want_inline_function_to_all_callers_p): Use it.
6820         (sum_callers, inline_to_all_callers): Break out from ...
6821         (ipa_inline): ... here.
6823 2013-09-06  Jan Hubicka  <jh@suse.cz>
6825         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6826         only when AVX is enabled.
6828 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6830         * config/aarch64/aarch64.md
6831         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6832         is fpsimd_<load/store>2.
6833         (load_pair<mode>): Likewise.
6834         (store_pair<mode>): Likewise.
6836 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6838         * config/arm/types.md (type): Add "mrs" type.
6839         * config/aarch64/aarch64.md
6840         (aarch64_load_tp_hard): Make type "mrs".
6841         * config/arm/arm.md
6842         (load_tp_hard): Make type "mrs".
6843         * config/arm/cortex-a15.md: Update with new attributes.
6844         * config/arm/cortex-a5.md: Update with new attributes.
6845         * config/arm/cortex-a53.md: Update with new attributes.
6846         * config/arm/cortex-a7.md: Update with new attributes.
6847         * config/arm/cortex-a8.md: Update with new attributes.
6848         * config/arm/cortex-a9.md: Update with new attributes.
6849         * config/arm/cortex-m4.md: Update with new attributes.
6850         * config/arm/cortex-r4.md: Update with new attributes.
6851         * config/arm/fa526.md: Update with new attributes.
6852         * config/arm/fa606te.md: Update with new attributes.
6853         * config/arm/fa626te.md: Update with new attributes.
6854         * config/arm/fa726te.md: Update with new attributes.
6856 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6858         * config/aarch64/aarch64.md
6859         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6860         (*movtf_aarch64): Likewise.
6861         * config/arm/arm.md
6862         (thumb1_movdi_insn): Use "multiple" for type where more than one
6863         instruction is used for a move.
6864         (*arm32_movhf): Likewise.
6865         (*thumb_movdf_insn): Likewise.
6867 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6869         * config/arm/types.md (type): Rename fcpys to fmov.
6870         * config/arm/vfp.md
6871         (*arm_movsi_vfp): Rename type fcpys as fmov.
6872         (*thumb2_movsi_vfp): Likewise
6873         (*movhf_vfp_neon): Likewise
6874         (*movhf_vfp): Likewise
6875         (*movsf_vfp): Likewise
6876         (*thumb2_movsf_vfp): Likewise
6877         (*movsfcc_vfp): Likewise
6878         (*thumb2_movsfcc_vfp): Likewise
6879         * config/aarch64/aarch64-simd.md
6880         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6881         * config/aarch64/aarch64.md
6882         (*movsi_aarch64): Replace type mov_reg with fmovs.
6883         (*movdi_aarch64): Likewise
6884         (*movsf_aarch64): Likewise
6885         (*movdf_aarch64): Likewise
6886         * config/arm/arm.c
6887         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6888         * config/arm/iwmmxt.md
6889         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6890         * config/arm/arm1020e.md: Update with new attributes.
6891         * config/arm/cortex-a15-neon.md: Update with new attributes.
6892         * config/arm/cortex-a5.md: Update with new attributes.
6893         * config/arm/cortex-a53.md: Update with new attributes.
6894         * config/arm/cortex-a7.md: Update with new attributes.
6895         * config/arm/cortex-a8-neon.md: Update with new attributes.
6896         * config/arm/cortex-a9.md: Update with new attributes.
6897         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6898         * config/arm/cortex-r4f.md: Update with new attributes.
6899         * config/arm/marvell-pj4.md: Update with new attributes.
6900         * config/arm/vfp11.md: Update with new attributes.
6902 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6904         * config/aarch64/aarch64.md
6905         (*madd<mode>): Fix type attribute.
6906         (*maddsi_uxtw): Likewise.
6907         (*msub<mode>): Likewise.
6908         (*msubsi_uxtw): Likewise.
6909         (<su_optab>maddsidi4): Likewise.
6910         (<su_optab>msubsidi4): Likewise.
6912 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6914         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6915         * config/arm/arm.md (core_cycles): Remove fdiv.
6916         * config/arm/vfp.md:
6917         (*sqrtsf2_vfp): Update for attribute changes.
6918         (*sqrtdf2_vfp): Likewise.
6919         * config/aarch64/aarch64.md:
6920         (sqrt<mode>2): Update for attribute changes.
6921         * config/arm/arm1020e.md: Update with new attributes.
6922         * config/arm/cortex-a15-neon.md: Update with new attributes.
6923         * config/arm/cortex-a5.md: Update with new attributes.
6924         * config/arm/cortex-a53.md: Update with new attributes.
6925         * config/arm/cortex-a7.md: Update with new attributes.
6926         * config/arm/cortex-a8-neon.md: Update with new attributes.
6927         * config/arm/cortex-a9.md: Update with new attributes.
6928         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6929         * config/arm/cortex-r4f.md: Update with new attributes.
6930         * config/arm/marvell-pj4.md: Update with new attributes.
6931         * config/arm/vfp11.md: Update with new attributes.
6933 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6935         * config/arm/types.md
6936         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6937         * config/aarch64/aarch64.md
6938         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6939         new attributes.
6940         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6941         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6942         (float<GPI:mode><GPF:mode>2): Likewise.
6943         * config/arm/vfp.md
6944         (*truncsisf2_vfp): Update with new attributes.
6945         (*truncsidf2_vfp): Likewise.
6946         (fixuns_truncsfsi2): Likewise.
6947         (fixuns_truncdfsi2): Likewise.
6948         (*floatsisf2_vfp): Likewise.
6949         (*floatsidf2_vfp): Likewise.
6950         (floatunssisf2): Likewise.
6951         (floatunssidf2): Likewise.
6952         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6953         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6954         * config/arm/arm1020e.md: Update with new attributes.
6955         * config/arm/cortex-a15-neon.md: Update with new attributes.
6956         * config/arm/cortex-a5.md: Update with new attributes.
6957         * config/arm/cortex-a53.md: Update with new attributes.
6958         * config/arm/cortex-a7.md: Update with new attributes.
6959         * config/arm/cortex-a8-neon.md: Update with new attributes.
6960         * config/arm/cortex-a9.md: Update with new attributes.
6961         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6962         * config/arm/cortex-r4f.md: Update with new attributes.
6963         * config/arm/marvell-pj4.md: Update with new attributes.
6964         * config/arm/vfp11.md: Update with new attributes.
6966 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6968         * config/aarch64/arm_neon.h
6969         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6970         (vqtbx<1,2,3,4><q>_s8): Likewise.
6972 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6974         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6975         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6976         (add<mode>3): Add type attribute.
6977         (add<mode>3): Likewise.
6978         (usadd<mode>3): Likewise.
6979         (ssadd<mode>3): Likewise.
6980         (sub<mode>3): Likewise.
6981         (sub<mode>3): Likewise.
6982         (ussub<mode>3): Likewise.
6983         (sssub<mode>3): Likewise.
6984         (ssmulsa3): Likewise.
6985         (usmulusa3): Likewise.
6986         (arm_usatsihi): Likewise.
6987         * config/arm/vfp.md
6988         (*movdi_vfp): Add types for all instructions.
6989         (*movdi_vfp_cortexa8): Likewise.
6990         (*movhf_vfp_neon): Likewise.
6991         (*movhf_vfp): Likewise.
6992         (*movdf_vfp): Likewise.
6993         (*thumb2_movdf_vfp): Likewise.
6994         (*thumb2_movdfcc_vfp): Likewise.
6995         * config/arm/arm.md: Add type attribute to all insn patterns.
6996         (*thumb1_adddi3): Add type attribute.
6997         (*arm_adddi3): Likewise.
6998         (*adddi_sesidi_di): Likewise.
6999         (*adddi_zesidi_di): Likewise.
7000         (*thumb1_addsi3): Likewise.
7001         (addsi3_compare0): Likewise.
7002         (*addsi3_compare0_scratch): Likewise.
7003         (*compare_negsi_si): Likewise.
7004         (cmpsi2_addneg): Likewise.
7005         (*addsi3_carryin_<optab>): Likewise.
7006         (*addsi3_carryin_alt2_<optab>): Likewise.
7007         (*addsi3_carryin_clobercc_<optab>): Likewise.
7008         (*subsi3_carryin): Likewise.
7009         (*subsi3_carryin_const): Likewise.
7010         (*subsi3_carryin_compare): Likewise.
7011         (*subsi3_carryin_compare_const): Likewise.
7012         (*arm_subdi3): Likewise.
7013         (*thumb_subdi3): Likewise.
7014         (*subdi_di_zesidi): Likewise.
7015         (*subdi_di_sesidi): Likewise.
7016         (*subdi_zesidi_di): Likewise.
7017         (*subdi_sesidi_di): Likewise.
7018         (*subdi_zesidi_ze): Likewise.
7019         (thumb1_subsi3_insn): Likewise.
7020         (*arm_subsi3_insn): Likewise.
7021         (*anddi3_insn): Likewise.
7022         (*anddi_zesidi_di): Likewise.
7023         (*anddi_sesdi_di): Likewise.
7024         (*ne_zeroextracts): Likewise.
7025         (*ne_zeroextracts): Likewise.
7026         (*ite_ne_zeroextr): Likewise.
7027         (*ite_ne_zeroextr): Likewise.
7028         (*anddi_notdi_di): Likewise.
7029         (*anddi_notzesidi): Likewise.
7030         (*anddi_notsesidi): Likewise.
7031         (andsi_notsi_si): Likewise.
7032         (thumb1_bicsi3): Likewise.
7033         (*iordi3_insn): Likewise.
7034         (*iordi_zesidi_di): Likewise.
7035         (*iordi_sesidi_di): Likewise.
7036         (*thumb1_iorsi3_insn): Likewise.
7037         (*xordi3_insn): Likewise.
7038         (*xordi_zesidi_di): Likewise.
7039         (*xordi_sesidi_di): Likewise.
7040         (*arm_xorsi3): Likewise.
7041         (*andsi_iorsi3_no): Likewise.
7042         (*smax_0): Likewise.
7043         (*smax_m1): Likewise.
7044         (*arm_smax_insn): Likewise.
7045         (*smin_0): Likewise.
7046         (*arm_smin_insn): Likewise.
7047         (*arm_umaxsi3): Likewise.
7048         (*arm_uminsi3): Likewise.
7049         (*minmax_arithsi): Likewise.
7050         (*minmax_arithsi_): Likewise.
7051         (*satsi_<SAT:code>): Likewise.
7052         (arm_ashldi3_1bit): Likewise.
7053         (arm_ashrdi3_1bit): Likewise.
7054         (arm_lshrdi3_1bit): Likewise.
7055         (*arm_negdi2): Likewise.
7056         (*thumb1_negdi2): Likewise.
7057         (*arm_negsi2): Likewise.
7058         (*thumb1_negsi2): Likewise.
7059         (*negdi_extendsid): Likewise.
7060         (*negdi_zero_extend): Likewise.
7061         (*arm_abssi2): Likewise.
7062         (*thumb1_abssi2): Likewise.
7063         (*arm_neg_abssi2): Likewise.
7064         (*thumb1_neg_abss): Likewise.
7065         (one_cmpldi2): Likewise.
7066         (extend<mode>di2): Likewise.
7067         (*compareqi_eq0): Likewise.
7068         (*arm_extendhisi2addsi): Likewise.
7069         (*arm_movdi): Likewise.
7070         (*thumb1_movdi_insn): Likewise.
7071         (*arm_movt): Likewise.
7072         (*thumb1_movsi_insn): Likewise.
7073         (pic_add_dot_plus_four): Likewise.
7074         (pic_add_dot_plus_eight): Likewise.
7075         (tls_load_dot_plus_eight): Likewise.
7076         (*thumb1_movhi_insn): Likewise.
7077         (*thumb1_movsf_insn): Likewise.
7078         (*movdf_soft_insn): Likewise.
7079         (*thumb_movdf_insn): Likewise.
7080         (cbranchsi4_insn): Likewise.
7081         (cbranchsi4_scratch): Likewise.
7082         (*negated_cbranchsi4): Likewise.
7083         (*tbit_cbranch): Likewise.
7084         (*tlobits_cbranch): Likewise.
7085         (*tstsi3_cbranch): Likewise.
7086         (*cbranchne_decr1): Likewise.
7087         (*addsi3_cbranch): Likewise.
7088         (*addsi3_cbranch_scratch): Likewise.
7089         (*arm_cmpdi_insn): Likewise.
7090         (*arm_cmpdi_unsig): Likewise.
7091         (*arm_cmpdi_zero): Likewise.
7092         (*thumb_cmpdi_zero): Likewise.
7093         (*deleted_compare): Likewise.
7094         (*mov_scc): Likewise.
7095         (*mov_negscc): Likewise.
7096         (*mov_notscc): Likewise.
7097         (*cstoresi_eq0_thumb1_insn): Likewise.
7098         (cstoresi_nltu_thumb1): Likewise.
7099         (cstoresi_ltu_thu): Likewise.
7100         (thumb1_addsi3_addgeu): Likewise.
7101         (*arm_jump): Likewise.
7102         (*thumb_jump): Likewise.
7103         (*check_arch2): Likewise.
7104         (arm_casesi_internal): Likewise.
7105         (thumb1_casesi_dispatch): Likewise.
7106         (*arm_indirect_jump): Likewise.
7107         (*thumb1_indirect_jump): Likewise.
7108         (nop): Likewise.
7109         (*and_scc): Likewise.
7110         (*ior_scc): Likewise.
7111         (*compare_scc): Likewise.
7112         (*cond_move): Likewise.
7113         (*cond_arith): Likewise.
7114         (*cond_sub): Likewise.
7115         (*cmp_ite0): Likewise.
7116         (*cmp_ite1): Likewise.
7117         (*cmp_and): Likewise.
7118         (*cmp_ior): Likewise.
7119         (*ior_scc_scc): Likewise.
7120         (*ior_scc_scc_cmp): Likewise.
7121         (*and_scc_scc): Likewise.
7122         (*and_scc_scc_cmp): Likewise.
7123         (*and_scc_scc_nod): Likewise.
7124         (*negscc): Likewise.
7125         (movcond_addsi): Likewise.
7126         (movcond): Likewise.
7127         (*ifcompare_plus_move): Likewise.
7128         (*if_plus_move): Likewise.
7129         (*ifcompare_move_plus): Likewise.
7130         (*if_move_plus): Likewise.
7131         (*ifcompare_arith_arith): Likewise.
7132         (*if_arith_arith): Likewise.
7133         (*ifcompare_arith_move): Likewise.
7134         (*if_arith_move): Likewise.
7135         (*ifcompare_move_arith): Likewise.
7136         (*if_move_arith): Likewise.
7137         (*ifcompare_move_not): Likewise.
7138         (*if_move_not): Likewise.
7139         (*ifcompare_not_move): Likewise.
7140         (*if_not_move): Likewise.
7141         (*ifcompare_shift_move): Likewise.
7142         (*if_shift_move): Likewise.
7143         (*ifcompare_move_shift): Likewise.
7144         (*if_move_shift): Likewise.
7145         (*ifcompare_shift_shift): Likewise.
7146         (*ifcompare_not_arith): Likewise.
7147         (*ifcompare_arith_not): Likewise.
7148         (*if_arith_not): Likewise.
7149         (*ifcompare_neg_move): Likewise.
7150         (*if_neg_move): Likewise.
7151         (*ifcompare_move_neg): Likewise.
7152         (*if_move_neg): Likewise.
7153         (prologue_thumb1_interwork): Likewise.
7154         (*cond_move_not): Likewise.
7155         (*sign_extract_onebit): Likewise.
7156         (*not_signextract_onebit): Likewise.
7157         (stack_tie): Likewise.
7158         (align_4): Likewise.
7159         (align_8): Likewise.
7160         (consttable_end): Likewise.
7161         (consttable_1): Likewise.
7162         (consttable_2): Likewise.
7163         (consttable_4): Likewise.
7164         (consttable_8): Likewise.
7165         (consttable_16): Likewise.
7166         (*thumb1_tablejump): Likewise.
7167         (prefetch): Likewise.
7168         (force_register_use): Likewise.
7169         (thumb_eh_return): Likewise.
7170         (load_tp_hard): Likewise.
7171         (load_tp_soft): Likewise.
7172         (tlscall): Likewise.
7173         (*arm_movtas_ze): Likewise.
7174         (*arm_rev): Likewise.
7175         (*arm_revsh): Likewise.
7176         (*arm_rev16): Likewise.
7177         * config/arm/thumb2.md
7178         (*thumb2_smaxsi3): Likewise.
7179         (*thumb2_sminsi3): Likewise.
7180         (*thumb32_umaxsi3): Likewise.
7181         (*thumb2_uminsi3): Likewise.
7182         (*thumb2_negdi2): Likewise.
7183         (*thumb2_abssi2): Likewise.
7184         (*thumb2_neg_abss): Likewise.
7185         (*thumb2_movsi_insn): Likewise.
7186         (tls_load_dot_plus_four): Likewise.
7187         (*thumb2_movhi_insn): Likewise.
7188         (*thumb2_mov_scc): Likewise.
7189         (*thumb2_mov_negs): Likewise.
7190         (*thumb2_mov_negs): Likewise.
7191         (*thumb2_mov_nots): Likewise.
7192         (*thumb2_mov_nots): Likewise.
7193         (*thumb2_movsicc_): Likewise.
7194         (*thumb2_movsfcc_soft_insn): Likewise.
7195         (*thumb2_indirect_jump): Likewise.
7196         (*thumb2_and_scc): Likewise.
7197         (*thumb2_ior_scc): Likewise.
7198         (*thumb2_ior_scc_strict_it): Likewise.
7199         (*thumb2_cond_move): Likewise.
7200         (*thumb2_cond_arith): Likewise.
7201         (*thumb2_cond_ari): Likewise.
7202         (*thumb2_cond_sub): Likewise.
7203         (*thumb2_negscc): Likewise.
7204         (*thumb2_movcond): Likewise.
7205         (thumb2_casesi_internal): Likewise.
7206         (thumb2_casesi_internal_pic): Likewise.
7207         (*thumb2_alusi3_short): Likewise.
7208         (*thumb2_mov<mode>_shortim): Likewise.
7209         (*thumb2_addsi_short): Likewise.
7210         (*thumb2_subsi_short): Likewise.
7211         (thumb2_addsi3_compare0): Likewise.
7212         (*thumb2_cbz): Likewise.
7213         (*thumb2_cbnz): Likewise.
7214         (*thumb2_one_cmplsi2_short): Likewise.
7215         (*thumb2_negsi2_short): Likewise.
7216         (*orsi_notsi_si): Likewise.
7217         * config/arm/arm1020e.md: Update with new attributes.
7218         * config/arm/arm1026ejs.md: Update with new attributes.
7219         * config/arm/arm1136jfs.md: Update with new attributes.
7220         * config/arm/arm926ejs.md: Update with new attributes.
7221         * config/arm/cortex-a15.md: Update with new attributes.
7222         * config/arm/cortex-a5.md: Update with new attributes.
7223         * config/arm/cortex-a53.md: Update with new attributes.
7224         * config/arm/cortex-a7.md: Update with new attributes.
7225         * config/arm/cortex-a8.md: Update with new attributes.
7226         * config/arm/cortex-a9.md: Update with new attributes.
7227         * config/arm/cortex-m4.md: Update with new attributes.
7228         * config/arm/cortex-r4.md: Update with new attributes.
7229         * config/arm/fa526.md: Update with new attributes.
7230         * config/arm/fa606te.md: Update with new attributes.
7231         * config/arm/fa626te.md: Update with new attributes.
7232         * config/arm/fa726te.md: Update with new attributes.
7234 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
7236         * config/aarch64/aarch64-simd.md
7237         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
7238         <vwx> iterator to ensure correct register choice.
7239         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
7240         (aarch64_sqdmull_n<mode>): Likewise.
7241         (aarch64_sqdmull2_n<mode>_internal): Likewise.
7242         * config/aarch64/arm_neon.h
7243         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
7244         (vml<as><q>_n_<su>16): Likewise.
7245         (vml<as>l_high_lane<q>_<su>16): Likewise.
7246         (vml<as>l_high_n_<su>16): Likewise.
7247         (vml<as>l_lane<q>_<su>16): Likewise.
7248         (vml<as>l_n_<su>16): Likewise.
7249         (vmul<q>_lane<q>_<su>16): Likewise.
7250         (vmul<q>_n_<su>16): Likewise.
7251         (vmull_lane<q>_<su>16): Likewise.
7252         (vmull_n_<su>16): Likewise.
7253         (vmull_high_lane<q>_<su>16): Likewise.
7254         (vmull_high_n_<su>16): Likewise.
7255         (vqrdmulh<q>_n_s16): Likewise.
7257 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
7259         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
7260         have the correct lane parameter.
7262 2013-09-06  Richard Biener <rguenther@suse.de>
7264         * cfganal.c (control_dependences::~control_dependences):
7265         Properly free all of the vector.
7267 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
7269         PR target/58269
7270         * config/i386/i386.c (ix86_conditional_register_usage):
7271         Proper initialize extended SSE registers.
7273 2013-09-06  Jan Hubicka  <jh@suse.cz>
7275         PR tree-optimization/58311
7276         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
7278 2013-09-06  Jan Hubicka  <jh@suse.cz>
7280         * Makefile.in (tree-sra.o): Update dependencies.
7281         * tree-sra.c: Include ipa-utils.h
7282         (scan_function): Use recursive_call_p.
7283         (has_caller_p): New function.
7284         (cgraph_for_node_and_aliases): Count also callers of aliases.
7286 2013-09-06  Jan Hubicka  <jh@suse.cz>
7288         PR middle-end/58094
7289         * cgraph.h (symtab_semantically_equivalent_p): Declare.
7290         * tree-tailcall.c: Include ipa-utils.h.
7291         (find_tail_calls): Use it.
7292         * ipa-pure-const.c (check_call): Likewise.
7293         * ipa-utils.c (recursive_call_p): New function.
7294         * ipa-utils.h (recursive_call_p): Dclare.
7295         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
7296         (symtab_semantically_equivalent_p): New function.
7297         * Makefile.in (tree-tailcall.o): Update dependencies.
7299 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
7301         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
7302         non-inlinable part.
7304 2013-09-06  Richard Biener  <rguenther@suse.de>
7306         * lto-streamer.h (lto_global_var_decls): Remove.
7307         * Makefile.in (OBJS): Remove lto-symtab.o.
7308         (lto-symtab.o): Remove.
7309         (GTFILES): Remove lto-symtab.c
7310         * lto-symtab.c: Move to lto/
7312 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7314         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
7315         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
7316         (UNSPEC_FPINT_RINT): New constant definitions.
7317         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
7318         definition with 2 attributes.
7319         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
7320         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
7321         definitions.
7323 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7325         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
7326         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
7327         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
7328         attribute to "z196_alone".
7329         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
7330         "zEC12_simple".
7332 2013-09-06  Richard Biener  <rguenther@suse.de>
7334         * basic-block.h (class control_dependences): New.
7335         * tree-ssa-dce.c (control_dependence_map): Remove.
7336         (cd): New global.
7337         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
7338         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
7339         find_pdom, find_control_dependence, find_all_control_dependences):
7340         Move to cfganal.c.
7341         (mark_control_dependent_edges_necessary,
7342         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
7343         tree_dce_done, perform_tree_ssa_dce): Adjust.
7344         * cfganal.c (set_control_dependence_map_bit,
7345         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
7346         find_all_control_dependences): Move from tree-ssa-dce.c and
7347         implement as methods of control_dependences class.
7348         (control_dependences::control_dependences): New.
7349         (control_dependences::~control_dependences): Likewise.
7350         (control_dependences::get_edges_dependent_on): Likewise.
7351         (control_dependences::get_edge): Likewise.
7353 2013-09-04  Jan Hubicka  <jh@suse.cz>
7355         * tree.c (types_same_for_odr): Drop overactive check.
7356         * ipa-devirt.c (hash_type_name): Likewise.
7358 2013-09-04  Jan Hubicka  <jh@suse.cz>
7360         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
7361         (analyze_functions): ... here.
7363 2013-09-04  Jan Hubicka  <jh@suse.cz>
7365         PR middle-end/58201
7366         * cgraphunit.c (analyze_functions): Clear AUX fields
7367         after processing; initialize assembler name has.
7369 2013-09-05  Jeff Law  <law@redhat.com>
7371         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
7372         from thread_around_empty_block.  Record threading path into PATH.
7373         Recurse if threading through the initial block is successful.
7374         (thread_across_edge): Corresponding changes to slightly simplify.
7376 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7378         * config/aarch64/aarch64.md
7379         (type): Remove frecpe, frecps, frecpx.
7380         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
7381         fix to be a TARGET_SIMD instruction.
7382         (aarch64_frecps): Remove.
7383         * config/aarch64/aarch64-simd.md
7384         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
7385         (aarch64_frecps<mode>): Handle all float/vector of float modes.
7387 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7388             Sofiane Naci  <sofiane.naci@arm.com>
7390         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
7391         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
7392         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
7393         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
7394         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
7395         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
7396         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
7397         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
7398         into "clz, "rbit".  Rename "shift" to "shift_imm".
7399         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
7400         changes.  Update for attribute changes all occurrences of arlo_* and
7401         shift* types.
7402         * config/arm/arm-fixed.md: Update for attribute changes
7403         all occurrences of arlo_* types.
7404         * config/arm/thumb2.md: Update for attribute changes all occurrences
7405         of arlo_* types.
7406         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
7407         (cortexa7_older_only): Likewise.
7408         (cortexa7_younger):  Likewise.
7409         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7410         (1020alu_shift_op): Likewise.
7411         (1020alu_shift_reg_op): Likewise.
7412         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7413         (alu_shift_op): Likewise.
7414         (alu_shift_reg_op): Likewise.
7415         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7416         (11_alu_shift_op): Likewise.
7417         (11_alu_shift_reg_op): Likewise.
7418         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7419         (9_alu_shift_reg_op): Likewise.
7420         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
7421         attribute changes.
7422         (cortex_a15_alu_shift): Likewise.
7423         (cortex_a15_alu_shift_reg): Likewise.
7424         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
7425         attribute changes.
7426         (cortex_a5_alu_shift): Likewise.
7427         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
7428         attribute changes.
7429         (cortex_a53_alu_shift): Likewise.
7430         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
7431         attribute changes.
7432         (cortex_a7_alu_reg): Likewise.
7433         (cortex_a7_alu_shift): Likewise.
7434         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
7435         attribute changes.
7436         (cortex_a8_alu_shift): Likewise.
7437         (cortex_a8_alu_shift_reg): Likewise.
7438         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7439         (cortex_a9_dp_shift): Likewise.
7440         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
7441         attribute changes.
7442         * config/arm/cortex-r4.md
7443         (cortex_r4_alu): Update for attribute changes.
7444         (cortex_r4_mov): Likewise.
7445         (cortex_r4_alu_shift_reg): Likewise.
7446         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7447         (526_alu_shift_op): Likewise.
7448         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7449         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7450         (626te_alu_shift_op): Likewise.
7451         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
7452         (726te_alu_shift_op): Likewise.
7453         (726te_alu_shift_reg_op): Likewise.
7454         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7455         (mp626_alu_shift_op): Likewise.
7456         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
7457         (pj4_alu_conds): Likewise.
7458         (pj4_shift): Likewise.
7459         (pj4_shift_conds): Likewise.
7460         (pj4_alu_shift): Likewise.
7461         (pj4_alu_shift_conds): Likewise.
7462         * config/aarch64/aarch64.md: Update for attribute change
7463         all occurrences of arlo_* and shift* types.
7465 2013-09-05  Mike Stump  <mikestump@comcast.net>
7467         * tree.h: Move documentation for tree_function_decl to tree-core.h
7468         with the declaration.
7470 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
7472         PR target/58139
7473         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
7474         looking for widest mode.
7476 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
7478         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
7480 2013-09-05  Richard Biener  <rguenther@suse.de>
7482         PR tree-optimization/58137
7483         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
7484         Do not create vectors of pointers.
7485         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
7486         types for the components of the vector initializer.
7487         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
7488         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
7490 2013-09-05  Martin Jambor  <mjambor@suse.cz>
7492         * ipa-prop.c (remove_described_reference): Accept missing references,
7493         return false if that hppens, otherwise return true.
7494         (cgraph_node_for_jfunc): New function.
7495         (try_decrement_rdesc_refcount): Likewise.
7496         (try_make_edge_direct_simple_call): Use them.
7497         (ipa_edge_removal_hook): Remove references from rdescs.
7498         (ipa_edge_duplication_hook): Clone rdescs and their references
7499         when the new edge has the same caller as the old one.
7500         * cgraph.c (cgraph_resolve_speculation): Remove speculative
7501         reference before removing any edges.
7503 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
7505         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
7506         initial store.
7507         * thumb2.md (thumb2_storewb_parisi): New pattern.
7509 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
7511         * config/aarch64/aarch64-option-extensions.def: Add
7512         AARCH64_OPT_EXTENSION of 'crc'.
7513         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
7514         (AARCH64_ISA_CRC): Ditto.
7515         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
7516         description of the CRC extension.
7518 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7520         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
7521         * config/rs6000/linux.h: Ditto.
7522         * alpha/linux.h: Ditto.
7523         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
7524         no_c99_libc_has_function.
7525         * config/c6x/uclinux-elf.h: Ditto.
7526         * config/lm32/uclinux-elf.h: Ditto.
7527         * config/m68k/uclinux.h: Ditto.
7528         * config/moxie/uclinux.h: Ditto.
7529         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
7530         (crisv32-*-linux*, cris-*-linux*): Ditto.
7531         * config/bfin/bfin.c: Include "tm_p.h".
7533 2013-09-05  Richard Biener  <rguenther@suse.de>
7535         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
7536         check for a definition without a basic-block.
7538 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
7539             Sofiane Naci <sofiane.naci@arm.com>
7541         * config/aarch64/aarch64.md
7542         (*movti_aarch64): Rename r_2_f and f_2_r.
7543         (*movsf_aarch64): Likewise.
7544         (*movdf_aarch64): Likewise.
7545         (*movtf_aarch64): Likewise.
7546         (aarch64_movdi_<mode>low): Likewise.
7547         (aarch64_movdi_<mode>high): Likewise.
7548         (aarch64_mov<mode>high_di): Likewise.
7549         (aarch64_mov<mode>low_di): Likewise.
7550         (aarch64_movtilow_tilow): Likewise.
7551         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
7552         values to config/arm/types.md
7553         (attribute "conds"): Update for attribute change.
7554         (anddi3_insn): Likewise.
7555         (iordi3_insn): Likewise.
7556         (xordi3_insn): Likewise.
7557         (one_cmpldi2): Likewise.
7558         * config/arm/types.md (type): Add Neon types.
7559         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
7560         use "type" attribute.
7561         (movmisalign<mode>_neon_store): Likewise.
7562         (movmisalign<mode>_neon_load): Likewise.
7563         (vec_set<mode>_internal): Likewise.
7564         (vec_setv2di_internal): Likewise.
7565         (vec_extract<mode>): Likewise.
7566         (vec_extractv2di): Likewise.
7567         (add<mode>3_neon): Likewise.
7568         (adddi3_neon): Likewise.
7569         (sub<mode>3_neon): Likewise.
7570         (subdi3_neon): Likewise.
7571         (mul<mode>3_neon): Likewise.
7572         (mul<mode>3add<mode>_neon): Likewise.
7573         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
7574         (fma<VCVTF:mode>4)): Likewise.
7575         (fma<VCVTF:mode>4_intrinsic): Likewise.
7576         (fmsub<VCVTF:mode>4)): Likewise.
7577         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
7578         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
7579         (ior<mode>3): Likewise.
7580         (and<mode>3): Likewise.
7581         (anddi3_neon): Likewise.
7582         (orn<mode>3_neon): Likewise.
7583         (orndi3_neon): Likewise.
7584         (bic<mode>3_neon): Likewise.
7585         (bicdi3_neon): Likewise.
7586         (xor<mode>3): Likewise.
7587         (one_cmpl<mode>2): Likewise.
7588         (abs<mode>2): Likewise.
7589         (neg<mode>2): Likewise.
7590         (umin<mode>3_neon): Likewise.
7591         (umax<mode>3_neon): Likewise.
7592         (smin<mode>3_neon): Likewise.
7593         (smax<mode>3_neon): Likewise.
7594         (vashl<mode>3): Likewise.
7595         (vashr<mode>3_imm): Likewise.
7596         (vlshr<mode>3_imm): Likewise.
7597         (ashl<mode>3_signed): Likewise.
7598         (ashl<mode>3_unsigned): Likewise.
7599         (neon_load_count): Likewise.
7600         (ashldi3_neon_noclobber): Likewise.
7601         (signed_shift_di3_neon): Likewise.
7602         (unsigned_shift_di3_neon): Likewise.
7603         (ashrdi3_neon_imm_noclobber): Likewise.
7604         (lshrdi3_neon_imm_noclobber): Likewise.
7605         (widen_ssum<mode>3): Likewise.
7606         (widen_usum<mode>3): Likewise.
7607         (quad_halves_<code>v4si): Likewise.
7608         (quad_halves_<code>v4sf): Likewise.
7609         (quad_halves_<code>v8hi): Likewise.
7610         (quad_halves_<code>v16qi): Likewise.
7611         (reduc_splus_v2di): Likewise.
7612         (neon_vpadd_internal<mode>): Likewise.
7613         (neon_vpsmin<mode>): Likewise.
7614         (neon_vpsmax<mode>): Likewise.
7615         (neon_vpumin<mode>): Likewise.
7616         (neon_vpumax<mode>): Likewise.
7617         (ss_add<mode>_neon): Likewise.
7618         (us_add<mode>_neon): Likewise.
7619         (ss_sub<mode>_neon): Likewise.
7620         (us_sub<mode>_neon): Likewise.
7621         (neon_vadd<mode>_unspec): Likewise.
7622         (neon_vaddl<mode>): Likewise.
7623         (neon_vaddw<mode>): Likewise.
7624         (neon_vhadd<mode>): Likewise.
7625         (neon_vqadd<mode>): Likewise.
7626         (neon_vaddhn<mode>): Likewise.
7627         (neon_vmul<mode>): Likewise.
7628         (neon_vmla<mode>): Likewise.
7629         (neon_vmlal<mode>): Likewise.
7630         (neon_vmls<mode>): Likewise.
7631         (neon_vmlsl<mode>): Likewise.
7632         (neon_vqdmulh<mode>): Likewise.
7633         (neon_vqdmlal<mode>): Likewise.
7634         (neon_vqdmlsl<mode>): Likewise.
7635         (neon_vmull<mode>): Likewise.
7636         (neon_vqdmull<mode>): Likewise.
7637         (neon_vsub<mode>_unspec): Likewise.
7638         (neon_vsubl<mode>): Likewise.
7639         (neon_vsubw<mode>): Likewise.
7640         (neon_vqsub<mode>): Likewise.
7641         (neon_vhsub<mode>): Likewise.
7642         (neon_vsubhn<mode>): Likewise.
7643         (neon_vceq<mode>): Likewise.
7644         (neon_vcge<mode>): Likewise.
7645         (neon_vcgeu<mode>): Likewise.
7646         (neon_vcgt<mode>): Likewise.
7647         (neon_vcgtu<mode>): Likewise.
7648         (neon_vcle<mode>): Likewise.
7649         (neon_vclt<mode>): Likewise.
7650         (neon_vcage<mode>): Likewise.
7651         (neon_vcagt<mode>): Likewise.
7652         (neon_vtst<mode>): Likewise.
7653         (neon_vabd<mode>): Likewise.
7654         (neon_vabdl<mode>): Likewise.
7655         (neon_vaba<mode>): Likewise.
7656         (neon_vabal<mode>): Likewise.
7657         (neon_vmax<mode>): Likewise.
7658         (neon_vmin<mode>): Likewise.
7659         (neon_vpaddl<mode>): Likewise.
7660         (neon_vpadal<mode>): Likewise.
7661         (neon_vpmax<mode>): Likewise.
7662         (neon_vpmin<mode>): Likewise.
7663         (neon_vrecps<mode>): Likewise.
7664         (neon_vrsqrts<mode>): Likewise.
7665         (neon_vqabs<mode>): Likewise.
7666         (neon_vqneg<mode>): Likewise.
7667         (neon_vcls<mode>): Likewise.
7668         (clz<mode>2): Likewise.
7669         (popcount<mode>2): Likewise.
7670         (neon_vrecpe): Likewise.
7671         (neon_vrsqrte): Likewise.
7672         (neon_vget_lane<mode>_sext_internal): Likewise.
7673         (neon_vget_lane<mode>_zext_internal): Likewise.
7674         (neon_vdup_n<mode>): Likewise.
7675         (neon_vdup_nv2di): Likewise.
7676         (neon_vdpu_lane<mode>_internal): Likewise.
7677         (neon_vswp<mode>): Likewise.
7678         (float<mode><V_cvtto>2): Likewise.
7679         (floatuns<mode><V_cvtto>2): Likewise.
7680         (fix_trunc<mode><V_cvtto>)2): Likewise
7681         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7682         (neon_vcvt<mode>): Likewise.
7683         (neon_vcvtv4sfv4hf): Likewise.
7684         (neon_vcvtv4hfv4sf): Likewise.
7685         (neon_vcvt_n<mode>): Likewise.
7686         (neon_vmovn<mode>): Likewise.
7687         (neon_vqmovn<mode>): Likewise.
7688         (neon_vqmovun<mode>): Likewise.
7689         (neon_vmovl<mode>): Likewise.
7690         (neon_vmul_lane<mode>): Likewise.
7691         (neon_vmull_lane<mode>): Likewise.
7692         (neon_vqdmull_lane<mode>): Likewise.
7693         (neon_vqdmulh_lane<mode>): Likewise.
7694         (neon_vmla_lane<mode>): Likewise.
7695         (neon_vmlal_lane<mode>): Likewise.
7696         (neon_vqdmlal_lane<mode>): Likewise.
7697         (neon_vmls_lane<mode>): Likewise.
7698         (neon_vmlsl_lane<mode>): Likewise.
7699         (neon_vqdmlsl_lane<mode>): Likewise.
7700         (neon_vext<mode>): Likewise.
7701         (neon_vrev64<mode>): Likewise.
7702         (neon_vrev32<mode>): Likewise.
7703         (neon_vrev16<mode>): Likewise.
7704         (neon_vbsl<mode>_internal): Likewise.
7705         (neon_vshl<mode>): Likewise.
7706         (neon_vqshl<mode>): Likewise.
7707         (neon_vshr_n<mode>): Likewise.
7708         (neon_vshrn_n<mode>): Likewise.
7709         (neon_vqshrn_n<mode>): Likewise.
7710         (neon_vqshrun_n<mode>): Likewise.
7711         (neon_vshl_n<mode>): Likewise.
7712         (neon_vqshl_n<mode>): Likewise.
7713         (neon_vqshlu_n<mode>): Likewise.
7714         (neon_vshll_n<mode>): Likewise.
7715         (neon_vsra_n<mode>): Likewise.
7716         (neon_vsri_n<mode>): Likewise.
7717         (neon_vsli_n<mode>): Likewise.
7718         (neon_vtbl1v8qi): Likewise.
7719         (neon_vtbl2v8qi): Likewise.
7720         (neon_vtbl3v8qi): Likewise.
7721         (neon_vtbl4v8qi): Likewise.
7722         (neon_vtbx1v8qi): Likewise.
7723         (neon_vtbx2v8qi): Likewise.
7724         (neon_vtbx3v8qi): Likewise.
7725         (neon_vtbx4v8qi): Likewise.
7726         (neon_vtrn<mode>_internal): Likewise.
7727         (neon_vzip<mode>_internal): Likewise.
7728         (neon_vuzp<mode>_internal): Likewise.
7729         (neon_vld1<mode>): Likewise.
7730         (neon_vld1_lane<mode>): Likewise.
7731         (neon_vld1_dup<mode>): Likewise.
7732         (neon_vld1_dupv2di): Likewise.
7733         (neon_vst1<mode>): Likewise.
7734         (neon_vst1_lane<mode>): Likewise.
7735         (neon_vld2<mode>): Likewise.
7736         (neon_vld2_lane<mode>): Likewise.
7737         (neon_vld2_dup<mode>): Likewise.
7738         (neon_vst2<mode>): Likewise.
7739         (neon_vst2_lane<mode>): Likewise.
7740         (neon_vld3<mode>): Likewise.
7741         (neon_vld3qa<mode>): Likewise.
7742         (neon_vld3qb<mode>): Likewise.
7743         (neon_vld3_lane<mode>): Likewise.
7744         (neon_vld3_dup<mode>): Likewise.
7745         (neon_vst3<mode>): Likewise.
7746         (neon_vst3qa<mode>): Likewise.
7747         (neon_vst3qb<mode>): Likewise.
7748         (neon_vst3_lane<mode>): Likewise.
7749         (neon_vld4<mode>): Likewise.
7750         (neon_vld4qa<mode>): Likewise.
7751         (neon_vld4qb<mode>): Likewise.
7752         (neon_vld4_lane<mode>): Likewise.
7753         (neon_vld4_dup<mode>): Likewise.
7754         (neon_vst4<mode>): Likewise.
7755         (neon_vst4qa<mode>): Likewise.
7756         (neon_vst4qb<mode>): Likewise.
7757         (neon_vst4_lane<mode>): Likewise.
7758         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7759         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7760         (neon_vec_<US>mult_lo_<mode>): Likewise.
7761         (neon_vec_<US>mult_hi_<mode>): Likewise.
7762         (neon_vec_<US>shiftl_<mode>): Likewise.
7763         (neon_unpack<US>_<mode>): Likewise.
7764         (neon_vec_<US>mult_<mode>): Likewise.
7765         (vec_pack_trunc_<mode>): Likewise.
7766         (neon_vec_pack_trunk_<mode>): Likewise.
7767         (neon_vabd<mode>_2): Likewise.
7768         (neon_vabd<mode>_3): Likewise.
7769         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7770         (thumb2_movsi_vfp): Likewise.
7771         (movdi_vfp): Likewise.
7772         (movdi_vfp_cortexa8): Likewise.
7773         (movhf_vfp_neon): Likewise.
7774         (movhf_vfp): Likewiwse.
7775         (movsf_vfp): Likewiwse.
7776         (thumb2_movsf_vfp): Likewiwse.
7777         (movdf_vfp): Likewise.
7778         (thumb2_movdf_vfp): Likewise.
7779         (movsfcc_vfp): Likewise.
7780         (thumb2_movsfcc_vfp): Likewise.
7781         (movdfcc_vfp): Likewise.
7782         (thumb2_movdfcc_vfp): Likewise.
7783         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7784         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7785         (v10_v2c): Likewise.
7786         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7787         attribute change.
7788         (cortex_a15_neon_int_2): Likewise.
7789         (cortex_a15_neon_int_3): Likewise.
7790         (cortex_a15_neon_int_4): Likewise.
7791         (cortex_a15_neon_int_5): Likewise.
7792         (cortex_a15_neon_vqneg_vqabs): Likewise.
7793         (cortex_a15_neon_vmov): Likewise.
7794         (cortex_a15_neon_vaba): Likewise.
7795         (cortex_a15_neon_vaba_qqq): Likewise.
7796         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7797         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7798         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7799         scalar_64_32_long_scalar): Likewise.
7800         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7801         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7802         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7803         lotype_qdd_64_32_long): Likewise.
7804         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7805         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7806         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7807         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7808         (cortex_a15_neon_shift_1): Likewise.
7809         (cortex_a15_neon_shift_2): Likewise.
7810         (cortex_a15_neon_shift_3): Likewise.
7811         (cortex_a15_neon_vshl_ddd): Likewise.
7812         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7813         (cortex_a15_neon_vsra_vrsra): Likewise.
7814         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7815         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7816         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7817         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7818         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7819         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7820         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7821         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7822         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7823         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7824         (cortex_a15_neon_bp_simple): Likewise.
7825         (cortex_a15_neon_bp_2cycle): Likewise.
7826         (cortex_a15_neon_bp_3cycle): Likewise.
7827         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7828         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7829         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7830         (cortex_a15_neon_vld2_4_regs): Likewise.
7831         (cortex_a15_neon_vld3_vld4): Likewise.
7832         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7833         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7834         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7835         (cortex_a15_neon_vst3_vst4): Likewise.
7836         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7837         (cortex_a15_neon_vld3_vld4_lane" 10
7838         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7839         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7840         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7841         (cortex_a15_neon_ldm_2): Likewise.0
7842         (cortex_a15_neon_stm_2): Likewise.
7843         (cortex_a15_neon_mcr): Likewise.
7844         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7845         (cortex_a15_neon_mrc): Likewise.
7846         (cortex_a15_neon_mrrc): Likewise.
7847         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7848         change.
7849         (cortex_a15_alu_shift): Likewise.
7850         (cortex_a15_alu_shift_reg): Likewise.
7851         (cortex_a15_mult32): Likewise.
7852         (cortex_a15_mult64): Likewise.
7853         (cortex_a15_block): Likewise.
7854         (cortex_a15_branch): Likewise.
7855         (cortex_a15_load1): Likewise.
7856         (cortex_a15_load3): Likewise.
7857         (cortex_a15_store1): Likewise.
7858         (cortex_a15_store3): Likewise.
7859         (cortex_a15_call): Likewise.
7860         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7861         (cortex_a5_f2r): Likewise.
7862         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7863         change.
7864         (cortex_a53_f2r): Likewise.
7865         * config/arm/cortex-a7.md
7866         (cortex_a7_branch): Update for attribute change.
7867         (cortex_a7_call): Likewise.
7868         (cortex_a7_alu_imm): Likewise.
7869         (cortex_a7_alu_reg): Likewise.
7870         (cortex_a7_alu_shift): Likewise.
7871         (cortex_a7_mul): Likewise.
7872         (cortex_a7_load1): Likewise.
7873         (cortex_a7_store1): Likewise.
7874         (cortex_a7_load2): Likewise.
7875         (cortex_a7_store2): Likewise.
7876         (cortex_a7_load3): Likewise.
7877         (cortex_a7_store3): Likewise.
7878         (cortex_a7_load4): Likewise.
7879         (cortex_a7_store4): Likewise.
7880         (cortex_a7_fpalu): Likewise.
7881         (cortex_a7_fconst): Likewise.
7882         (cortex_a7_fpmuls): Likewise.
7883         (cortex_a7_neon_mul): Likewise.
7884         (cortex_a7_fpmacs): Likewise.
7885         (cortex_a7_neon_mla: Likewise.
7886         (cortex_a7_fpmuld: Likewise.
7887         (cortex_a7_fpmacd: Likewise.
7888         (cortex_a7_fpfmad: Likewise.
7889         (cortex_a7_fdivs: Likewise.
7890         (cortex_a7_fdivd: Likewise.
7891         (cortex_a7_r2f: Likewise.
7892         (cortex_a7_f2r: Likewise.
7893         (cortex_a7_f_flags: Likewise.
7894         (cortex_a7_f_loads: Likewise.
7895         (cortex_a7_f_loadd: Likewise.
7896         (cortex_a7_f_stores: Likewise.
7897         (cortex_a7_f_stored: Likewise.
7898         (cortex_a7_neon): Likewise.
7899         * config/arm/cortex-a8-neon.md
7900         (cortex_a8_neon_mrc): Update for attribute change.
7901         (cortex_a8_neon_mrrc): Likewise.
7902         (cortex_a8_neon_int_1): Likewise.
7903         (cortex_a8_neon_int_2): Likewise.
7904         (cortex_a8_neon_int_3): Likewise.
7905         (cortex_a8_neon_int_4): Likewise.
7906         (cortex_a8_neon_int_5): Likewise.
7907         (cortex_a8_neon_vqneg_vqabs): Likewise.
7908         (cortex_a8_neon_vmov): Likewise.
7909         (cortex_a8_neon_vaba): Likewise.
7910         (cortex_a8_neon_vaba_qqq): Likewise.
7911         (cortex_a8_neon_vsma): Likewise.
7912         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7913         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7914         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7915         long_scalar): Likewise.
7916         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7917         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7918         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7919         long_scalar_qdd_64_32_long): Likewise.
7920         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7921         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7922         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7923         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7924         (cortex_a8_neon_shift_1): Likewise.
7925         (cortex_a8_neon_shift_2): Likewise.
7926         (cortex_a8_neon_shift_3): Likewise.
7927         (cortex_a8_neon_vshl_ddd): Likewise.
7928         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7929         (cortex_a8_neon_vsra_vrsra): Likewise.
7930         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7931         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7932         (cortex_a8_neon_fp_vsum): Likewise.
7933         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7934         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7935         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7936         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7937         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7938         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7939         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7940         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7941         (cortex_a8_neon_bp_simple): Likewise.
7942         (cortex_a8_neon_bp_2cycle): Likewise.
7943         (cortex_a8_neon_bp_3cycle): Likewise.
7944         (cortex_a8_neon_ldr): Likewise.
7945         (cortex_a8_neon_str): Likewise.
7946         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7947         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7948         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7949         (cortex_a8_neon_vld2_4_regs): Likewise.
7950         (cortex_a8_neon_vld3_vld4): Likewise.
7951         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7952         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7953         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7954         (cortex_a8_neon_vst3_vst4): Likewise.
7955         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7956         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7957         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7958         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7959         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7960         (cortex_a8_neon_mcr): Likewise.
7961         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7962         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7963         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7964         change.
7965         (ca9_neon_mrrc): Likewise.
7966         (cortex_a9_neon_int_1): Likewise.
7967         (cortex_a9_neon_int_2): Likewise.
7968         (cortex_a9_neon_int_3): Likewise.
7969         (cortex_a9_neon_int_4): Likewise.
7970         (cortex_a9_neon_int_5): Likewise.
7971         (cortex_a9_neon_vqneg_vqabs): Likewise.
7972         (cortex_a9_neon_vmov): Likewise.
7973         (cortex_a9_neon_vaba): Likewise.
7974         (cortex_a9_neon_vaba_qqq): Likewise.
7975         (cortex_a9_neon_vsma): Likewise.
7976         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7977         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7978         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7979         long_scalar): Likewise.
7980         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7981         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7982         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7983         long_scalar_qdd_64_32_long): Likewise.
7984         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7985         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7986         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7987         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7988         (cortex_a9_neon_shift_1): Likewise.
7989         (cortex_a9_neon_shift_2): Likewise.
7990         (cortex_a9_neon_shift_3): Likewise.
7991         (cortex_a9_neon_vshl_ddd): Likewise.
7992         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7993         (cortex_a9_neon_vsra_vrsra): Likewise.
7994         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7995         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7996         (cortex_a9_neon_fp_vsum): Likewise.
7997         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7998         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7999         (cortex_a9_neon_fp_vmla_ddd): Likewise.
8000         (cortex_a9_neon_fp_vmla_qqq): Likewise.
8001         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
8002         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
8003         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
8004         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
8005         (cortex_a9_neon_bp_simple): Likewise.
8006         (cortex_a9_neon_bp_2cycle): Likewise.
8007         (cortex_a9_neon_bp_3cycle): Likewise.
8008         (cortex_a9_neon_ldr): Likewise.
8009         (cortex_a9_neon_str): Likewise.
8010         (cortex_a9_neon_vld1_1_2_regs): Likewise.
8011         (cortex_a9_neon_vld1_3_4_regs): Likewise.
8012         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
8013         (cortex_a9_neon_vld2_4_regs): Likewise.
8014         (cortex_a9_neon_vld3_vld4): Likewise.
8015         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
8016         (cortex_a9_neon_vst1_3_4_regs): Likewise.
8017         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
8018         (cortex_a9_neon_vst3_vst4): Likewise.
8019         (cortex_a9_neon_vld1_vld2_lane): Likewise.
8020         (cortex_a9_neon_vld3_vld4_lane): Likewise.
8021         (cortex_a9_neon_vst1_vst2_lane): Likewise.
8022         (cortex_a9_neon_vst3_vst4_lane): Likewise.
8023         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
8024         (cortex_a9_neon_mcr): Likewise.
8025         (cortex_a9_neon_mcr_2_mcrr): Likewise.
8026         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
8027         (cortex_a9_fps): Likewise.
8028         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
8029         change.
8030         (cortex_m4_fmuls): Likewise.
8031         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
8032         change.
8033         (cortex_r4_mrc): Likewise.
8034         * config/arm/iterators.md: Update comment referring to neon_type.
8035         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
8036         (iwmmxt_movsi_insn): Likewise.
8037         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
8038         attribute change.
8039         (pj4_core_to_vfp): Likewise.
8040         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
8041         attribute change.
8042         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
8043         (vfp_fstore): Likewise.
8044         * doc/md.texi: Change references to neon_type to refer to type.
8046 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
8048         * tree.h (DECL_BUILT_IN): Fix typo in comment.
8050 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
8052         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
8053         lglobl if not weak.
8055 2013-09-04  Easwaran Raman  <eraman@google.com>
8057         PR middle-end/57370
8058         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
8059         (build_and_add_sum): Use it.
8060         (appears_later_in_bb): Simplify code.
8062 2013-09-04  Teresa Johnson  <tejohnson@google.com>
8064         * dumpfile.c (dump_finish): Don't close stderr/stdout.
8066 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
8068         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
8070 2013-09-04  Jan Hubicka  <jh@suse.cz>
8072         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
8073         * ipa-devirt.c: Include diganostic.h
8074         (odr_type_d): Add types and types_set.
8075         (hash_type_name): Work for types with vtables during LTO.
8076         (odr_hasher::remove): Fix comment; destroy types_set.
8077         (add_type_duplicate): New function,
8078         (get_odr_type): Use it.
8079         (dump_type_inheritance_graph): Dump type duplicates.
8080         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
8081         graph.
8082         * tree.c (types_same_for_odr): Give exact answers on types with
8083         virtual tables.
8085 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
8087         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
8088         explaining their differences.
8090 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
8092         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
8094 2013-09-03  Jeff Law  <law@redhat.com>
8096         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
8097         when not threading through a joiner block.  Pass joiner/no joiner
8098         state to register_jump_thread.
8099         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
8100         state from argument rather than implying on path length.
8101         Dump the entire jump thread path into debugging dump.
8102         * tree-flow.h (register_jump_thread): Update prototype.
8104 2013-08-29  Xinliang David Li  <davidxl@google.com>
8106         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
8107         Remove a trivial gcc_assert.
8109 2013-08-29  Xinliang David Li  <davidxl@google.com>
8111         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
8112         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
8113         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
8114         Delay base decl alignment adjustment.
8115         * tree-vectorizer.c (vect_destroy_datarefs): New function.
8116         * tree-vectorizer.h: New data structure.
8117         (set_dr_misalignment): New function.
8118         (dr_misalignment): Ditto.
8119         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
8120         (vectorizable_load): Ditto.
8121         (ensure_base_align): New function.
8122         (vectorize_loops): Add dbg_cnt support.
8123         (execute_vect_slp): Ditto.
8124         * dbgcnt.def: New debug counter.
8125         * Makefile: New dependency.
8127 2013-09-03  Meador Inge  <meadori@codesourcery.com>
8129         Revert:
8131         2013-08-30  Meador Inge  <meadori@codesourcery.com>
8133         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
8135 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
8137         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
8138         function descriptor.
8140 2013-09-03  Richard Biener  <rguenther@suse.de>
8142         * tree-affine.c (add_elt_to_tree): Fix association issue,
8143         avoid useless converts and make sure to always return a
8144         properly typed result.
8146 2013-09-03  Richard Biener  <rguenther@suse.de>
8148         PR middle-end/57656
8149         * fold-const.c (negate_expr_p): Fix division case.
8150         (negate_expr): Likewise.
8152 2013-09-03  Richard Biener  <rguenther@suse.de>
8154         PR lto/58285
8155         * tree-streamer-out.c: Include tm.h.
8156         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
8158 2013-09-03  Jan Hubicka  <jh@suse.cz>
8160         * tree-profile.c (tree_profiling): Cleanup CFG when done.
8162 2013-09-03  Alan Modra  <amodra@gmail.com>
8164         * config.gcc (powerpc*-*-linux*): Add support for little-endian
8165         multilibs to big-endian target and vice versa.
8166         * config/rs6000/t-linux64: Use := assignment on all vars.
8167         (MULTILIB_EXTRA_OPTS): Remove fPIC.
8168         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
8169         * config/rs6000/t-linux64le: New file.
8170         * config/rs6000/t-linux64bele: New file.
8171         * config/rs6000/t-linux64lebe: New file.
8173 2013-09-02  Jan Hubicka  <jh@suse.cz>
8175         * ipa-inline-transform.c (inline_transform): Do not
8176         optimize_inline_calls when not optimizing.
8178 2013-09-02  Jan Hubicka  <jh@suse.cz>
8180         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
8181         duplicated nodes for assembler names.
8182         * symtab.c (symtab_unregister_node): Do not attempt to unlink
8183         hard registers from assembler name hash.
8185 2013-09-02  Jan Hubicka  <jh@suse.cz>
8187         * ipa-split.c (execute_split_functions): Split externally visible
8188         functions called once.
8190 2013-09-02  Martin Jambor  <mjambor@suse.cz>
8192         PR ipa/58106
8193         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
8194         linked list.  When finding the correct duplicate, also consider also
8195         the caller in additon to its inlined_to node.
8197 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
8199         * config/aarch64/aarch64-simd-builtins.def
8200         (dup_lane_scalar): Remove.
8201         * config/aarch64/aarch64-simd.md
8202         (aarch64_simd_dup): Add 'w->w' alternative.
8203         (aarch64_dup_lane<mode>): Allow for VALL.
8204         (aarch64_dup_lane_scalar<mode>): Remove.
8205         (aarch64_dup_lane_<vswap_width_name><mode>): New.
8206         (aarch64_get_lane_signed<mode>): Add w->w altenative.
8207         (aarch64_get_lane_unsigned<mode>): Likewise.
8208         (aarch64_get_lane<mode>): Likewise.
8209         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
8210         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
8211         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
8212         (VCON): Change container of V2SF.
8213         (vswap_width_name): Likewise.
8214         * config/aarch64/arm_neon.h
8215         (__aarch64_vdup_lane_any): New.
8216         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
8217         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
8218         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
8220 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
8222         PR middle-end/56382
8223         * expr.c (emit_move_complex): Do not move complex FP values as parts if
8224         the source or the destination is a single hard register.
8226 2013-09-02  Richard Biener  <rguenther@suse.de>
8228         PR middle-end/57511
8229         * tree-scalar-evolution.c (instantiate_scev_name): Allow
8230         non-linear SCEVs.
8232 2013-09-02  Richard Biener  <rguenther@suse.de>
8234         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
8235         arithmetic to sizetype.
8237 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
8239         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
8240         Find auto-increment use both before and after candidate.
8242 2013-09-02  Marek Polacek  <polacek@redhat.com>
8244         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
8246 2013-09-01  Jan Hubicka  <jh@suse.cz>
8248         * Makefile.in: Add ipa-profile.o
8249         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
8250         * cgraph.c (struct cgraph_propagate_frequency_data,
8251         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
8252         ipa-profile.c; replace cgraph_ by ipa_ prefix.
8253         * cgraph.h (cgraph_propagate_frequency): Remove.
8254         * ipa-inline-analysis.c: Include ipa-utils.h;
8255         drop duplicated cfgloop.h.
8256         (inline_update_callee_summaries): Update.
8257         * ipa-profile.c: New file.
8258         * ipa-utils.h (ipa_propagate_frequency): Declare.
8259         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
8260         data-streamer.h, value-prof.h.
8261         (symtab_remove_unreachable_nodes): Update profile.
8262         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
8263         account_time_size, cmp_counts, dump_histogram,
8264         ipa_profile_generate_summary, ipa_profile_write_summary,
8265         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
8266         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
8267         Move to ipa-profile.c.
8269 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
8271         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
8273 2013-09-01  Jan Hubicka  <jh@suse.cz>
8275         * common.opt (fdevirtualize-speculatively): New function.
8276         * invoke.texi (fdevirtualize-speculatively): Document.
8277         * ipa-devirt.c: Include ipa-inline.h
8278         (likely_target_p): New function.
8279         (ipa_devirt): New function.
8280         (gate_ipa_devirt): New function.
8281         (pass_data_ipa_devirt): New static var.
8282         (pass_ipa_devirt): Likewise.
8283         (make_pass_ipa_devirt): New function.
8284         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
8285         (common_handle_option): Disable devirtualization when
8286         value range profiling is available.
8287         * passes.def (pass_ipa_devirt): Add.
8288         * timever.def (TV_IPA_DEVIRT): New timevar.
8289         * tree-pass.h (make_pass_ipa_devirt):
8291 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
8293         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
8294         include sanitize(undefined).
8296 2013-08-31  Diego Novillo  <dnovillo@google.com>
8298         * Makefile.in (TREE_CORE_H): Define.
8299         (TREE_H): Use.
8300         (GTFILES): Add tree-core.h.
8301         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
8302         size the array.
8303         * tree-core.h: New file.
8304         Move all data structures, enum, typedefs, global
8305         declarations and constants from ...
8306         * tree.h: ... here.
8308 2013-08-31  Jan Hubicka  <jh@suse.cz>
8310         * bulitins.c (expand_builtin): Do not early exit for gcov
8311         instrumented functions.
8313 2013-08-31  Marek Polacek  <polacek@redhat.com>
8315         * ubsan.c: Include tm_p.h.
8317 2013-08-31  Jan Hubicka  <jh@suse.cz>
8319         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
8320         warning.
8322         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
8323         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
8324         * tree-cfg.c (verify_gimple_label): ... here.
8325         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
8326         (ipa_merge_profiles): New function.
8327         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
8328         (lto_input_function_body): Likewise.
8329         * ipa-utils.h (ipa_merge_profiles): Declare.
8330         * lto-streamer.h (lto_input_function_body): Update prototype.
8331         (emit_label_in_global_context_p): Remove.
8332         * lto-symtab.c: Include ipa-utils.h
8333         (lto_cgraph_replace_node): Use ipa_merge_profiles.
8335 2013-08-31  Jan Hubicka  <jh@suse.cz>
8337         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
8339 2013-08-31  Jan Hubicka  <jh@suse.cz>
8341         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
8343 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
8345         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
8346         "cmp" RTX before signed_comparison_operator check to account
8347         for "code" changes.
8349 2013-08-30  Jan Hubicka  <jh@suse.cz>
8351         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
8352         (detect_type_change_1): Rename to ...
8353         (detect_type_change): ... this one; early return on non-polymorphic
8354         types.
8355         (detect_type_change_ssa): Add comp_type parameter; update
8356         use of detect_type_change.
8357         (compute_complex_assign_jump_func): Add param_type parameter;
8358         update use of detect_type_change_ssa.
8359         (compute_complex_ancestor_jump_func): Likewise.
8360         (ipa_get_callee_param_type): New function.
8361         (ipa_compute_jump_functions_for_edge): Compute parameter type;
8362         update calls to the jump function computation functions.
8364 2013-08-30  Teresa Johnson  <tejohnson@google.com>
8365             Steven Bosscher  <steven@gcc.gnu.org>
8367         * cfgrtl.c (fixup_new_cold_bb): New routine.
8368         (commit_edge_insertions): Invoke fixup_partitions.
8369         (find_partition_fixes): New routine.
8370         (fixup_partitions): Ditto.
8371         (verify_hot_cold_block_grouping): Update comments.
8372         (rtl_verify_edges): Invoke find_partition_fixes.
8373         (rtl_verify_bb_pointers): Update comments.
8374         (rtl_verify_bb_layout): Ditto.
8375         * basic-block.h (probably_never_executed_edge_p): Declare.
8376         (fixup_partitions): Ditto.
8377         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
8378         * bb-reorder.c (sanitize_hot_paths): New function.
8379         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
8380         sanitize_hot_paths.
8381         * predict.c (probably_never_executed_edge_p): New routine.
8382         * cfg.c (check_bb_profile): Add partition insanity warnings.
8384 2013-08-30  Meador Inge  <meadori@codesourcery.com>
8386         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
8388 2013-08-30  Marek Polacek  <polacek@redhat.com>
8390         * Makefile.in (ubsan.o): Add.
8391         (c-family/c-ubsan.o): Add.
8392         (builtins.o): Add ubsan.h dependency.
8393         * ubsan.h: New file.
8394         * ubsan.c: New file.
8395         * common.opt: Add -fsanitize=undefined option.
8396         (flag_sanitize): Add variable.
8397         (fsanitize=): Add option.  Add Driver.
8398         (fsanitize=thread): Remove option.
8399         (fsanitize=address): Likewise.
8400         (static-libubsan): New option.
8401         * doc/invoke.texi: Document the new flag and -static-libubsan.
8402         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
8403         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
8404         * builtin-attrs.def (ATTR_COLD): Define.
8405         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
8406         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
8407         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
8408         * flag-types.h (sanitize_code): New enum.
8409         * opts.c (common_handle_option): Parse command line arguments
8410         of -fsanitize=.  Add -fsanitize=unreachable option.
8411         * varasm.c (get_variable_section): Adjust.
8412         (assemble_noswitch_variable): Likewise.
8413         (assemble_variable): Likewise.
8414         (output_constant_def_contents): Likewise.
8415         (categorize_decl_for_section): Likewise.
8416         (place_block_symbol): Likewise.
8417         (output_object_block): Likewise.
8418         * builtins.def: Likewise.
8419         * toplev.c (compile_file): Likewise.
8420         (process_options): Likewise.
8421         * cppbuiltin.c: Likewise.
8422         * tsan.c (tsan_pass): Likewise.
8423         (tsan_gate): Likewise.
8424         (tsan_gate_O0): Likewise.
8425         * cfgexpand.c (partition_stack_vars): Likewise.
8426         (expand_stack_vars): Likewise.
8427         (defer_stack_allocation): Likewise.
8428         (expand_used_vars): Likewise.
8429         * cfgcleanup.c (old_insns_match_p): Likewise.
8430         * asan.c (asan_finish_file): Likewise.
8431         (asan_instrument): Likewise.
8432         (gate_asan): Likewise.
8433         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
8434         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
8435         (asan_global_struct): Use pointer_sized_int_node instead
8436         calling build_nonstandard_integer_type.
8437         (initialize_sanitizer_builtins): Likewise.
8438         (asan_finish_file): Likewise.
8439         * gcc.c: Document %{%:function(args):X}.
8440         (static_spec_functions): Add sanitize.
8441         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
8442         store funcval != NULL there.
8443         (do_spec_1): Adjust handle_spec_function caller.
8444         (handle_braces): Allow %:function(args) as condition.
8445         (sanitize_spec_function): New function.
8446         (ADD_STATIC_LIBUBSAN_LIBS): Define.
8447         (LIBUBSAN_SPEC): Likewise.
8448         (LIBUBSAN_EARLY_SPEC): Likewise.
8449         (SANITIZER_SPEC): Handle libubsan.
8450         (SANITIZER_EARLY_SPEC): Likewise.
8451         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
8452         instead of fsanitize=address.
8453         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
8454         instead of fsanitize=address*.
8455         * builtins.c: Include ubsan.h.
8456         (fold_builtin_0): Instrument __builtin_unreachable.
8457         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
8458         instead of flag_asan.
8459         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
8460         (pointer_sized_int_node): Define.
8461         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
8463 2013-08-30  Mike Stump  <mikestump@comcast.net>
8465         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
8466         with RE patterns.
8468 2013-08-29  Jan Hubicka  <jh@suse.cz>
8470         * cgraph.c (cgraph_function_body_availability): Handle weakref
8471         correctly.
8472         * passes.def: Remove pass_fixup_cfg.
8473         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
8474         track when we need to remove functions.
8475         (gate_ipa_inline): Execute inlining always; add comment why.
8476         (pass_data_ipa_inline): Remove TODO_remove_functions.
8477         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
8478         do not produce summaries.
8479         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
8480         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
8481         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
8482         constant pool and vtable.
8484 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
8486         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
8487         New arm_neon.h's internal macros to specify 64-bit constants.
8488         Avoid using stdint.h's macros.
8490 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
8492         * recog.c (verify_changes): Verify that changes[i].old is non-zero
8493         before applying REG_P.
8495 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
8497         PR tree-optimization/58277
8498         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
8499         after seeing too many stmts with vdef in between dombb and current
8500         bb, invalidate everything.
8502 2013-08-30  Richard Biener  <rguenther@suse.de>
8504         * fold-const.c (fold_single_bit_test): Fix overflow test.
8506 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
8508         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
8509         and which can live in a register, always retrieve the value on entry.
8510         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
8511         passed by invisible reference specially.
8512         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
8513         (vt_add_function_parameter): Correctly deal with a parameter passed by
8514         invisible reference.
8516 2013-08-30  Jan Hubicka  <jh@suse.cz>
8518         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
8520 2013-08-30  Richard Biener  <rguenther@suse.de>
8522         PR tree-optimization/58228
8523         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
8524         allow invariant loads in nested loop vectorization.
8526 2013-08-30  Richard Biener  <rguenther@suse.de>
8528         PR tree-optimization/58223
8529         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
8530         (has_anti_or_output_dependence): ... this and adjust to also
8531         look for output dependences.
8532         (mark_nodes_having_upstream_mem_writes): Adjust.
8533         (rdg_flag_uses): Likewise.
8535 2013-08-30  Richard Biener  <rguenther@suse.de>
8537         PR tree-optimization/58010
8538         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
8539         assert that we have a loop-closed PHI.
8541 2013-08-29  Jan Hubicka  <jh@suse.cz>
8543         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
8544         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
8545         * lto-section-in.c (lto_free_function_in_decl_state): New function.
8546         (lto_free_function_in_decl_state_for_node): New function.
8548 2013-08-29  Xinliang David Li  <davidxl@google.com>
8550         * loop-unroll.c (report_unroll_peel): Minor message change.
8551         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
8552         Emit alignment peeling message with default -fopt-info.
8553         (vect_loop_versioning): Emit loop version info message.
8554         * tree-vectorizer.c (vectorize_loops): Minor message change.
8555         (execute_vect_slp): Ditto.
8557 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
8559         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
8560         of the clone from the DECL_NAME of the original function.
8562 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
8564         * passes.c (register_pass): Add overload.
8565         * tree-pass.h (register_pass): Forward declare it.  Add comment.
8567 2013-08-29  Jan Hubicka  <jh@suse.cz>
8569         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
8570         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
8571         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
8572         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8573         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
8574         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
8575         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
8576         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
8578 2013-08-29  Teresa Johnson  <tejohnson@google.com>
8580         * dumpfile.c (dump_loc): Output column number.
8581         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
8582         * doc/invoke.texi: Document optall -fopt-info flag.
8583         * profile.c (read_profile_edge_counts): Use new dump framework.
8584         (compute_branch_probabilities): Ditto.
8585         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
8586         when pass not in any opt group.
8587         * pass_manager.h (pass_manager::get_pass_profile): New method.
8588         * value-prof.c (check_counter): Use new dump framework.
8589         (check_ic_target): Ditto.
8590         * coverage.c (get_coverage_counts): Ditto.
8591         (coverage_init): Setup new dump framework.
8593 2013-08-29  Richard Biener  <rguenther@suse.de>
8595         PR tree-optimization/58246
8596         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
8597         handle the dominance check inside a basic-block.
8599 2013-08-29  Richard Biener  <rguenther@suse.de>
8601         PR middle-end/57287
8602         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
8603         of default defs that appear in abnormal PHI nodes.
8605 2013-08-29  Richard Biener  <rguenther@suse.de>
8607         PR tree-optimization/57685
8608         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
8609         single-use operands to avoid exponential complexity.
8611 2013-08-28  Dehao Chen  <dehao@google.com>
8613         * ipa-inline.c (edge_badness): Fix integer underflow.
8615 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
8617         * gtm-builtins.def (_ITM_free): Declare leaf.
8619 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8621         PR target/58067
8622         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
8623         (*tls_local_dynamic_base_64_largepic): Likewise.
8624         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
8625         Remove predicate from call operand.
8626         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
8627         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
8629 2013-08-28  Jeff Law  <law@redhat.com>
8631         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
8632         checks for the number of predecessors and successors allowed.
8633         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
8634         which require copying a joiner block if there is a request which
8635         is a subpath that requires no joiner block copying.
8637 2013-08-28  Jan Hubicka  <jh@suse.cz>
8639         * lto-streamer-out.c (DFS_write_tree_body): Drop
8640         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8641         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8642         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8643         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8644         Do not read DECL_ERROR_ISSUED.
8645         (unpack_ts_decl_with_vis_value_fields): Do not read
8646         DECL_DEFER_OUTPUT.
8647         (lto_input_ts_binfo_tree_pointers): Do not read
8648         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8649         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8650         write DECL_ERROR_ISSUED..
8651         (pack_ts_decl_with_vis_value_fields): Do not write
8652         DECL_DEFER_OUTPUT.
8653         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8654         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8655         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8656         * tree.h (tree_decl_common): Update comment.
8657         (DECL_ERROR_ISSUED): Remove.
8659 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8661         PR middle-end/58257
8662         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8664 2013-08-28  Jan Hubicka  <jh@suse.cz>
8666         * builtins.def (free): Declare leaf.
8668 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8670         * gdbhooks.py: New.
8671         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8672         * configure: Regenerate.
8674 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8676         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8677         (ipa_ancestor_jf_data): Likewise.
8678         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8679         (ipa_get_jf_pass_through_type_preserved): New function.
8680         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8681         (ipa_get_jf_ancestor_type_preserved): New function.
8682         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8683         (ipa_get_jf_ancestor_result): Likewise.
8684         (propagate_vals_accross_pass_through): Use
8685         ipa_get_jf_pass_through_result to do all the value mappings.
8686         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8687         type_preserved flag.
8688         (ipa_set_jf_cst_copy): New function.
8689         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8690         (ipa_set_jf_arith_pass_through): Likewise.
8691         (ipa_set_ancestor_jf): Likewise.
8692         (compute_complex_assign_jump_func): Set type_preserved instead of
8693         punting.
8694         (ipa_compute_jump_functions_for_edge): Likewise.
8695         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8696         (update_jump_functions_after_inlining): Update type_preserved.
8697         Explicitely create jump functions when combining one with pass_through.
8698         (ipa_write_jump_function): Stream the type_preserved flags.
8699         (ipa_read_jump_function): Likewise.
8701 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8702             Aldy Hernandez  <aldyh@redhat.com>
8704         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8705         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8706         * function.h (struct function): Add has_force_vect_loops and
8707         has_simduid_loops.
8708         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8709         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8710         handle it.
8711         * gimple.def: Update CLAUSES comments.
8712         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8713         (gimple_build_omp_for): Add argument to prototype.
8714         (gimple_omp_for_kind): New.
8715         (gimple_omp_for_set_kind): New.
8716         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8717         GOVD_DATA_SHARE_CLASS.
8718         (enum omp_region_type): Add ORT_SIMD.
8719         (gimple_add_tmp_var): Handle ORT_SIMD.
8720         (gimplify_var_or_parm_decl): Same.
8721         (is_gimple_stmt): Same.
8722         (omp_firstprivatize_variable): Same.
8723         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8724         (omp_notice_variable): Handle ORT_SIMD.
8725         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8726         (omp_check_private): Handle ORT_SIMD.
8727         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8728         OMP_CLAUSE_SAFELEN.
8729         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8730         Handle OMP_CLAUSE_LASTPRIVATE.
8731         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8732         OMP_CLAUSE_SAFELEN.
8733         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8734         (gimplify_expr): Handle OMP_SIMD.
8735         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8736         (expand_GOMP_SIMD_VF): New.
8737         (expand_GOMP_SIMD_LAST_LANE): New.
8738         * internal-fn.def (GOMP_SIMD_LANE): New.
8739         (GOMP_SIMD_VF): New.
8740         (GOMP_SIMD_LAST_LANE): New.
8741         * omp-low.c: Include target.h.
8742         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8743         OMP_CLAUSE_SAFELEN.
8744         (check_omp_nesting_restrictions): Same.
8745         (omp_max_vf): New.
8746         (lower_rec_simd_input_clauses): New.
8747         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8748         OMP_CLAUSE_LINEAR.
8749         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8750         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8751         (expand_omp_build_assign): New.
8752         (expand_omp_for_init_counts): New.
8753         (expand_omp_for_init_vars): New.
8754         (extract_omp_for_update_vars): New.
8755         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8756         and rewrite accordingly.
8757         (expand_omp_simd): New.
8758         (expand_omp_for): Use expand_omp_simd.
8759         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8760         (lower_omp_for): Do not lower the body.
8761         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8762         in their own loops.
8763         * tree-flow.h (find_omp_clause): Remove prototype.
8764         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8765         forcing vectorization of the loop, or if flag_tree_vectorize.
8766         (gate_tree_if_conversion): Similarly.
8767         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8768         gimple_build_omp_for.
8769         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8770         * tree-parloops (create_parallel_loop): Pass kind argument to
8771         gimple_build_omp_for.
8772         * tree-pretty-print.c (dump_omp_clause): Add cases for
8773         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8774         OMP_CLAUSE__SIMDUID_.
8775         (dump_generic_node): Handle OMP_SIMD.
8776         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8777         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8778         unroll OMP_SIMD loops here.
8779         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8780         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8781         loop->safelen.
8782         (vect_analyze_data_refs): Handle simd loops.
8783         * tree-vect-loop.c (vectorizable_live_operation): Handle
8784         IFN_GOMP_SIMD*.
8785         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8786         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8787         (vectorizable_load): Same.
8788         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8789         (struct simduid_to_vf): New.
8790         (simduid_to_vf::hash): New.
8791         (simduid_to-vf::equal): New.
8792         (struct simd_array_to_simduid): New.
8793         (simd_array_to_simduid::hash): New.
8794         (simd_array_to_simduid::equal): New.
8795         (adjust_simduid_builtins): New.
8796         (struct note_simd_array_uses_struct): New.
8797         (note_simd_array_uses_cb): New.
8798         (note_simd_array_uses): New.
8799         (vectorize_loops): Handle simd hints and adjust simd builtins
8800         accordingly.
8801         * tree-vectorizer.h (struct _stmt_vec_info): Add
8802         simd_lane_access_p field.
8803         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8804         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8805         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8806         (omp_clause_code_name): Same.
8807         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8808         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8809         * tree.def (OMP_SIMD): New entry.
8810         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8811         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8812         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8813         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8814         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8815         (OMP_CLAUSE_LINEAR_STEP): New.
8816         (OMP_CLAUSE_SAFELEN_EXPR): New.
8817         (OMP_CLAUSE__SIMDUID__DECL): New.
8818         (find_omp_clause): New prototype.
8820 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8822         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8823         Haswell processor detection.
8825 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8827         PR target/57927
8828         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8829         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8830         AVX2 capable processors.
8832 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8834         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8835         of vget_low_* with implementations in terms of other intrinsics.
8837 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8839         PR middle-end/57219
8840         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8841         values to -1, 0 and 1.
8843 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8845         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8846         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8847         (<optab><mode>3_insn): Remove.
8848         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8849         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8850         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8851         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8852         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8853         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8854         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8855         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8856         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8857         (ror<mode>3_insn): Likewise.
8858         * config/aarch64/predicates.md (aarch64_simd_register): New.
8860 2013-08-27  Richard Biener  <rguenther@suse.de>
8862         PR tree-optimization/57521
8863         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8864         one edge is non-critical.
8865         (find_phi_replacement_condition): Make sure to use a non-critical
8866         edge.  Cleanup and remove old bug workarounds.
8867         (bb_postdominates_preds): Remove.
8868         (if_convertible_loop_p_1): Do not compute post-dominators.
8869         (combine_blocks): Do not free post-dominators.
8870         (main_tree_if_conversion): Likewise.
8871         (pass_data_if_conversion): Add TODO_verify_ssa.
8873 2013-08-27  DJ Delorie  <dj@redhat.com>
8875         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8877 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8879         * function.c (assign_parm_find_data_types): Set passed_mode and
8880         nominal_mode to the TYPE_MODE of nominal_type for the built
8881         pointer type in case of the struct-pass-by-reference.
8883 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8885         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8886         (UINT16_TYPE): Change default to "unsigned int".
8888         * config/avr/avr.opt (mfract-convert-truncate): New option.
8889         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8890         is set, round negative fractional integers according to n1169
8891         when converting to integer types.
8893 2013-08-26  Jan Hubicka  <jh@suse.cz>
8895         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8896         methods can not be called indirectly when their address is not taken.
8898 2013-08-26  Jan Hubicka  <jh@suse.cz>
8900         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8901         ctor_for_folding.
8903 2013-08-26  Jan Hubicka  <jh@suse.cz>
8905         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8906         can be unshared.
8908 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8910         * reload.c (find_valid_class): Allow classes that do not include
8911         FIRST_PSEUDO_REGISTER - 1.
8913 2013-08-26  Jan Hubicka  <jh@suse.cz>
8915         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8916         fix edge count/frequency when speculation failed; fix type check
8917         for the direct call.
8919 2013-08-26  Jan Hubicka  <jh@suse.cz>
8921         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8923 2013-08-26  Jan Hubicka  <jh@suse.cz>
8925         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8926         to be changed by edge redirection.
8928 2013-08-26  Jan Hubicka  <jh@suse.cz>
8930         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8931         formating; add sanity check.
8932         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8933         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8934         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8935         (resolve_noninline_speculation): Update callee keys, too.
8937 2013-08-26  Jan Hubicka  <jh@suse.cz>
8939         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8940         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8942 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8944         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8945         into proper place.
8947 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8949         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8950         (x86_64_elf_select_section): Ditto.
8951         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8952         arguments.
8953         (ix86_pass_by_reference): Ditto.
8954         (output_set_got): Ditto.
8955         (ix86_unary_operator_ok): Ditto.
8956         (ix86_expand_builtin): Ditto.
8958 2013-08-23  Jan Hubicka  <jh@suse.cz>
8960         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8962 2013-08-23  Jan Hubicka  <jh@suse.cz>
8964         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8965         (tree_decl_with_vis): Add FINAL field.
8967 2013-08-23  Jeff Law  <law@redhat.com>
8969         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8970         the debugging dump when the expression is fully redundant.
8972 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8974         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8975         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8976         * pretty-print.c (pp_formatted_text_data): Likewise.
8977         (pp_write_text_to_stream): Likewise.
8978         (pp_write_text_as_dot_label_to_stream): Likewise.
8979         (pp_append_r): Likewise.
8980         (pp_format): Likewise.
8981         (pp_flush): Likewise.
8982         (pp_clear_output_area): Likewise.
8983         (pp_append_text): Likewise.
8984         (pp_formatted_text): Likewise.
8985         (pp_remaining_character_count_for_line): Likewise.
8986         (pp_newline): Likewise.
8987         (pp_character): Likewise.
8988         (output_buffer::~output_buffer): Define.
8989         (pretty_printer::~pretty_printer): Destruct output buffer.
8990         * pretty-print.h (output_buffer::~output_buffer): Declare.
8991         (pretty_printer::~pretty_printer): Declare virtual.
8993 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8995         PR other/57324
8996         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8997         HOST_WIDE_INT_M1U): New macros.
8998         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8999         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
9000         unsigned -1 for lshift.
9001         * cse.c (cse_insn): Likewise.
9002         * double-int.c (rshift_double, lshift_double): Likewise.
9003         * builtins.c (fold_builtin_bitop): Likewise.
9004         * combine.c (force_to_mode): Likewise.
9005         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
9006         * simplify-rtx.c (simplify_const_unary_operation,
9007         simplify_const_binary_operation): Likewise.
9008         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
9009         check_va_list_escapes): Likewise.
9010         * rtlanal.c (nonzero_bits1): Likewise.
9011         * expmed.c (expand_smod_pow2): Likewise.
9012         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
9014 2013-08-23  Jan Hubicka  <jh@suse.cz>
9016         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
9017         as having address taken.
9019 2013-08-23  Jan Hubicka  <jh@suse.cz>
9021         * ipa-utils.h (method_class_type): Declare.
9022         * ipa-devirt.c (method_class_type): Export.
9024         * cgraphunit.c (analyze_functions): Do basic devirtualization;
9025         do not walk base classes of anonymous types.
9027 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
9029         PR rtl-optimization/58220
9030         PR regression/58221
9031         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
9032         handle SEQUENCE insns properly.
9034 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9036         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
9037         definition.
9038         (pp_newline_and_indent): Likewise.
9039         (pp_separate_with): Likewise.
9040         * pretty-print.c (pp_newline_and_flush): Define.
9041         (pp_newline_and_indent): Likewise.
9042         (pp_separate_with): Likewise.
9044 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
9046         PR target/58218
9047         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
9048         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
9050 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
9052         * config/i386/predicates.md (ext_sse_reg_operand): New.
9053         * config/i386/i386.md (*movti_internal): Use
9054         predicate to determine if EVEX is needed.
9055         (*movsi_internal): Ditto.
9056         (*movdf_internal): Ditto.
9057         (*movsf_internal): Ditto.
9058         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
9060 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
9062         PR tree-optimization/58209
9063         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
9064         (find_tail_calls): Give up for pointer result types if m is non-NULL.
9065         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
9066         emit POINTER_PLUS_EXPR.
9067         (create_tailcall_accumulator): For pointer result type accumulate in
9068         sizetype type.
9070 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
9072         * configure.ac: Add backslashes missing from the last change.
9073         * configure: Regenerate.
9075 2013-08-22  Jan Hubicka  <jh@susue.cz>
9077         * ipa.c (function_and_variable_visibility): First remember function
9078         was global and then make it local.
9080 2013-08-22  Julian Brown  <julian@codesourcery.com>
9082         * configure.ac: Add aarch64 to list of arches which use "nop" in
9083         debug_line test.
9084         * configure: Regenerate.
9086 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9088         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
9089         gnu_libc_has_function.
9090         * config/s390/tpf.h: Likewise.
9092 2013-08-22  Jan Hubicka  <jh@susue.cz>
9094         * timevar.c (validate_phases): Add cast.
9096 2013-08-22  Jan Hubicka  <jh@susue.cz>
9098         * timevar.c (validate_phases): Use size_t for memory.
9099         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
9101 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9103         * pretty-print.h (output_buffer::output_buffer): Declare.
9104         (pretty_printer::pretty_printer): Likewise.
9105         (pp_construct): Remove.
9106         * pretty-print.c (output_buffer::output_buffer): Define.
9107         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
9108         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
9109         (print_gimple_expr): Likewise.
9110         (print_gimple_seq): Likewise.
9111         (gimple_dump_bb): Likewise.
9112         * sched-vis.c (dump_value_slim): Likewise.
9113         (dump_insn_slim): Likewise.
9114         (dump_rtl_slim): Likewise.
9115         (str_pattern_slim): Likewise.
9116         * tree-mudflap.c (mf_varname_tree): Likewise.
9117         * graph.c (print_graph_cfg): Likewise.
9118         (start_graph_dump): Likewise.
9119         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
9120         placement-new.
9121         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
9122         pretty printer initialization.
9123         * coretypes.h (diagnostic_context): Remove superflous type alias
9124         declaration.
9125         (pretty_printer): Likewise.  Declare directly as a class.
9126         (pretty_print_info): Remove declaration as class.
9127         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
9128         and pp_clear_output_area.
9129         (asan_add_global): Likewise.
9131 2013-08-22  Jan Hubicka  <jh@suse.cz>
9133         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
9134         * ipa-utils.h (update_type_inheritance_graph): Declare.
9135         (possible_polymorphic_call_target_p): Declare.
9136         (possible_polymorphic_call_target_p): New.
9137         * ipa-devirt.c: Update toplevel comments.
9138         (cached_polymorphic_call_targets): Move up.
9139         (odr_type_d): Move ID down.
9140         (polymorphic_type_binfo_p): Update comment.
9141         (odr_hasher::remove): Likewise;
9142         (get_odr_type): Set anonymous_namespace.
9143         (dump_odr_type): Dump it.
9144         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
9145         (maybe_record_node): Record node in cached_polymorphic_call_targets.
9146         (record_binfo): Add comment.
9147         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
9148         built.
9149         (devirt_node_removal_hook): Do not iCE when cache is freed.
9150         (possible_polymorphic_call_target_p): New predicate.
9151         (update_type_inheritance_graph): New function.
9153 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9154             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
9155             Sergey Lega  <sergey.s.lega@intel.com>
9156             Anna Tikhonova  <anna.tikhonova@intel.com>
9157             Ilya Tocar  <ilya.tocar@intel.com>
9158             Andrey Turetskiy  <andrey.turetskiy@intel.com>
9159             Ilya Verbin  <ilya.verbin@intel.com>
9160             Kirill Yukhin  <kirill.yukhin@intel.com>
9161             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
9163         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
9164         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
9165         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
9166         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
9167         (OPTION_MASK_ISA_AVX2_UNSET): Update.
9168         (OPTION_MASK_ISA_AVX512F_UNSET): New.
9169         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
9170         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
9171         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
9172         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
9173         OPT_mavx512pf, OPT_mavx512er cases.
9174         * config/i386/constraints.md (v): New constraint.
9175         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
9176         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
9177         (bit_AVX512CD): New.
9178         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9179         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
9180         * config/i386/i386-c.c (ix86_target_macros_internal):
9181         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
9182         __AVX512PF__.
9183         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
9184         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
9185         * config/i386/i386.c (regclass_map, dbx_register_map)
9186         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
9187         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
9188         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
9189         -mavx512pf options.
9190         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
9191         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
9192         -mavx512pf options.  Fix formatting.
9193         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
9194         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
9195         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
9196         -mavx512cd, -mavx512pf options.
9197         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
9198         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
9199         (ix86_preferred_output_reload_class): Replace SSE_REGS with
9200         ALL_SSE_REGS.
9201         (ix86_hard_regno_mode_ok): Support 512-bit registers.
9202         (ix86_set_reg_reg_cost): Ditto.
9203         (x86_order_regs_for_local_alloc): Ditto.
9204         (MAX_VECT_LEN): Extend to 64-byte.
9205         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
9206         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
9207         (TARGET_AVX512ER, TARGET_AVX512CD): New.
9208         (BIGGEST_ALIGNMENT): Extend to 512-bits.
9209         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
9210         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
9211         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
9212         (SSE_REG_MODE_P): Support new modes.
9213         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
9214         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
9215         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
9216         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
9217         (REG_CLASS_CONTENTS): Add new registers.
9218         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
9219         (EXT_REX_SSE_REGNO_P): New.
9220         (HI_REGISTER_NAMES): Add new registers.
9221         * config/i386/i386.md: Define constants for new registers.
9222         (mode): Add new 512-bit modes.
9223         (prefix): Support evex prefix.
9224         (isa): Support avx512f, noavx512f, fma_avx512f.
9225         (ssemodesuffix): Add new 512-bit modes.
9226         (movxi): New.
9227         (*movxi_internal_avx512f): Ditto.
9228         (*movdi_internal): Replace constraint "x" with the new constraint "v".
9229         Support MODE_XI.
9230         (*movsi_internal): Likewise.
9231         (*movdf_internal): Likewise.
9232         (*movsf_internal): Likewise.
9233         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
9234         (<code><mode>3): Likewise.
9235         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
9236         New.
9237         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
9238         with the new constraint "v".
9239         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
9240         modes.
9241         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
9242         with the new constraint "v".
9243         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
9244         (<sse2>_storedqu<avxsizesuffix>): Likewise.
9245         (*<plusminus_insn><mode>3): Likewise.
9246         (<sse>_vm<plusminus_insn><mode>3): Likewise.
9247         (*mul<mode>3): Likewise.
9248         (<sse>_vmmul<mode>3): Likewise.
9249         (<sse>_div<mode>3): Likewise.
9250         (<sse>_vmdiv<mode>3): Likewise.
9251         (<sse>_sqrt<mode>2): Likewise.
9252         (<sse>_vmsqrt<mode>2): Likewise.
9253         (*<code><mode>3_finite): Likewise.
9254         (*<code><mode>3) <smaxmin>: Likewise.
9255         (<sse>_vm<code><mode>3): Likewise.
9256         (*<code><mode>3) <any_logic>: Likewise.
9257         (*fma_fmadd_<mode>): Likewise.
9258         (*fma_fmsub_<mode>): Likewise.
9259         (*fma_fnmadd_<mode>): Likewise.
9260         (*fma_fnmsub_<mode>): Likewise.
9261         (*fma_fmaddsub_<mode>): Likewise.
9262         (*fma_fmsubadd_<mode>): Likewise.
9263         (*fmai_fmadd_<mode>): Likewise.
9264         (*fmai_fmsub_<mode>): Likewise.
9265         (*fmai_fnmadd_<mode>): Likewise.
9266         (*fmai_fnmsub_<mode>): Likewise.
9267         (sse_cvtsi2ss): Likewise.
9268         (sse_cvtsi2ssq): Likewise.
9269         (sse_cvtss2si): Likewise.
9270         (sse_cvtss2si_2): Likewise.
9271         (sse_cvtss2siq): Likewise.
9272         (sse_cvtss2siq_2): Likewise.
9273         (sse_cvttss2si): Likewise.
9274         (sse_cvtss2siq_2): Likewise.
9275         (float<sseintvecmodelower><mode>2): Likewise.
9276         (sse2_cvtsd2si_2): Likewise.
9277         (sse2_cvtsd2siq_2): Likewise.
9278         (*<plusminus_insn><mode>3): Likewise.
9279         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
9280         (*<sse4_1_avx2>_mul<mode>3): Likewise.
9281         (ashr<mode>3): Likewise.
9282         (<shift_insn><mode>3): Likewise.
9283         (avx2_<code><mode>3): Likewise.
9284         (*avx2_<code><mode>3): Likewise.
9285         (*andnot<mode>3): Likewise.
9286         (*<code><mode>3) <any_logic>: Likewise.
9287         (abs<mode>2): Likewise.
9288         (avx2_permvar<mode>): Likewise.
9289         (avx2_perm<mode>_1): Likewise.
9290         (*avx_vpermilp<mode>): Likewise.
9291         (avx_vpermilvar<mode>3): Likewise.
9292         (avx2_ashrv<mode>): Likewise.
9293         (avx2_<shift_insn>v<mode>): Likewise.
9294         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
9295         -mavx512cd.
9296         * doc/rtl.texi: Document XImode.
9298 2013-08-21  Jeff Law  <law@redhat.com>
9300         * tree-flow.h (register_jump_thread): Pass vector of edges
9301         instead of each important edge.
9302         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
9303         thread path into a vector and pass that to register_jump_thread.
9304         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
9305         passed in edge vector to the current 3-edge form.
9307         Revert:
9308         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
9310         * dce.c (fini_dce): Call df_analyze again just in case
9311         delete_unmarked_insns removed anything.
9313 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
9315         * reload.h (struct reg_equivs): Rename to ..
9316         (struct reg_equivs_s): .. this.
9318 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
9320         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
9322 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9324         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
9326 2013-08-21  Jeff Law  <law@redhat.com>
9328         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
9329         simplify assignments too.  If the RHS collapses to a singleton
9330         range, then return the value for the range.
9332 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
9334         * config/i386/sse.md (V16): Rename to...
9335         (VMOVE): this.
9336         (mov<mode>): Update iterator name.
9337         (*mov<mode>_internal): Ditto.
9338         (push<mode>1): Ditto.
9339         (movmisalign<mode>): Ditto.
9341 2013-08-20  Jan Hubicka  <jh@suse.cz>
9343         PR bootstrap/58186
9344         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
9345         entry for direct edges.
9346         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
9348 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9350         Revert my last two changes, r201865 and r201864:
9352         Revert r201865:
9353         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9355         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
9356         instances can own GC refs.
9358         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
9359         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
9360         (gcc::context::gt_pch_nx): Likewise.
9361         (gcc::context::gt_pch_nx):  Likewise.
9362         * ggc.h (gt_ggc_mx <T>): New.
9363         (gt_pch_nx_with_op <T>): New.
9364         (gt_pch_nx <T>): New.
9365         * passes.c (opt_pass::gt_ggc_mx): New.
9366         (opt_pass::gt_pch_nx): New.
9367         (opt_pass::gt_pch_nx_with_op): New.
9368         (pass_manager::gt_ggc_mx): New.
9369         (pass_manager::gt_pch_nx): New.
9370         (pass_manager::gt_pch_nx_with_op): New.
9371         (pass_manager::operator new): Use
9372         ggc_internal_cleared_alloc_stat rather than xcalloc.
9373         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
9374         (pass_manager::gt_ggc_mx): New.
9375         (pass_manager::gt_pch_nx): New.
9376         (pass_manager::gt_pch_nx_with_op): New.
9377         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
9378         (opt_pass::operator new): New.
9379         (opt_pass::gt_ggc_mx): New.
9380         (opt_pass::gt_pch_nx): New.
9381         (opt_pass::gt_pch_nx_with_op): New.
9383         Revert r201864:
9384         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9386         * Makefile.in (GTFILES): Add context.h.
9387         * context.c (gcc::context::operator new): New.
9388         (gcc::context::gt_ggc_mx): New.
9389         (gcc::context::gt_pch_nx): New.
9390         (gcc::context::gt_pch_nx): New.
9391         * context.h (gcc::context): Add GTY((user)) marking.
9392         (gcc::context::operator new): New.
9393         (gcc::context::gt_ggc_mx): New.
9394         (gcc::context::gt_pch_nx): New.
9395         (gcc::context::gt_pch_nx): New.
9396         (g): Add GTY marking.
9397         (gt_ggc_mx (gcc::context *)): New.
9398         (gt_pch_nx (gcc::context *)): New.
9399         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
9400         void *cookie)): New.
9401         * gengtype.c (open_base_files) <ifiles>: Add context.h.
9403 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
9405         * dce.c (fini_dce): Call df_analyze again just in case
9406         delete_unmarked_insns removed anything.
9408 2013-08-20  Teresa Johnson  <tejohnson@google.com>
9410         PR rtl-optimizations/57451
9411         * final.c (reemit_insn_block_notes): Prevent lexical blocks
9412         from crossing split section boundaries.
9414 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
9416         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
9417         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
9418         with MULTLIB_DEFAULTS.
9420 2013-08-20  Nick Clifton  <nickc@redhat.com>
9422         * target.def (narrow_volatile_bitfield): Note that the default
9423         value is false, not !TARGET_STRICT_ALIGN.
9424         * doc/tm.texi: Regenerate.
9426 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
9428         Fix LIB_SPEC for systems without libpthread.
9430         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
9431         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
9432         for Android.
9433         * config/i386/linux-common.h: Likewise.
9434         * config/mips/linux-common.h: Likewise.
9436 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
9438         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
9439         checks.
9441 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9443         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
9444         instances can own GC refs.
9446         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
9447         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
9448         (gcc::context::gt_pch_nx): Likewise.
9449         (gcc::context::gt_pch_nx):  Likewise.
9450         * ggc.h (gt_ggc_mx <T>): New.
9451         (gt_pch_nx_with_op <T>): New.
9452         (gt_pch_nx <T>): New.
9453         * passes.c (opt_pass::gt_ggc_mx): New.
9454         (opt_pass::gt_pch_nx): New.
9455         (opt_pass::gt_pch_nx_with_op): New.
9456         (pass_manager::gt_ggc_mx): New.
9457         (pass_manager::gt_pch_nx): New.
9458         (pass_manager::gt_pch_nx_with_op): New.
9459         (pass_manager::operator new): Use
9460         ggc_internal_cleared_alloc_stat rather than xcalloc.
9461         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
9462         (pass_manager::gt_ggc_mx): New.
9463         (pass_manager::gt_pch_nx): New.
9464         (pass_manager::gt_pch_nx_with_op): New.
9465         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
9466         (opt_pass::operator new): New.
9467         (opt_pass::gt_ggc_mx): New.
9468         (opt_pass::gt_pch_nx): New.
9469         (opt_pass::gt_pch_nx_with_op): New.
9471 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
9473         * Makefile.in (GTFILES): Add context.h.
9474         * context.c (gcc::context::operator new): New.
9475         (gcc::context::gt_ggc_mx): New.
9476         (gcc::context::gt_pch_nx): New.
9477         (gcc::context::gt_pch_nx): New.
9478         * context.h (gcc::context): Add GTY((user)) marking.
9479         (gcc::context::operator new): New.
9480         (gcc::context::gt_ggc_mx): New.
9481         (gcc::context::gt_pch_nx): New.
9482         (gcc::context::gt_pch_nx): New.
9483         (g): Add GTY marking.
9484         (gt_ggc_mx (gcc::context *)): New.
9485         (gt_pch_nx (gcc::context *)): New.
9486         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
9487         void *cookie)): New.
9488         * gengtype.c (open_base_files) <ifiles>: Add context.h.
9490 2013-08-20  Alan Modra  <amodra@gmail.com>
9492         PR target/57865
9493         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
9494         (rs6000_emit_epilogue): Likewise.
9496 2013-08-19  Dehao Chen  <dehao@google.com>
9498         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
9500 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
9501             Jakub Jelinek  <jakub@redhat.com>
9503         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
9504         (BUILT_IN_FABSD64): Likewise.
9505         (BUILT_IN_FABSD128): Likewise.
9506         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
9507         (fold_builtin_1): Likewise.
9508         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
9509         destination and source operands.
9510         (*abstd2_fpr): Likewise.
9511         (*nabstd2_fpr): Likewise.
9513 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
9515         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
9516         JUMP_P and INSN_P.
9518 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
9520         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
9521         incomplete.
9523 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9525         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
9526         * builtins.c (default_libc_has_function): New.
9527         (gnu_libc_has_function): Ditto.
9528         (no_c99_libc_has_function): Ditto.
9529         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
9530         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
9531         (fold_builtin_sincos): Likewise.
9532         (fold_builtin_cexp): Likewise.
9533         * builtins.def (DEF_C94_BUILTIN): Likewise.
9534         (DEF_C99_BUILTIN): Likewise.
9535         (DEF_C99_C90RES_BUILTIN): Likewise.
9536         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
9537         definitions to using this define.
9538         * config/darwin-protos.h (darwin_libc_has_function): New.
9539         * config/darwin.c (darwin_libc_has_function): Ditto.
9540         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
9541         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
9542         * config/darwin.h: Ditto.
9543         * config/elfos.h: Ditto.
9544         * config/freebsd.h: Ditto.
9545         * config/i386/cygming.h: Ditto.
9546         * config/i386/djgpp.h: Ditto.
9547         * config/i386/i386-interix.h: Ditto.
9548         * config/microblaze/microblaze.h: Ditto.
9549         * config/mmix/mmix.h: Ditto.
9550         * config/gnu-user.h: Ditto.
9551         * config/ia64/hpux.h: Ditto.
9552         * config/pa/pa-hpux.h: Ditto.
9553         * config/pdp11/pdp11.h: Ditto.
9554         * config/picochip/picochip.h: Ditto.
9555         * config/linux.h: Ditto.
9556         * config/netbsd.h: Ditto.
9557         * config/openbsd.h: Ditto.
9558         * config/rs6000/aix43.h: Ditto.
9559         * config/rs6000/aix51.h: Ditto.
9560         * config/rs6000/aix52.h: Ditto.
9561         * config/rs6000/aix53.h: Ditto.
9562         * config/rs6000/aix61.h: Ditto.
9563         * config/rs6000/darwin.h: Ditto.
9564         * config/rs6000/linux.h: Ditto.
9565         * config/rs6000/linux64.h: Ditto.
9566         * config/s390/tpf.h: Ditto.
9567         * config/sol2-10.h: Ditto.
9568         * config/sol2.h: Ditto.
9569         * config/vms/vms.h: Ditto.
9570         * config/vxworks.h: Ditto.
9571         * config/linux-android.c (linux_android_libc_has_function):
9572         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
9573         * config/linux-protos.h (linux_android_libc_has_function):
9574         New declaration.
9575         * config/i386/i386.c (ix86_libc_has_function): New.
9576         * config/i386/i386-protos.h
9577         (ix86_libc_has_function): New declaration.
9578         * config/i386/i386.md
9579         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
9580         ("isinf<mode>2): Likewise.
9581         * convert.c (convert_to_integer): Using new target hook
9582         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
9583         TARGET_C99_FUNCTIONS.
9584         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
9585         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
9586         * coretypes.h (function_class): New enum for different
9587         classes of functions.
9588         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
9589         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
9590         (TARGET_HAS_SINCOS): Likewise.
9591         (TARGET_LIBC_HAS_FUNCTION): New.
9592         * doc/tm.texi: Regenerated.
9593         * targhooks.h (default_libc_has_function): New declaration.
9594         (no_c99_libc_has_function): Ditto.
9595         (gnu_libc_has_function): Ditto.
9596         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
9597         and TARGET_HAS_SINCOS.
9599 2013-08-18  Jan Hubicka  <jh@suse.cz>
9601         * Makeifle-in (ipa-devirt.o): New.
9602         (GTFILES): Add ipa-utils.h and ipa-devirt.c
9603         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
9604         (analyze_functions): Look into possible targets of polymorphic call.
9605         * dumpfile.c (dump_files): Add type-inheritance dump.
9606         * dumpfile.h (TDI_inheritance): New.
9607         * ipa-devirt.c: New file.
9608         * ipa-utils.h (odr_type_d): Forward declare.
9609         (odr_type): New type.
9610         (build_type_inheritance_graph): Declare.
9611         (possible_polymorphic_call_targets): Declare and introduce inline
9612         variant when only edge is pased.
9613         (dump_possible_polymorphic_call_targets): Likewise.
9614         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
9615         * tree.c (type_in_anonymous_namespace_p): Break out from ...
9616         (types_same_for_odr): ... here.
9617         * tree.h (type_in_anonymous_namespace_p): Declare.
9619 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
9621         PR tree-optimization/58006
9622         * tree-parloops.c (take_address_of): Don't ICE if get_name
9623         returns NULL.
9624         (eliminate_local_variables_stmt): Remove clobber stmts.
9626 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
9628         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
9629         error message is issued for an alias to undefined symbol.
9631 2013-08-18  Jan Hubicka  <jh@suse.cz>
9633         * cgraph.c (cgraph_create_indirect_edge): Discover
9634         polymorphic calls and record basic info into indirect_info.
9635         * gimple-fold.c (gimple_fold_call): When doing BINFO based
9636         devirtualization, ignore objc function calls.
9637         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9638         call with no parm index info.
9639         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9640         * tree.c (virtual_method_call_p): New function.
9641         * tree.h (virtual_method_call_p): Declare.
9643 2013-08-16  Jan Hubicka  <jh@suse.cz>
9645         PR middle-end/58179
9646         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9648 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9650         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9651         attribute.
9653 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9655         * gengtype.c (type_for_name): Add special-case support for
9656         locating types within the "gcc::" namespace.
9657         (open_base_files): Emit a "using namespace gcc" directive.
9659 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9661         PR target/58160
9662         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9663         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9665         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9666         array instead of each individual operand as a separate argument.
9667         (emit_fusion_gpr_load): Likewise.
9668         (expand_fusion_gpr_load): Add new function declaration.
9670         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9671         signature to have the operands passed as an array, instead of as
9672         separate arguments.  Allow ZERO_EXTEND to be in the memory
9673         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9674         depend on the register live/dead flags when peepholes are run.
9675         (expand_fusion_gpr_load): New function to be called from the
9676         peephole2 pass, to change the register that addis sets to be the
9677         target register.
9678         (emit_fusion_gpr_load): Change the calling signature to have the
9679         operands passed as an array, instead of as separate arguments.
9680         Allow ZERO_EXTEND to be in the memory address, and also
9681         SIGN_EXTEND if -mpower8-fusion-sign.
9683         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9684         unspec enumeration.
9685         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9686         adjust the register addis loads up in the peephole2 pass.  Do not
9687         depend on the register live/dead state when the peephole pass is done.
9689 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9691         * gengtype.c (create_user_defined_type): Ensure that the kind
9692         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9693         declaration is seen before the GTY((user)) marking.
9695 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9697         PR target/58105
9698         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9700 2013-08-16  Jan Hubicka  <jh@suse.cz>
9702         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9703         arugment expected_type.
9704         (gimple_fold_call): Use it.
9705         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9706         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9707         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9708         (try_make_edge_direct_virtual_call): Likewise.
9709         * tree.c (obj_type_ref_class): New.
9710         * tree.h (obj_type_ref_class): Use it.
9712 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9714         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9715         (rtl_slim_pp): Likewise.
9716         (init_rtl_slim_pretty_print): Likewise.
9717         (dump_value_slim):  Don't call it.  Use local pretty printer.
9718         (dump_insn_slim): Likewise.
9719         (dump_rtl_slim): Likewise.
9720         (str_pattern_slim): Likewise.
9721         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9722         Simplify.
9724 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9726         PR tree-optimization/58164
9727         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9728         walk gimple_goto_dest of GIMPLE_GOTO.
9730         PR tree-optimization/58165
9731         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9732         bi_call must be the last stmt in a bb, don't split_block, instead
9733         use fallthru edge from it and give up if there is none.
9734         Release conds vector when returning early.
9736 2013-08-14  Xinliang David Li  <davidxl@google.com>
9738         * config/i386/i386.c (ix86_option_override_internal):
9739         Remove unused variable and field.
9741 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9743         PR target/57949
9744         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9745         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9746         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9747         and Linux, correct BLKmode alignment when 128-bit alignment is
9748         required and compatibility flag is not set.
9749         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9750         for zero-size arguments when compatibility flag is not set.
9752 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9754         PR tree-optimization/58145
9755         * tree-sra.c (build_ref_for_offset): If prev_base has
9756         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9758 2013-08-14  Xinliang David Li  <davidxl@google.com>
9760         * config/i386/i386.c (ix86_option_override_internal):
9761         Fix uninitialized variable error.
9763 2013-08-14  Xinliang David Li  <davidxl@google.com>
9765         * config/i386/i386.opt: Define two new options.
9766         * config/i386/x86-tune.def: Add arch selector field in macros.
9767         * config/i386/i386.h: Adjust macro definition.
9768         * config/i386/i386.c (ix86_option_override_internal):
9769         Refactor the code.
9770         (parse_mtune_ctrl_str): New function.
9771         (set_ix86_tune_features): New function.
9772         (ix86_function_specific_restore): Call the new helper function.
9774 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9776         PR rtl-optimization/57662
9777         * sel-sched.c (code_motion_process_successors): When the current insn
9778         is removed after the recursive traversal, break from the loop.
9779         Add comments and debug printouts.
9781 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9782             Alexandre Oliva  <aoliva@redhat.com>
9784         PR target/58067
9785         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9786         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9787         there also UNSPEC_PLTOFF.
9789 2013-08-14  Marek Polacek  <polacek@redhat.com>
9791         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9792         NUM_CONDITIONS bit positions.
9794 2013-08-13  Cary Coutant  <ccoutant@google.com>
9796         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9797         (attr_checksum): Hash vector contents instead of pointer.
9798         (attr_checksum_ordered): Likewise.
9800 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9802         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9803         when Pmode != word_mode.  Add length_address attribute.
9804         (sse3_monitor_<mode>): Merge from sse3_monitor and
9805         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9806         Pmode != word_mode.  Update insn length attribute.
9807         * config/i386/i386.c (ix86_option_override_internal): Update
9808         ix86_gen_monitor selection for merged sse3_monitor insn.
9810 2013-08-13  Julian Brown  <julian@codesourcery.com>
9812         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9813         perform invalid legitimization on greater-than-word-size modes for
9814         TARGET_E500_DOUBLE.
9816 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9818         * ira.c (setup_class_translate_array): Use aclass instead of cl
9819         for classes not fully covered by allocno classes.
9821 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9823         PR tree-optimization/57661
9824         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9825         * tree-inline.c (tree_function_versioning): Initialize it.
9826         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9827         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9828         that is not being copied.
9830         PR sanitizer/56417
9831         * asan.c (instrument_strlen_call): Fix typo in comment.
9832         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9834 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9836         * config/mips/mips.md (prefetch): Use lw instead of ld on
9837         loongson in 32bit mode.
9839 2013-08-13  Nick Clifton  <nickc@redhat.com>
9841         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9843 2013-08-13  Jan Hubicka  <jh@suse.cz>
9845         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9846         introduced edge; fix typo in sanity check.
9847         (cgraph_resolve_speculation): Export; improve diagnostic.
9848         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9849         speculation at type mismatch.
9850         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9851         (cgraph_resolve_speculation): Declare.
9852         (symtab_can_be_discarded): New function.
9853         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9854         * ipa-inline-transform.c (speculation_removed): New global var.
9855         (clone_inlined_nodes): See if speculation can be removed.
9856         (inline_call): If speculations was removed, we growths may not match.
9857         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9858         (speculation_useful_p): New function.
9859         (resolve_noninline_speculation): New function.
9860         (inline_small_functions): Resolve useless speculations.
9861         * ipa-inline.h (speculation_useful_p): Declare
9862         * ipa.c (can_replace_by_local_alias): Simplify.
9863         (ipa_profile): Produce speculative calls in non-lto, too;
9864         add simple cost model; produce local aliases.
9866 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9868         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9869         PASS_MANAGER_H.
9871 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9873         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9875 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9877         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9878         iterator instead of X87MODEF.
9880 2013-08-12  Perez Read  <netfirewall@gmail.com>
9882         PR target/58132
9883         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9884         operand 0 for intel asm alternative.
9885         (*movabs<mode>_2): Ditto for operand 1.
9887 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9889         * config/aarch64/arm_none.h
9890         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9892 2013-08-12  Nick Clifton  <nickc@redhat.com>
9894         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9896 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9898         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9899         expand for QI/HImode operand to produce more effictive code for
9900         unsigned char(short) --> float(double) conversion.
9902 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9904         * doc/invoke.texi: Mention that -ftls-model does not force the final
9905         model.
9907 2013-08-12  Marek Polacek  <polacek@redhat.com>
9908             Marc Glisse  <marc.glisse@inria.fr>
9910         PR tree-optimization/57980
9911         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9912         when creating -1 constant.
9914 2013-08-10  Jan Hubicka  <jh@suse.cz>
9916         Workaround binutils PR14342.
9917         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9918         (gimple_init_edge_profiler): Likewise.
9919         (gimple_gen_ic_func_profiler): Likewise.
9921 2013-08-09  Jan Hubicka  <jh@suse.cz>
9923         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9925 2013-08-09  Xinliang David Li  <davidxl@google.com>
9927         * config/i386/stringop.def: New file.
9928         * config/i386/stringop.opt: New file.
9929         * config/i386/i386-opts.h: Include stringopt.def.
9930         * config/i386/i386.opt: Include stringopt.opt.
9931         * config/i386/i386.c (ix86_option_override_internal):
9932         Override default size based stringop inline strategies with options.
9933         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9934         New function.
9936 2013-08-09  Jan Hubicka  <jh@suse.cz>
9938         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9940 2013-08-09  Jan Hubicka  <jh@suse.cz>
9942         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9943         CGRAPH_FREQ_MAX.
9944         (dump_cgraph_node): Dump profile-id.
9945         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9946         and common_target_probability.
9947         * lto-cgraph.c (lto_output_edge): Stream common targets.
9948         (lto_output_node): Stream profile ids.
9949         (input_node): Stream profile ids.
9950         (input_edge): Stream common targets.
9951         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9952         * ipa.c: Include value-prof.h
9953         (ipa_profile_generate_summary): Turn indirect call statement histograms
9954         into common targets.
9955         (ipa_profile): Turn common targets into speculative edges.
9957 2013-08-09  Jan Hubicka  <jh@suse.cz>
9959         * cgraph.h (cgraph_node): Add profile_id.
9960         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9961         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9962         (del_node_map): Update.
9963         (find_func_by_funcdef_no): Replace by ...
9964         (find_func_by_profile_id): ... this one.
9965         (gimple_ic_transform): Do not remove useful histograms when
9966         speculation is not done; dump info when indirect call removal
9967         can happen at LTO.
9968         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9969         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9970         (__gcov_indirect_call_profiler_v2): .. this one.
9971         * profile.h (init_node_map): Update.
9972         * coverage.c (coverage_compute_profile_id): New function.
9973         * coverage.h (coverage_compute_profile_id): Declare.
9974         * tree-profile.c (init_ic_make_global_vars): Make
9975         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9976         (gimple_init_edge_profiler): Update prototype of
9977         __gcov_indirect_call_profiler.
9978         (gimple_gen_ic_func_profiler): Simplify.
9979         (tree_profiling): Use init_node_map
9981 2013-08-09  Jan Hubicka  <jh@suse.cz>
9983         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9984         non-speculative refs.
9985         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9986         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9987         (cgraph_set_call_stmt): Likewise.
9988         (cgraph_create_edge_1): Fix release checking compilatoin;
9989         clear lto_stmt_uid.
9990         (cgraph_free_edge): Free indirect info.
9991         (cgraph_turn_edge_to_speculative): New function.
9992         (cgraph_speculative_call_info): New function.
9993         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9994         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9995         (dump_cgraph_node): Dump speculation.
9996         (verify_edge_count_and_frequency): Accept speculative edges.
9997         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9998         (verify_cgraph_node): Handle speculation.
9999         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
10000         (cgraph_set_call_stmt): Update prototype.
10001         (cgraph_make_edge_direct): Update prototype.
10002         (cgraph_speculative_call_info): Declare.
10003         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
10004         to change; update call of ipa_find_references.
10005         * ipa-ref.c (ipa_record_reference): Fix return value; clear
10006         lto_stmt_uid and speculative flags.
10007         (ipa_dump_references): Dump speculation.
10008         (ipa_clone_references): Clone speculative flag.
10009         (ipa_clone_referring): Likewise.
10010         (ipa_clone_ref): New function.
10011         (ipa_find_reference): Look into lto_stmt_uids
10012         (ipa_clear_stmts_in_references): Do not clear speculative calls.
10013         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
10014         (ipa_find_reference): Update declaration.
10015         (ipa_clone_ref): Declare.
10016         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
10017         stream speculative flag.
10018         (lto_output_ref): Stream statements uids and speculation.
10019         (input_ref): Likewise.
10020         (input_edge): Stream speuclation.
10021         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
10022         (cgraph_set_call_stmt_including_clones): Handle speculation.
10023         * ipa-inline.c (heap_edge_removal_hook): New function.
10024         (inline_small_functions): Register it.
10025         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
10026         also initialize refs.
10027         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
10028         edge to change.
10029         (try_make_edge_direct_simple_call): Likewise.
10030         (try_make_edge_direct_simple_call): Likewise.
10031         (update_indirect_edges_after_inlining): Likewise.
10032         (remove_described_reference): Look proper lto_stmt_uid.
10033         (propagate_controlled_uses): Likewise.
10034         (propagate_controlled_uses): Liekwise.
10035         * tree-inline.c (copy_bb): Copy speculative edges.
10036         (redirect_all_calls): New function.
10037         (copy_cfg_body): Do redirection after loop info is updated.
10038         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
10040 2013-08-09  Jan Hubicka  <jh@suse.cz>
10042         * lto-streamer-out.c (output_function): Renumber PHIs.
10043         * lto-streamer-in.c (input_function): Likewise.
10045 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
10047         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
10048         (get_lane_unsigned): Likewise.
10049         (dup_lane_scalar): Likewise.
10050         (get_lane): enable for VALL.
10051         * config/aarch64/aarch64-simd.md
10052         (aarch64_dup_lane_scalar<mode>): Remove.
10053         (aarch64_get_lane_signed<mode>): Likewise.
10054         (aarch64_get_lane_unsigned<mode>): Likewise.
10055         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
10056         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
10057         (aarch64_get_lane<mode>): Enable for all vector modes.
10058         (aarch64_get_lanedi): Remove misleading constraints.
10059         * config/aarch64/arm_neon.h
10060         (__aarch64_vget_lane_any): Define.
10061         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
10062         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
10063         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
10064         * config/aarch64/iterators.md (VDQQH): New.
10065         (VDQQHS): Likewise.
10066         (vwcore): Likewise.
10068 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
10070         * configure.ac: Add GAS check for LEON instructions on SPARC.
10071         * configure: Regenerate.
10072         * config.in: Likewise.
10073         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
10074         sparc*-*-* block.
10075         * config/sparc/sparc.opt (LEON, LEON3): New masks.
10076         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
10077         for LEON or LEON3.
10078         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
10079         (AS_LEON_FLAG): New macro.
10080         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
10081         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
10082         Deal with LEON and LEON3 for the memory model.
10083         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
10084         (atomic_compare_and_swap<mode>_1): Likewise.
10085         (*atomic_compare_and_swap<mode>_1): Likewise.
10087 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10089         * config/arm/neon.md (vcond): Fix floating-point vector
10090         comparisons against 0.
10092 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
10094         * lra-constraints.c (emit_spill_move): Remove assert.
10095         (process_alt_operands): Add more debugging
10096         output.  Increase reject for spilling into memory.  Decrease
10097         reject for reloading scratch.
10098         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
10100 2013-08-08  Steve Ellcey  <sellcey@mips.com>
10102         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
10103         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
10104         micromips incompatible.  Add nan2008.
10105         (MULTILIB_DIRNAMES): Add nan2008.
10106         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
10107         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
10108         and micromips incompatible.  Add nan2008.
10109         (MULTILIB_DIRNAMES): Add nan2008.
10110         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
10112 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
10114         PR rtl-optimization/58079
10115         * combine.c (combine_simplify_rtx): Avoid using SUBST if
10116         simplify_comparison has widened a comparison with an integer.
10118 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10120         * config/arm/neon.md (movmisalign<mode>): Disable when we
10121         don't allow unaligned accesses.
10122         (*movmisalign<mode>_neon_store): Likewise.
10123         (*movmisalign<mode>_neon_load): Likewise.
10124         (*movmisalign<mode>_neon_store): Likewise.
10125         (*movmisalign<mode>_neon_load): Likewise.
10127 2013-08-08  Jan Hubicka  <jh@suse.cz>
10129         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
10130         (make_pass_rebuild_cgraph_edges): Also clear references.
10131         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
10132         * ipa-inline-transform.c (inline_transform): Remove all references
10133         after inlining.
10134         * cgraphunit.c (expand_function): Remove all references after
10135         expansion.
10136         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
10137         (ipa_find_reference): Rewrite to iterator.
10138         (remove_stmt_references): Likewise.
10139         (ipa_clear_stmts_in_references): New function.
10140         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
10141         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
10142         clear references.
10143         * ipa-split.c (split_function): Remove references in split function.
10145 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
10147         PR target/57431
10148         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
10149         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
10151 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
10153         PR target/56979
10154         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
10155         suggested mode for the assignment isn't compatible with the
10156         registers required.
10158 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
10160         PR target/58065
10161         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
10163 2013-08-07  Xinliang David Li  <davidxl@google.com>
10165         * config/i386/i386.opt: New option -mtune-ctrl=.
10166         * config/i386/x86-tune.def: New file.
10167         * config/i386/i386.h: include x86-tune.def.
10168         * config/i386/i386.c (ix86_option_override_internal):
10169         Parsing -mtune-ctrl= option and set tune features.
10171 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
10173         PR other/12081
10174         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
10175         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
10176         to gen_2arg_fn_t.
10178 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
10180         * rtl.h (update_alignments): Declare.
10181         * final.c (grow_label_align): New function extracted from...
10182         (shorten_branches): ...here.  Call it.
10183         (update_alignments): New function.
10184         * reorg.c (sibling_labels): New variable.
10185         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
10186         the new label along with it onto the sibling_labels vector.
10187         (fill_simple_delay_slots): Adjust call to get_label_before.
10188         (fill_slots_from_thread): Likewise.
10189         (relax_delay_slots): Likewise.
10190         (make_return_insns): Likewise.
10191         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
10193 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
10195         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
10196         document its semantics.
10197         (diagnostic_report_diagnostic): Adjust accordingly.
10199 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
10201         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
10202         (sparc_option_override): ...and port to new C++ pass API.
10203         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
10205 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
10207         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
10209 2013-08-06  Caroline Tice  <cmtice@google.com>
10211         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
10212         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
10213         * tree-pass.h: Add pass_vtable_verify.
10214         * varasm.c (assemble_variable): Add code to properly set the comdat
10215         section and name for the .vtable_map_vars section.
10216         (assemble_vtyv_preinit_initializer): New function.
10217         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
10218         LINK_ONCE flag.
10219         * output.h: Add function decl for assemble_vtv_preinit_initializer.
10220         * vtable-verify.c: New file.
10221         * vtable-verify.h: New file.
10222         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
10223         initialiation levels.
10224         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
10225         * passes.def: Insert pass_vtable_verify.
10226         * aclocal.m4: Reorder includes.
10227         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
10228         -fvtv-counts options.
10229         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
10230         as appropriate, if -fvtable-verify=... is used.
10231         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
10232         -fvtable-verify=... is used.
10233         * Makefile.in (OBJS):  Add vtable-verify.o to list.
10234         (vtable-verify.o): Add new build rule.
10235         (GTFILES): Add vtable-verify.c to list.
10236         * common.opt (fvtable-verify=): New flag.
10237         (vtv_priority): Values for fvtable-verify= flag.
10238         (fvtv-counts): New flag.
10239         (fvtv-debug): New flag.
10240         * tree.h (save_vtable_map_decl): New extern function decl.
10242 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
10244         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
10245         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
10246         (pass_data_rl78_devirt): ...new pass_data instance and...
10247         (make_pass_rl78_devirt): ...new function.
10248         (rl78_asm_file_start): Port pass registration to new C++ API.
10250 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
10252         * coretypes.h (rtl_opt_pass): Add.
10253         (gcc::context): Add.
10254         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
10255         (epiphany_init): Port to new C++ pass API.
10256         (epiphany_optimize_mode_switching): Likewise.
10257         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
10258         (pass_manager::get_pass_mode_switching): New.
10259         (pass_manager::get_pass_peephole2): New.
10260         * mode-switching.c (pass_mode_switching): Add clone method.
10261         * recog.c (pass_peephole2): Add clone method.
10262         (pass_split_all_insns): Add clone method.
10264 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
10266         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
10267         (mips_option_override): ...here, porting to new C++ API for passes.
10269 2013-08-06  Jan Hubicka  <jh@suse.cz>
10271         * cgraph.c (cgraph_get_body): New function based on lto.c
10272         implementation.
10273         * cgraph.h (cgraph_get_body): Declare.
10274         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
10275         LTO paths.
10276         * cgraphunit.c (expand_function): Get body prior expanding.
10277         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
10278         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
10279         really need.
10280         * passes.c (do_per_function_toporder): Get body.
10281         * tree-inline.c (expand_call_inline): Get body prior inlining it.
10282         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
10284 2013-08-06  Martin Jambor  <mjambor@suse.cz>
10286         PR fortran/57987
10287         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
10288         is not re-finalized.  Rename second parameter to no_collect.
10290 2013-08-06  Martin Jambor  <mjambor@suse.cz>
10292         PR middle-end/58041
10293         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
10294         MEM_REF has proper alignment information.
10296 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
10298         PR other/12081
10299         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
10300         class insn_gen_fn.
10301         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
10302         rtx (*) (rtx, ...) with insn_gen_fn.
10303         * genoutput.c (output_insn_data): Cast gen_? function pointers to
10304         insn_gen_fn::stored_funcptr.  Add initializer braces.
10306 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10308         Rewrite how instances of passes are cloned to remove assumptions
10309         about their sizes (thus allowing pass subclasses to have
10310         additional data fields, albeit non-GC-managed ones at this point).
10312         * passes.c (make_pass_instance): Now that passes have clone
10313         methods, rewrite this function to eliminate XNEW and memcpy
10314         calls that used hardcoded sizes.  Since this function no longer
10315         creates pass instances, rename it to...
10316         (add_pass_instance): ...this.  Document the old way that passes were
10317         numbered and flagged, and rework this function to continue using it.
10318         (next_pass_1): Add an initial_pass argument for use by
10319         add_pass_instance.
10320         (position_pass): When adding multiple instances of a pass, use
10321         the pass's clone method, rather than relying on the XNEW/memcpy
10322         within the former make_pass_instance (now add_pass_instance).
10323         (pass_manager::pass_manager): When invoking next_pass_1, also supply
10324         the initial instance of the current pass within the pass manager.
10326 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10328         This is the automated part of the conversion of passes from C
10329         structs to C++ classes.
10331         Patch autogenerated by refactor_passes.py from
10332         https://github.com/davidmalcolm/gcc-refactoring-scripts
10333         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
10335         * asan.c (pass_asan): Convert from a global struct to a subclass of
10336         gimple_opt_pass along with...
10337         (pass_data_asan): ...new pass_data instance and...
10338         (make_pass_asan): ...new function.
10339         (pass_asan_O0): Convert from a global struct to a subclass of
10340         gimple_opt_pass along with...
10341         (pass_data_asan_O0): ...new pass_data instance and...
10342         (make_pass_asan_O0): ...new function.
10343         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
10344         subclass of rtl_opt_pass along with...
10345         (pass_data_inc_dec): ...new pass_data instance and...
10346         (make_pass_inc_dec): ...new function.
10347         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
10348         a subclass of rtl_opt_pass along with...
10349         (pass_data_reorder_blocks): ...new pass_data instance and...
10350         (make_pass_reorder_blocks): ...new function.
10351         (pass_duplicate_computed_gotos): Convert from a global struct to a
10352         subclass of rtl_opt_pass along with...
10353         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
10354         (make_pass_duplicate_computed_gotos): ...new function.
10355         (pass_partition_blocks): Convert from a global struct to a subclass of
10356         rtl_opt_pass along with...
10357         (pass_data_partition_blocks): ...new pass_data instance and...
10358         (make_pass_partition_blocks): ...new function.
10359         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
10360         struct to a subclass of rtl_opt_pass along with...
10361         (pass_data_branch_target_load_optimize1): ...new pass_data instance
10362         and...
10363         (make_pass_branch_target_load_optimize1): ...new function.
10364         (pass_branch_target_load_optimize2): Convert from a global struct to a
10365         subclass of rtl_opt_pass along with...
10366         (pass_data_branch_target_load_optimize2): ...new pass_data instance
10367         and...
10368         (make_pass_branch_target_load_optimize2): ...new function.
10369         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
10370         of rtl_opt_pass along with...
10371         (pass_data_jump): ...new pass_data instance and...
10372         (make_pass_jump): ...new function.
10373         (pass_jump2): Convert from a global struct to a subclass of
10374         rtl_opt_pass along with...
10375         (pass_data_jump2): ...new pass_data instance and...
10376         (make_pass_jump2): ...new function.
10377         * cfgexpand.c (pass_expand): Convert from a global struct to a
10378         subclass of rtl_opt_pass along with...
10379         (pass_data_expand): ...new pass_data instance and...
10380         (make_pass_expand): ...new function.
10381         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
10382         of rtl_opt_pass along with...
10383         (pass_data_free_cfg): ...new pass_data instance and...
10384         (make_pass_free_cfg): ...new function.
10385         (pass_into_cfg_layout_mode): Convert from a global struct to a
10386         subclass of rtl_opt_pass along with...
10387         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
10388         (make_pass_into_cfg_layout_mode): ...new function.
10389         (pass_outof_cfg_layout_mode): Convert from a global struct to a
10390         subclass of rtl_opt_pass along with...
10391         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
10392         (make_pass_outof_cfg_layout_mode): ...new function.
10393         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
10394         struct to a subclass of gimple_opt_pass along with...
10395         (pass_data_build_cgraph_edges): ...new pass_data instance and...
10396         (make_pass_build_cgraph_edges): ...new function.
10397         (pass_rebuild_cgraph_edges): Convert from a global struct to a
10398         subclass of gimple_opt_pass along with...
10399         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
10400         (make_pass_rebuild_cgraph_edges): ...new function.
10401         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
10402         subclass of gimple_opt_pass along with...
10403         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
10404         and...
10405         (make_pass_remove_cgraph_callee_edges): ...new function.
10406         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
10407         struct to a subclass of rtl_opt_pass along with...
10408         (pass_data_stack_adjustments): ...new pass_data instance and...
10409         (make_pass_stack_adjustments): ...new function.
10410         * combine.c (pass_combine): Convert from a global struct to a subclass
10411         of rtl_opt_pass along with...
10412         (pass_data_combine): ...new pass_data instance and...
10413         (make_pass_combine): ...new function.
10414         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
10415         global struct to a subclass of rtl_opt_pass along with...
10416         (pass_data_compare_elim_after_reload): ...new pass_data instance
10417         and...
10418         (make_pass_compare_elim_after_reload): ...new function.
10419         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
10420         of rtl_opt_pass along with...
10421         (pass_data_rtl_cprop): ...new pass_data instance and...
10422         (make_pass_rtl_cprop): ...new function.
10423         * cse.c (pass_cse): Convert from a global struct to a subclass of
10424         rtl_opt_pass along with...
10425         (pass_data_cse): ...new pass_data instance and...
10426         (make_pass_cse): ...new function.
10427         (pass_cse2): Convert from a global struct to a subclass of
10428         rtl_opt_pass along with...
10429         (pass_data_cse2): ...new pass_data instance and...
10430         (make_pass_cse2): ...new function.
10431         (pass_cse_after_global_opts): Convert from a global struct to a
10432         subclass of rtl_opt_pass along with...
10433         (pass_data_cse_after_global_opts): ...new pass_data instance and...
10434         (make_pass_cse_after_global_opts): ...new function.
10435         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
10436         of rtl_opt_pass along with...
10437         (pass_data_ud_rtl_dce): ...new pass_data instance and...
10438         (make_pass_ud_rtl_dce): ...new function.
10439         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
10440         rtl_opt_pass along with...
10441         (pass_data_fast_rtl_dce): ...new pass_data instance and...
10442         (make_pass_fast_rtl_dce): ...new function.
10443         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
10444         a subclass of rtl_opt_pass along with...
10445         (pass_data_df_initialize_opt): ...new pass_data instance and...
10446         (make_pass_df_initialize_opt): ...new function.
10447         (pass_df_initialize_no_opt): Convert from a global struct to a
10448         subclass of rtl_opt_pass along with...
10449         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
10450         (make_pass_df_initialize_no_opt): ...new function.
10451         (pass_df_finish): Convert from a global struct to a subclass of
10452         rtl_opt_pass along with...
10453         (pass_data_df_finish): ...new pass_data instance and...
10454         (make_pass_df_finish): ...new function.
10455         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
10456         rtl_opt_pass along with...
10457         (pass_data_rtl_dse1): ...new pass_data instance and...
10458         (make_pass_rtl_dse1): ...new function.
10459         (pass_rtl_dse2): Convert from a global struct to a subclass of
10460         rtl_opt_pass along with...
10461         (pass_data_rtl_dse2): ...new pass_data instance and...
10462         (make_pass_rtl_dse2): ...new function.
10463         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
10464         subclass of rtl_opt_pass along with...
10465         (pass_data_dwarf2_frame): ...new pass_data instance and...
10466         (make_pass_dwarf2_frame): ...new function.
10467         * except.c (pass_set_nothrow_function_flags): Convert from a global
10468         struct to a subclass of rtl_opt_pass along with...
10469         (pass_data_set_nothrow_function_flags): ...new pass_data instance
10470         and...
10471         (make_pass_set_nothrow_function_flags): ...new function.
10472         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
10473         subclass of rtl_opt_pass along with...
10474         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
10475         and...
10476         (make_pass_convert_to_eh_region_ranges): ...new function.
10477         * final.c (pass_compute_alignments): Convert from a global struct to a
10478         subclass of rtl_opt_pass along with...
10479         (pass_data_compute_alignments): ...new pass_data instance and...
10480         (make_pass_compute_alignments): ...new function.
10481         (pass_final): Convert from a global struct to a subclass of
10482         rtl_opt_pass along with...
10483         (pass_data_final): ...new pass_data instance and...
10484         (make_pass_final): ...new function.
10485         (pass_shorten_branches): Convert from a global struct to a subclass of
10486         rtl_opt_pass along with...
10487         (pass_data_shorten_branches): ...new pass_data instance and...
10488         (make_pass_shorten_branches): ...new function.
10489         (pass_clean_state): Convert from a global struct to a subclass of
10490         rtl_opt_pass along with...
10491         (pass_data_clean_state): ...new pass_data instance and...
10492         (make_pass_clean_state): ...new function.
10493         * function.c (pass_instantiate_virtual_regs): Convert from a global
10494         struct to a subclass of rtl_opt_pass along with...
10495         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
10496         (make_pass_instantiate_virtual_regs): ...new function.
10497         (pass_leaf_regs): Convert from a global struct to a subclass of
10498         rtl_opt_pass along with...
10499         (pass_data_leaf_regs): ...new pass_data instance and...
10500         (make_pass_leaf_regs): ...new function.
10501         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
10502         subclass of rtl_opt_pass along with...
10503         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
10504         and...
10505         (make_pass_thread_prologue_and_epilogue): ...new function.
10506         (pass_match_asm_constraints): Convert from a global struct to a
10507         subclass of rtl_opt_pass along with...
10508         (pass_data_match_asm_constraints): ...new pass_data instance and...
10509         (make_pass_match_asm_constraints): ...new function.
10510         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
10511         subclass of rtl_opt_pass along with...
10512         (pass_data_rtl_fwprop): ...new pass_data instance and...
10513         (make_pass_rtl_fwprop): ...new function.
10514         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
10515         rtl_opt_pass along with...
10516         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
10517         (make_pass_rtl_fwprop_addr): ...new function.
10518         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
10519         rtl_opt_pass along with...
10520         (pass_data_rtl_pre): ...new pass_data instance and...
10521         (make_pass_rtl_pre): ...new function.
10522         (pass_rtl_hoist): Convert from a global struct to a subclass of
10523         rtl_opt_pass along with...
10524         (pass_data_rtl_hoist): ...new pass_data instance and...
10525         (make_pass_rtl_hoist): ...new function.
10526         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
10527         subclass of gimple_opt_pass along with...
10528         (pass_data_lower_cf): ...new pass_data instance and...
10529         (make_pass_lower_cf): ...new function.
10530         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
10531         from a global struct to a subclass of gimple_opt_pass along with...
10532         (pass_data_strength_reduction): ...new pass_data instance and...
10533         (make_pass_strength_reduction): ...new function.
10534         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
10535         of rtl_opt_pass along with...
10536         (pass_data_rtl_ifcvt): ...new pass_data instance and...
10537         (make_pass_rtl_ifcvt): ...new function.
10538         (pass_if_after_combine): Convert from a global struct to a subclass of
10539         rtl_opt_pass along with...
10540         (pass_data_if_after_combine): ...new pass_data instance and...
10541         (make_pass_if_after_combine): ...new function.
10542         (pass_if_after_reload): Convert from a global struct to a subclass of
10543         rtl_opt_pass along with...
10544         (pass_data_if_after_reload): ...new pass_data instance and...
10545         (make_pass_if_after_reload): ...new function.
10546         * init-regs.c (pass_initialize_regs): Convert from a global struct to
10547         a subclass of rtl_opt_pass along with...
10548         (pass_data_initialize_regs): ...new pass_data instance and...
10549         (make_pass_initialize_regs): ...new function.
10550         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
10551         of ipa_opt_pass_d along with...
10552         (pass_data_ipa_cp): ...new pass_data instance and...
10553         (make_pass_ipa_cp): ...new function.
10554         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
10555         global struct to a subclass of gimple_opt_pass along with...
10556         (pass_data_inline_parameters): ...new pass_data instance and...
10557         (make_pass_inline_parameters): ...new function.
10558         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
10559         subclass of gimple_opt_pass along with...
10560         (pass_data_early_inline): ...new pass_data instance and...
10561         (make_pass_early_inline): ...new function.
10562         (pass_ipa_inline): Convert from a global struct to a subclass of
10563         ipa_opt_pass_d along with...
10564         (pass_data_ipa_inline): ...new pass_data instance and...
10565         (make_pass_ipa_inline): ...new function.
10566         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
10567         struct to a subclass of gimple_opt_pass along with...
10568         (pass_data_local_pure_const): ...new pass_data instance and...
10569         (make_pass_local_pure_const): ...new function.
10570         (pass_ipa_pure_const): Convert from a global struct to a subclass of
10571         ipa_opt_pass_d along with...
10572         (pass_data_ipa_pure_const): ...new pass_data instance and...
10573         (make_pass_ipa_pure_const): ...new function.
10574         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
10575         to a subclass of ipa_opt_pass_d along with...
10576         (pass_data_ipa_reference): ...new pass_data instance and...
10577         (make_pass_ipa_reference): ...new function.
10578         * ipa-split.c (pass_split_functions): Convert from a global struct to
10579         a subclass of gimple_opt_pass along with...
10580         (pass_data_split_functions): ...new pass_data instance and...
10581         (make_pass_split_functions): ...new function.
10582         (pass_feedback_split_functions): Convert from a global struct to a
10583         subclass of gimple_opt_pass along with...
10584         (pass_data_feedback_split_functions): ...new pass_data instance and...
10585         (make_pass_feedback_split_functions): ...new function.
10586         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
10587         global struct to a subclass of simple_ipa_opt_pass along with...
10588         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
10589         instance and...
10590         (make_pass_ipa_function_and_variable_visibility): ...new function.
10591         (pass_ipa_free_inline_summary): Convert from a global struct to a
10592         subclass of simple_ipa_opt_pass along with...
10593         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
10594         (make_pass_ipa_free_inline_summary): ...new function.
10595         (pass_ipa_whole_program_visibility): Convert from a global struct to a
10596         subclass of ipa_opt_pass_d along with...
10597         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
10598         and...
10599         (make_pass_ipa_whole_program_visibility): ...new function.
10600         (pass_ipa_profile): Convert from a global struct to a subclass of
10601         ipa_opt_pass_d along with...
10602         (pass_data_ipa_profile): ...new pass_data instance and...
10603         (make_pass_ipa_profile): ...new function.
10604         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
10605         ipa_opt_pass_d along with...
10606         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
10607         (make_pass_ipa_cdtor_merge): ...new function.
10608         * ira.c (pass_ira): Convert from a global struct to a subclass of
10609         rtl_opt_pass along with...
10610         (pass_data_ira): ...new pass_data instance and...
10611         (make_pass_ira): ...new function.
10612         (pass_reload): Convert from a global struct to a subclass of
10613         rtl_opt_pass along with...
10614         (pass_data_reload): ...new pass_data instance and...
10615         (make_pass_reload): ...new function.
10616         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
10617         subclass of rtl_opt_pass along with...
10618         (pass_data_cleanup_barriers): ...new pass_data instance and...
10619         (make_pass_cleanup_barriers): ...new function.
10620         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
10621         of rtl_opt_pass along with...
10622         (pass_data_loop2): ...new pass_data instance and...
10623         (make_pass_loop2): ...new function.
10624         (pass_rtl_loop_init): Convert from a global struct to a subclass of
10625         rtl_opt_pass along with...
10626         (pass_data_rtl_loop_init): ...new pass_data instance and...
10627         (make_pass_rtl_loop_init): ...new function.
10628         (pass_rtl_loop_done): Convert from a global struct to a subclass of
10629         rtl_opt_pass along with...
10630         (pass_data_rtl_loop_done): ...new pass_data instance and...
10631         (make_pass_rtl_loop_done): ...new function.
10632         (pass_rtl_move_loop_invariants): Convert from a global struct to a
10633         subclass of rtl_opt_pass along with...
10634         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
10635         (make_pass_rtl_move_loop_invariants): ...new function.
10636         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10637         rtl_opt_pass along with...
10638         (pass_data_rtl_unswitch): ...new pass_data instance and...
10639         (make_pass_rtl_unswitch): ...new function.
10640         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10641         subclass of rtl_opt_pass along with...
10642         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10643         and...
10644         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10645         (pass_rtl_doloop): Convert from a global struct to a subclass of
10646         rtl_opt_pass along with...
10647         (pass_data_rtl_doloop): ...new pass_data instance and...
10648         (make_pass_rtl_doloop): ...new function.
10649         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10650         a subclass of rtl_opt_pass along with...
10651         (pass_data_lower_subreg): ...new pass_data instance and...
10652         (make_pass_lower_subreg): ...new function.
10653         (pass_lower_subreg2): Convert from a global struct to a subclass of
10654         rtl_opt_pass along with...
10655         (pass_data_lower_subreg2): ...new pass_data instance and...
10656         (make_pass_lower_subreg2): ...new function.
10657         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10658         struct to a subclass of ipa_opt_pass_d along with...
10659         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10660         (make_pass_ipa_lto_gimple_out): ...new function.
10661         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10662         of ipa_opt_pass_d along with...
10663         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10664         (make_pass_ipa_lto_finish_out): ...new function.
10665         * mode-switching.c (pass_mode_switching): Convert from a global struct
10666         to a subclass of rtl_opt_pass along with...
10667         (pass_data_mode_switching): ...new pass_data instance and...
10668         (make_pass_mode_switching): ...new function.
10669         * modulo-sched.c (pass_sms): Convert from a global struct to a
10670         subclass of rtl_opt_pass along with...
10671         (pass_data_sms): ...new pass_data instance and...
10672         (make_pass_sms): ...new function.
10673         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10674         subclass of gimple_opt_pass along with...
10675         (pass_data_expand_omp): ...new pass_data instance and...
10676         (make_pass_expand_omp): ...new function.
10677         (pass_lower_omp): Convert from a global struct to a subclass of
10678         gimple_opt_pass along with...
10679         (pass_data_lower_omp): ...new pass_data instance and...
10680         (make_pass_lower_omp): ...new function.
10681         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10682         of gimple_opt_pass along with...
10683         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10684         (make_pass_diagnose_omp_blocks): ...new function.
10685         * passes.c (pass_early_local_passes): Convert from a global struct to
10686         a subclass of simple_ipa_opt_pass along with...
10687         (pass_data_early_local_passes): ...new pass_data instance and...
10688         (make_pass_early_local_passes): ...new function.
10689         (pass_all_early_optimizations): Convert from a global struct to a
10690         subclass of gimple_opt_pass along with...
10691         (pass_data_all_early_optimizations): ...new pass_data instance and...
10692         (make_pass_all_early_optimizations): ...new function.
10693         (pass_all_optimizations): Convert from a global struct to a subclass
10694         of gimple_opt_pass along with...
10695         (pass_data_all_optimizations): ...new pass_data instance and...
10696         (make_pass_all_optimizations): ...new function.
10697         (pass_all_optimizations_g): Convert from a global struct to a subclass
10698         of gimple_opt_pass along with...
10699         (pass_data_all_optimizations_g): ...new pass_data instance and...
10700         (make_pass_all_optimizations_g): ...new function.
10701         (pass_rest_of_compilation): Convert from a global struct to a subclass
10702         of rtl_opt_pass along with...
10703         (pass_data_rest_of_compilation): ...new pass_data instance and...
10704         (make_pass_rest_of_compilation): ...new function.
10705         (pass_postreload): Convert from a global struct to a subclass of
10706         rtl_opt_pass along with...
10707         (pass_data_postreload): ...new pass_data instance and...
10708         (make_pass_postreload): ...new function.
10709         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10710         subclass of rtl_opt_pass along with...
10711         (pass_data_gcse2): ...new pass_data instance and...
10712         (make_pass_gcse2): ...new function.
10713         * postreload.c (pass_postreload_cse): Convert from a global struct to
10714         a subclass of rtl_opt_pass along with...
10715         (pass_data_postreload_cse): ...new pass_data instance and...
10716         (make_pass_postreload_cse): ...new function.
10717         * predict.c (pass_profile): Convert from a global struct to a subclass
10718         of gimple_opt_pass along with...
10719         (pass_data_profile): ...new pass_data instance and...
10720         (make_pass_profile): ...new function.
10721         (pass_strip_predict_hints): Convert from a global struct to a subclass
10722         of gimple_opt_pass along with...
10723         (pass_data_strip_predict_hints): ...new pass_data instance and...
10724         (make_pass_strip_predict_hints): ...new function.
10725         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10726         of rtl_opt_pass along with...
10727         (pass_data_peephole2): ...new pass_data instance and...
10728         (make_pass_peephole2): ...new function.
10729         (pass_split_all_insns): Convert from a global struct to a subclass of
10730         rtl_opt_pass along with...
10731         (pass_data_split_all_insns): ...new pass_data instance and...
10732         (make_pass_split_all_insns): ...new function.
10733         (pass_split_after_reload): Convert from a global struct to a subclass
10734         of rtl_opt_pass along with...
10735         (pass_data_split_after_reload): ...new pass_data instance and...
10736         (make_pass_split_after_reload): ...new function.
10737         (pass_split_before_regstack): Convert from a global struct to a
10738         subclass of rtl_opt_pass along with...
10739         (pass_data_split_before_regstack): ...new pass_data instance and...
10740         (make_pass_split_before_regstack): ...new function.
10741         (pass_split_before_sched2): Convert from a global struct to a subclass
10742         of rtl_opt_pass along with...
10743         (pass_data_split_before_sched2): ...new pass_data instance and...
10744         (make_pass_split_before_sched2): ...new function.
10745         (pass_split_for_shorten_branches): Convert from a global struct to a
10746         subclass of rtl_opt_pass along with...
10747         (pass_data_split_for_shorten_branches): ...new pass_data instance
10748         and...
10749         (make_pass_split_for_shorten_branches): ...new function.
10750         * ree.c (pass_ree): Convert from a global struct to a subclass of
10751         rtl_opt_pass along with...
10752         (pass_data_ree): ...new pass_data instance and...
10753         (make_pass_ree): ...new function.
10754         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10755         subclass of rtl_opt_pass along with...
10756         (pass_data_stack_regs): ...new pass_data instance and...
10757         (make_pass_stack_regs): ...new function.
10758         (pass_stack_regs_run): Convert from a global struct to a subclass of
10759         rtl_opt_pass along with...
10760         (pass_data_stack_regs_run): ...new pass_data instance and...
10761         (make_pass_stack_regs_run): ...new function.
10762         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10763         subclass of rtl_opt_pass along with...
10764         (pass_data_cprop_hardreg): ...new pass_data instance and...
10765         (make_pass_cprop_hardreg): ...new function.
10766         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10767         subclass of rtl_opt_pass along with...
10768         (pass_data_reginfo_init): ...new pass_data instance and...
10769         (make_pass_reginfo_init): ...new function.
10770         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10771         of rtl_opt_pass along with...
10772         (pass_data_regmove): ...new pass_data instance and...
10773         (make_pass_regmove): ...new function.
10774         * regrename.c (pass_regrename): Convert from a global struct to a
10775         subclass of rtl_opt_pass along with...
10776         (pass_data_regrename): ...new pass_data instance and...
10777         (make_pass_regrename): ...new function.
10778         * reorg.c (pass_delay_slots): Convert from a global struct to a
10779         subclass of rtl_opt_pass along with...
10780         (pass_data_delay_slots): ...new pass_data instance and...
10781         (make_pass_delay_slots): ...new function.
10782         (pass_machine_reorg): Convert from a global struct to a subclass of
10783         rtl_opt_pass along with...
10784         (pass_data_machine_reorg): ...new pass_data instance and...
10785         (make_pass_machine_reorg): ...new function.
10786         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10787         of rtl_opt_pass along with...
10788         (pass_data_sched): ...new pass_data instance and...
10789         (make_pass_sched): ...new function.
10790         (pass_sched2): Convert from a global struct to a subclass of
10791         rtl_opt_pass along with...
10792         (pass_data_sched2): ...new pass_data instance and...
10793         (make_pass_sched2): ...new function.
10794         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10795         to a subclass of rtl_opt_pass along with...
10796         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10797         (make_pass_stack_ptr_mod): ...new function.
10798         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10799         to a subclass of rtl_opt_pass along with...
10800         (pass_data_rtl_store_motion): ...new pass_data instance and...
10801         (make_pass_rtl_store_motion): ...new function.
10802         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10803         of gimple_opt_pass along with...
10804         (pass_data_tracer): ...new pass_data instance and...
10805         (make_pass_tracer): ...new function.
10806         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10807         to a subclass of gimple_opt_pass along with...
10808         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10809         (make_pass_diagnose_tm_blocks): ...new function.
10810         (pass_lower_tm): Convert from a global struct to a subclass of
10811         gimple_opt_pass along with...
10812         (pass_data_lower_tm): ...new pass_data instance and...
10813         (make_pass_lower_tm): ...new function.
10814         (pass_tm_init): Convert from a global struct to a subclass of
10815         gimple_opt_pass along with...
10816         (pass_data_tm_init): ...new pass_data instance and...
10817         (make_pass_tm_init): ...new function.
10818         (pass_tm_mark): Convert from a global struct to a subclass of
10819         gimple_opt_pass along with...
10820         (pass_data_tm_mark): ...new pass_data instance and...
10821         (make_pass_tm_mark): ...new function.
10822         (pass_tm_edges): Convert from a global struct to a subclass of
10823         gimple_opt_pass along with...
10824         (pass_data_tm_edges): ...new pass_data instance and...
10825         (make_pass_tm_edges): ...new function.
10826         (pass_tm_memopt): Convert from a global struct to a subclass of
10827         gimple_opt_pass along with...
10828         (pass_data_tm_memopt): ...new pass_data instance and...
10829         (make_pass_tm_memopt): ...new function.
10830         (pass_ipa_tm): Convert from a global struct to a subclass of
10831         simple_ipa_opt_pass along with...
10832         (pass_data_ipa_tm): ...new pass_data instance and...
10833         (make_pass_ipa_tm): ...new function.
10834         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10835         subclass of gimple_opt_pass along with...
10836         (pass_data_call_cdce): ...new pass_data instance and...
10837         (make_pass_call_cdce): ...new function.
10838         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10839         subclass of gimple_opt_pass along with...
10840         (pass_data_build_cfg): ...new pass_data instance and...
10841         (make_pass_build_cfg): ...new function.
10842         (pass_split_crit_edges): Convert from a global struct to a subclass of
10843         gimple_opt_pass along with...
10844         (pass_data_split_crit_edges): ...new pass_data instance and...
10845         (make_pass_split_crit_edges): ...new function.
10846         (pass_warn_function_return): Convert from a global struct to a
10847         subclass of gimple_opt_pass along with...
10848         (pass_data_warn_function_return): ...new pass_data instance and...
10849         (make_pass_warn_function_return): ...new function.
10850         (pass_warn_function_noreturn): Convert from a global struct to a
10851         subclass of gimple_opt_pass along with...
10852         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10853         (make_pass_warn_function_noreturn): ...new function.
10854         (pass_warn_unused_result): Convert from a global struct to a subclass
10855         of gimple_opt_pass along with...
10856         (pass_data_warn_unused_result): ...new pass_data instance and...
10857         (make_pass_warn_unused_result): ...new function.
10858         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10859         a subclass of gimple_opt_pass along with...
10860         (pass_data_merge_phi): ...new pass_data instance and...
10861         (make_pass_merge_phi): ...new function.
10862         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10863         a subclass of gimple_opt_pass along with...
10864         (pass_data_lower_complex): ...new pass_data instance and...
10865         (make_pass_lower_complex): ...new function.
10866         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10867         gimple_opt_pass along with...
10868         (pass_data_lower_complex_O0): ...new pass_data instance and...
10869         (make_pass_lower_complex_O0): ...new function.
10870         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10871         subclass of gimple_opt_pass along with...
10872         (pass_data_lower_eh): ...new pass_data instance and...
10873         (make_pass_lower_eh): ...new function.
10874         (pass_refactor_eh): Convert from a global struct to a subclass of
10875         gimple_opt_pass along with...
10876         (pass_data_refactor_eh): ...new pass_data instance and...
10877         (make_pass_refactor_eh): ...new function.
10878         (pass_lower_resx): Convert from a global struct to a subclass of
10879         gimple_opt_pass along with...
10880         (pass_data_lower_resx): ...new pass_data instance and...
10881         (make_pass_lower_resx): ...new function.
10882         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10883         of gimple_opt_pass along with...
10884         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10885         (make_pass_lower_eh_dispatch): ...new function.
10886         (pass_cleanup_eh): Convert from a global struct to a subclass of
10887         gimple_opt_pass along with...
10888         (pass_data_cleanup_eh): ...new pass_data instance and...
10889         (make_pass_cleanup_eh): ...new function.
10890         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10891         to a subclass of simple_ipa_opt_pass along with...
10892         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10893         (make_pass_ipa_lower_emutls): ...new function.
10894         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10895         a subclass of gimple_opt_pass along with...
10896         (pass_data_if_conversion): ...new pass_data instance and...
10897         (make_pass_if_conversion): ...new function.
10898         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10899         subclass of gimple_opt_pass along with...
10900         (pass_data_build_ssa): ...new pass_data instance and...
10901         (make_pass_build_ssa): ...new function.
10902         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10903         global struct to a subclass of gimple_opt_pass along with...
10904         (pass_data_loop_distribution): ...new pass_data instance and...
10905         (make_pass_loop_distribution): ...new function.
10906         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10907         subclass of gimple_opt_pass along with...
10908         (pass_data_mudflap_1): ...new pass_data instance and...
10909         (make_pass_mudflap_1): ...new function.
10910         (pass_mudflap_2): Convert from a global struct to a subclass of
10911         gimple_opt_pass along with...
10912         (pass_data_mudflap_2): ...new pass_data instance and...
10913         (make_pass_mudflap_2): ...new function.
10914         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10915         subclass of gimple_opt_pass along with...
10916         (pass_data_mudflap_1): ...new pass_data instance and...
10917         (make_pass_mudflap_1): ...new function.
10918         (pass_mudflap_2): Convert from a global struct to a subclass of
10919         gimple_opt_pass along with...
10920         (pass_data_mudflap_2): ...new pass_data instance and...
10921         (make_pass_mudflap_2): ...new function.
10922         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10923         gimple_opt_pass along with...
10924         (pass_data_nrv): ...new pass_data instance and...
10925         (make_pass_nrv): ...new function.
10926         (pass_return_slot): Convert from a global struct to a subclass of
10927         gimple_opt_pass along with...
10928         (pass_data_return_slot): ...new pass_data instance and...
10929         (make_pass_return_slot): ...new function.
10930         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10931         to a subclass of gimple_opt_pass along with...
10932         (pass_data_object_sizes): ...new pass_data instance and...
10933         (make_pass_object_sizes): ...new function.
10934         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10935         global struct to a subclass of gimple_opt_pass along with...
10936         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10937         and...
10938         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10939         (pass_fixup_cfg): Convert from a global struct to a subclass of
10940         gimple_opt_pass along with...
10941         (pass_data_fixup_cfg): ...new pass_data instance and...
10942         (make_pass_fixup_cfg): ...new function.
10943         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10944         (make_pass_mudflap_1): ...new function.
10945         (pass_mudflap_2): Replace declaration with that of...
10946         (make_pass_mudflap_2): ...new function.
10947         (pass_asan): Replace declaration with that of...
10948         (make_pass_asan): ...new function.
10949         (pass_asan_O0): Replace declaration with that of...
10950         (make_pass_asan_O0): ...new function.
10951         (pass_tsan): Replace declaration with that of...
10952         (make_pass_tsan): ...new function.
10953         (pass_tsan_O0): Replace declaration with that of...
10954         (make_pass_tsan_O0): ...new function.
10955         (pass_lower_cf): Replace declaration with that of...
10956         (make_pass_lower_cf): ...new function.
10957         (pass_refactor_eh): Replace declaration with that of...
10958         (make_pass_refactor_eh): ...new function.
10959         (pass_lower_eh): Replace declaration with that of...
10960         (make_pass_lower_eh): ...new function.
10961         (pass_lower_eh_dispatch): Replace declaration with that of...
10962         (make_pass_lower_eh_dispatch): ...new function.
10963         (pass_lower_resx): Replace declaration with that of...
10964         (make_pass_lower_resx): ...new function.
10965         (pass_build_cfg): Replace declaration with that of...
10966         (make_pass_build_cfg): ...new function.
10967         (pass_early_tree_profile): Replace declaration with that of...
10968         (make_pass_early_tree_profile): ...new function.
10969         (pass_cleanup_eh): Replace declaration with that of...
10970         (make_pass_cleanup_eh): ...new function.
10971         (pass_sra): Replace declaration with that of...
10972         (make_pass_sra): ...new function.
10973         (pass_sra_early): Replace declaration with that of...
10974         (make_pass_sra_early): ...new function.
10975         (pass_early_ipa_sra): Replace declaration with that of...
10976         (make_pass_early_ipa_sra): ...new function.
10977         (pass_tail_recursion): Replace declaration with that of...
10978         (make_pass_tail_recursion): ...new function.
10979         (pass_tail_calls): Replace declaration with that of...
10980         (make_pass_tail_calls): ...new function.
10981         (pass_tree_loop): Replace declaration with that of...
10982         (make_pass_tree_loop): ...new function.
10983         (pass_tree_loop_init): Replace declaration with that of...
10984         (make_pass_tree_loop_init): ...new function.
10985         (pass_lim): Replace declaration with that of...
10986         (make_pass_lim): ...new function.
10987         (pass_tree_unswitch): Replace declaration with that of...
10988         (make_pass_tree_unswitch): ...new function.
10989         (pass_predcom): Replace declaration with that of...
10990         (make_pass_predcom): ...new function.
10991         (pass_iv_canon): Replace declaration with that of...
10992         (make_pass_iv_canon): ...new function.
10993         (pass_scev_cprop): Replace declaration with that of...
10994         (make_pass_scev_cprop): ...new function.
10995         (pass_empty_loop): Replace declaration with that of...
10996         (make_pass_empty_loop): ...new function.
10997         (pass_record_bounds): Replace declaration with that of...
10998         (make_pass_record_bounds): ...new function.
10999         (pass_graphite): Replace declaration with that of...
11000         (make_pass_graphite): ...new function.
11001         (pass_graphite_transforms): Replace declaration with that of...
11002         (make_pass_graphite_transforms): ...new function.
11003         (pass_if_conversion): Replace declaration with that of...
11004         (make_pass_if_conversion): ...new function.
11005         (pass_loop_distribution): Replace declaration with that of...
11006         (make_pass_loop_distribution): ...new function.
11007         (pass_vectorize): Replace declaration with that of...
11008         (make_pass_vectorize): ...new function.
11009         (pass_slp_vectorize): Replace declaration with that of...
11010         (make_pass_slp_vectorize): ...new function.
11011         (pass_complete_unroll): Replace declaration with that of...
11012         (make_pass_complete_unroll): ...new function.
11013         (pass_complete_unrolli): Replace declaration with that of...
11014         (make_pass_complete_unrolli): ...new function.
11015         (pass_parallelize_loops): Replace declaration with that of...
11016         (make_pass_parallelize_loops): ...new function.
11017         (pass_loop_prefetch): Replace declaration with that of...
11018         (make_pass_loop_prefetch): ...new function.
11019         (pass_iv_optimize): Replace declaration with that of...
11020         (make_pass_iv_optimize): ...new function.
11021         (pass_tree_loop_done): Replace declaration with that of...
11022         (make_pass_tree_loop_done): ...new function.
11023         (pass_ch): Replace declaration with that of...
11024         (make_pass_ch): ...new function.
11025         (pass_ccp): Replace declaration with that of...
11026         (make_pass_ccp): ...new function.
11027         (pass_phi_only_cprop): Replace declaration with that of...
11028         (make_pass_phi_only_cprop): ...new function.
11029         (pass_build_ssa): Replace declaration with that of...
11030         (make_pass_build_ssa): ...new function.
11031         (pass_build_alias): Replace declaration with that of...
11032         (make_pass_build_alias): ...new function.
11033         (pass_build_ealias): Replace declaration with that of...
11034         (make_pass_build_ealias): ...new function.
11035         (pass_dominator): Replace declaration with that of...
11036         (make_pass_dominator): ...new function.
11037         (pass_dce): Replace declaration with that of...
11038         (make_pass_dce): ...new function.
11039         (pass_dce_loop): Replace declaration with that of...
11040         (make_pass_dce_loop): ...new function.
11041         (pass_cd_dce): Replace declaration with that of...
11042         (make_pass_cd_dce): ...new function.
11043         (pass_call_cdce): Replace declaration with that of...
11044         (make_pass_call_cdce): ...new function.
11045         (pass_merge_phi): Replace declaration with that of...
11046         (make_pass_merge_phi): ...new function.
11047         (pass_split_crit_edges): Replace declaration with that of...
11048         (make_pass_split_crit_edges): ...new function.
11049         (pass_pre): Replace declaration with that of...
11050         (make_pass_pre): ...new function.
11051         (pass_profile): Replace declaration with that of...
11052         (make_pass_profile): ...new function.
11053         (pass_strip_predict_hints): Replace declaration with that of...
11054         (make_pass_strip_predict_hints): ...new function.
11055         (pass_lower_complex_O0): Replace declaration with that of...
11056         (make_pass_lower_complex_O0): ...new function.
11057         (pass_lower_complex): Replace declaration with that of...
11058         (make_pass_lower_complex): ...new function.
11059         (pass_lower_vector): Replace declaration with that of...
11060         (make_pass_lower_vector): ...new function.
11061         (pass_lower_vector_ssa): Replace declaration with that of...
11062         (make_pass_lower_vector_ssa): ...new function.
11063         (pass_lower_omp): Replace declaration with that of...
11064         (make_pass_lower_omp): ...new function.
11065         (pass_diagnose_omp_blocks): Replace declaration with that of...
11066         (make_pass_diagnose_omp_blocks): ...new function.
11067         (pass_expand_omp): Replace declaration with that of...
11068         (make_pass_expand_omp): ...new function.
11069         (pass_expand_omp_ssa): Replace declaration with that of...
11070         (make_pass_expand_omp_ssa): ...new function.
11071         (pass_object_sizes): Replace declaration with that of...
11072         (make_pass_object_sizes): ...new function.
11073         (pass_strlen): Replace declaration with that of...
11074         (make_pass_strlen): ...new function.
11075         (pass_fold_builtins): Replace declaration with that of...
11076         (make_pass_fold_builtins): ...new function.
11077         (pass_stdarg): Replace declaration with that of...
11078         (make_pass_stdarg): ...new function.
11079         (pass_early_warn_uninitialized): Replace declaration with that of...
11080         (make_pass_early_warn_uninitialized): ...new function.
11081         (pass_late_warn_uninitialized): Replace declaration with that of...
11082         (make_pass_late_warn_uninitialized): ...new function.
11083         (pass_cse_reciprocals): Replace declaration with that of...
11084         (make_pass_cse_reciprocals): ...new function.
11085         (pass_cse_sincos): Replace declaration with that of...
11086         (make_pass_cse_sincos): ...new function.
11087         (pass_optimize_bswap): Replace declaration with that of...
11088         (make_pass_optimize_bswap): ...new function.
11089         (pass_optimize_widening_mul): Replace declaration with that of...
11090         (make_pass_optimize_widening_mul): ...new function.
11091         (pass_warn_function_return): Replace declaration with that of...
11092         (make_pass_warn_function_return): ...new function.
11093         (pass_warn_function_noreturn): Replace declaration with that of...
11094         (make_pass_warn_function_noreturn): ...new function.
11095         (pass_cselim): Replace declaration with that of...
11096         (make_pass_cselim): ...new function.
11097         (pass_phiopt): Replace declaration with that of...
11098         (make_pass_phiopt): ...new function.
11099         (pass_forwprop): Replace declaration with that of...
11100         (make_pass_forwprop): ...new function.
11101         (pass_phiprop): Replace declaration with that of...
11102         (make_pass_phiprop): ...new function.
11103         (pass_tree_ifcombine): Replace declaration with that of...
11104         (make_pass_tree_ifcombine): ...new function.
11105         (pass_dse): Replace declaration with that of...
11106         (make_pass_dse): ...new function.
11107         (pass_nrv): Replace declaration with that of...
11108         (make_pass_nrv): ...new function.
11109         (pass_rename_ssa_copies): Replace declaration with that of...
11110         (make_pass_rename_ssa_copies): ...new function.
11111         (pass_sink_code): Replace declaration with that of...
11112         (make_pass_sink_code): ...new function.
11113         (pass_fre): Replace declaration with that of...
11114         (make_pass_fre): ...new function.
11115         (pass_check_data_deps): Replace declaration with that of...
11116         (make_pass_check_data_deps): ...new function.
11117         (pass_copy_prop): Replace declaration with that of...
11118         (make_pass_copy_prop): ...new function.
11119         (pass_vrp): Replace declaration with that of...
11120         (make_pass_vrp): ...new function.
11121         (pass_uncprop): Replace declaration with that of...
11122         (make_pass_uncprop): ...new function.
11123         (pass_return_slot): Replace declaration with that of...
11124         (make_pass_return_slot): ...new function.
11125         (pass_reassoc): Replace declaration with that of...
11126         (make_pass_reassoc): ...new function.
11127         (pass_rebuild_cgraph_edges): Replace declaration with that of...
11128         (make_pass_rebuild_cgraph_edges): ...new function.
11129         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
11130         (make_pass_remove_cgraph_callee_edges): ...new function.
11131         (pass_build_cgraph_edges): Replace declaration with that of...
11132         (make_pass_build_cgraph_edges): ...new function.
11133         (pass_local_pure_const): Replace declaration with that of...
11134         (make_pass_local_pure_const): ...new function.
11135         (pass_tracer): Replace declaration with that of...
11136         (make_pass_tracer): ...new function.
11137         (pass_warn_unused_result): Replace declaration with that of...
11138         (make_pass_warn_unused_result): ...new function.
11139         (pass_diagnose_tm_blocks): Replace declaration with that of...
11140         (make_pass_diagnose_tm_blocks): ...new function.
11141         (pass_lower_tm): Replace declaration with that of...
11142         (make_pass_lower_tm): ...new function.
11143         (pass_tm_init): Replace declaration with that of...
11144         (make_pass_tm_init): ...new function.
11145         (pass_tm_mark): Replace declaration with that of...
11146         (make_pass_tm_mark): ...new function.
11147         (pass_tm_memopt): Replace declaration with that of...
11148         (make_pass_tm_memopt): ...new function.
11149         (pass_tm_edges): Replace declaration with that of...
11150         (make_pass_tm_edges): ...new function.
11151         (pass_split_functions): Replace declaration with that of...
11152         (make_pass_split_functions): ...new function.
11153         (pass_feedback_split_functions): Replace declaration with that of...
11154         (make_pass_feedback_split_functions): ...new function.
11155         (pass_strength_reduction): Replace declaration with that of...
11156         (make_pass_strength_reduction): ...new function.
11157         (pass_ipa_lower_emutls): Replace declaration with that of...
11158         (make_pass_ipa_lower_emutls): ...new function.
11159         (pass_ipa_function_and_variable_visibility): Replace declaration with
11160         that of...
11161         (make_pass_ipa_function_and_variable_visibility): ...new function.
11162         (pass_ipa_tree_profile): Replace declaration with that of...
11163         (make_pass_ipa_tree_profile): ...new function.
11164         (pass_early_local_passes): Replace declaration with that of...
11165         (make_pass_early_local_passes): ...new function.
11166         (pass_ipa_whole_program_visibility): Replace declaration with that
11167         of...
11168         (make_pass_ipa_whole_program_visibility): ...new function.
11169         (pass_ipa_lto_gimple_out): Replace declaration with that of...
11170         (make_pass_ipa_lto_gimple_out): ...new function.
11171         (pass_ipa_increase_alignment): Replace declaration with that of...
11172         (make_pass_ipa_increase_alignment): ...new function.
11173         (pass_ipa_inline): Replace declaration with that of...
11174         (make_pass_ipa_inline): ...new function.
11175         (pass_ipa_free_lang_data): Replace declaration with that of...
11176         (make_pass_ipa_free_lang_data): ...new function.
11177         (pass_ipa_free_inline_summary): Replace declaration with that of...
11178         (make_pass_ipa_free_inline_summary): ...new function.
11179         (pass_ipa_cp): Replace declaration with that of...
11180         (make_pass_ipa_cp): ...new function.
11181         (pass_ipa_reference): Replace declaration with that of...
11182         (make_pass_ipa_reference): ...new function.
11183         (pass_ipa_pure_const): Replace declaration with that of...
11184         (make_pass_ipa_pure_const): ...new function.
11185         (pass_ipa_pta): Replace declaration with that of...
11186         (make_pass_ipa_pta): ...new function.
11187         (pass_ipa_lto_finish_out): Replace declaration with that of...
11188         (make_pass_ipa_lto_finish_out): ...new function.
11189         (pass_ipa_tm): Replace declaration with that of...
11190         (make_pass_ipa_tm): ...new function.
11191         (pass_ipa_profile): Replace declaration with that of...
11192         (make_pass_ipa_profile): ...new function.
11193         (pass_ipa_cdtor_merge): Replace declaration with that of...
11194         (make_pass_ipa_cdtor_merge): ...new function.
11195         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
11196         of...
11197         (make_pass_cleanup_cfg_post_optimizing): ...new function.
11198         (pass_init_datastructures): Replace declaration with that of...
11199         (make_pass_init_datastructures): ...new function.
11200         (pass_fixup_cfg): Replace declaration with that of...
11201         (make_pass_fixup_cfg): ...new function.
11202         (pass_expand): Replace declaration with that of...
11203         (make_pass_expand): ...new function.
11204         (pass_instantiate_virtual_regs): Replace declaration with that of...
11205         (make_pass_instantiate_virtual_regs): ...new function.
11206         (pass_rtl_fwprop): Replace declaration with that of...
11207         (make_pass_rtl_fwprop): ...new function.
11208         (pass_rtl_fwprop_addr): Replace declaration with that of...
11209         (make_pass_rtl_fwprop_addr): ...new function.
11210         (pass_jump): Replace declaration with that of...
11211         (make_pass_jump): ...new function.
11212         (pass_jump2): Replace declaration with that of...
11213         (make_pass_jump2): ...new function.
11214         (pass_lower_subreg): Replace declaration with that of...
11215         (make_pass_lower_subreg): ...new function.
11216         (pass_cse): Replace declaration with that of...
11217         (make_pass_cse): ...new function.
11218         (pass_fast_rtl_dce): Replace declaration with that of...
11219         (make_pass_fast_rtl_dce): ...new function.
11220         (pass_ud_rtl_dce): Replace declaration with that of...
11221         (make_pass_ud_rtl_dce): ...new function.
11222         (pass_rtl_dce): Replace declaration with that of...
11223         (make_pass_rtl_dce): ...new function.
11224         (pass_rtl_dse1): Replace declaration with that of...
11225         (make_pass_rtl_dse1): ...new function.
11226         (pass_rtl_dse2): Replace declaration with that of...
11227         (make_pass_rtl_dse2): ...new function.
11228         (pass_rtl_dse3): Replace declaration with that of...
11229         (make_pass_rtl_dse3): ...new function.
11230         (pass_rtl_cprop): Replace declaration with that of...
11231         (make_pass_rtl_cprop): ...new function.
11232         (pass_rtl_pre): Replace declaration with that of...
11233         (make_pass_rtl_pre): ...new function.
11234         (pass_rtl_hoist): Replace declaration with that of...
11235         (make_pass_rtl_hoist): ...new function.
11236         (pass_rtl_store_motion): Replace declaration with that of...
11237         (make_pass_rtl_store_motion): ...new function.
11238         (pass_cse_after_global_opts): Replace declaration with that of...
11239         (make_pass_cse_after_global_opts): ...new function.
11240         (pass_rtl_ifcvt): Replace declaration with that of...
11241         (make_pass_rtl_ifcvt): ...new function.
11242         (pass_into_cfg_layout_mode): Replace declaration with that of...
11243         (make_pass_into_cfg_layout_mode): ...new function.
11244         (pass_outof_cfg_layout_mode): Replace declaration with that of...
11245         (make_pass_outof_cfg_layout_mode): ...new function.
11246         (pass_loop2): Replace declaration with that of...
11247         (make_pass_loop2): ...new function.
11248         (pass_rtl_loop_init): Replace declaration with that of...
11249         (make_pass_rtl_loop_init): ...new function.
11250         (pass_rtl_move_loop_invariants): Replace declaration with that of...
11251         (make_pass_rtl_move_loop_invariants): ...new function.
11252         (pass_rtl_unswitch): Replace declaration with that of...
11253         (make_pass_rtl_unswitch): ...new function.
11254         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
11255         (make_pass_rtl_unroll_and_peel_loops): ...new function.
11256         (pass_rtl_doloop): Replace declaration with that of...
11257         (make_pass_rtl_doloop): ...new function.
11258         (pass_rtl_loop_done): Replace declaration with that of...
11259         (make_pass_rtl_loop_done): ...new function.
11260         (pass_web): Replace declaration with that of...
11261         (make_pass_web): ...new function.
11262         (pass_cse2): Replace declaration with that of...
11263         (make_pass_cse2): ...new function.
11264         (pass_df_initialize_opt): Replace declaration with that of...
11265         (make_pass_df_initialize_opt): ...new function.
11266         (pass_df_initialize_no_opt): Replace declaration with that of...
11267         (make_pass_df_initialize_no_opt): ...new function.
11268         (pass_reginfo_init): Replace declaration with that of...
11269         (make_pass_reginfo_init): ...new function.
11270         (pass_inc_dec): Replace declaration with that of...
11271         (make_pass_inc_dec): ...new function.
11272         (pass_stack_ptr_mod): Replace declaration with that of...
11273         (make_pass_stack_ptr_mod): ...new function.
11274         (pass_initialize_regs): Replace declaration with that of...
11275         (make_pass_initialize_regs): ...new function.
11276         (pass_combine): Replace declaration with that of...
11277         (make_pass_combine): ...new function.
11278         (pass_if_after_combine): Replace declaration with that of...
11279         (make_pass_if_after_combine): ...new function.
11280         (pass_ree): Replace declaration with that of...
11281         (make_pass_ree): ...new function.
11282         (pass_partition_blocks): Replace declaration with that of...
11283         (make_pass_partition_blocks): ...new function.
11284         (pass_match_asm_constraints): Replace declaration with that of...
11285         (make_pass_match_asm_constraints): ...new function.
11286         (pass_regmove): Replace declaration with that of...
11287         (make_pass_regmove): ...new function.
11288         (pass_split_all_insns): Replace declaration with that of...
11289         (make_pass_split_all_insns): ...new function.
11290         (pass_fast_rtl_byte_dce): Replace declaration with that of...
11291         (make_pass_fast_rtl_byte_dce): ...new function.
11292         (pass_lower_subreg2): Replace declaration with that of...
11293         (make_pass_lower_subreg2): ...new function.
11294         (pass_mode_switching): Replace declaration with that of...
11295         (make_pass_mode_switching): ...new function.
11296         (pass_sms): Replace declaration with that of...
11297         (make_pass_sms): ...new function.
11298         (pass_sched): Replace declaration with that of...
11299         (make_pass_sched): ...new function.
11300         (pass_ira): Replace declaration with that of...
11301         (make_pass_ira): ...new function.
11302         (pass_reload): Replace declaration with that of...
11303         (make_pass_reload): ...new function.
11304         (pass_clean_state): Replace declaration with that of...
11305         (make_pass_clean_state): ...new function.
11306         (pass_branch_prob): Replace declaration with that of...
11307         (make_pass_branch_prob): ...new function.
11308         (pass_value_profile_transformations): Replace declaration with that
11309         of...
11310         (make_pass_value_profile_transformations): ...new function.
11311         (pass_postreload_cse): Replace declaration with that of...
11312         (make_pass_postreload_cse): ...new function.
11313         (pass_gcse2): Replace declaration with that of...
11314         (make_pass_gcse2): ...new function.
11315         (pass_split_after_reload): Replace declaration with that of...
11316         (make_pass_split_after_reload): ...new function.
11317         (pass_branch_target_load_optimize1): Replace declaration with that
11318         of...
11319         (make_pass_branch_target_load_optimize1): ...new function.
11320         (pass_thread_prologue_and_epilogue): Replace declaration with that
11321         of...
11322         (make_pass_thread_prologue_and_epilogue): ...new function.
11323         (pass_stack_adjustments): Replace declaration with that of...
11324         (make_pass_stack_adjustments): ...new function.
11325         (pass_peephole2): Replace declaration with that of...
11326         (make_pass_peephole2): ...new function.
11327         (pass_if_after_reload): Replace declaration with that of...
11328         (make_pass_if_after_reload): ...new function.
11329         (pass_regrename): Replace declaration with that of...
11330         (make_pass_regrename): ...new function.
11331         (pass_cprop_hardreg): Replace declaration with that of...
11332         (make_pass_cprop_hardreg): ...new function.
11333         (pass_reorder_blocks): Replace declaration with that of...
11334         (make_pass_reorder_blocks): ...new function.
11335         (pass_branch_target_load_optimize2): Replace declaration with that
11336         of...
11337         (make_pass_branch_target_load_optimize2): ...new function.
11338         (pass_leaf_regs): Replace declaration with that of...
11339         (make_pass_leaf_regs): ...new function.
11340         (pass_split_before_sched2): Replace declaration with that of...
11341         (make_pass_split_before_sched2): ...new function.
11342         (pass_compare_elim_after_reload): Replace declaration with that of...
11343         (make_pass_compare_elim_after_reload): ...new function.
11344         (pass_sched2): Replace declaration with that of...
11345         (make_pass_sched2): ...new function.
11346         (pass_stack_regs): Replace declaration with that of...
11347         (make_pass_stack_regs): ...new function.
11348         (pass_stack_regs_run): Replace declaration with that of...
11349         (make_pass_stack_regs_run): ...new function.
11350         (pass_df_finish): Replace declaration with that of...
11351         (make_pass_df_finish): ...new function.
11352         (pass_compute_alignments): Replace declaration with that of...
11353         (make_pass_compute_alignments): ...new function.
11354         (pass_duplicate_computed_gotos): Replace declaration with that of...
11355         (make_pass_duplicate_computed_gotos): ...new function.
11356         (pass_variable_tracking): Replace declaration with that of...
11357         (make_pass_variable_tracking): ...new function.
11358         (pass_free_cfg): Replace declaration with that of...
11359         (make_pass_free_cfg): ...new function.
11360         (pass_machine_reorg): Replace declaration with that of...
11361         (make_pass_machine_reorg): ...new function.
11362         (pass_cleanup_barriers): Replace declaration with that of...
11363         (make_pass_cleanup_barriers): ...new function.
11364         (pass_delay_slots): Replace declaration with that of...
11365         (make_pass_delay_slots): ...new function.
11366         (pass_split_for_shorten_branches): Replace declaration with that of...
11367         (make_pass_split_for_shorten_branches): ...new function.
11368         (pass_split_before_regstack): Replace declaration with that of...
11369         (make_pass_split_before_regstack): ...new function.
11370         (pass_convert_to_eh_region_ranges): Replace declaration with that
11371         of...
11372         (make_pass_convert_to_eh_region_ranges): ...new function.
11373         (pass_shorten_branches): Replace declaration with that of...
11374         (make_pass_shorten_branches): ...new function.
11375         (pass_set_nothrow_function_flags): Replace declaration with that of...
11376         (make_pass_set_nothrow_function_flags): ...new function.
11377         (pass_dwarf2_frame): Replace declaration with that of...
11378         (make_pass_dwarf2_frame): ...new function.
11379         (pass_final): Replace declaration with that of...
11380         (make_pass_final): ...new function.
11381         (pass_rtl_seqabstr): Replace declaration with that of...
11382         (make_pass_rtl_seqabstr): ...new function.
11383         (pass_release_ssa_names): Replace declaration with that of...
11384         (make_pass_release_ssa_names): ...new function.
11385         (pass_early_inline): Replace declaration with that of...
11386         (make_pass_early_inline): ...new function.
11387         (pass_inline_parameters): Replace declaration with that of...
11388         (make_pass_inline_parameters): ...new function.
11389         (pass_update_address_taken): Replace declaration with that of...
11390         (make_pass_update_address_taken): ...new function.
11391         (pass_convert_switch): Replace declaration with that of...
11392         (make_pass_convert_switch): ...new function.
11393         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
11394         to a subclass of simple_ipa_opt_pass along with...
11395         (pass_data_ipa_tree_profile): ...new pass_data instance and...
11396         (make_pass_ipa_tree_profile): ...new function.
11397         * tree-sra.c (pass_sra_early): Convert from a global struct to a
11398         subclass of gimple_opt_pass along with...
11399         (pass_data_sra_early): ...new pass_data instance and...
11400         (make_pass_sra_early): ...new function.
11401         (pass_sra): Convert from a global struct to a subclass of
11402         gimple_opt_pass along with...
11403         (pass_data_sra): ...new pass_data instance and...
11404         (make_pass_sra): ...new function.
11405         (pass_early_ipa_sra): Convert from a global struct to a subclass of
11406         gimple_opt_pass along with...
11407         (pass_data_early_ipa_sra): ...new pass_data instance and...
11408         (make_pass_early_ipa_sra): ...new function.
11409         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
11410         subclass of gimple_opt_pass along with...
11411         (pass_data_ccp): ...new pass_data instance and...
11412         (make_pass_ccp): ...new function.
11413         (pass_fold_builtins): Convert from a global struct to a subclass of
11414         gimple_opt_pass along with...
11415         (pass_data_fold_builtins): ...new pass_data instance and...
11416         (make_pass_fold_builtins): ...new function.
11417         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
11418         subclass of gimple_opt_pass along with...
11419         (pass_data_copy_prop): ...new pass_data instance and...
11420         (make_pass_copy_prop): ...new function.
11421         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
11422         global struct to a subclass of gimple_opt_pass along with...
11423         (pass_data_rename_ssa_copies): ...new pass_data instance and...
11424         (make_pass_rename_ssa_copies): ...new function.
11425         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
11426         subclass of gimple_opt_pass along with...
11427         (pass_data_dce): ...new pass_data instance and...
11428         (make_pass_dce): ...new function.
11429         (pass_dce_loop): Convert from a global struct to a subclass of
11430         gimple_opt_pass along with...
11431         (pass_data_dce_loop): ...new pass_data instance and...
11432         (make_pass_dce_loop): ...new function.
11433         (pass_cd_dce): Convert from a global struct to a subclass of
11434         gimple_opt_pass along with...
11435         (pass_data_cd_dce): ...new pass_data instance and...
11436         (make_pass_cd_dce): ...new function.
11437         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
11438         subclass of gimple_opt_pass along with...
11439         (pass_data_dominator): ...new pass_data instance and...
11440         (make_pass_dominator): ...new function.
11441         (pass_phi_only_cprop): Convert from a global struct to a subclass of
11442         gimple_opt_pass along with...
11443         (pass_data_phi_only_cprop): ...new pass_data instance and...
11444         (make_pass_phi_only_cprop): ...new function.
11445         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
11446         subclass of gimple_opt_pass along with...
11447         (pass_data_dse): ...new pass_data instance and...
11448         (make_pass_dse): ...new function.
11449         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
11450         a subclass of gimple_opt_pass along with...
11451         (pass_data_forwprop): ...new pass_data instance and...
11452         (make_pass_forwprop): ...new function.
11453         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
11454         struct to a subclass of gimple_opt_pass along with...
11455         (pass_data_tree_ifcombine): ...new pass_data instance and...
11456         (make_pass_tree_ifcombine): ...new function.
11457         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
11458         subclass of gimple_opt_pass along with...
11459         (pass_data_ch): ...new pass_data instance and...
11460         (make_pass_ch): ...new function.
11461         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
11462         subclass of gimple_opt_pass along with...
11463         (pass_data_tree_loop): ...new pass_data instance and...
11464         (make_pass_tree_loop): ...new function.
11465         (pass_tree_loop_init): Convert from a global struct to a subclass of
11466         gimple_opt_pass along with...
11467         (pass_data_tree_loop_init): ...new pass_data instance and...
11468         (make_pass_tree_loop_init): ...new function.
11469         (pass_lim): Convert from a global struct to a subclass of
11470         gimple_opt_pass along with...
11471         (pass_data_lim): ...new pass_data instance and...
11472         (make_pass_lim): ...new function.
11473         (pass_tree_unswitch): Convert from a global struct to a subclass of
11474         gimple_opt_pass along with...
11475         (pass_data_tree_unswitch): ...new pass_data instance and...
11476         (make_pass_tree_unswitch): ...new function.
11477         (pass_predcom): Convert from a global struct to a subclass of
11478         gimple_opt_pass along with...
11479         (pass_data_predcom): ...new pass_data instance and...
11480         (make_pass_predcom): ...new function.
11481         (pass_vectorize): Convert from a global struct to a subclass of
11482         gimple_opt_pass along with...
11483         (pass_data_vectorize): ...new pass_data instance and...
11484         (make_pass_vectorize): ...new function.
11485         (pass_graphite): Convert from a global struct to a subclass of
11486         gimple_opt_pass along with...
11487         (pass_data_graphite): ...new pass_data instance and...
11488         (make_pass_graphite): ...new function.
11489         (pass_graphite_transforms): Convert from a global struct to a subclass
11490         of gimple_opt_pass along with...
11491         (pass_data_graphite_transforms): ...new pass_data instance and...
11492         (make_pass_graphite_transforms): ...new function.
11493         (pass_check_data_deps): Convert from a global struct to a subclass of
11494         gimple_opt_pass along with...
11495         (pass_data_check_data_deps): ...new pass_data instance and...
11496         (make_pass_check_data_deps): ...new function.
11497         (pass_iv_canon): Convert from a global struct to a subclass of
11498         gimple_opt_pass along with...
11499         (pass_data_iv_canon): ...new pass_data instance and...
11500         (make_pass_iv_canon): ...new function.
11501         (pass_scev_cprop): Convert from a global struct to a subclass of
11502         gimple_opt_pass along with...
11503         (pass_data_scev_cprop): ...new pass_data instance and...
11504         (make_pass_scev_cprop): ...new function.
11505         (pass_record_bounds): Convert from a global struct to a subclass of
11506         gimple_opt_pass along with...
11507         (pass_data_record_bounds): ...new pass_data instance and...
11508         (make_pass_record_bounds): ...new function.
11509         (pass_complete_unroll): Convert from a global struct to a subclass of
11510         gimple_opt_pass along with...
11511         (pass_data_complete_unroll): ...new pass_data instance and...
11512         (make_pass_complete_unroll): ...new function.
11513         (pass_complete_unrolli): Convert from a global struct to a subclass of
11514         gimple_opt_pass along with...
11515         (pass_data_complete_unrolli): ...new pass_data instance and...
11516         (make_pass_complete_unrolli): ...new function.
11517         (pass_parallelize_loops): Convert from a global struct to a subclass
11518         of gimple_opt_pass along with...
11519         (pass_data_parallelize_loops): ...new pass_data instance and...
11520         (make_pass_parallelize_loops): ...new function.
11521         (pass_loop_prefetch): Convert from a global struct to a subclass of
11522         gimple_opt_pass along with...
11523         (pass_data_loop_prefetch): ...new pass_data instance and...
11524         (make_pass_loop_prefetch): ...new function.
11525         (pass_iv_optimize): Convert from a global struct to a subclass of
11526         gimple_opt_pass along with...
11527         (pass_data_iv_optimize): ...new pass_data instance and...
11528         (make_pass_iv_optimize): ...new function.
11529         (pass_tree_loop_done): Convert from a global struct to a subclass of
11530         gimple_opt_pass along with...
11531         (pass_data_tree_loop_done): ...new pass_data instance and...
11532         (make_pass_tree_loop_done): ...new function.
11533         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
11534         struct to a subclass of gimple_opt_pass along with...
11535         (pass_data_cse_reciprocals): ...new pass_data instance and...
11536         (make_pass_cse_reciprocals): ...new function.
11537         (pass_cse_sincos): Convert from a global struct to a subclass of
11538         gimple_opt_pass along with...
11539         (pass_data_cse_sincos): ...new pass_data instance and...
11540         (make_pass_cse_sincos): ...new function.
11541         (pass_optimize_bswap): Convert from a global struct to a subclass of
11542         gimple_opt_pass along with...
11543         (pass_data_optimize_bswap): ...new pass_data instance and...
11544         (make_pass_optimize_bswap): ...new function.
11545         (pass_optimize_widening_mul): Convert from a global struct to a
11546         subclass of gimple_opt_pass along with...
11547         (pass_data_optimize_widening_mul): ...new pass_data instance and...
11548         (make_pass_optimize_widening_mul): ...new function.
11549         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
11550         subclass of gimple_opt_pass along with...
11551         (pass_data_phiopt): ...new pass_data instance and...
11552         (make_pass_phiopt): ...new function.
11553         (pass_cselim): Convert from a global struct to a subclass of
11554         gimple_opt_pass along with...
11555         (pass_data_cselim): ...new pass_data instance and...
11556         (make_pass_cselim): ...new function.
11557         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
11558         subclass of gimple_opt_pass along with...
11559         (pass_data_phiprop): ...new pass_data instance and...
11560         (make_pass_phiprop): ...new function.
11561         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
11562         subclass of gimple_opt_pass along with...
11563         (pass_data_pre): ...new pass_data instance and...
11564         (make_pass_pre): ...new function.
11565         (pass_fre): Convert from a global struct to a subclass of
11566         gimple_opt_pass along with...
11567         (pass_data_fre): ...new pass_data instance and...
11568         (make_pass_fre): ...new function.
11569         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
11570         subclass of gimple_opt_pass along with...
11571         (pass_data_reassoc): ...new pass_data instance and...
11572         (make_pass_reassoc): ...new function.
11573         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
11574         subclass of gimple_opt_pass along with...
11575         (pass_data_sink_code): ...new pass_data instance and...
11576         (make_pass_sink_code): ...new function.
11577         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
11578         subclass of gimple_opt_pass along with...
11579         (pass_data_strlen): ...new pass_data instance and...
11580         (make_pass_strlen): ...new function.
11581         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
11582         struct to a subclass of gimple_opt_pass along with...
11583         (pass_data_build_alias): ...new pass_data instance and...
11584         (make_pass_build_alias): ...new function.
11585         (pass_build_ealias): Convert from a global struct to a subclass of
11586         gimple_opt_pass along with...
11587         (pass_data_build_ealias): ...new pass_data instance and...
11588         (make_pass_build_ealias): ...new function.
11589         (pass_ipa_pta): Convert from a global struct to a subclass of
11590         simple_ipa_opt_pass along with...
11591         (pass_data_ipa_pta): ...new pass_data instance and...
11592         (make_pass_ipa_pta): ...new function.
11593         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
11594         subclass of gimple_opt_pass along with...
11595         (pass_data_uncprop): ...new pass_data instance and...
11596         (make_pass_uncprop): ...new function.
11597         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
11598         global struct to a subclass of gimple_opt_pass along with...
11599         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
11600         (make_pass_late_warn_uninitialized): ...new function.
11601         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
11602         to a subclass of gimple_opt_pass along with...
11603         (pass_data_init_datastructures): ...new pass_data instance and...
11604         (make_pass_init_datastructures): ...new function.
11605         (pass_early_warn_uninitialized): Convert from a global struct to a
11606         subclass of gimple_opt_pass along with...
11607         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
11608         (make_pass_early_warn_uninitialized): ...new function.
11609         (pass_update_address_taken): Convert from a global struct to a
11610         subclass of gimple_opt_pass along with...
11611         (pass_data_update_address_taken): ...new pass_data instance and...
11612         (make_pass_update_address_taken): ...new function.
11613         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
11614         struct to a subclass of gimple_opt_pass along with...
11615         (pass_data_release_ssa_names): ...new pass_data instance and...
11616         (make_pass_release_ssa_names): ...new function.
11617         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
11618         subclass of gimple_opt_pass along with...
11619         (pass_data_stdarg): ...new pass_data instance and...
11620         (make_pass_stdarg): ...new function.
11621         * tree-switch-conversion.c (pass_convert_switch): Convert from a
11622         global struct to a subclass of gimple_opt_pass along with...
11623         (pass_data_convert_switch): ...new pass_data instance and...
11624         (make_pass_convert_switch): ...new function.
11625         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
11626         to a subclass of gimple_opt_pass along with...
11627         (pass_data_tail_recursion): ...new pass_data instance and...
11628         (make_pass_tail_recursion): ...new function.
11629         (pass_tail_calls): Convert from a global struct to a subclass of
11630         gimple_opt_pass along with...
11631         (pass_data_tail_calls): ...new pass_data instance and...
11632         (make_pass_tail_calls): ...new function.
11633         * tree-vect-generic.c (pass_lower_vector): Convert from a global
11634         struct to a subclass of gimple_opt_pass along with...
11635         (pass_data_lower_vector): ...new pass_data instance and...
11636         (make_pass_lower_vector): ...new function.
11637         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11638         gimple_opt_pass along with...
11639         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11640         (make_pass_lower_vector_ssa): ...new function.
11641         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11642         to a subclass of gimple_opt_pass along with...
11643         (pass_data_slp_vectorize): ...new pass_data instance and...
11644         (make_pass_slp_vectorize): ...new function.
11645         (pass_ipa_increase_alignment): Convert from a global struct to a
11646         subclass of simple_ipa_opt_pass along with...
11647         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11648         (make_pass_ipa_increase_alignment): ...new function.
11649         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11650         gimple_opt_pass along with...
11651         (pass_data_vrp): ...new pass_data instance and...
11652         (make_pass_vrp): ...new function.
11653         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11654         subclass of simple_ipa_opt_pass along with...
11655         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11656         (make_pass_ipa_free_lang_data): ...new function.
11657         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11658         gimple_opt_pass along with...
11659         (pass_data_tsan): ...new pass_data instance and...
11660         (make_pass_tsan): ...new function.
11661         (pass_tsan_O0): Convert from a global struct to a subclass of
11662         gimple_opt_pass along with...
11663         (pass_data_tsan_O0): ...new pass_data instance and...
11664         (make_pass_tsan_O0): ...new function.
11665         * var-tracking.c (pass_variable_tracking): Convert from a global
11666         struct to a subclass of rtl_opt_pass along with...
11667         (pass_data_variable_tracking): ...new pass_data instance and...
11668         (make_pass_variable_tracking): ...new function.
11669         * web.c (pass_web): Convert from a global struct to a subclass of
11670         rtl_opt_pass along with...
11671         (pass_data_web): ...new pass_data instance and...
11672         (make_pass_web): ...new function.
11673         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11674         declaration with that of...
11675         (make_pass_mode_switch_use): ...new function.
11676         (pass_resolve_sw_modes): Replace declaration with that of...
11677         (make_pass_resolve_sw_modes): ...new function.
11678         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11679         from a global struct to a subclass of rtl_opt_pass along with...
11680         (pass_data_mode_switch_use): ...new pass_data instance and...
11681         (make_pass_mode_switch_use): ...new function.
11682         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11683         from a global struct to a subclass of rtl_opt_pass along with...
11684         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11685         (make_pass_resolve_sw_modes): ...new function.
11686         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11687         struct to a subclass of rtl_opt_pass along with...
11688         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11689         (make_pass_insert_vzeroupper): ...new function.
11690         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11691         global struct to a subclass of rtl_opt_pass along with...
11692         (pass_data_work_around_errata): ...new pass_data instance and...
11693         (make_pass_work_around_errata): ...new function.
11694         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11695         struct to a subclass of rtl_opt_pass along with...
11696         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11697         (make_pass_mips_machine_reorg2): ...new function.
11699 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11701         * passes.c (pass_manager::operator new): New.
11703 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11705         Handwritten part of conversion of passes to C++ classes.
11707         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11708         (toplev.o): Add dep on PASS_MANAGER_H.
11709         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11710         of early local pases to reflect this moving from a global to a
11711         member of gcc::pass_manager.
11712         (cgraph_add_new_function): Likewise.
11713         * lto-cgraph.c (lto_output_node): Update for conversion of
11714         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11715         * passes.c (opt_pass::clone): New.
11716         (opt_pass::gate): New.
11717         (opt_pass::execute): New.
11718         (opt_pass::opt_pass): New.
11719         (pass_manager::execute_early_local_passes): New.
11720         (pass_manager::execute_pass_mode_switching): new.
11721         (finish_optimization_passes): Convert to...
11722         (pass_manager::finish_optimization_passes): ...this.
11723         (finish_optimization_passes): Update for conversion of passes to
11724         C++ classes.
11725         (register_dump_files_1): Use has_gate since we cannot portably
11726         check a vtable entry against NULL.
11727         (dump_one_pass): Likewise.
11728         (ipa_write_summaries_2): Likewise.
11729         (ipa_write_optimization_summaries_1): Likewise.
11730         (ipa_read_summaries_1): Likewise.
11731         (ipa_read_optimization_summaries_1): Likewise.
11732         (execute_ipa_stmt_fixups): Likewise.
11733         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11734         pass-creation functions rather than wiring up globals, and
11735         storing the results in fields of pass_manager generated using
11736         pass-instances.def.
11737         (pass_manager::dump_profile_report): Update for conversion of
11738         passes to C++ classes.
11739         (pass_manager::execute_ipa_summary_passes): Likewise.
11740         (execute_one_ipa_transform_pass): Likewise.
11741         (execute_one_pass): Use has_gate and has_execute since we cannot
11742         portably check a vtable entry against NULL.
11743         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11744         (pass_manager): Use pass-instances.def to add fields for the
11745         various pass instances.
11746         * toplev.c (finalize): Update for move of
11747         finish_optimization_passes to a method of gcc::pass_manager.
11748         * toplev.h (finish_optimization_passes): Move to method of class
11749         pass_manager.
11750         * tree-pass.h (struct pass_data): New.
11751         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11752         (opt_pass::gate): Convert to virtual function.
11753         (opt_pass::~opt_pass): New.
11754         (opt_pass::clone): New.
11755         (opt_pass::execute): Convert to virtual function.
11756         (opt_pass::opt_pass): New.
11757         (opt_pass::ctxt_): new.
11758         (gimple_opt_pass): Convert to subclass of opt_pass.
11759         (gimple_opt_pass::gimple_opt_pass): New.
11760         (rtl_opt_pass): Convert to subclass of opt_pass.
11761         (rtl_opt_pass::rtl_opt_pass): New.
11762         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11763         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11764         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11765         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11766         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11767         invocation of pass_mode_switching to reflect this moving from a
11768         global to a member of gcc::pass_manager.
11769         (ix86_option_override): Rework how pass_insert_vzeroupper is
11770         added to the pass_manager to reflect autogenerated changes.
11771         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11773 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11775         PR rtl-optimization/57708
11776         * recog.c (peep2_find_free_register): Validate all regs in a
11777         multi-reg mode.
11779 2013-08-05  Jan Hubicka  <jh@suse.cz>
11781         PR lto/57602
11782         * cgraph.c (verify_cgraph_node): Accept local flags from other
11783         partitions.
11784         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11785         (function_and_variable_visibility): Likewise.
11786         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11788 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11790         * graph.c (init_graph_slim_pretty_print): Remove.
11791         (print_graph_cfg): Do not call it.  Use local pretty printer.
11792         (start_graph_dump): Likewise.
11794 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11796         * gimple-pretty-print.c (buffer): Remove.
11797         (initialized): Likewise.
11798         (maybe_init_pretty_print): Likewise.
11799         (print_gimple_stmt): Do not call it.  Use non-static local
11800         pretty_printer variable.
11801         (print_gimple_expr): Likewise.
11802         (print_gimple_seq): Likewise.
11803         (gimple_dump_bb): Likewise.
11805 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11807         * asan.c (asan_pp): Remove.
11808         (asan_pp_initialized): Likewise.
11809         (asan_pp_initialize): Likewise.
11810         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11811         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11812         (asan_add_global): Likewise.
11814 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11816         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11817         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11818         * pretty-print.c (pp_formatted_text_data): Likewise.
11819         (pp_indent): Rename from pp_base_indent.
11820         (pp_format): Rename from pp_base_format.
11821         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11822         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11823         (pp_flush): Rename from pp_base_flush.
11824         (pp_set_line_maximum_length): Rename from
11825         pp_base_set_line_maximum_length.
11826         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11827         (pp_set_prefix): Rename from pp_base_set_prefix.
11828         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11829         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11830         (pp_append_text): Rename from pp_base_append_text.
11831         (pp_formatted_text): Rename from pp_base_formatted_text.
11832         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11833         (pp_remaining_character_count_for_line): Rename from
11834         pp_base_remaining_character_count_for_line.
11835         (pp_newline): Rename from pp_base_newline.
11836         (pp_character): Rename from pp_base_character.
11837         (pp_string): Rename from pp_base_string.
11838         (pp_maybe_space): Rename from pp_base_maybe_space.
11839         * asan.c (asan_pp_string): Adjust.
11840         (asan_emit_stack_protection): Likewise.
11841         (asan_add_global): Likewise.
11842         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11843         * tree-mudflap.c (mf_varname_tree): Likewise.
11844         * tree-pretty-print.c (pp_tree_identifier): Rename from
11845         pp_base_tree_identifier.
11846         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11847         Declare as function.
11849 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11851         * pretty-print.h (pp_bar_bar): New.
11852         (pp_ampersand_ampersand): Likewise.
11853         (pp_less_equal): Likewise.
11854         (pp_greater_equal): Likewise.
11855         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11856         printer functions instead of pp_string or operators and punctuators.
11857         (dump_gimple_call): Likewise.
11858         (dump_gimple_omp_for): Likewise.
11859         (dump_gimple_transaction): Likewise.
11860         (dump_gimple_phi): Likewise.
11861         (pp_gimple_stmt_1): Likewise.
11862         * sched-vis.c (print_insn): Likewise.
11863         * tree-mudflap.c (mf_varname_tree): Likewise.
11864         * tree-pretty-print.c (dump_block_node): Likewise.
11865         (dump_generic_node): Likewise.
11867 2013-08-02  Jan Hubicka  <jh@suse.cz>
11869         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11870         boundaries.
11871         * lto-streamer-out.c (tree_is_indexable): Results decls and
11872         parm decls are not indexable.
11873         (DFS_write_tree_body): Do not follow args and results.
11874         (hash_tree): Likewise.
11875         (output_functions): Rearrange so struct function is needed
11876         only when real body is output; be able to also ouptut abstract
11877         functions; output DECL_ARGUMENTS and DECL_RESULT.
11878         (lto_output): When not in WPA, ale store abstract functions.
11879         (write_symbol): Do not care about RESULT_DECL.
11880         (output_symbol_p): Handle correctly sbtract decls.
11881         * lto-streamer-in.c (input_function): Rearrange so struct
11882         function can be NULL at entry; allow streaming of
11883         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11884         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11885         sanity check during LTO.
11886         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11887         RESULT_DECl and DECL_ARGUMENTS.
11888         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11889         Likewise.
11891 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11893         * pretty-print.h (pp_underscore): New.
11894         (pp_comma): Tidy.
11895         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11896         printer functions instead of pp_character.
11897         (dump_binary_rhs): Likewise.
11898         (dump_ternary_rhs): Likewise.
11899         (dump_gimple_call_args): Likewise.
11900         (pp_points_to_solution): Likewise.
11901         (dump_gimple_call): Likewise.
11902         (dump_gimple_switch): Likewise.
11903         (dump_gimple_cond): Likewise.
11904         (dump_gimple_bind): Likewise.
11905         (dump_gimple_try): Likewise.
11906         (dump_gimple_omp_for): Likewise.
11907         (dump_gimple_omp_continue): Likewise.
11908         (dump_gimple_omp_single): Likewise.
11909         (dump_gimple_omp_sections): Likewise.
11910         (dump_gimple_omp_block): Likewise.
11911         (dump_gimple_omp_critical): Likewise.
11912         (dump_gimple_transaction): Likewise.
11913         (dump_gimple_asm): Likewise.
11914         (dump_gimple_phi): Likewise.
11915         (dump_gimple_omp_parallel): Likewise.
11916         (dump_gimple_omp_task): Likewise.
11917         (dump_gimple_omp_atomic_load): Likewise.
11918         (dump_gimple_omp_atomic_store): Likewise.
11919         (dump_gimple_mem_ops): Likewise.
11920         (pp_gimple_stmt_1): Likewise.
11921         (pp_cfg_jump): Likewise.
11922         (dump_implicit_edges): Likewise.
11923         (gimple_dump_bb_for_graph): Likewise.
11924         * graph.c (draw_cfg_node): Likewise.
11925         * langhooks.c (lhd_print_error_function): Likewise.
11926         * sched-vis.c (print_exp): Likewise.
11927         (print_value): Likewise.
11928         (print_pattern): Likewise.
11929         (print_insn): Likewise.
11930         (rtl_dump_bb_for_graph): Likewise.
11931         * tree-pretty-print.c (dump_function_declaration): Likewise.
11932         (dump_array_domain): Likewise.
11933         (dump_omp_clause): Likewise.
11934         (dump_location): Likewise.
11935         (dump_generic_node): Likewise.
11936         (print_struct_decl): Likewise.
11937         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11939 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11941         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11942         candidate table when replacing a candidate statement.
11943         (replace_rhs_if_not_dup): Likewise.
11944         (replace_one_candidate): Likewise.
11946 2013-08-02  Jan Hubicka  <jh@suse.cz>
11947             Martin Liska  <marxin.liska@gmail.com>
11949         * cgraphunit.c (add_new_function): Fix logic when adding from
11950         late IPA pass.
11951         (assemble_thunk): Rename to ...
11952         (expand_thunk); .. this one; export; get it working with
11953         general functions; make produced gimple valid.
11954         * cgraph.h (expand_thunk): Declare.
11956 2013-08-02  Jan Hubicka  <jh@suse.cz>
11958         * ipa-cp.c (gather_context_independent_values): Use
11959         ipa_get_param_move_cost.
11960         (get_replacement_map): Remove PARAM; move parameter folding
11961         into tree-inline.c
11962         (create_specialized_node): Update.
11963         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11964         assert that we have gimple body; update move_cost.
11965         (count_formal_params): Assert that we have gimple body.
11966         (ipa_dump_param): New function.
11967         (ipa_alloc_node_params): Break out from ...
11968         (ipa_initialize_node_params): ... here.
11969         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11970         (ipa_write_node_info): Stream move costs.
11971         (ipa_read_node_info): Read move costs.
11972         (ipa_update_after_lto_read): Do not recompute node params.
11973         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11974         (ipa_get_param): Check we are not in WPA.
11975         (ipa_get_param_move_cost): New.
11976         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11977         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11978         parm numbers to be present.
11980 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11982         PR rtl-optimization/58048
11983         * lra-constraints.c (process_alt_operands): Don't check asm
11984         operand on register.
11986 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11988         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11989         the implied StoreLoad barrier for atomic operations if before.
11991 2013-08-02  Jan Hubicka  <jh@suse.cz>
11992             Martin Liska  <marxin.liska@gmail.com>
11994         * cgraph.c (cgraph_function_body_availability): Do not check
11995         cgraph flags.
11996         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11997         symtab_node_availability): Declare.
11998         * ipa.c (can_replace_by_local_alias): New.
11999         (function_and_variable_visibility): Use it.
12000         * symtab.c (symtab_for_node_and_aliases,
12001         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
12003 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
12005         PR rtl-optimization/57963
12006         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
12007         (lra_constraints): Use them.
12009 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
12011         * config/arm/types.md (define_attr "type"): Add "load_acq"
12012         and "store_rel".
12013         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
12014         changes.
12015         (cortex_a53_store1): Likewise.
12017 2013-08-01  Jan Hubicka  <jh@suse.cz>
12019         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
12020         partitions are not needed.
12022 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
12024         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
12025         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
12026         MAYBE_NON_Q_CLASS_P where appropriate.
12028 2013-08-01  Jan Hubicka  <jh@suse.cz>
12030         * cgraph.h (release_function_body): Declare.
12031         * tree.c (free_lang_data_in_decl): Free, parameters and return values
12032         of unused delcarations.
12034 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12036         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
12037         RTL form when subtracting a constant.
12039 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12041         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
12042         Generate canonical plus rtx with negated immediate instead of minus
12043         where appropriate.
12044         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
12046 2013-08-01  Jan Hubicka  <jh@suse.cz>
12048         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
12049         (cgraph_release_function_body): Likewise.
12050         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
12051         * cgraph.h (cgrpah_node): Rename abstract_and_needed
12052         to used_as_abstract_origin.
12053         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
12054         symbols used as abstract origins.
12055         * cgraphunit.c (analyze_functions): Update.
12056         * ipa.c (symtab_remove_unreachable_nodes): Recompute
12057         used_as_abstract_origin.
12058         * tree-inline.c (tree_function_versioning): Update
12059         used_as_abstract_origin; be ready for DECL_RESULT and
12060         DECL_ARGUMENTS to be NULL.
12062         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
12063         for abstract functions.
12064         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
12065         real symbols.
12067 2013-08-01  Jan Hubicka  <jh@suse.cz>
12069         * profile.c (compute_value_histograms): Fix thinko.
12071 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
12073         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
12074         aarch-common-protos.h to extra_headers.
12075         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
12076         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
12077         * config/aarch64/t-aarch64 (aarch-common.o): Define.
12079 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
12081         * config/aarch64/aarch64.md (define_attr "type"): Delete.
12082         Include "../arm/types.md".  Define "type" attribute for all patterns.
12083         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
12084         attribute changes.
12086 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
12088         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
12089         to support power8 load fusion.
12090         (fusion_gpr_mem_load): Likewise.
12092         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
12094         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
12095         declarations for power8 load fusion.
12096         (emit_fusion_gpr_load): Likewise.
12098         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
12099         tuning for power8, turn on fusion mode by default.  Turn on sign
12100         extending fusion mode if normal fusion mode is on, and we are at
12101         -O2 or -O3.
12102         (fusion_gpr_load_p): New function, return true if we can fuse an
12103         addis instruction with a dependent load to a GPR.
12104         (emit_fusion_gpr_load): Emit the instructions for power8 load
12105         fusion to GPRs.
12107         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
12108         (VSX load fusion peepholes): New peepholes to fuse together an
12109         addi instruction with a VSX load instruction.
12111         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
12112         peepholes to fuse an addis instruction with a load to a GPR base
12113         register.  If we are supporting sign extending fusions, convert
12114         sign extending loads to zero extending loads and add an explicit
12115         sign extension.
12117 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
12119         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
12120         aarch-common-protos.h to extra_headers.
12121         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
12122         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
12123         (arm_early_store_addr_dep): Likewise.
12124         (arm_no_early_alu_shift_dep): Likewise.
12125         (arm_no_early_alu_shift_value_dep): Likewise.
12126         (arm_no_early_mul_dep): Likewise.
12127         (arm_no_early_store_addr_dep): Likewise.
12128         (arm_mac_accumulator_is_mul_result): Likewise.
12129         (arm_mac_accumulator_is_result): Likewise.
12130         * config/arm/aarch-common.c: ... here.  New file.
12131         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
12132         here to ...
12133         (arm_early_store_addr_dep): Likewise.
12134         (arm_no_early_alu_shift_dep): Likewise.
12135         (arm_no_early_alu_shift_value_dep): Likewise.
12136         (arm_no_early_mul_dep): Likewise.
12137         (arm_no_early_store_addr_dep): Likewise.
12138         (arm_mac_accumulator_is_mul_result): Likewise.
12139         (arm_mac_accumulator_is_result): Likewise.
12140         * config/arm/aarch-common-protos.h: ... here.  New file.
12141         * config/arm/t-arm (aarch-common.o): Define.
12143 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
12145         * config/arm/arm.md: Include new file "types.md".
12146         (define_attr "type"): Move from here to ...
12147         (define_attr "mul32"): Likewise.
12148         (define_attr "mul64"): Likewise.
12149         * config/arm/types.md: ... here.  New file.
12151 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12153         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
12154         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
12156 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12158         * gen-pass-instances.awk: Fix offset of substr().
12160 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
12162         * Makefile.in (pass-instances.def): New.
12163         (passes.o): Replace dependency on passes.def with one on
12164         pass-instances.def
12166         * gen-pass-instances.awk: New.
12168         * passes.c (pass_manager::pass_manager): Use pass-instances.def
12169         rather than passes.def, updating local definition of NEXT_PASS
12170         macro to add an extra NUM parameter (currently unused).
12172 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
12174         * Makefile.in (PASS_MANAGER_H): New.
12175         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
12176         (passes.o): Likewise.
12177         (statistics.o): Likewise.
12178         (cgraphunit.o): Likewise.
12179         (context.o): Depend on PASS_MANAGER_H.
12181         * pass_manager.h: New.
12183         * cgraphunit.c (cgraph_add_new_function): Update for moves
12184         of globals to fields of pass_manager.
12185         (analyze_function): Likewise.
12186         (expand_function): Likewise.
12187         (ipa_passes): Likewise.
12188         (compile): Likewise.
12190         * context.c (context::context): New.
12191         * context.h  (context::context): New.
12192         (context::get_passes): New.
12193         (context::passes_): New.
12195         * lto-cgraph.c (input_node): Update for moves of globals to
12196         fields of pass_manager.
12198         * passes.c (all_passes): Remove, in favor of a field of the
12199         same name within the new class pass_manager.
12200         (all_small_ipa_passes): Likewise.
12201         (all_lowering_passes): Likewise.
12202         (all_regular_ipa_passes): Likewise.
12203         (all_late_ipa_passes): Likewise.
12204         (all_lto_gen_passes): Likewise.
12205         (passes_by_id): Likewise.
12206         (passes_by_id_size): Likewise.
12207         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
12208         the new class pass_manager.
12209         (set_pass_for_id): Convert to...
12210         (pass_manager::set_pass_for_id): ...method.
12211         (get_pass_for_id): Convert to...
12212         (pass_manager::get_pass_for_id): ...method.
12213         (register_one_dump_file): Move body of implementation into...
12214         (pass_manager::register_one_dump_file): ...here.
12215         (register_dump_files_1): Convert to...
12216         (pass_manager::register_dump_files_1): ...method.
12217         (register_dump_files): Convert to...
12218         (pass_manager::register_dump_files): ...method.
12219         (create_pass_tab): Update for moves of globals to fields of
12220         pass_manager.
12221         (dump_passes): Move body of implementation into...
12222         (pass_manager::dump_passes): ...here.
12223         (register_pass): Move body of implementation into...
12224         (pass_manager::register_pass): ...here.
12225         (init_optimization_passes): Convert into...
12226         (pass_manager::pass_manager): ...constructor for new
12227         pass_manager class, and initialize the pass_lists array.
12228         (check_profile_consistency): Update for moves of globals to
12229         fields of pass_manager.
12230         (dump_profile_report): Move body of implementation into...
12231         (pass_manager::dump_profile_report): ...here.
12232         (ipa_write_summaries_1): Update for moves of pass lists from
12233         being globals to fields of pass_manager.
12234         (ipa_write_optimization_summaries): Likewise.
12235         (ipa_read_summaries):  Likewise.
12236         (ipa_read_optimization_summaries): Likewise.
12237         (execute_all_ipa_stmt_fixups): Likewise.
12239         * statistics.c (statistics_fini): Update for moves of globals to
12240         fields of pass_manager.
12242         * toplev.c (general_init): Replace call to
12243         init_optimization_passes with construction of the pass_manager
12244         instance.
12246         * tree-pass.h (all_passes): Remove, in favor of a field of the
12247         same name within the new class pass_manager.
12248         (all_small_ipa_passes): Likewise.
12249         (all_lowering_passes): Likewise.
12250         (all_regular_ipa_passes): Likewise.
12251         (all_lto_gen_passes): Likewise.
12252         (all_late_ipa_passes): Likewise.
12253         (passes_by_id): Likewise.
12254         (passes_by_id_size): Likewise.
12255         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
12256         the new class pass_manager.
12257         (get_pass_for_id): Remove.
12259 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
12261         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
12262         configs.
12264 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
12266         * arm.md (mulhi3): New expand pattern.
12268 2013-07-30  Jan Hubicka  <jh@suse.cz>
12269             Martin Liska  <marxin.liska@gmail.com>
12271         * profile.c (compute_value_histograms): Do not ICE when
12272         there is mismatch only on some counters.
12274 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
12276         PR rtl-optimization/57637
12277         * function.c (move_insn_for_shrink_wrap): Also check the
12278         GEN set of the LIVE problem for the liveness analysis
12279         if it exists, otherwise give up.
12281 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12283         PR tree-optimization/57993
12284         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
12285         replaced statement in the candidate table.
12286         (phi_add_costs): Return infinite cost when the hidden basis does
12287         not dominate all phis on which the candidate is dependent.
12288         (replace_one_candidate): Record replaced statement in the
12289         candidate table.
12291 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
12293         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
12294         (ashlv2si3): New expander.
12295         (*ashlv2si3_i): New define_insn_and_split.
12296         * predicates.md (float_operation): Allow patterns with three
12297         basic sub-patterns.
12299         PR rtl-optimization/58021
12300         * mode-switching.c (create_pre_exit): Always split off preceding
12301         insns if we are not at the basic block head.
12303 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
12305         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
12306         (UCLIBC_DYNAMIC_LINKER): New macro.
12307         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
12308         `-mnan=2008'.
12309         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
12310         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
12311         `-mnan=2008'.
12312         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
12313         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
12314         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
12315         for SF and DF modes.  Use ieee_quad_format for TF mode.
12316         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
12317         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
12318         (mips_option_override): Handle `-mnan=legacy'.
12319         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
12320         `-mabs=2008' and `-mnan=2008'.
12321         (OPTION_DEFAULT_SPECS): Add "nan" default.
12322         (ASM_SPEC): Handle `-mnan='.
12323         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
12324         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
12325         comment accordingly.
12326         (neg<mode>2): Likewise.
12327         * config/mips/mips.opt (mabs, mnan): New options.
12328         * doc/install.texi (Configuration): Document `--with-nan=' option.
12329         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
12330         `-mnan=' options.
12331         (MIPS Options): Document them.
12332         * config.gcc <mips*-*-*>: Handle `--with-nan='.
12333         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
12334         * configure: Regenerate.
12335         * config.in: Regenerate.
12337 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
12339         * config/i386/i386.md (float post-reload splitters): Do not check
12340         for subregs of SSE registers.
12342 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
12343             H.J. Lu  <hongjiu.lu@intel.com>
12345         PR target/57954
12346         PR target/57988
12347         * config/i386/i386.md (post-reload splitter
12348         to avoid partial SSE reg dependency stalls): New pattern.
12350 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
12352         * config/s390/s390.md ("movcc"): Swap load and store instructions.
12354 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
12356         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12357         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
12359 2013-07-26  Cary Coutant  <ccoutant@google.com>
12361         * dwarf2out.c (die_checksum_ordered): Don't include template
12362         instantiations in signature.
12363         (is_template_parameter): New function.
12364         (is_template_instantiation): New function.
12365         (generate_skeleton_bottom_up): Don't include template instantiations
12366         in type unit DIE.
12367         (generate_skeleton): Likewise.
12368         (break_out_comdat_types): Move recursive call to break out nested
12369         types earlier.
12370         (prune_unused_types_mark_generic_parms_dies): Call
12371         is_template_parameter.
12373 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
12375         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
12376         uses vector registers.
12377         * config/aarch64/iterators.md: Add attributes rtn and vas.
12379 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12380             Richard Earnshaw  <richard.earnshaw@arm.com>
12382         * combine.c (simplify_comparison): Re-canonicalize operands
12383         where appropriate.
12384         * config/arm/arm.md (movcond_addsi): New splitter.
12386 2013-07-25  Sterling Augustine  <saugustine@google.com>
12388         * dwarf2out.c (size_of_pubnames): Move code to...
12389         (include_pubname_in_output): ...here.  New.
12390         (want_pubnames): Rearrange.
12391         (output_pubnames): Call include_pubname_in_output.  Move assertion.
12393 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
12395         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
12397 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
12399         PR target/38836
12400         * doc/extend.texi: Remove obsolete builtins. Fix
12401         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
12403 2013-07-25  Jan Hubicka  <jh@suse.cz>
12405         * cgraph.c (release_function_body): Break out from ...
12406         (cgraph_release_function_body): ... this one; also release DECL_RESULT
12407         and DECL_ARGUMENTS.
12408         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
12409         old_tree in the map.
12410         (create_specialized_node): Update.
12411         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
12412         into index.
12413         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
12414         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
12415         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
12416         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
12417         DECL_RESULT.
12419 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12421         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
12422         addsi3_carryin_alt2_<optab>): Correct output template.
12424 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12426         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
12427         Adjust for arm_restrict_it.
12428         Remove trailing whitespace.
12430 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
12432         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
12433         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
12435         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
12437 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
12439         PR rtl-optimization/57960
12440         * lra-constraints.c (process_alt_operands): Use the right mode
12441         when checking strict_low.
12443 2013-07-25  Jan Hubicka  <jh@suse.cz>
12445         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
12446         * cgraph.c (cgraph_remove_node): Do not release function body
12447         when in cgraph streaming.
12448         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
12449         in other partitions are not considered reachable; fix handling of
12450         clones.
12452 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12454         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
12456 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12458         PR target/19599
12459         PR target/57731
12460         PR target/57837
12461         * config/arm/arm.md ("*sibcall_insn): Replace use of
12462         Ss with US. Adjust output for v5 and v4t.
12463         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
12465         * config/arm/constraints.md ("Ss"): Rename to US.
12467 2013-07-25  Terry Guo  <terry.guo@arm.com>
12469         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
12470         shift_add/shift_sub0/shift_sub1 RTXs.
12472 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
12473             Anton Blanchard  <anton@au1.ibm.com>
12475         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
12476         (altivec_vpks<VI_char>ss): Likewise.
12477         (altivec_vpks<VI_char>us): Likewise.
12478         (altivec_vpku<VI_char>us): Likewise.
12479         (altivec_vpku<VI_char>um): Likewise.
12481 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
12483         Introduce context class.
12485         * Makefile.in (CONTEXT_H): New.
12486         (OBJS): Add context.o.
12487         (toplev.o): Add CONTEXT_H to dependencies.
12488         (context.o): New.
12490         * toplev.c (general_init): Create the singleton gcc::context instance.
12492         * context.c: New.
12494         * context.h: New.
12496 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
12498         PR rtl-optimization/57968
12499         * mode-switching.c (create_pre_exit): Allow instructions that
12500         don't set a return register to need a non-exit mode.
12502 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12503             Anton Blanchard  <anton@au1.ibm.com>
12505         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
12506         operands to vperm for little endian.
12507         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
12508         of lvsl to create the control mask for a vperm for little endian.
12510 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12511             Anton Blanchard  <anton@au1.ibm.com>
12513         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
12514         two operands for little-endian.
12516 2013-07-23  Steve Ellcey  <sellcey@mips.com>
12518         * config/mips/mips.c (mips_case_values_threshold): New.
12519         (TARGET_CASE_VALUES_THRESHOLD): Define.
12521 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12522             Anton Blanchard  <anton@au1.ibm.com>
12524         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
12525         selection of field for vector splat in little endian mode.
12527 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
12529         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
12530         expanders to rs6000.md.
12531         (ior<mode>3): Likewise.
12532         (and<mode>3): Likewise.
12533         (one_cmpl<mode>2): Likewise.
12534         (nor<mode>3): Likewise.
12535         (andc<mode>3): Likewise.
12536         (eqv<mode>3): Likewise.
12537         (nand<mode>3): Likewise.
12538         (orc<mode>3): Likewise.
12540         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
12541         declaration.
12543         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
12544         to split multi-word logical operations.
12545         (rs6000_split_logical_di): Likewise.
12546         (rs6000_split_logical): Likewise.
12548         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
12549         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
12550         and allow TImode operations in 32-bit.
12551         (vsx_and<mode>3_64bit): Likewise.
12552         (vsx_ior<mode>3_32bit): Likewise.
12553         (vsx_ior<mode>3_64bit): Likewise.
12554         (vsx_xor<mode>3_32bit): Likewise.
12555         (vsx_xor<mode>3_64bit): Likewise.
12556         (vsx_one_cmpl<mode>2_32bit): Likewise.
12557         (vsx_one_cmpl<mode>2_64bit): Likewise.
12558         (vsx_nor<mode>3_32bit): Likewise.
12559         (vsx_nor<mode>3_64bit): Likewise.
12560         (vsx_andc<mode>3_32bit): Likewise.
12561         (vsx_andc<mode>3_64bit): Likewise.
12562         (vsx_eqv<mode>3_32bit): Likewise.
12563         (vsx_eqv<mode>3_64bit): Likewise.
12564         (vsx_nand<mode>3_32bit): Likewise.
12565         (vsx_nand<mode>3_64bit): Likewise.
12566         (vsx_orc<mode>3_32bit): Likewise.
12567         (vsx_orc<mode>3_64bit): Likewise.
12569         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
12570         logical types in GPRs.
12572         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
12573         logical insns to rs6000.md, and allow TImode operations in
12574         32-bit.
12575         (altivec_ior<mode>3): Likewise.
12576         (altivec_xor<mode>3): Likewise.
12577         (altivec_one_cmpl<mode>2): Likewise.
12578         (altivec_nor<mode>3): Likewise.
12579         (altivec_andc<mode>3): Likewise.
12581         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
12582         attributes for moving the 128-bit logical operations into
12583         rs6000.md.
12584         (BOOL_REGS_OUTPUT): Likewise.
12585         (BOOL_REGS_OP1): Likewise.
12586         (BOOL_REGS_OP2): Likewise.
12587         (BOOL_REGS_UNARY): Likewise.
12588         (BOOL_REGS_AND_CR0): Likewise.
12589         (one_cmpl<mode>2): Add support for DI logical operations on
12590         32-bit, splitting the operations to 32-bit.
12591         (anddi3): Likewise.
12592         (iordi3): Likewise.
12593         (xordi3): Likewise.
12594         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
12595         changes to combine the 32/64-bit code, allow logical operations on
12596         TI mode in 32-bit, and to use similar match_operator patterns like
12597         scalar mode uses.  Combine the Altivec and VSX code for logical
12598         operations, and move it here.
12599         (ior<mode>3, 128-bit types): Likewise.
12600         (xor<mode>3, 128-bit types): Likewise.
12601         (one_cmpl<mode>3, 128-bit types): Likewise.
12602         (nor<mode>3, 128-bit types): Likewise.
12603         (andc<mode>3, 128-bit types): Likewise.
12604         (eqv<mode>3, 128-bit types): Likewise.
12605         (nand<mode>3, 128-bit types): Likewise.
12606         (orc<mode>3, 128-bit types): Likewise.
12607         (and<mode>3_internal): Likewise.
12608         (bool<mode>3_internal): Likewise.
12609         (boolc<mode>3_internal1): Likewise.
12610         (boolc<mode>3_internal2): Likewise.
12611         (boolcc<mode>3_internal1): Likewise.
12612         (boolcc<mode>3_internal2): Likewise.
12613         (eqv<mode>3_internal1): Likewise.
12614         (eqv<mode>3_internal2): Likewise.
12615         (one_cmpl1<mode>3_internal): Likewise.
12617 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12619         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12620         Rename flag_stack_usage to flag_stack_usage_info.
12622 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
12624         * config/microblaze/sync.md: New file.
12625         * config/microblaze/microblaze.md: Include sync.md
12626         * config/microblaze/microblaze.c: Add print_operand 'y'.
12627         * config/microblaze/constraints.md: Add memory_contraint
12628         'Q' which is a single register.
12630 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
12632         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
12634 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12636         * reload.c (find_reloads): Exit loop once we find this operand
12637         cannot be reloaded somehow for this alternative.
12639         * reload.c (find_reloads): Exit loop once we find a hard register.
12641         * rtlanal.c (computed_jump_p): Exit loop once we find label
12642         reference is used.
12644         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12646         * cfgloopmanip.c (remove_path): Exit loop after setting
12647         irred_invalidated.
12649         * gensupport.c (subst_dup): Avoid loop if code is not
12650         MATCH_DUP nor MATCH_OP_DUP.
12652 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12654         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12656 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12658         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12659         true for SP_REGNUM if mode == ptr_mode.
12660         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12661         with value R0_REGNUM + 31.
12663 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12665         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12666         pad pointer-typed argument downward.
12668 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12670         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12671         and __ILP32__ when the ILP32 model is in use.
12673 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12675         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12676         (aarch64_load_symref_appropriately): In the case of
12677         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12678         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12679         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12680         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12681         if 'mode' doesn't equal to 'ptr_mode'.
12682         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12683         'vcall_offset'; change to call aarch64_emit_move differently depending
12684         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12685         to calculate the upper bound of 'vcall_offset'.
12686         (aarch64_cannot_force_const_mem): Change to also return true if
12687         mode != ptr_mode.
12688         (aarch64_legitimize_reload_address): In the case of large
12689         displacements, add new local variable 'xmode' and an assertion
12690         based on it; change to use 'xmode' to generate the new rtx and
12691         reload.
12692         (aarch64_asm_trampoline_template): Change to generate the template
12693         differently depending on TARGET_ILP32 or not; change to use
12694         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12695         (aarch64_trampoline_size): Removed.
12696         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12697         and replace immediate literals with it.  Change to use 'ptr_mode'
12698         instead of 'DImode' and call convert_memory_address if the mode
12699         of 'fnaddr' doesn't equal to 'ptr_mode'.
12700         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12701         to output symbol.
12702         (aarch64_elf_asm_destructor): Likewise.
12703         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12704         on TARGET_ILP32 instead of aarch64_trampoline_size.
12705         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12706         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12707         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12708         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12709         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12710         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12711         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12712         depending on the value of 'mode'.
12713         (add_losym_<mode>): New.
12714         (ldr_got_small_<mode>): New, based on ldr_got_small.
12715         (ldr_got_small): Remove.
12716         (ldr_got_small_sidi): New.
12717         * config/aarch64/iterators.md (P): New.
12718         (PTR): Change to 'ptr_mode' in the condition.
12720 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12722         * config.gcc (aarch64*-*-*): Support --with-abi.
12723         (aarch64*-*-elf): Support --with-multilib-list.
12724         (aarch64*-*-linux*): Likewise.
12725         (supported_defaults): Add abi to aarch64*-*-*.
12726         * configure.ac: Mention AArch64 for --with-multilib-list.
12727         * configure: Re-generated.
12728         * config/aarch64/biarchilp32.h: New file.
12729         * config/aarch64/biarchlp64.h: New file.
12730         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12731         (ABI_SPEC): Ditto.
12732         (MULTILIB_DEFAULTS): Ditto.
12733         (DRIVER_SELF_SPECS): Ditto.
12734         (ASM_SPEC): Update to also substitute -mabi.
12735         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12736         file whose name depends on -mabi= and -mbig-endian.
12737         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12738         TARGET_ILP32.
12739         (POINTER_SIZE): New define.
12740         (POINTERS_EXTEND_UNSIGNED): Ditto.
12741         (enum aarch64_abi_type): New enumeration tag.
12742         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12743         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12744         (TARGET_ILP32): New define.
12745         * config/aarch64/aarch64.opt (mabi): New.
12746         (aarch64_abi): New.
12747         (ilp32, lp64): New values for -mabi.
12748         * config/aarch64/t-aarch64 (comma): New define.
12749         (MULTILIB_OPTIONS): Ditto.
12750         (MULTILIB_DIRNAMES): Ditto.
12751         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12752         * doc/invoke.texi: Document -mabi for AArch64.
12754 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12756         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12758 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12759             Anton Blanchard  <anton@au1.ibm.com>
12761         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12762         endianness when selecting field to splat.
12764 2013-07-22  Eric Christopher  <echristo@gmail.com>
12766         * dwarf2out.c (die_odr_checksum): New function to use
12767         CHECKSUM_ macros and ULEB128 for DIE tag.
12768         (generate_type_signature): Use.
12770 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12772         * config.gcc (sparc*-*-*): Accept leon3 processor.
12773         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12774         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12775         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12776         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12777         (mfix-ut699): Adjust comment.
12778         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12779         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12780         (CPP_CPU_SPEC): Likewise.
12781         (ASM_CPU_SPEC): Likewise.
12782         * config/sparc/sparc.c (leon3_cost): New constant.
12783         (sparc_option_override): Add leon3 support.
12784         (mem_ref): New function.
12785         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12786         (sparc_do_work_around_errata): Look into the instruction in the delay
12787         slot and adjust accordingly.  Add fix for the data cache nullify issues
12788         of the UT699.  Change insertion position for the NOP.
12789         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12790         (leon3_load): New reservation.
12791         (leon_store): Bump latency to 2.
12792         (grfpu): New automaton.
12793         (grfpu_alu): New unit.
12794         (grfpu_ds): Likewise.
12795         (leon_fp_alu): Adjust.
12796         (leon_fp_mult): Delete.
12797         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12798         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12799         * config/sparc/sparc.md (cpu): Add leon3.
12800         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12801         (swapsi): Likewise.
12802         (atomic_test_and_set): Likewise.
12803         (ldstub): Likewise.
12805 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12807         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12808         default for R5900 targets.
12809         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12810         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12811         * config/mips/mips.c (mips_option_override): Report an error for
12812         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12813         for -march=r5900 -mhard-float.
12815 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12817         * df-problems.c (can_move_insns_across): Exit loop once we
12818         find a non-fixed, non-global register.
12820         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12821         setting can_throw.
12823         * omega.c (omega_eliminate_red): Break after setting red_found.
12824         (omega_problem_has_red_equations): Similarly after setting found.
12825         (omega_query_variable): Similarly after setting coupled.
12827 2013-07-22  Marek Polacek  <polacek@redhat.com>
12829         * gimplify.c: Don't include gimple.h twice.
12831 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12833         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12834         instead of TARGET_THUMB1.
12835         (Pz): New constraint.
12836         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12837         encodings.
12838         (compare_negsi_si): Likewise.
12839         (compare_addsi2_op0): Likewise.
12840         (compare_addsi2_op1): Likewise.
12841         (addsi3_carryin_<optab>): Likewise.
12842         (addsi3_carryin_alt2_<optab>): Likewise.
12843         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12844         for arm_restrict_it.
12845         (subsi3_carryin): Likewise.
12846         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12847         (minmax_arithsi): Disable for arm_restrict_it.
12848         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12849         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12850         (satsi_<SAT:code>_shift): Likewise.
12851         (arm_shiftsi3): Add alternative for 16-bit encoding.
12852         (arm32_movhf): Disable for arm_restrict_it.
12853         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12854         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12856 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12858         * config/arm/arm.md (attribute "insn"): Delete.
12859         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12860         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12861         (not_shiftsi): Update for attribute change.
12862         (not_shiftsi_compare0): Likewise.
12863         (not_shiftsi_compare0_scratch): Likewise.
12864         (arm_one_cmplsi2): Likewise.
12865         (thumb1_one_cmplsi2): Likewise.
12866         (notsi_compare0): Likewise.
12867         (notsi_compare0_scratch): Likewise.
12868         (thumb1_movdi_insn): Likewise.
12869         (arm_movsi_insn): Likewise.
12870         (movhi_insn_arch4): Likewise.
12871         (movhi_bytes): Likewise.
12872         (arm_movqi_insn): Likewise.
12873         (thumb1_movqi_insn): Likewise.
12874         (arm32_movhf): Likewise.
12875         (thumb1_movhf): Likewise.
12876         (arm_movsf_soft_insn): Likewise.
12877         (thumb1_movsf_insn): Likewise.
12878         (thumb_movdf_insn): Likewise.
12879         (movsicc_insn): Likewise.
12880         (movsfcc_soft_insn): Likewise.
12881         (and_scc): Likewise.
12882         (cond_move): Likewise.
12883         (if_move_not): Likewise.
12884         (if_not_move): Likewise.
12885         (if_shift_move): Likewise.
12886         (if_move_shift): Likewise.
12887         (if_shift_shift): Likewise.
12888         (if_not_arith): Likewise.
12889         (if_arith_not): Likewise.
12890         (cond_move_not): Likewise.
12891         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12892         (neon_mov<mode>): Likewise.
12893         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12894         (thumb2_movsi_vfp): Likewise.
12895         (movsf_vfp): Likewise.
12896         (thumb2_movsf_vfp): Likewise.
12897         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12898         change.
12899         (cortexa7_older_only): Likewise.
12900         (cortexa7_younger): Likewise.
12901         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12902         (1020alu_shift_op): Likewise.
12903         (1020alu_shift_reg_op): Likewise.
12904         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12905         (alu_shift_op): Likewise.
12906         (alu_shift_reg_op): Likewise.
12907         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12908         (11_alu_shift_op): Likewise.
12909         (11_alu_shift_reg_op): Likewise.
12910         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12911         (9_alu_shift_reg_op): Likewise.
12912         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12913         change.
12914         (cortex_a15_alu_shift): Likewise.
12915         (cortex_a15_alu_shift_reg): Likewise.
12916         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12917         (cortex_a5_alu_shift): Likewise.
12918         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12919         change.
12920         (cortex_a53_alu_shift): Likewise.
12921         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12922         change.
12923         (cortex_a7_alu_reg): Likewise.
12924         (cortex_a7_alu_shift): Likewise.
12925         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12926         (cortex_a8_alu_shift): Likewise.
12927         (cortex_a8_alu_shift_reg): Likewise.
12928         (cortex_a8_mov): Likewise.
12929         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12930         (cortex_a9_dp_shift): Likewise.
12931         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12932         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12933         (cortex_r4_mov): Likewise.
12934         (cortex_r4_alu_shift): Likewise.
12935         (cortex_r4_alu_shift_reg): Likewise.
12936         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12937         (526_alu_shift_op): Likewise.
12938         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12939         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12940         (626te_alu_shift_op): Likewise.
12941         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12942         (726te_alu_op): Likewise.
12943         (726te_alu_shift_op): Likewise.
12944         (726te_alu_shift_reg_op): Likewise.
12945         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12946         (mp626_alu_shift_op): Likewise.
12947         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12948         (pj4_alu_e1_conds): Likewise.
12949         (pj4_alu): Likewise.
12950         (pj4_alu_conds): Likewise.
12951         (pj4_shift): Likewise.
12952         (pj4_shift_conds): Likewise.
12953         (pj4_alu_shift): Likewise.
12954         (pj4_alu_shift_conds): Likewise.
12956 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12958         * config/arm/predicates.md (shiftable_operator_strict_it):
12959         New predicate.
12960         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12961         Disable cond_exec version for arm_restrict_it.
12962         (thumb2_smaxsi3): Convert to generate cond_exec.
12963         (thumb2_sminsi3): Likewise.
12964         (thumb32_umaxsi3): Likewise.
12965         (thumb2_uminsi3): Likewise.
12966         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12967         (thumb2_neg_abssi2): Likewise.
12968         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12969         (thumb2_movsicc_insn): Adjust alternatives.
12970         (thumb2_mov_negscc): Disable for arm_restrict_it.
12971         (thumb2_mov_negscc_strict_it): New pattern.
12972         (thumb2_mov_notscc_strict_it): New pattern.
12973         (thumb2_mov_notscc): Disable for arm_restrict_it.
12974         (thumb2_ior_scc): Likewise.
12975         (thumb2_ior_scc_strict_it): New pattern.
12976         (thumb2_cond_move): Adjust for arm_restrict_it.
12977         (thumb2_cond_arith): Disable for arm_restrict_it.
12978         (thumb2_cond_arith_strict_it): New pattern.
12979         (thumb2_cond_sub): Adjust for arm_restrict_it.
12980         (thumb2_movcond): Likewise.
12981         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12982         (thumb2_zero_extendhisi2_v6): Likewise.
12983         (thumb2_zero_extendqisi2_v6): Likewise.
12984         (orsi_notsi_si): Likewise.
12985         (orsi_not_shiftsi_si): Likewise.
12987 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12989         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12990         instruction sequence is 1 byte shorter.
12992 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12994         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12995         it is not needed after split.
12997 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12999         PR target/51784
13000         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
13001         second label for nonlocal goto receivers. Don't output pic base labels
13002         unless we're producing PIC; mark that action unreachable().
13003         (ix86_save_reg): If the function contains a nonlocal label, save the
13004         PIC base reg.
13005         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
13006         * config/darwin.c (emitted_pic_label_num): New GTY.
13007         (update_pic_label_number_if_needed): New.
13008         (machopic_output_function_base_name): Adjust for nonlocal receiver
13009         case.
13010         (machopic_should_output_picbase_label): New.
13011         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
13012         (nonlocal_goto_receiver): New insn and split.
13014 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
13016         * config/aarch64/aarch64-builtins.c
13017         (aarch64_fold_builtin): Fold abs in all modes.
13018         * config/aarch64/aarch64-simd-builtins.def
13019         (abs): Enable for all modes.
13020         * config/aarch64/arm_neon.h
13021         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
13022         (vabs_f64): Add missing intrinsic.
13024 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
13026         * config/aarch64/arm_neon.h (vabs_s64): New function
13028 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
13030         PR target/57516
13031         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
13032         * config/avr/avr.md (adjust_len): Add `round'.
13033         * config/avr/avr-protos.h (avr_out_round): New prototype.
13034         (avr_out_plus): Add `out_label' argument.
13035         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
13036         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
13037         Handle the case where `insn' is just a pattern.
13038         (avr_out_bitop): Handle the case where `insn' is just a pattern.
13039         (avr_out_round): New function.
13040         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
13042 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
13044         * config/microblaze/microblaze.c (microblaze_expand_prologue):
13045         Add check for flag_stack_usage to handle -fstack-usage support
13047 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
13049         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
13050         interaction for new Power8 flags and VSX.
13052 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
13054         PR middle-end/57698
13055         * tree-inline.c (expand_call_inline): Emit errors during
13056         early_inlining only if optimization is not turned on.
13058 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
13060         * passes.def: New.
13062         * passes.c (init_optimization_passes): Move the construction of
13063         the pass hierarchy into a new passes.def file.
13065         * Makefile.in (passes.o): Add dependency on passes.def.
13067 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
13069         * passes.c (init_optimization_passes): Introduce macros for
13070         constructing the tree of passes (INSERT_PASSES_AFTER,
13071         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
13072         TERMINATE_PASS_LIST).
13074 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
13075             Wei Mi  <wmi@google.com>
13077         PR rtl-optimization/57878
13078         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
13079         top.
13080         (reload_pseudo_compare_func): Check nregs first for reload
13081         pseudos.
13083 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
13085         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
13087 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
13089         * read-rtl.c (validate_const_int): Once an invalid character is
13090         seen, quit the loop.
13092         * gengtype.c (write_roots): Similarly once we find the "deletable"
13093         or "if_marked" option.
13095 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
13097         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
13098         "xtab" and "sat".  Move value "clz" from here to ...
13099         (attriubte "type"): ... here.
13100         (satsi_<SAT:code>): Delete "insn" attribute.
13101         (satsi_<SAT:code>_shift): Likewise.
13102         (arm_zero_extendqisi2addsi): Likewise.
13103         (arm_extendqisi2addsi): Likewise.
13104         (clzsi2): Update for attribute changes.
13105         (rbitsi2): Likewise.
13106         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
13107         attribute.
13108         (arm_usatsihi): Likewise.
13109         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
13111 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
13113         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
13114         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
13115         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
13116         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
13117         in alphabetical order.
13118         (attribute "core_cycles"): Update for attribute changes.
13119         (arm_addsi3): Likewise.
13120         (addsi3_compare0): Likewise.
13121         (addsi3_compare0_scratch): Likewise.
13122         (addsi3_compare_op1): Likewise.
13123         (addsi3_compare_op2): Likewise.
13124         (compare_addsi2_op0): Likewise.
13125         (compare_addsi2_op1): Likewise.
13126         (addsi3_carryin_shift_<optab>): Likewise.
13127         (subsi3_carryin_shift): Likewise.
13128         (rsbsi3_carryin_shift): Likewise.
13129         (arm_subsi3_insn): Likewise.
13130         (subsi3_compare0): Likewise.
13131         (subsi3_compare): Likewise.
13132         (arm_andsi3_insn): Likewise.
13133         (thumb1_andsi3_insn): Likewise.
13134         (andsi3_compare0): Likewise.
13135         (andsi3_compare0_scratch): Likewise.
13136         (zeroextractsi_compare0_scratch
13137         (andsi_not_shiftsi_si): Likewise.
13138         (iorsi3_insn): Likewise.
13139         (iorsi3_compare0): Likewise.
13140         (iorsi3_compare0_scratch): Likewise.
13141         (arm_xorsi3): Likewise.
13142         (thumb1_xorsi3_insn): Likewise.
13143         (xorsi3_compare0): Likewise.
13144         (xorsi3_compare0_scratch): Likewise.
13145         (satsi_<SAT:code>_shift): Likewise.
13146         (rrx): Likewise.
13147         (arm_shiftsi3): Likewise.
13148         (shiftsi3_compare0): Likewise.
13149         (not_shiftsi): Likewise.
13150         (not_shiftsi_compare0): Likewise.
13151         (not_shiftsi_compare0_scratch): Likewise.
13152         (arm_one_cmplsi2): Likewise.
13153         (thumb_one_complsi2): Likewise.
13154         (notsi_compare0): Likewise.
13155         (notsi_compare0_scratch): Likewise.
13156         (thumb1_zero_extendhisi2): Likewise.
13157         (arm_zero_extendhisi2): Likewise.
13158         (arm_zero_extendhisi2_v6): Likewise.
13159         (arm_zero_extendhisi2addsi): Likewise.
13160         (thumb1_zero_extendqisi2): Likewise.
13161         (thumb1_zero_extendqisi2_v6): Likewise.
13162         (arm_zero_extendqisi2): Likewise.
13163         (arm_zero_extendqisi2_v6): Likewise.
13164         (arm_zero_extendqisi2addsi): Likewise.
13165         (thumb1_extendhisi2): Likewise.
13166         (arm_extendhisi2): Likewise.
13167         (arm_extendhisi2_v6): Likewise.
13168         (arm_extendqisi): Likewise.
13169         (arm_extendqisi_v6): Likewise.
13170         (arm_extendqisi2addsi): Likewise.
13171         (thumb1_extendqisi2): Likewise.
13172         (thumb1_movdi_insn): Likewise.
13173         (arm_movsi_insn): Likewise.
13174         (movsi_compare0): Likewise.
13175         (movhi_insn_arch4): Likewise.
13176         (movhi_bytes): Likewise.
13177         (arm_movqi_insn): Likewise.
13178         (thumb1_movqi_insn): Likewise.
13179         (arm32_movhf): Likewise.
13180         (thumb1_movhf): Likewise.
13181         (arm_movsf_soft_insn): Likewise.
13182         (thumb1_movsf_insn): Likewise.
13183         (movdf_soft_insn): Likewise.
13184         (thumb_movdf_insn): Likewise.
13185         (arm_cmpsi_insn): Likewise.
13186         (cmpsi_shiftsi): Likewise.
13187         (cmpsi_shiftsi_swp): Likewise.
13188         (arm_cmpsi_negshiftsi_si): Likewise.
13189         (movsicc_insn): Likewise.
13190         (movsfcc_soft_insn): Likewise.
13191         (arith_shiftsi): Likewise.
13192         (arith_shiftsi_compare0
13193         (arith_shiftsi_compare0_scratch
13194         (sub_shiftsi): Likewise.
13195         (sub_shiftsi_compare0
13196         (sub_shiftsi_compare0_scratch
13197         (and_scc): Likewise.
13198         (cond_move): Likewise.
13199         (if_plus_move): Likewise.
13200         (if_move_plus): Likewise.
13201         (if_move_not): Likewise.
13202         (if_not_move): Likewise.
13203         (if_shift_move): Likewise.
13204         (if_move_shift): Likewise.
13205         (if_shift_shift): Likewise.
13206         (if_not_arith): Likewise.
13207         (if_arith_not): Likewise.
13208         (cond_move_not): Likewise.
13209         (thumb1_ashlsi3): Set type attribute.
13210         (thumb1_ashrsi3): Likewise.
13211         (thumb1_lshrsi3): Likewise.
13212         (thumb1_rotrsi3): Likewise.
13213         (shiftsi3_compare0_scratch): Likewise.
13214         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
13215         (neon_mov<mode>): Likewise.
13216         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
13217         attribute changes.
13218         (thumb2_movsi_insn): Likewise.
13219         (thumb2_cmpsi_neg_shiftsi): Likewise.
13220         (thumb2_extendqisi_v6): Likewise.
13221         (thumb2_zero_extendhisi2_v6): Likewise.
13222         (thumb2_zero_extendqisi2_v6): Likewise.
13223         (thumb2_shiftsi3_short): Likewise.
13224         (thumb2_addsi3_compare0_scratch): Likewise.
13225         (orsi_not_shiftsi_si): Likewise.
13226         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
13227         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
13228         changes.
13229         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
13230         (1020alu_shift_op): Likewise.
13231         (1020alu_shift_reg_op): Likewise.
13232         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
13233         (alu_shift_op): Likewise.
13234         (alu_shift_reg_op): Likewise.
13235         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
13236         (11_alu_shift_op): Likewise.
13237         (11_alu_shift_reg_op): Likewise.
13238         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
13239         (9_alu_shift_reg_op): Likewise.
13240         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
13241         changes.
13242         (cortex_a15_alu_shift): Likewise.
13243         (cortex_a15_alu_shift_reg): Likewise.
13244         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
13245         changes.
13246         (cortex_a5_alu_shift): Likewise.
13247         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
13248         changes.
13249         (cortex_a53_alu_shift): Likewise.
13250         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
13251         changes.
13252         (cortex_a7_alu_reg): Likewise.
13253         (cortex_a7_alu_shift): Likewise.
13254         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
13255         changes.
13256         (cortex_a8_alu_shift): Likewise.
13257         (cortex_a8_alu_shift_reg): Likewise.
13258         (cortex_a8_mov): Likewise.
13259         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
13260         (cortex_a9_dp_shift): Likewise.
13261         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
13262         changes.
13263         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
13264         changes.
13265         (cortex_r4_mov): Likewise.
13266         (cortex_r4_alu_shift): Likewise.
13267         (cortex_r4_alu_shift_reg): Likewise.
13268         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
13269         (526_alu_shift_op): Likewise.
13270         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
13271         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
13272         (626te_alu_shift_op): Likewise.
13273         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
13274         (726te_alu_op): Likewise.
13275         (726te_alu_shift_op): Likewise.
13276         (726te_alu_shift_reg_op): Likewise.
13277         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
13278         (mp626_alu_shift_op): Likewise.
13279         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
13280         (pj4_alu_e1_conds): Likewise.
13281         (pj4_alu): Likewise.
13282         (pj4_alu_conds): Likewise.
13283         (pj4_shift): Likewise.
13284         (pj4_shift_conds): Likewise.
13285         (pj4_alu_shift): Likewise.
13286         (pj4_alu_shift_conds): Likewise.
13287         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
13288         changes.
13289         (cortexa7_older_only): Likewise.
13290         (cortexa7_younger): Likewise.
13292 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
13294         * ipa-pure-const.c (generate_summary): Rename to...
13295         (pure_const_generate_summary): ... this.
13297 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
13299         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
13301 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
13303         PR target/57909
13304         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
13305         usage in HI mode.
13307 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13309         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
13310         enabled without -march=zEC12.
13311         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
13312         flags to be set.
13314 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
13316         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
13317         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
13318         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
13319         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
13320         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
13321         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
13322         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
13323         ISA_HAS_FP4.
13324         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
13325         and ISA_HAS_NMADD3_NMSUB3.
13326         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
13327         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
13328         (nmsub4<mode>, nmsub3<mode>): Likewise.
13329         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
13331 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
13333         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
13334         TARGET_MIPS5400 checking.
13336 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
13337             Peter Bergner  <bergner@vnet.ibm.com>
13339         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
13340         registers in the comment.
13341         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
13342         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
13343         rather than FIRST_PSEUDO_REGISTERS.
13345 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
13347         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
13348         enable extra ISA flags with TARGET_HTM.
13350 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
13352         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
13353         Fix comment typos.
13355 2013-07-15  Cong Hou  <congh@google.com>
13357         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
13358         in compare function for sorting.
13360 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
13362         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
13363         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
13364         * config/rs6000/rs6000.opt: Add -mhtm option.
13365         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
13366         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
13367         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
13368         __HTM__ if the HTM instructions are available.
13369         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
13370         htm_spr_reg_operand): New define_predicates.
13371         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
13372         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
13373         Include htm.md.
13374         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
13375         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
13376         HTM builtin functions.
13377         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
13378         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
13379         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
13380         (rs6000_builtin_mask_calculate): Likewise.
13381         (rs6000_option_override_internal): Likewise.
13382         (bdesc_htm): Add new HTM builtin support.
13383         (htm_spr_num): New function.
13384         (htm_spr_regno): Likewise.
13385         (rs6000_htm_spr_icode): Likewise.
13386         (htm_expand_builtin): Likewise.
13387         (htm_init_builtins): Likewise.
13388         (rs6000_expand_builtin): Add support for HTM builtin functions.
13389         (rs6000_init_builtins): Likewise.
13390         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
13391         option.
13392         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
13393         (TARGET_HTM, MASK_HTM): Define macros.
13394         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
13395         (FIXED_REGISTERS): Likewise.
13396         (CALL_USED_REGISTERS): Likewise.
13397         (CALL_REALLY_USED_REGISTERS): Likewise.
13398         (REG_ALLOC_ORDER): Likewise.
13399         (enum reg_class): Likewise.
13400         (REG_CLASS_NAMES): Likewise.
13401         (REG_CLASS_CONTENTS): Likewise.
13402         (REGISTER_NAMES): Likewise.
13403         (ADDITIONAL_REGISTER_NAMES): Likewise.
13404         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
13405         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
13406         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
13407         * config/rs6000/htm.md: New file.
13408         * config/rs6000/htmintrin.h: New file.
13409         * config/rs6000/htmxlintrin.h: New file.
13411 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13413         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13414         Define SYMBOL_TINY_GOT, update comment.
13415         * config/aarch64/aarch64.c
13416         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
13417         (aarch64_expand_mov_immediate): Likewise.
13418         (aarch64_print_operand): Likewise.
13419         (aarch64_classify_symbol): Likewise.
13420         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
13421         (ldr_got_tiny): Define.
13423 2013-07-13  Tobias Grosser  <tobias@grosser.es>
13425         PR tree-optimization/54094
13426         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
13427           scheduling dimension for the parallelism check from the polyhedral
13428           information in the AST.
13429         * graphite-dependences.c (carries_deps): Do not assume the schedule is
13430           in 2D + 1 form.
13432 2013-07-13  Jason Merrill  <jason@redhat.com>
13434         * print-tree.c (debug_vec_tree): Use debug_raw.
13435         (debug_raw (vec<tree, va_gc> &)): New.
13436         (debug_raw (vec<tree, va_gc> *)): New.
13437         * tree.h: Declare them.
13439 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
13441         * ifcvt.c (ifcvt_after_combine): New static variable.
13442         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
13443         for size.
13444         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
13445         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
13446         rest_of_handle_if_after_reload): Pass new argument for if_convert.
13448 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
13450         * config/mips/mips.c (mips_expand_call): Remove empty statement.
13452 2013-07-12  Michael Matz  <matz@suse.de>
13454         PR middle-end/55771
13455         * convert.c (convert_to_real): Reject non-float inner types.
13457 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
13459         * config/aarch64/aarch64-protos.h
13460         (aarch64_simd_immediate_valid_for_move): Remove.
13461         * config/aarch64/aarch64.c (simd_immediate_info): New member.
13462         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
13463         cases.
13464         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
13466 2013-07-11  Steve Ellcey  <sellcey@mips.com>
13468         * config/mips/mips.c (mips_conditional_register_usage): Do not
13469         use t[0-7] registers in MIPS16 mode when optimizing for size.
13471 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
13473         * config/i386/i386.c (dispatch_function_versions): Fix array
13474         indexing of function_version_info to match actual_versions.
13476 2013-07-11  Teresa Johnson  <tejohnson@google.com>
13478         * vec.h (struct va_gc): Move release out-of-line.
13479         (va_gc::release): Call ggc_free on released vec.
13481 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13483         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
13484         Require GOT register as additional operand in UNSPEC.
13485         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
13486         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
13487         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
13488         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
13489         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
13490         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
13491         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
13493 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
13495         PR target/57631
13496         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
13497         name seen by assembler/linker rather if available.
13499 2013-07-11  Andreas Schwab  <schwab@suse.de>
13501         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
13503 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
13505         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
13507 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
13509         * doc/tm.texi.in: Move hook documentation to ....
13510         * target.def: ... here.
13512         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
13513         text on @hook line.
13514         * doc/tm.texi: Regenerate.
13516 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
13518         PR c++/57869
13519         * doc/invoke.texi: Document -Wconditionally-supported.
13521 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13523         PR target/57844
13524         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
13525         of my_fp.
13527 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13529         PR target/57506
13530         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
13531         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
13532         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
13533         Remove duplicate devices.
13534         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
13535         * config/avr/t-multilib: Regenerate.
13536         * config/avr/avr-tables.opt: Regenerate.
13537         * doc/avr-mmcu.texi: Regenerate.
13539 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
13541         PR target/56987
13542         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
13544 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
13546         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
13547         the cost of MULT when optimizing for size.
13549 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
13551         * config/cr16/cr16-protos.h: Don't include target.h.
13553 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
13555         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
13556         adjust register size for TDmode and TFmode for VSX registers.
13558 2013-07-08  Kai Tietz  <ktietz@redhat.com>
13560         PR target/56892
13561         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
13562         hook_bool_const_tree_true.
13564 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13566         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
13567         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
13568         * config/s390/s390.md: Define FPR*_REGNUM constants.
13569         Fix FPR2_REGNUM constant (18 -> 17).
13570         ("*trunc<BFP:mode><DFP_ALL:mode>2")
13571         ("*trunc<DFP_ALL:mode><BFP:mode>2")
13572         ("trunc<BFP:mode><DFP_ALL:mode>2")
13573         ("trunc<DFP_ALL:mode><BFP:mode>2")
13574         ("*extend<BFP:mode><DFP_ALL:mode>2")
13575         ("*extend<DFP_ALL:mode><BFP:mode>2")
13576         ("extend<BFP:mode><DFP_ALL:mode>2")
13577         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
13578         FPR4_REGNUM.
13580 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
13582         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
13584 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13586         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
13587         and cfun_fpr_bit_p to cfun_fpr_save_p.
13588         (s390_frame_area, s390_register_info, s390_frame_info)
13589         (s390_emit_prologue, s390_emit_epilogue)
13590         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
13591         register numbers.
13592         * config/s390/s390.h: Define *_REGNUM macros for floating point
13593         register numbers.
13595 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
13597         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
13599 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
13601         PR rtl-optimization/57786
13602         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
13603         and break out of the loop when it is set to false.
13605 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
13607         PR target/57819
13608         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
13609         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
13610         (const_int 63)) 0)).
13611         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
13612         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
13613         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
13615         PR rtl-optimization/57829
13616         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
13617         mask bits outside of mode are just sign-extension from mode to HWI.
13619 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
13621         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
13622         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
13623         adjust_address instead of change_address to keep info about alignment.
13624         (emit_strmov): Remove.
13625         (emit_memmov): New function.
13626         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
13627         (expand_movmem_epilogue): Likewise and return updated rtx for
13628         destination.
13629         (expand_constant_movmem_prologue): Likewise and return updated rtx for
13630         destination and source.
13631         (decide_alignment): Refactor, handle vector_loop.
13632         (ix86_expand_movmem): Likewise.
13633         (ix86_expand_setmem): Likewise.
13634         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
13636 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13638         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13639         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13641 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13643         * config/i386/sse.md (sse_movlhps): Change alternative 3
13644         of operand 2 to "m".
13646 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13648         PR target/57807
13649         * config/i386/sse.md (iptr): New mode attribute.
13650         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13651         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13652         (<sse>_vmmul<mode>3): Ditto.
13653         (<sse>_vmdiv<mode>3): Ditto.
13654         (sse_vmrcpv4sf2): Ditto.
13655         (<sse>_vmsqrt<mode>2): Ditto.
13656         (sse_vmrsqrtv4sf2): Ditto.
13657         (<sse>_vm<code><mode>3): Ditto.
13658         (avx_vmcmp<mode>3): Ditto.
13659         (<sse>_vmmaskcmp<mode>3): Ditto.
13660         (<sse>_comi): Ditto.
13661         (<sse>_ucomi): Ditto.
13662         (*xop_vmfrcz_<mode>): Ditto.
13663         (*fmai_fmadd_<mode>): Ditto.
13664         (*fmai_fmsub_<mode>): Ditto.
13665         (*fmai_fnmadd_<mode>): Ditto.
13666         (*fmai_fnmsub_<mode>): Ditto.
13667         (*fma4i_vmfmadd_<mode>): Ditto.
13668         (*fma4i_vmfmsub_<mode>): Ditto.
13669         (*fma4i_vmfnmadd_<mode>): Ditto.
13670         (*fma4i_vmfnmsub_<mode>): Ditto.
13671         (*xop_vmfrcz_<mode>): Ditto.
13672         (sse_cvtps2pi): Ditto.
13673         (sse_cvttps2pi): Ditto.
13674         (sse_cvtss2si): Ditto.
13675         (sse_cvtss2si_2): Ditto.
13676         (sse_cvtss2siq_2): Ditto.
13677         (sse_cvttss2si): Ditto.
13678         (sse_cvttss2siq): Ditto.
13679         (sse_cvtsd2si): Ditto.
13680         (sse_cvtsd2si_2): Ditto.
13681         (sse_cvtsd2siq_2): Ditto.
13682         (sse_cvttsd2si): Ditto.
13683         (sse_cvttsd2siq): Ditto.
13684         (sse_cvtsd2ss): Ditto.
13685         (sse_cvtss2sd): Ditto.
13686         (avx2_pbroadcast<mode>): Ditto.
13687         (avx2_pbroadcast<mode>_1): Ditto.
13688         (*avx_vperm_broadcast_v4sf): Ditto.
13690         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13691         (sse_movlhps): Ditto.
13692         (sse_storehps): Ditto.
13693         (sse_loadhps): Ditto.
13694         (sse_storelps): Ditto.
13695         (sse_loadlps): Ditto.
13696         (*vec_concatv4sf): Ditto.
13697         (*vec_interleave_highv2df): Ditto.
13698         (*vec_interleave_lowv2df): Ditto.
13699         (*vec_extractv2df_1_sse): Ditto.
13700         (*vec_extractv2df_0_sse): Ditto.
13701         (sse2_storelpd): Ditto.
13702         (sse2_loadlpd): Ditto.
13703         (sse2_movsd): Ditto.
13704         (*vec_concatv4si): Ditto.
13705         (vec_concatv2di): Ditto.
13707         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13708         for Intel asm dialect.
13709         (mmx_punpcklwd): Ditto.
13710         (mmx_punpckldq): Ditto.
13712         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13713         for intel assembler dialect.
13715 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13717         PR target/29776
13718         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13719         for BUILT_IN_C{LZ,LRSB}*.
13720         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13721         * tree-vrp.c (extract_range_basic): Handle
13722         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13723         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13724         fall thru to code calling set_value*.
13725         * builtins.c (expand_builtin): Remove *IMAX cases.
13726         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13727         if width is bigger than 2*HWI.
13729 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13731         PR rtl-optimization/55342
13732         * lra-int.h (lra_subreg_reload_pseudos): New.
13733         * lra.c: Add undoing optional reloads to the block diagram.
13734         (lra_subreg_reload_pseudos): New.
13735         (lra_optional_reload_pseudos): Change comments.
13736         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13737         lra_optional_reload_pseudos after undo transformations.
13738         * lra-assigns.c (pseudo_prefix_title): New.
13739         (lra_setup_reg_renumber): Use it.
13740         (spill_for): Ditto.  Check subreg reload pseudos too.
13741         (assign_by_spills): Consider subreg reload pseudos too.
13742         * lra-constraints.c (simplify_operand_subreg): Use
13743         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13744         (curr_insn_transform): Recognize and do optional reloads.
13745         (undo_optional_reloads): New.
13746         (lra_undo_inheritance): Call undo_optional_reloads.
13748 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13750         * tree-complex.c (expand_complex_operations_1): Fix typo.
13752 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13754         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13755         (tune_params): New member 'const vec_costs'.
13756         * config/aarch64/aarch64.c (generic_vector_cost): New.
13757         (generic_tunings): New member 'generic_vector_cost'.
13758         (aarch64_builtin_vectorization_cost): New.
13759         (aarch64_add_stmt_cost): New.
13760         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13761         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13763 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13765         PR target/57777
13766         * config/i386/predicates.md (vsib_address_operand): Disallow
13767         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13769 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13771         PR middle-end/55030
13772         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13773         expand_builtin_setjmp_receiver.
13774         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13775         with NULL for the label parameter.
13776         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13777         the frame-pointer.  Adjust comments.
13778         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13779         only if LABEL is non-NULL.
13781 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13783         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13784         (ARM_ABI_AAPCS64): Ditto.
13785         (arm_abi): Ditto.
13786         (ARM_DEFAULT_ABI): Ditto.
13788 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13790         * config/aarch64/aarch64-builtins.c
13791         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13792         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13793         (st1): Likewise.
13794         * config/aarch64/aarch64-simd.md
13795         (aarch64_ld1<VALL:mode>): New.
13796         (aarch64_st1<VALL:mode>): Likewise.
13797         * config/aarch64/arm_neon.h
13798         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13800 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13802         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13803         target ISA is AVX.
13804         (ix86_option_override_internal):Turn on all -mavx target flags by
13805         default as they are dependent on AVX anyway.
13807 2013-07-02  Cary Coutant  <ccoutant@google.com>
13809         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13810         deterministic hash.
13811         (loc_checksum_ordered): Likewise.
13812         (hash_loc_operands): Remove inline keyword.
13814 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13816         PR tree-optimization/57741
13817         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13818         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13819         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13820         Allow REAL_CST step_exprs if flag_associative_math.
13821         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13823 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13825         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13827 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13829         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13831 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13833         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13834         encoding.
13835         (iorsi3_insn): Likewise.
13836         (arm_xorsi3): Likewise.
13838 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13840         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13841         to ...
13842         (attribute "type"): ... here, and prefix with "wmmx_".
13843         (attribute "core_cycles"): Update for attribute changes.
13844         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13845         (tbcstv4hi): Likewise.
13846         (tbcstv2si): Likewise.
13847         (iwmmxt_iordi3): Likewise.
13848         (iwmmxt_xordi3): Likewise.
13849         (iwmmxt_anddi3): Likewise.
13850         (iwmmxt_nanddi3): Likewise.
13851         (iwmmxt_arm_movdi): Likewise.
13852         (iwmmxt_movsi_insn): Likewise.
13853         (mov<mode>_internal): Likewise.
13854         (and<mode>3_iwmmxt): Likewise.
13855         (ior<mode>3_iwmmxt): Likewise.
13856         (xor<mode>3_iwmmxt): Likewise.
13857         (add<mode>3_iwmmxt): Likewise.
13858         (ssaddv8qi3): Likewise.
13859         (ssaddv4hi3): Likewise.
13860         (ssaddv2si3): Likewise.
13861         (usaddv8qi3): Likewise.
13862         (usaddv4hi3): Likewise.
13863         (usaddv2si3): Likewise.
13864         (sub<mode>3_iwmmxt): Likewise.
13865         (sssubv8qi3): Likewise.
13866         (sssubv4hi3): Likewise.
13867         (sssubv2si3): Likewise.
13868         (ussubv8qi3): Likewise.
13869         (ussubv4hi3): Likewise.
13870         (ussubv2si3): Likewise.
13871         (mulv4hi3_iwmmxt): Likewise.
13872         (smulv4hi3_highpart): Likewise.
13873         (umulv4hi3_highpart): Likewise.
13874         (iwmmxt_wmacs): Likewise.
13875         (iwmmxt_wmacsz): Likewise.
13876         (iwmmxt_wmacu): Likewise.
13877         (iwmmxt_wmacuz): Likewise.
13878         (iwmmxt_clrdi): Likewise.
13879         (iwmmxt_clrv8qi): Likewise.
13880         (iwmmxt_clr4hi): Likewise.
13881         (iwmmxt_clr2si): Likewise.
13882         (iwmmxt_uavgrndv8qi3): Likewise.
13883         (iwmmxt_uavgrndv4hi3): Likewise.
13884         (iwmmxt_uavgv8qi3): Likewise.
13885         (iwmmxt_uavgv4hi3): Likewise.
13886         (iwmmxt_tinsrb): Likewise.
13887         (iwmmxt_tinsrh): Likewise.
13888         (iwmmxt_tinsrw): Likewise.
13889         (iwmmxt_textrmub): Likewise.
13890         (iwmmxt_textrmsb): Likewise.
13891         (iwmmxt_textrmuh): Likewise.
13892         (iwmmxt_textrmsh): Likewise.
13893         (iwmmxt_textrmw): Likewise.
13894         (iwmxxt_wshufh): Likewise.
13895         (eqv8qi3): Likewise.
13896         (eqv4hi3): Likewise.
13897         (eqv2si3): Likewise.
13898         (gtuv8qi3): Likewise.
13899         (gtuv4hi3): Likewise.
13900         (gtuv2si3): Likewise.
13901         (gtv8qi3): Likewise.
13902         (gtv4hi3): Likewise.
13903         (gtv2si3): Likewise.
13904         (smax<mode>3_iwmmxt): Likewise.
13905         (umax<mode>3_iwmmxt): Likewise.
13906         (smin<mode>3_iwmmxt): Likewise.
13907         (umin<mode>3_iwmmxt): Likewise.
13908         (iwmmxt_wpackhss): Likewise.
13909         (iwmmxt_wpackwss): Likewise.
13910         (iwmmxt_wpackdss): Likewise.
13911         (iwmmxt_wpackhus): Likewise.
13912         (iwmmxt_wpackwus): Likewise.
13913         (iwmmxt_wpackdus): Likewise.
13914         (iwmmxt_wunpckihb): Likewise.
13915         (iwmmxt_wunpckihh): Likewise.
13916         (iwmmxt_wunpckihw): Likewise.
13917         (iwmmxt_wunpckilb): Likewise.
13918         (iwmmxt_wunpckilh): Likewise.
13919         (iwmmxt_wunpckilw): Likewise.
13920         (iwmmxt_wunpckehub): Likewise.
13921         (iwmmxt_wunpckehuh): Likewise.
13922         (iwmmxt_wunpckehuw): Likewise.
13923         (iwmmxt_wunpckehsb): Likewise.
13924         (iwmmxt_wunpckehsh): Likewise.
13925         (iwmmxt_wunpckehsw): Likewise.
13926         (iwmmxt_wunpckelub): Likewise.
13927         (iwmmxt_wunpckeluh): Likewise.
13928         (iwmmxt_wunpckeluw): Likewise.
13929         (iwmmxt_wunpckelsb): Likewise.
13930         (iwmmxt_wunpckelsh): Likewise.
13931         (iwmmxt_wunpckelsw): Likewise.
13932         (ror<mode>3): Likewise.
13933         (ashr<mode>3_iwmmxt): Likewise.
13934         (lshr<mode>3_iwmmxt): Likewise.
13935         (ashl<mode>3_iwmmxt): Likewise.
13936         (ror<mode>3_di): Likewise.
13937         (ashr<mode>3_di): Likewise.
13938         (lshr<mode>3_di): Likewise.
13939         (ashl<mode>3_di): Likewise.
13940         (iwmmxt_wmadds): Likewise.
13941         (iwmmxt_wmaddu): Likewise.
13942         (iwmmxt_tmia): Likewise.
13943         (iwmmxt_tmiaph): Likewise.
13944         (iwmmxt_tmiabb): Likewise.
13945         (iwmmxt_tmiatb): Likewise.
13946         (iwmmxt_tmiabt): Likewise.
13947         (iwmmxt_tmiatt): Likewise.
13948         (iwmmxt_tmovmskb): Likewise.
13949         (iwmmxt_tmovmskh): Likewise.
13950         (iwmmxt_tmovmskw): Likewise.
13951         (iwmmxt_waccb): Likewise.
13952         (iwmmxt_wacch): Likewise.
13953         (iwmmxt_waccw): Likewise.
13954         (iwmmxt_waligni): Likewise.
13955         (iwmmxt_walignr): Likewise.
13956         (iwmmxt_walignr0): Likewise.
13957         (iwmmxt_walignr1): Likewise.
13958         (iwmmxt_walignr2): Likewise.
13959         (iwmmxt_walignr3): Likewise.
13960         (iwmmxt_wsadb): Likewise.
13961         (iwmmxt_wsadh): Likewise.
13962         (iwmmxt_wsadbz): Likewise.
13963         (iwmmxt_wsadhz): Likewise.
13964         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13965         (iwmmxt_wabsdiffb): Likewise.
13966         (iwmmxt_wabsdiffh): Likewise.
13967         (iwmmxt_wabsdiffw): Likewise.
13968         (iwmmxt_waddsubhx): Likewise
13969         (iwmmxt_wsubaddhx): Likewise.
13970         (addc<mode>3): Likewise.
13971         (iwmmxt_avg4): Likewise.
13972         (iwmmxt_avg4r): Likewise.
13973         (iwmmxt_wmaddsx): Likewise.
13974         (iwmmxt_wmaddux): Likewise.
13975         (iwmmxt_wmaddsn): Likewise.
13976         (iwmmxt_wmaddun): Likewise.
13977         (iwmmxt_wmulwsm): Likewise.
13978         (iwmmxt_wmulwum): Likewise.
13979         (iwmmxt_wmulsmr): Likewise.
13980         (iwmmxt_wmulumr): Likewise.
13981         (iwmmxt_wmulwsmr): Likewise.
13982         (iwmmxt_wmulwumr): Likewise.
13983         (iwmmxt_wmulwl): Likewise.
13984         (iwmmxt_wqmulm): Likewise.
13985         (iwmmxt_wqmulwm): Likewise.
13986         (iwmmxt_wqmulmr): Likewise.
13987         (iwmmxt_wqmulwmr): Likewise.
13988         (iwmmxt_waddbhusm): Likewise.
13989         (iwmmxt_waddbhusl): Likewise.
13990         (iwmmxt_wqmiabb): Likewise.
13991         (iwmmxt_wqmiabt): Likewise.
13992         (iwmmxt_wqmiatb): Likewise.
13993         (iwmmxt_wqmiatt): Likewise.
13994         (iwmmxt_wqmiabbn): Likewise.
13995         (iwmmxt_wqmiabtn): Likewise.
13996         (iwmmxt_wqmiatbn): Likewise.
13997         (iwmmxt_wqmiattn): Likewise.
13998         (iwmmxt_wmiabb): Likewise.
13999         (iwmmxt_wmiabt): Likewise.
14000         (iwmmxt_wmiatb): Likewise.
14001         (iwmmxt_wmiatt): Likewise.
14002         (iwmmxt_wmiabbn): Likewise.
14003         (iwmmxt_wmiabtn): Likewise.
14004         (iwmmxt_wmiatbn): Likewise.
14005         (iwmmxt_wmiattn): Likewise.
14006         (iwmmxt_wmiawbb): Likewise.
14007         (iwmmxt_wmiawbt): Likewise.
14008         (iwmmxt_wmiawtb): Likewise.
14009         (iwmmxt_wmiawtt): Likewise.
14010         (iwmmxt_wmiawbbn): Likewise.
14011         (iwmmxt_wmiawbtn): Likewise.
14012         (iwmmxt_wmiawtbn): Likewise.
14013         (iwmmxt_wmiawttn): Likewise.
14014         (iwmmxt_wmerge): Likewise.
14015         (iwmmxt_tandc<mode>3): Likewise.
14016         (iwmmxt_torc<mode>3): Likewise.
14017         (iwmmxt_torvsc<mode>3): Likewise.
14018         (iwmmxt_textrc<mode>3): Likewise.
14019         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
14020         (wmmxt_pack): Likewise.
14021         (wmmxt_mult_c1): Likewise.
14022         (wmmxt_mult_c2): Likewise.
14023         (wmmxt_alu_c1): Likewise.
14024         (wmmxt_alu_c2): Likewise.
14025         (wmmxt_alu_c3): Likewise.
14026         (wmmxt_transfer_c1): Likewise.
14027         (wmmxt_transfer_c2): Likewise.
14028         (wmmxt_transfer_c3): Likewise.
14029         (marvell_f_iwmmxt_wstr): Likewise.
14030         (marvell_f_iwmmxt_wldr): Likewise.
14032 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
14034         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
14036 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
14038         Revert:
14039         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
14040         * lra-constraints.c (need_for_split_p): Check call used hard regs
14041         living through calls.
14043         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
14044         call used regs for call insn.
14046 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
14048         PR target/57736
14049         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
14050         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
14052 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
14054         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
14056 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
14058         * lra-constraints.c (need_for_split_p): Check call used hard regs
14059         living through calls.
14061 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
14063         PR target/57744
14064         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
14065         to tie with any other modes.  Eliminate Altivec vector mode tests,
14066         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
14067         code, to return 0 if testing MODE2 for a condition, if we've
14068         already tested MODE1 for the same condition.
14070 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
14072         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
14073         layout.
14075 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
14077         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
14078         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
14080 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
14082         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
14083         Define.
14084         (aarch64_symbolic_constant_p): Remove.
14085         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
14086         static.  Fix line length and white space.
14087         (aarch64_symbolic_constant_p): Remove.
14088         * config/aarch64/predicates.md (aarch64_valid_symref):
14089         Use aarch64_classify_symbol_expression.
14091 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14093         * config/arm/constraints.md (Ts): New constraint.
14094         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
14095         16-bit encodings.
14096         (compare_scc): Use "Ts" constraint for operand 0.
14097         (ior_scc_scc): Likewise.
14098         (and_scc_scc): Likewise.
14099         (and_scc_scc_nodom): Likewise.
14100         (ior_scc_scc_cmp): Likewise for operand 7.
14101         (and_scc_scc_cmp): Likewise.
14102         * config/arm/thumb2.md (thumb2_movsi_insn):
14103         Add alternatives for 16-bit encodings.
14104         (thumb2_movhi_insn): Likewise.
14105         (thumb2_movsicc_insn): Likewise.
14106         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
14107         (thumb2_negscc): Use "Ts" constraint.
14108         Move mvn instruction outside cond_exec block.
14109         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
14110         for 16-bit encodings.
14112 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14114         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
14115         encoding.
14116         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
14117         (mulsi3subsi): Likewise.
14118         (mulsidi3adddi): Likewise.
14119         (mulsidi3_v6): Likewise.
14120         (umulsidi3_v6): Likewise.
14121         (umulsidi3adddi_v6): Likewise.
14122         (smulsi3_highpart_v6): Likewise.
14123         (umulsi3_highpart_v6): Likewise.
14124         (mulhisi3tb): Likewise.
14125         (mulhisi3bt): Likewise.
14126         (mulhisi3tt): Likewise.
14127         (maddhisi4): Likewise.
14128         (maddhisi4tb): Likewise.
14129         (maddhisi4tt): Likewise.
14130         (maddhidi4): Likewise.
14131         (maddhidi4tb): Likewise.
14132         (maddhidi4tt): Likewise.
14133         (zeroextractsi_compare0_scratch): Likewise.
14134         (insv_zero): Likewise.
14135         (insv_t2): Likewise.
14136         (anddi_notzesidi_di): Likewise.
14137         (anddi_notsesidi_di): Likewise.
14138         (andsi_notsi_si): Likewise.
14139         (iordi_zesidi_di): Likewise.
14140         (xordi_zesidi_di): Likewise.
14141         (andsi_iorsi3_notsi): Likewise.
14142         (smax_0): Likewise.
14143         (smax_m1): Likewise.
14144         (smin_0): Likewise.
14145         (not_shiftsi): Likewise.
14146         (unaligned_loadsi): Likewise.
14147         (unaligned_loadhis): Likewise.
14148         (unaligned_loadhiu): Likewise.
14149         (unaligned_storesi): Likewise.
14150         (unaligned_storehi): Likewise.
14151         (extv_reg): Likewise.
14152         (extzv_t2): Likewise.
14153         (divsi3): Likewise.
14154         (udivsi3): Likewise.
14155         (arm_zero_extendhisi2addsi): Likewise.
14156         (arm_zero_extendqisi2addsi): Likewise.
14157         (compareqi_eq0): Likewise.
14158         (arm_extendhisi2_v6): Likewise.
14159         (arm_extendqisi2addsi): Likewise.
14160         (arm_movt): Likewise.
14161         (thumb2_ldrd): Likewise.
14162         (thumb2_ldrd_base): Likewise.
14163         (thumb2_ldrd_base_neg): Likewise.
14164         (thumb2_strd): Likewise.
14165         (thumb2_strd_base): Likewise.
14166         (thumb2_strd_base_neg): Likewise.
14167         (arm_negsi2): Add alternative for 16-bit encoding.
14168         (arm_one_cmplsi2): Likewise.
14170 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14172         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
14173         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
14174         (movdfcc): Likewise.
14175         * config/arm/vfp.md (*thumb2_movsf_vfp):
14176         Disable predication for arm_restrict_it.
14177         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
14178         (*thumb2_movdfcc_vfp): Likewise.
14179         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
14180         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
14181         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
14182         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
14183         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
14184         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
14185         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
14186         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
14187         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
14188         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
14189         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
14190         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
14191         Disable predication for arm_restrict_it.
14193 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
14195         * config/i386/bmiintrin.h (_bextr_u32): New.
14196         (_bextr_u64): Ditto.
14198 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
14200         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
14201         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
14202         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
14203         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
14204         * config/mips/n32-elf.h: ...this new file.
14206 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
14208         PR target/57224
14209         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
14210         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
14212 2013-06-27  Catherine Moore  <clm@codesourcery.com>
14214         * config/mips/mips-tables.opt: Regenerate.
14215         * config/mips/mips-cpus.def: Add m14ke and m14kec.
14216         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
14217         * doc/invoke.texi: Add -m14kc.
14219 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
14221         PR target/57623
14222         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
14223         constraints of operand 1 and 2.
14225         PR target/57623
14226         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
14227         to match RTL canonicalization.  Swap predicates and
14228         constraints of operand 1 and 2.
14230 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
14232         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
14233         Process OP_INOUT regs for splitting too.
14235 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
14237         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
14238         decl before the loop, initialize to NULL.
14239         (vectorizable_load): Initialize ptr_incr to NULL.
14241 2013-06-27  Martin Jambor  <mjambor@suse.cz>
14243         PR lto/57208
14244         * ipa-ref.h (ipa_maybe_record_reference): Declare.
14245         * ipa-ref.c (ipa_maybe_record_reference): New function.
14246         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
14247         * ipa-cp.c (create_specialized_node): Record potential references from
14248         aggvals.
14249         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
14251 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
14253         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
14254         parameter 'mode' of type 'enum machine_mode mode'; change to pass
14255         'mode' to force_reg.
14256         (aarch64_add_offset): Update calls to aarch64_force_temporary.
14257         (aarch64_expand_mov_immediate): Likewise.
14259 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
14261         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
14262         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
14264 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14266         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
14267         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
14268         (struct machine_function): Add tbegin_p.
14269         (s390_canonicalize_comparison): Fold CC mode compares to
14270         conditional jump if possible.
14271         (s390_emit_jump): Return the emitted jump.
14272         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
14273         Handle CCRAWmode compares.
14274         (s390_option_override): Default to -mhtm if available.
14275         (s390_reg_clobbered_rtx): Handle floating point regs as well.
14276         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
14277         FPRs instead of df_regs_ever_live_p.
14278         (s390_optimize_nonescaping_tx): New function.
14279         (s390_init_frame_layout): Extend clobbered_regs array to cover
14280         FPRs as well.
14281         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
14282         (s390_expand_tbegin): New function.
14283         (enum s390_builtin): New enum definition.
14284         (code_for_builtin): New array definition.
14285         (s390_init_builtins): New function.
14286         (s390_expand_builtin): New function.
14287         (TARGET_INIT_BUILTINS): Define.
14288         (TARGET_EXPAND_BUILTIN): Define.
14289         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
14290         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
14291         (s390_alc_comparison): Likewise.
14292         * config/s390/s390-modes.def: Add CCRAWmode.
14293         * config/s390/s390.h (processor_flags): Add PF_TX.
14294         (TARGET_CPU_HTM): Define macro.
14295         (TARGET_HTM): Define macro.
14296         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
14297         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
14298         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
14299         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
14300         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
14301         (TBEGIN_MASK, TBEGINC_MASK): New constants.
14302         ("*cc_to_int"): Move up.
14303         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
14304         constants other than 0.
14305         ("*ccraw_to_int"): New insn and splitter definition.
14306         ("tbegin", "tbegin_nofloat", "tbegin_retry")
14307         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
14308         ("tx_assist"): New expander.
14309         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
14310         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
14311         * config/s390/s390.opt: Add -mhtm option.
14312         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
14313         * config/s390/htmxlintrin.h: New file.
14314         * config/s390/htmintrin.h: New file.
14315         * config/s390/s390intrin.h: New file.
14316         * doc/extend.texi: Document htm builtins.
14317         * config.gcc: Add the new header files to extra_headers.
14319 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
14321         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
14322         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
14324 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
14325             Pat Haugen  <pthaugen@us.ibm.com>
14326             Peter Bergner  <bergner@vnet.ibm.com>
14328         * config/rs6000/power8.md: New.
14329         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
14330         setting for power8 entry.
14331         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
14332         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
14333         test for Power4/Power5 only.
14334         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
14335         support.
14336         (force_new_group): Adjust comment.
14337         * config/rs6000/rs6000.md: Include power8.md.
14339 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
14341         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
14342         * config/arm/arm-protos.h (arm_max_conditional_execute): New
14343         declaration.
14344         (tune_params): Update comment.
14345         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
14346         (arm_max_conditional_execute): New function.
14347         (thumb2_final_prescan_insn): Use max_insn_skipped and
14348         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
14350 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
14352         PR tree-optimization/57705
14353         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
14354         SSA_NAME step, provided that it is not defined inside the loop.
14355         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
14356         (get_initial_def_for_induction): Handle SSA_NAME IV step.
14358 2013-06-25  Martin Jambor  <mjambor@suse.cz>
14360         PR middle-end/57670
14361         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
14362         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
14363         calls in the dump.
14364         (ipa_note_param_call): Initialize member_ptr flag.
14365         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
14366         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
14367         (ipa_write_indirect_edge_info): Stream member_ptr flag.
14368         (ipa_read_indirect_edge_info): Likewise.
14370 2013-06-25  Richard Biener  <rguenther@suse.de>
14372         PR middle-end/56977
14373         * passes.c (init_optimization_passes): Move pass_fold_builtins
14374         and pass_dce earlier with -Og.
14376 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
14378         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
14379         <BIT_FIELD_REF>: Remove trailing TAB.
14380         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
14381         remove blank line.
14383 2013-06-24  Martin Jambor  <mjambor@suse.cz>
14385         PR tree-optimization/57358
14386         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
14387         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
14388         (ipa_analyze_params_uses): Generate pessimistic info when true.
14390 2013-06-24  Martin Jambor  <mjambor@suse.cz>
14392         PR tree-optimization/57539
14393         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
14394         global.inlined_to of the new node to it.  All callers changed.
14395         * ipa-inline-transform.c (clone_inlined_nodes): New variable
14396         inlining_into, pass it to cgraph_clone_node.
14397         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
14398         ipa_free_edge_args_substructures.
14399         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
14400         rdesc linked list.  Do not assert rdesc edges have inlined caller.
14401         Assert we have found an rdesc in the rdesc list.
14403 2013-06-24  Richard Biener  <rguenther@suse.de>
14405         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
14406         (pointer_set_lookup): Declare.
14407         (class pointer_map): New template class implementing a
14408         generic pointer to T map.
14409         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
14410         pointer_map<T>::contains, pointer_map<T>::insert,
14411         pointer_map<T>::traverse): New functions.
14412         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
14413         (pointer_set_lookup): New function.
14414         (pointer_set_contains): Use pointer_set_lookup.
14415         (pointer_set_insert): Likewise.
14416         (insert_aux): Remove.
14417         (struct pointer_map_t): Embed a pointer_set_t.
14418         (pointer_map_create): Adjust.
14419         (pointer_map_destroy): Likewise.
14420         (pointer_map_contains): Likewise.
14421         (pointer_map_insert): Likewise.
14422         (pointer_map_traverse): Likewise.
14423         * tree-streamer.h (struct streamer_tree_cache_d): Use a
14424         pointer_map<unsigned> instead of a pointer_map_t.
14425         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
14426         (streamer_tree_cache_lookup): Likewise.
14427         (streamer_tree_cache_create): Likewise.
14428         (streamer_tree_cache_delete): Likewise.
14429         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
14430         pointer_map<unsigned> instead of a pointer_map_t.
14431         (lto_init_tree_ref_encoder): Adjust.
14432         (lto_destroy_tree_ref_encoder): Likewise.
14433         * lto-section-out.c (lto_output_decl_index): Likewise.
14434         (lto_record_function_out_decl_state): Likewise.
14435         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
14437 2013-06-24  Richard Biener  <rguenther@suse.de>
14439         PR tree-optimization/57488
14440         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
14442 2013-06-24  Alan Modra  <amodra@gmail.com>
14444         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
14445         (gen_easy_altivec_constant): Likewise.
14446         * config/rs6000/predicates.md (easy_vector_constant_add_self,
14447         easy_vector_constant_msb): Likewise.
14449 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
14451         PR target/57688
14452         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
14453         add missing return true.
14455 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
14457         PR target/52483
14458         * config/sh/predicates.md (general_extend_operand): Invoke
14459         general_movsrc_operand for memory operands.
14460         (general_movsrc_operand): Allow reg+reg addressing, do not use
14461         general_operand for memory operands.
14463 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
14465         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
14466         when current target options does not apply.
14467         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
14468         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
14469         * config/i386/bmiintrin.h: Pass appropriate target
14470         attributes to header.
14471         * config/i386/mmintrin.h: Ditto.
14472         * config/i386/nmmintrin.h: Ditto.
14473         * config/i386/avx2intrin.h: Ditto.
14474         * config/i386/fxsrintrin.h: Ditto.
14475         * config/i386/tbmintrin.h: Ditto.
14476         * config/i386/xsaveintrin.h: Ditto.
14477         * config/i386/f16cintrin.h: Ditto.
14478         * config/i386/xtestintrin.h: Ditto.
14479         * config/i386/xsaveoptintrin.h: Ditto.
14480         * config/i386/bmi2intrin.h: Ditto.
14481         * config/i386/lzcntintrin.h: Ditto.
14482         * config/i386/smmintrin.h: Ditto.
14483         * config/i386/wmmintrin.h: Ditto.
14484         * config/i386/x86intrin.h: Remove all header include guards.
14485         * config/i386/prfchwintrin.h: Ditto.
14486         * config/i386/pmmintrin.h: Ditto.
14487         * config/i386/tmmintrin.h: Ditto.
14488         * config/i386/xmmintrin.h: Ditto.
14489         * config/i386/popcntintrin.h: Ditto.
14490         * config/i386/rdseedintrin.h: Ditto.
14491         * config/i386/ammintrin.h: Ditto.
14492         * config/i386/emmintrin.h: Ditto.
14493         * config/i386/immintrin.h: Remove all header include guards.
14494         * config/i386/fma4intrin.h: Ditto.
14495         * config/i386/lwpintrin.h: Ditto.
14496         * config/i386/xopintrin.h: Ditto.
14497         * config/i386/ia32intrin.h: Ditto.
14498         * config/i386/avxintrin.h: Ditto.
14499         * config/i386/rtmintrin.h: Ditto.
14500         * config/i386/fmaintrin.h: Ditto.
14501         * config/i386/mm3dnow.h: Ditto.
14503 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
14505         * common/config/i386/i386-common.c: Handle LZCNT.
14507 2013-06-22  Andi Kleen  <ak@linux.intel.com>
14509         * doc/extend.texi: Use __atomic_store_n instead of
14510         __atomic_store in HLE example.
14512 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
14514         * config/sh/sh.c: Remove <cstdlib> workaround.
14516 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14518         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
14520 2013-06-21  Andi Kleen  <ak@linux.intel.com>
14522         * doc/extend.texi: Document that __atomic_clear and
14523           __atomic_test_and_set should only be used with bool.
14525 2013-06-20  Jan Hubicka  <jh@suse.cz>
14527         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
14528         types_same_for_odr.
14529         * tree.c (decls_same_for_odr): New function.
14530         (same_for_edr): New function.
14531         (types_same_for_odr): New function.
14532         (get_binfo_at_offset): Use it.
14533         * tree.h (types_same_for_odr): Declare.
14535 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
14536             Jason Merrill  <jason@redhat.com>
14538         * system.h: Include <cstdlib> as well as <stdlib.h>.
14540 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
14542         PR target/57655
14543         * config/i386/i386.c (construct_container): Report error if
14544         long double is used with disabled x87 float returns.
14546 2013-06-20  Jan Hubicka  <jh@suse.cz>
14548         * lto-cgraph.c (input_symtab): Do not set cgraph state.
14550 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
14552         PR rtl-optimization/57425
14553         PR rtl-optimization/57569
14554         * alias.c (write_dependence_p): Remove parameters mem_mode and
14555         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
14556         Changed all callers.
14557         (canon_anti_dependence): Get comments and semantics in sync.
14558         Add parameter mem_canonicalized.  Changed all callers.
14559         * rtl.h (canon_anti_dependence): Update prototype.
14561 2013-06-20  Richard Biener  <rguenther@suse.de>
14563         * data-streamer-in.c (streamer_read_uhwi): Optimize single
14564         byte case, inline streamer_read_uchar and defer section
14565         overrun check.
14567 2013-06-20  Richard Biener  <rguenther@suse.de>
14569         PR tree-optimization/57584
14570         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
14571         SSA names into the expanded expression that take part in
14572         abnormal coalescing.
14574 2013-06-19  Sharad Singhai  <singhai@google.com>
14576         * gcov.c (print_usage): Handle new option.
14577         (process_args): Ditto.
14578         (get_gcov_intermediate_filename): New function.
14579         (output_intermediate_file): New function.
14580         (output_gcov_file): New function
14581         (generate_results): Handle new option.
14582         (release_function): Relase demangled name.
14583         (read_graph_file): Handle demangled name.
14584         (output_lines): Ditto.
14585         * doc/gcov.texi: Document gcov intermediate format.
14587 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
14589         PR bootstrap/57604
14590         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
14591         (lra_emit_add): Use the functions.  Add comment about Y as an
14592         address segment.
14594 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
14596         PR driver/57652
14597         * collect2.c (collect_atexit): New.
14598         (collect_exit): Delete.
14599         (main): Register collect_atexit with atexit.
14600         (collect_wait): Change collect_exit to exit.
14601         (do_wait): Same.
14602         * collect2.h (collect_exit): Delete.
14603         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
14605 2013-06-19  Wei Mi  <wmi@google.com>
14607         PR rtl-optimization/57518
14608         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
14609         if regno is used in paradoxical subreg.
14610         (update_equiv_regs): Check pdx_subregs[regno] before
14611         set a reg to be equivalent with a mem.
14613 2013-06-19  Matthias Klose  <doko@ubuntu.com>
14615         PR driver/57651
14616         * file-find.h (find_a_file): Add a mode parameter.
14617         * file-find.c (find_a_file): Likewise.
14618         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
14619         with X_OK for the executables.
14620         * collect2.c (main): Call find_a_file with X_OK.
14622 2013-06-19  Steve Ellcey  <sellcey@mips.com>
14624         PR target/56942
14625         * config/mips/mips.md (casesi_internal_mips16_<mode>):
14626         Use NEXT_INSN instead of next_real_insn.
14628 2013-06-19  Jan Hubicka  <jh@suse.cz>
14630         * cgraph.h (const_value_known_p): Replace by ...
14631         (ctor_for_folding): .. this one.
14632         * cgraphunit.c (process_function_and_variable_attributes): Use it.
14633         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
14634         * expr.c (expand_expr_real_1): Likewise.
14635         (string_constant): Likewise.
14636         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14637         * ipa.c (process_references): Likewise.
14638         (symtab_remove_unreachable_nodes): Likewise.
14639         * ipa-inline-analysis.c (param_change_prob): Likewise.
14640         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14641         (get_base_constructor): Likwise.
14642         * varpool.c (varpool_remove_node): Likewise.
14643         (varpool_remove_initializer): LIkewise.
14644         (dump_varpool_node): LIkwise.
14645         (const_value_known_p): Rewrite to ...
14646         (ctor_for_folding): ... this one.
14648 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14650         PR driver/57651
14651         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14652         PERSONALITY in $PATH derived prefixes.
14654 2013-06-19  Jeff Law  <law@redhat.com>
14656         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14657         in comment.
14659         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14660         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14661         booleans.
14663 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14665         * config/arm/vfp.md: Move VFP instruction classification documentation
14666         to ...
14667         * config/arm/arm.md: ... here.  Update instruction classification
14668         documentation.
14670 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14672         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14673         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14674         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14676 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14678         PR target/57609
14679         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14680         with NEXT_INSN.  Use tablejump_p to check for jump table data
14681         insns.
14683 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14685         PR c++/56544
14686         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14687         that now in C++ the value is correct per the C++ standards.
14689 2013-06-19  Richard Biener  <rguenther@suse.de>
14691         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14692         for global context.
14694 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14696         Revert:
14697         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14699         PR target/57609
14700         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14701         with next_active_insn.
14703 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14705         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14706         functions are inlined during failures to flag an error.
14707         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14708         in early inline pass.
14710 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14712         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14713         in comments.
14715 2013-06-18  Julian Brown  <julian@codesourcery.com>
14717         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14718         Permit virtual register pre-reload if !strict.
14719         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14720         change.
14721         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14722         prototype.
14723         * config/arm/neon.md (movmisalign<mode>): Use
14724         neon_perm_struct_or_reg_operand instead of
14725         neon_struct_or_register_operand.
14726         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14727         neon_permissive_struct_operand instead of neon_struct_operand.
14728         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14729         neon_vector_mem_operand.
14730         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14731         neon_vector_mem_operand.
14732         (neon_permissive_struct_operand): New.
14733         (neon_struct_or_register_operand): Rename to...
14734         (neon_perm_struct_or_reg_operand): This. Adjust call to
14735         neon_vector_mem_operand.
14737 2013-06-18  Richard Biener  <rguenther@suse.de>
14739         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14740         * lto-streamer.h: Include pointer-set.h.
14741         (struct lto_decl_slot): Remove.
14742         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14743         Remove next_index entry.
14744         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14745         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14746         (lto_init_tree_ref_encoder): Adjust.
14747         (lto_destroy_tree_ref_encoder): Likewise.
14748         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14749         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14750         (lto_output_decl_index): Adjust.
14751         (lto_new_out_decl_state): Likewise.
14752         (lto_record_function_out_decl_state): Likewise.
14753         * lto-streamer-out.c (copy_function): Likewise.
14755 2013-06-18  Richard Biener  <rguenther@suse.de>
14757         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14758         * cgraphunit.c: Include cfgloop.h.
14759         (init_lowered_empty_function): Initialize the loop tree.
14760         (assemble_thunk): Insert new BBs into loops.
14762 2013-06-18  Richard Biener  <rguenther@suse.de>
14764         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14765         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14766         the map from cache entry to cache index optional.
14767         (streamer_tree_cache_replace_tree): Adjust accordingly.
14768         (streamer_tree_cache_append): Likewise.
14769         (streamer_tree_cache_delete): Likewise.
14770         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14771         streamer cache map from cache entry to cache index.
14772         * lto-streamer-out.c (create_output_block): Adjust.
14774 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14776         * config/arm/arm.md (attribute "insn"): Move multiplication and
14777         division attributes to...
14778         (attribute "type"): ... here.  Remove mult.
14779         (attribute "mul32"): New attribute.
14780         (attribute "mul64"): Add umaal.
14781         (*arm_mulsi3): Update attributes.
14782         (*arm_mulsi3_v6): Likewise.
14783         (*thumb_mulsi3): Likewise.
14784         (*thumb_mulsi3_v6): Likewise.
14785         (*mulsi3_compare0): Likewise.
14786         (*mulsi3_compare0_v6): Likewise.
14787         (*mulsi_compare0_scratch): Likewise.
14788         (*mulsi_compare0_scratch_v6): Likewise.
14789         (*mulsi3addsi): Likewise.
14790         (*mulsi3addsi_v6): Likewise.
14791         (*mulsi3addsi_compare0): Likewise.
14792         (*mulsi3addsi_compare0_v6): Likewise.
14793         (*mulsi3addsi_compare0_scratch): Likewise.
14794         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14795         (*mulsi3subsi): Likewise.
14796         (*mulsidi3adddi): Likewise.
14797         (*mulsi3addsi_v6): Likewise.
14798         (*mulsidi3adddi_v6): Likewise.
14799         (*mulsidi3_nov6): Likewise.
14800         (*mulsidi3_v6): Likewise.
14801         (*umulsidi3_nov6): Likewise.
14802         (*umulsidi3_v6): Likewise.
14803         (*umulsidi3adddi): Likewise.
14804         (*umulsidi3adddi_v6): Likewise.
14805         (*smulsi3_highpart_nov6): Likewise.
14806         (*smulsi3_highpart_v6): Likewise.
14807         (*umulsi3_highpart_nov6): Likewise.
14808         (*umulsi3_highpart_v6): Likewise.
14809         (mulhisi3): Likewise.
14810         (*mulhisi3tb): Likewise.
14811         (*mulhisi3bt): Likewise.
14812         (*mulhisi3tt): Likewise.
14813         (maddhisi4): Likewise.
14814         (*maddhisi4tb): Likewise.
14815         (*maddhisi4tt): Likewise.
14816         (maddhidi4): Likewise.
14817         (*maddhidi4tb): Likewise.
14818         (*maddhidi4tt): Likewise.
14819         (divsi3): Likewise.
14820         (udivsi3): Likewise.
14821         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14822         (thumb2_mulsi_short_compare0): Likewise.
14823         (thumb2_mulsi_short_compare0_scratch): Likewise.
14824         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14825         (1020mult2): Likewise.
14826         (1020mult3): Likewise.
14827         (1020mult4): Likewise.
14828         (1020mult5): Likewise.
14829         (1020mult6): Likewise.
14830         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14831         change.
14832         (cortex_a15_mult64): Likewise.
14833         (cortex_a15_sdiv): Likewise.
14834         (cortex_a15_udiv): Likewise.
14835         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14836         (mult2): Likewise.
14837         (mult3): Likewise.
14838         (mult4): Likewise.
14839         (mult5): Likewise.
14840         (mult6): Likewise.
14841         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14842         (pj4_ir_div): Likewise.
14843         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14844         (11_mult2): Likewise.
14845         (11_mult3): Likewise.
14846         (11_mult4): Likewise.
14847         (11_mult5): Likewise.
14848         (11_mult6): Likewise.
14849         (11_mult7): Likewise.
14850         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14851         (cortex_a8_mla): Likewise.
14852         (cortex_a8_mull): Likewise.
14853         (cortex_a8_smulwy): Likewise.
14854         (cortex_a8_smlald): Likewise.
14855         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14856         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14857         (cortex_r4_mul_3): Likewise.
14858         (cortex_r4_mla_4): Likewise.
14859         (cortex_r4_mla_3): Likewise.
14860         (cortex_r4_smlald): Likewise.
14861         (cortex_r4_mull): Likewise.
14862         (cortex_r4_sdiv): Likewise.
14863         (cortex_r4_udiv): Likewise.
14864         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14865         (cortex_a7_idiv): Likewise.
14866         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14867         (9_mult2): Likewise.
14868         (9_mult3): Likewise.
14869         (9_mult4): Likewise.
14870         (9_mult5): Likewise.
14871         (9_mult6): Likewise.
14872         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14873         (cortex_a53_sdiv): Likewise.
14874         (cortex_a53_udiv): Likewise.
14875         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14876         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14877         (mp626_mult2): Likewise.
14878         (mp626_mult3): Likewise.
14879         (mp626_mult4): Likewise.
14880         * config/arm/fa526.md (526_mult1): Update attribute change.
14881         (526_mult2): Likewise.
14882         * config/arm/arm-generic.md (mult): Update attribute change.
14883         (mult_ldsched_strongarm): Likewise.
14884         (mult_ldsched): Likewise.
14885         (multi_cycle): Likewise.
14886         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14887         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14888         (606te_mult2): Likewise.
14889         (606te_mult3): Likewise.
14890         (606te_mult4): Likewise.
14891         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14892         (cortex_a9_mac16): Likewise.
14893         (cortex_a9_multiply): Likewise.
14894         (cortex_a9_mac): Likewise.
14895         (cortex_a9_multiply_long): Likewise.
14896         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14897         (626te_mult2): Likewise.
14898         (626te_mult3): Likewise.
14899         (626te_mult4): Likewise.
14901 2013-06-18  Richard Biener  <rguenther@suse.de>
14903         PR lto/57334
14904         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14906 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14908         PR target/57609
14909         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14910         with next_active_insn.
14912 2013-06-18  Alan Modra  <amodra@gmail.com>
14914         * config/rs6000/rs6000.h (enum data_align): New.
14915         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14916         (DATA_ABI_ALIGNMENT): Define.
14917         (CONSTANT_ALIGNMENT): Correct comment.
14918         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14919         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14921 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14923         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14924         ATTRIBUTE_UNUSED marking.
14926 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14928         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14929         alternative and update.
14930         (aarch64_dup_lanedi): Delete.
14931         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14932         * config/aarch64/aarch64-simd-builtins.def: Update.
14934 2013-06-17  Richard Biener  <rguenther@suse.de>
14936         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14937         (lto_input_scc): Declare.
14938         (lto_input_tree_1): Likewise.
14939         (struct lto_stats_d): Add num_tree_bodies_output and
14940         num_pickle_refs_output.
14941         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14942         (lto_read_tree_1): Split out from ...
14943         (lto_read_tree): ... this.
14944         (lto_input_scc): New function.
14945         (lto_input_tree_1): Split out from ...
14946         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14947         (lto_data_in_create): Create the streamer cache without hashes.
14948         * lto-streamer-out.c (create_output_block): Create the streamer
14949         cache with hashes when not doing WPA.
14950         (lto_write_tree_1): Split out from ...
14951         (lto_write_tree): ... this.
14952         (get_symbol_initial_value): New function.
14953         (lto_output_tree_1): Split out from ...
14954         (lto_output_tree): ... this.  Write trees as series of SCCs
14955         using a DFS walk via DFS_write_tree.
14956         (struct sccs, struct scc_entry): New types.
14957         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14958         (DFS_write_tree_body): New function.
14959         (DFS_write_tree): Likewise.
14960         (hash_tree): Likewise.
14961         (scc_entry_compare): Likewise.
14962         (hash_scc): Likewise.
14963         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14964         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14965         TREE_CHAIN as regular reference.
14966         (streamer_read_integer_cst): Remove.
14967         (streamer_get_pickled_tree): Adjust.
14968         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14969         of DECL_EXTERNALs in BLOCK_VARS for now.
14970         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14971         reference.
14972         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14973         Add hash value argument and record that if hashes are recorded
14974         in the cache.
14975         (streamer_tree_cache_insert_1): Adjust.
14976         (streamer_tree_cache_insert): Likewise.
14977         (streamer_tree_cache_insert_at): Rename to ...
14978         (streamer_tree_cache_replace_tree): ... this and adjust.
14979         (streamer_tree_cache_append): Adjust.
14980         (record_common_node): Likewise.
14981         (streamer_tree_cache_create): Add argument whether to
14982         record hash values together with trees.
14983         (streamer_tree_cache_delete): Adjust.
14984         * tree-streamer.h (struct streamer_tree_cache_d): Add
14985         vector of hashes.
14986         (streamer_read_integer_cst): Remove.
14987         (streamer_tree_cache_insert): Adjust.
14988         (streamer_tree_cache_append): Likewise.
14989         (streamer_tree_cache_insert_at): Rename to ...
14990         (streamer_tree_cache_replace_tree): ... this and adjust.
14991         (streamer_tree_cache_create): Add argument whether to record hashes.
14992         (streamer_tree_cache_get): Rename to ...
14993         (streamer_tree_cache_get_tree): ... this.
14994         (streamer_tree_cache_get_hash): New function.
14995         * tree.c (cache_integer_cst): New function.
14996         * tree.h (cache_integer_cst): Declare.
14997         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14998         * lto-symtab.c (lto_varpool_replace_node): Only release
14999         DECL_INITIAL of non-prevailing decls.
15000         * varpool.c (varpool_remove_initializer): Do not release
15001         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
15003 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
15005         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
15006         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
15007         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
15008         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
15009         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
15010         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
15011         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
15012         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
15013         instead of TARGET_64BIT.
15014         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
15015         Require ISA_HAS_<D>DIV.
15017 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
15019         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
15020         (mips*-*-linux*): Move default with_llsc setting to where other
15021         defaults are set.
15022         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
15023         with_arch block.
15024         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
15025         Likewise.  Remove default with_tune setting.  Move default float
15026         setting to its own block.  Handle with_llsc in the same block as above.
15028 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
15030         PR rtl-optimization/57425
15031         PR rtl-optimization/57569
15032         * alias.c (write_dependence_p): Add new parameters mem_mode,
15033         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
15034         Changed all callers.
15035         (canon_anti_dependence): New function.
15036         * cse.c (check_dependence): Use canon_anti_dependence.
15037         * cselib.c (cselib_invalidate_mem): Likewise.
15038         * rtl.h (canon_anti_dependence): Declare.
15040 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
15042         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
15043         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
15044         ".set mips3" for 64-bit targets.
15046 2013-06-15  Dehao Chen  <dehao@google.com>
15048         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
15049         * gimple-low.c (gimple_check_call_matching_types): Likewise.
15050         (gimple_check_call_args): Likewise.
15051         * value-prof.c (check_ic_target): Likewise.
15052         * ipa-inline.c (early_inliner): Likewise.
15053         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
15054         * cgraph.c (cgraph_create_edge_1): Likewise.
15055         (cgraph_make_edge_direct): Likewise.
15057 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
15059         PR target/57615
15060         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
15061         rs6000_output_move_128bit to handle emitting quad memory
15062         operations.  Set attribute length to 8 bytes.
15064 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
15066         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
15067         New pattern.
15068         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
15069         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
15070         (aarch64_<su>mlsl<mode>): Likewise.
15072 2013-06-14  Mike Stump  <mikestump@comcast.net>
15074         * Makefile.in (TARGET_H): Add insn-codes.h.
15076 2013-06-14  Alan Modra  <amodra@gmail.com>
15078         PR middle-end/57134
15079         PR middle-end/57586
15080         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
15081         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
15082         bitfield expansion when EXPAND_MEMORY.
15083         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
15085 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
15087         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
15088         test for clearing quad memory on 32-bit later.
15090 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
15092         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
15093         (fold_negate_expr): Likewise.
15094         (fold_real_zero_addition_p): Handle vectors.
15095         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
15097 2013-06-14  Alan Modra  <amodra@gmail.com>
15099         * varasm.c (force_const_mem): Revert 2013-06-07 change.
15101 2013-06-13  Jan Hubicka  <jh@suse.cz>
15103         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15104         Local comdats are not externally visible.
15105         * symtab.c (dump_symtab_base): Dump externally visible.
15106         (verify_symtab_base): Verify back links in the symtab hash.
15108 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
15110         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
15111         CONVERT_EXPR as equal nodes.
15113 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
15115         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
15117 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
15119         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
15120         Generalize to complex and vector.
15121         * tree.c (build_all_ones_cst): New function.
15122         * tree.h (build_all_ones_cst): Declare it.
15124 2013-06-13  Alan Modra  <amodra@gmail.com>
15126         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
15127         * config/rs6000/rs6000.md (signbittf2): New insn.
15128         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
15129         (abstf2_internal, cmptf_internal2): Likewise.
15130         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
15132 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
15133             Pat Haugen  <pthaugen@us.ibm.com>
15134             Peter Bergner  <bergner@vnet.ibm.com>
15136         * config/rs6000/rs6000.c (emit_load_locked): Add support for
15137         power8 byte, half-word, and quad-word atomic instructions.
15138         (emit_store_conditional): Likewise.
15139         (rs6000_expand_atomic_compare_and_swap): Likewise.
15140         (rs6000_expand_atomic_op): Likewise.
15142         * config/rs6000/sync.md (larx): Add new modes for power8.
15143         (stcx): Likewise.
15144         (AINT): New mode iterator to include TImode as well as normal
15145         integer modes on power8.
15146         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
15147         that VSX registers are not considered.  Use AINT mode iterator
15148         instead of INT1 to allow inclusion of quad word atomic operations
15149         on power8.
15150         (load_locked<mode>): Likewise.
15151         (store_conditional<mode>): Likewise.
15152         (atomic_compare_and_swap<mode>): Likewise.
15153         (atomic_exchange<mode>): Likewise.
15154         (atomic_nand<mode>): Likewise.
15155         (atomic_fetch_<fetchop_name><mode>): Likewise.
15156         (atomic_nand_fetch<mode>): Likewise.
15157         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
15158         each type.
15159         (ATOMIC): On power8, add QImode, HImode modes.
15160         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
15161         modes that promote to SImode.
15162         (load_lockedti): Convert TImode arguments to PTImode, so that we
15163         get a guaranteed even/odd register pair.
15164         (load_lockedpti): Likewise.
15165         (store_conditionalti): Likewise.
15166         (store_conditionalpti): Likewise.
15168         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
15169         atomic load/store instructions.
15170         (HSI): Likewise.
15172 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
15174         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
15175         loads.
15176         (insn_count): New attribute, with most cases extracted from...
15177         (length): ...here.  Redefine most cases in terms of insn_count.
15178         (single_insn): Delete.
15179         (can_delay): Use insn_count to check for single instructions.
15180         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
15181         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
15182         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
15183         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
15184         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
15185         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
15186         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
15187         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
15188         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
15189         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
15190         rather than "length".
15191         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
15192         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
15193         Use "insn_count" rather than "length".
15194         * config/mips/mips-dsp.md
15195         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
15196         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
15197         length attributes.
15199 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
15201         PR tree-optimization/57361
15202         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
15204 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
15206         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
15207         to split.
15208         (aarch64_simd_combine<mode>): New instruction expansion.
15209         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
15210         function prototype.
15211         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
15212         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
15214 2013-06-12  Jan Hubicka  <jh@suse.cz>
15216         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
15217         decl has when in streaming stage.
15218         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
15219         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
15221 2013-06-12  Roland Stigge  <stigge@antcom.de>
15223         PR target/57578
15224         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
15226 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
15228         PR tree-optimization/57537
15229         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
15230         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
15232 2013-06-12  Richard Biener  <rguenther@suse.de>
15234         * data-streamer.h (streamer_write_char_stream): CSE
15235         obs->current_pointer.
15236         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
15237         streamer_write_char_stream manually and optimize the resulting loop.
15238         (streamer_write_hwi_stream): Likewise.
15240 2013-06-12  Jan Hubicka  <jh@suse.cz>
15242         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
15243         * cgraph.h (varpool_create_empty_node): Declare.
15244         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
15245         duplicated nodes.
15246         * symtab.c (symtab_unregister_node): Be lax about missin entries
15247         in node hash.
15248         (symtab_get_node): Update comment.
15249         * varpool.c (varpool_create_empty_node): Break out from ...
15250         (varpool_node_for_decl): ... here.
15251         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
15253 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
15255         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
15256         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
15257         part.  Use straight-line flow at the end.
15258         <COMPONENT_REF>: Remove superfluous else.
15259         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
15261 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
15263         PR target/56564
15264         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
15265         target hook even for !TREE_PUBLIC decls.  If no resolution info
15266         is available, return false for common and external decls.
15268 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15270         * config/rl78/constraints.md (U): New constraint.
15271         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
15272         valloc attribute.
15274 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
15276         PR target/57589
15277         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
15278         to allow returning address to AT_PLATFORM name.
15280 2013-06-11  Jan Hubicka  <jh@suse.cz>
15282         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
15283         * cgraph.h (symtab_node_base): Add weakref flag.
15284         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
15285         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
15286         (output_weakrefs): Use weakref flag.
15287         * fold-const.c (simple_operand_p): Handle WEAK.
15288         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
15289         * ipa.c (varpool_externally_visible_p): Drop weakref.
15290         (function_and_variable_visibility): Update comment; fix weakref
15291         sanity checks; do not clear DECL_WEAK on them.
15292         * lto-cgraph.c (lto_output_node): update.
15293         (lto_output_varpool_node): Update.
15294         (input_overwrite_node): Update.
15295         (input_node): Update.
15296         (input_varpool_node): Update.
15297         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
15298         (lto_symtab_merge_symbols): Add sanity check.
15299         (lto_symtab_prevailing_decl): Do not special case weakrefs.
15300         * passes.c (rest_of_decl_compilation): Set static flag, too.
15301         * symtab.c (dump_symtab_base): Dump weakref.
15302         (verify_symtab_base): Sanity check weakrefs.
15303         (symtab_make_decl_local): Remove duplicated code.
15304         (symtab_alias_ultimate_target): Simplify.
15305         * varpool.c (varpool_create_variable_alias): Set weakref flag.
15307 2013-06-11  Tom de Vries  <tom@codesourcery.com>
15309         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
15310         sequence_vect == NULL.
15312 2013-06-11  DJ Delorie  <dj@redhat.com>
15314         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
15315         (rl78_unwind_word_mode): New.
15317 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
15319         * final.c (debug_prefix_maps): Make static.
15321 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
15323         * function.c (initial_trampoline): Remove stray copy.
15325 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
15327         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
15329 2013-06-11  Martin Jambor  <mjambor@suse.cz>
15331         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
15332         within bounds at the beginning of the function.
15334 2013-06-11  Alan Modra  <amodra@gmail.com>
15336         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
15337         reporting.
15338         (get_named_section): Don't NULL !DECL_P decl.
15340 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
15342         * doc/invoke.texi (core-avx2): Document.
15343         (slm): Likewise.
15344         (atom): Updated with MOVBE.
15346 2013-06-11  Richard Biener  <rguenther@suse.de>
15348         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
15350 2013-06-11  Anton Blanchard  <anton@samba.org>
15352         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
15353         correct shift value in little-endian mode.
15355 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
15357         PR target/56564
15358         * varasm.c (get_variable_align): Move #endif to the right place.
15360 2013-06-10  Cary Coutant  <ccoutant@google.com>
15362         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
15363         for hash so that hash table traversal order is deterministic.
15365 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
15366             Pat Haugen  <pthaugen@us.ibm.com>
15367             Peter Bergner  <bergner@vnet.ibm.com>
15369         * config/rs6000/vector.md (GPR move splitter): Do not split moves
15370         of vectors in GPRS if they are direct moves or quad word load or
15371         store moves.
15373         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
15374         declaration.
15375         (direct_move_p): Likewise.
15376         (quad_load_store_p): Likewise.
15378         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
15379         classes into bins based on the physical register type.
15380         (reg_class_to_reg_type): Likewise.
15381         (IS_STD_REG_TYPE): Likewise.
15382         (IS_FP_VECT_REG_TYPE): Likewise.
15383         (reload_fpr_gpr): Arrays to determine what insn to use if we can
15384         use direct move instructions.
15385         (reload_gpr_vsx): Likewise.
15386         (reload_vsx_gpr): Likewise.
15387         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
15388         information that is a simplification of register classes.  Also
15389         precalculate direct move reload helpers.
15390         (direct_move_p): New function to return true if the operation can
15391         be done as a direct move instruciton.
15392         (quad_load_store_p): New function to return true if the operation
15393         is a quad memory operation.
15394         (rs6000_legitimize_address): If quad memory, only allow register
15395         indirect for TImode addresses.
15396         (rs6000_legitimate_address_p): Likewise.
15397         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
15398         (rs6000_reload_register_type): Likewise.
15399         (register_to_reg_type): Return register type.
15400         (rs6000_secondary_reload_simple_move): New helper function for
15401         secondary reload and secondary memory needed to identify anything
15402         that is a simple move, and does not need reloading.
15403         (rs6000_secondary_reload_direct_move): New helper function for
15404         secondary reload to identify cases that can be done with several
15405         instructions via the direct move instructions.
15406         (rs6000_secondary_reload_move): New helper function for secondary
15407         reload to identify moves between register types that can be done.
15408         (rs6000_secondary_reload): Add support for quad memory operations
15409         and for direct move.
15410         (rs6000_secondary_memory_needed): Likewise.
15411         (rs6000_debug_secondary_memory_needed): Change argument names.
15412         (rs6000_output_move_128bit): New function to return the move to
15413         use for 128-bit moves, including knowing about the various
15414         limitations of quad memory operations.
15416         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
15417         memory operations.  call rs6000_output_move_128bit for the actual
15418         instruciton(s) to generate.
15419         (vsx_movti_64bit): Likewise.
15421         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
15422         (UNSPEC_P8V_MTVSRWZ): Likewise.
15423         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
15424         (UNSPEC_P8V_MTVSRD): Likewise.
15425         (UNSPEC_P8V_XXPERMDI): Likewise.
15426         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
15427         (UNSPEC_FUSION_GPR): Likewise.
15428         (FMOVE128_GPR): New iterator for direct move.
15429         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
15430         (f32_sv): Likewise.
15431         (f32_dm): Likewise.
15432         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
15433         loads and direct move instructions.
15434         (zero_extendsidi2_lfiwzx): Likewise.
15435         (extendsidi2_lfiwax): Likewise.
15436         (extendsidi2_nocell): Likewise.
15437         (floatsi<mode>2_lfiwax): Likewise.
15438         (lfiwax): Likewise.
15439         (floatunssi<mode>2_lfiwzx): Likewise.
15440         (lfiwzx): Likewise.
15441         (fix_trunc<mode>_stfiwx): Likewise.
15442         (fixuns_trunc<mode>_stfiwx): Likewise.
15443         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
15444         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
15445         (parity<mode>2_cmpb): Set length/type attr.
15446         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
15447         for 'mr.' to fast_compare.
15448         (bpermd_<mode>): Change type attr to popcnt.
15449         (p8_fmrgow_<mode>): New insns for power8 direct move support.
15450         (p8_mtvsrwz_1): Likewise.
15451         (p8_mtvsrwz_2): Likewise.
15452         (reload_fpr_from_gpr<mode>): Likewise.
15453         (p8_mtvsrd_1): Likewise.
15454         (p8_mtvsrd_2): Likewise.
15455         (p8_xxpermdi_<mode>): Likewise.
15456         (reload_vsx_from_gpr<mode>): Likewise.
15457         (reload_vsx_from_gprsf): Likewise.
15458         (p8_mfvsrd_3_<mode>): LIkewise.
15459         (reload_gpr_from_vsx<mode>): Likewise.
15460         (reload_gpr_from_vsxsf): Likewise.
15461         (p8_mfvsrd_4_disf): Likewise.
15462         (multi-word GPR splits): Do not split direct moves or quad memory
15463         operations.
15465 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
15467         * tree-into-ssa.c (interesting_blocks): Make static.
15469 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
15471         PR target/56564
15472         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
15473         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
15474         Use DATA_ABI_ALIGNMENT for that case instead if defined.
15475         (get_variable_align): New function.
15476         (get_variable_section, emit_bss, emit_common,
15477         assemble_variable_contents, place_block_symbol): Use
15478         get_variable_align instead of DECL_ALIGN.
15479         (assemble_noswitch_variable): Add align argument, use it
15480         instead of DECL_ALIGN.
15481         (assemble_variable): Adjust caller.  Use get_variable_align
15482         instead of DECL_ALIGN.
15483         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
15484         caller.
15485         (DATA_ABI_ALIGNMENT): Define.
15486         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
15487         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
15488         opt is false, only return the psABI mandated alignment increase.
15489         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
15490         (DATA_ABI_ALIGNMENT): ... this.
15491         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
15492         (DATA_ABI_ALIGNMENT): ... this.
15493         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
15494         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
15495         (DATA_ABI_ALIGNMENT): ... this.
15496         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
15497         * doc/tm.texi: Regenerated.
15499 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
15501         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
15502         cmp_code to construct REG_EQUAL note.
15504 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
15506         PR target/57568
15507         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
15508         that operands[2] doesn't overlap with operands[0].
15510 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
15511             Jan Hubicka  <jh@suse.cz>
15513         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
15514         hack to mark symbols as used.
15516 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
15518         PR rtl-optimization/57559
15519         * lra-constraints.c (process_alt_operands): Don't discourage
15520         memory with known offset for offsetable memory constraint.
15521         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
15523 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
15525         * varasm.c (struct oc_local_state): Reorder fields.
15526         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
15527         and adjust accordingly.
15528         (output_constructor): Reorder initialization code and adjust call to
15529         output_constructor_bitfield.
15531 2013-06-07  Jan Hubicka  <jh@suse.cz>
15533         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
15535 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
15537         * tree-object-size.c (unknown): Make const.
15539 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15541         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
15542         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
15543         for last alternative in the cpu_facility attribute.
15545 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15547         PR target/56315
15548         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
15549         (xordi3): Change operand 2 constraint to arm_xordi_operand.
15550         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
15551         * config/arm/constraints.md (Dg): New constraint.
15552         * config/arm/neon.md (xordi3_neon): Remove.
15553         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
15554         * config/arm/predicates.md (arm_xordi_operand): New predicate.
15556 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15558         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
15559         Clean up alternatives.
15561 2013-06-07  Alan Modra  <amodra@gmail.com>
15563         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
15564         va_list_gpr_size.
15566 2013-06-07  Alan Modra  <amodra@gmail.com>
15568         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
15570 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15572         * config/arm/constraints.md (Df): New constraint.
15573         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
15574         Correct length attribute for last two alternatives.
15576 2013-06-07  Alan Modra  <amodra@gmail.com>
15578         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
15579         override user -mfp-in-toc.
15580         (offsettable_ok_by_alignment): Consider just the current access
15581         rather than the whole object, unless BLKmode.  Handle
15582         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
15583         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
15584         for -mcmodel=medium.
15585         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
15586         override user -mfp-in-toc or -msum-in-toc.  Default to
15587         -mno-fp-in-toc for -mcmodel=medium.
15589 2013-06-06  DJ Delorie  <dj@redhat.com>
15591         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
15592         TARGET_VALID_POINTER_MODE.
15594 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
15595             Pat Haugen  <pthaugen@us.ibm.com>
15596             Peter Bergner  <bergner@vnet.ibm.com>
15598         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
15599         Document new power8 builtins.
15601         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
15602         condition code register, to allow 128-bit logical operations to be
15603         done in the VSX or GPR registers.
15604         (nor<mode>3): Use the canonical form for nor.
15605         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
15606         vclz*, and vpopcnt* vector instructions.
15607         (nand<mode>3): Likewise.
15608         (orc<mode>3): Likewise.
15609         (clz<mode>2): LIkewise.
15610         (popcount<mode>2): Likewise.
15612         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
15613         that only the GPRs are recognized.
15615         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15616         support for new power8 builtins.
15618         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
15619         builtin functions.
15620         (xscvdpspn): Likewise.
15621         (vclz): Likewise.
15622         (vclzb): Likewise.
15623         (vclzh): Likewise.
15624         (vclzw): Likewise.
15625         (vclzd): Likewise.
15626         (vpopcnt): Likewise.
15627         (vpopcntb): Likewise.
15628         (vpopcnth): Likewise.
15629         (vpopcntw): Likewise.
15630         (vpopcntd): Likewise.
15631         (vgbbd): Likewise.
15632         (vmrgew): Likewise.
15633         (vmrgow): Likewise.
15634         (eqv): Likewise.
15635         (eqv_v16qi3): Likewise.
15636         (eqv_v8hi3): Likewise.
15637         (eqv_v4si3): Likewise.
15638         (eqv_v2di3): Likewise.
15639         (eqv_v4sf3): Likewise.
15640         (eqv_v2df3): Likewise.
15641         (nand): Likewise.
15642         (nand_v16qi3): Likewise.
15643         (nand_v8hi3): Likewise.
15644         (nand_v4si3): Likewise.
15645         (nand_v2di3): Likewise.
15646         (nand_v4sf3): Likewise.
15647         (nand_v2df3): Likewise.
15648         (orc): Likewise.
15649         (orc_v16qi3): Likewise.
15650         (orc_v8hi3): Likewise.
15651         (orc_v4si3): Likewise.
15652         (orc_v2di3): Likewise.
15653         (orc_v4sf3): Likewise.
15654         (orc_v2df3): Likewise.
15656         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15657         allow power8 quad mode in 64-bit.
15658         (rs6000_builtin_vectorized_function): Add support to vectorize
15659         ISA 2.07 count leading zeros, population count builtins.
15660         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15661         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15662         (builtin_function_type): Add vgbbd builtin function which takes an
15663         unsigned argument.
15664         (altivec_expand_vec_perm_const): Add support for new power8 merge
15665         instructions.
15667         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15668         that does not include TImdoe for use with 32-bit.
15669         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15670         instructions.
15671         (UNSPEC_VSX_CVDPSPN): Likewise.
15672         (vsx_xscvdpspn): Likewise.
15673         (vsx_xscvspdpn): Likewise.
15674         (vsx_xscvdpspn_scalar): Likewise.
15675         (vsx_xscvspdpn_directmove): Likewise.
15676         (vsx_and<mode>3): Split logical operations into 32-bit and
15677         64-bit. Add support to do logical operations on TImode as well as
15678         VSX vector types.  Allow logical operations to be done in either
15679         VSX registers or in general purpose registers in 64-bit mode.  Add
15680         splitters if GPRs were used. For AND, add clobber of CCmode to
15681         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15682         encoding.
15683         (vsx_and<mode>3_32bit): Likewise.
15684         (vsx_and<mode>3_64bit): Likewise.
15685         (vsx_ior<mode>3): Likewise.
15686         (vsx_ior<mode>3_32bit): Likewise.
15687         (vsx_ior<mode>3_64bit): Likewise.
15688         (vsx_xor<mode>3): Likewise.
15689         (vsx_xor<mode>3_32bit): Likewise.
15690         (vsx_xor<mode>3_64bit): Likewise.
15691         (vsx_one_cmpl<mode>2): Likewise.
15692         (vsx_one_cmpl<mode>2_32bit): Likewise.
15693         (vsx_one_cmpl<mode>2_64bit): Likewise.
15694         (vsx_nor<mode>3): Likewise.
15695         (vsx_nor<mode>3_32bit): Likewise.
15696         (vsx_nor<mode>3_64bit): Likewise.
15697         (vsx_andc<mode>3): Likewise.
15698         (vsx_andc<mode>3_32bit): Likewise.
15699         (vsx_andc<mode>3_64bit): Likewise.
15700         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15701         and xxlorc instructions.
15702         (vsx_eqv<mode>3_64bit): Likewise.
15703         (vsx_nand<mode>3_32bit): Likewise.
15704         (vsx_nand<mode>3_64bit): Likewise.
15705         (vsx_orc<mode>3_32bit): Likewise.
15706         (vsx_orc<mode>3_64bit): Likewise.
15708         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15710         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15711         instruction.
15712         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15713         (p8_vmrgow): Likewise.
15714         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15715         GPRs to be split under VSX.
15716         (p8v_clz<mode>2): Add power8 count leading zero support.
15717         (p8v_popcount<mode>2): Add power8 population count support.
15718         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15719         support.
15721         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15722         instruction.
15724         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15725         builtin functions.
15726         (vec_nand): Likewise.
15727         (vec_vclz): Likewise.
15728         (vec_vclzb): Likewise.
15729         (vec_vclzd): Likewise.
15730         (vec_vclzh): Likewise.
15731         (vec_vclzw): Likewise.
15732         (vec_vgbbd): Likewise.
15733         (vec_vmrgew): Likewise.
15734         (vec_vmrgow): Likewise.
15735         (vec_vpopcnt): Likewise.
15736         (vec_vpopcntb): Likewise.
15737         (vec_vpopcntd): Likewise.
15738         (vec_vpopcnth): Likewise.
15739         (vec_vpopcntw): Likewise.
15741 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15743         PR rtl-optimization/57468
15744         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15745         spilled pseudos.
15747 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15749         PR rtl-optimization/57459
15750         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15751         type when setting live regs.
15753 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15755         * config/s390/s390.opt (mlra): New option.
15756         * config/s390/s390.c (s390_decompose_address): Check displacement
15757         for all registers for LRA.
15758         (s390_secondary_reload): Don't used secondary reloads for LRA.
15759         (s390_lra_p): New function.
15760         (TARGET_LRA_P): Define.
15761         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15762         of attribute cpu_facility to zarch for the last alternative.
15763         (*cmpmem_short): Ditto.
15765 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15767         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15768         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15769         (arm_expand_prologue): Likewise.
15771 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15773         PR c++/53743
15774         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15775         as this is now done by redirect_edge_and_branch_force.
15776         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15777         barriers, and fix interaction with splitting.
15778         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15779         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15780         reflect changes made in the routine.
15781         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15782         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15783         since this is called in cfglayout mode, and replace partition fixup
15784         with assert as that is now done by force_nonfallthru_and_redirect.
15785         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15786         already be marked with region crossing note.
15787         (insert_section_boundary_note): Make non-static, gate on flag
15788         has_bb_partition, rewrite to also check for multiple partitions.
15789         (rest_of_handle_reorder_blocks): Remove call to
15790         insert_section_boundary_note, now done later during free_cfg.
15791         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15792         * bb-reorder.h (insert_section_boundary_note): Declare.
15793         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15794         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15795         invoke insert_section_boundary_note.
15796         (try_redirect_by_replacing_jump): Remove unnecessary
15797         check for region crossing note.
15798         (fixup_partition_crossing): New function.
15799         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15800         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15801         in non-cfglayout mode.
15802         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15803         remove old code that tried to do this. Emit barrier correctly
15804         when we are in cfglayout mode.
15805         (last_bb_in_partition): New function.
15806         (rtl_split_edge): Correctly fixup partition boundaries.
15807         (commit_one_edge_insertion): Remove old code that tried to
15808         fixup region crossing edge since this is now handled in
15809         split_block, and set up insertion point correctly since
15810         block may now end in a jump.
15811         (verify_hot_cold_block_grouping): Guard against checking when not in
15812         linearized RTL mode.
15813         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15814         notes.
15815         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15816         rtl_verify_flow_info, so not called in cfglayout mode.
15817         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15818         (fixup_reorder_chain): Remove old code that attempted to fixup region
15819         crossing note as this is now handled in force_nonfallthru_and_redirect.
15820         (duplicate_insn_chain): Don't duplicate switch section notes.
15821         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15822         note.
15823         * basic-block.h (emit_barrier_after_bb): Declare.
15825 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15827         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15828         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15829         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15831 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15833         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15834         where appropriate.
15835         * config/arm/ldmstm.md: Regenerate.
15837 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15839         * config/arm/sync.md (atomic_loaddi_1):
15840         Disable predication for arm_restrict_it.
15841         (arm_load_exclusive<mode>): Likewise.
15842         (arm_load_exclusivesi): Likewise.
15843         (arm_load_exclusivedi): Likewise.
15844         (arm_load_acquire_exclusive<mode>): Likewise.
15845         (arm_load_acquire_exclusivesi): Likewise.
15846         (arm_load_acquire_exclusivedi): Likewise.
15847         (arm_store_exclusive<mode>): Likewise.
15848         (arm_store_exclusive<mode>): Likewise.
15849         (arm_store_release_exclusivedi): Likewise.
15850         (arm_store_release_exclusive<mode>): Likewise.
15852 2013-06-06  Richard Biener  <rguenther@suse.de>
15854         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15855         after LTO_null.
15856         (lto_tag_is_tree_code_p): Adjust.
15857         (lto_tag_is_gimple_code_p): Likewise.
15858         (lto_gimple_code_to_tag): Likewise.
15859         (lto_tag_to_gimple_code): Likewise.
15860         (lto_tree_code_to_tag): Likewise.
15861         (lto_tag_to_tree_code): Likewise.
15862         * data-streamer.h (streamer_write_hwi_in_range): Use
15863         uhwi streaming to stream the normalized range.
15864         (streamer_read_hwi_in_range): Likewise.
15866 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15868         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15869         (predicable_short_it): Likewise.
15870         (predicated): Likewise.
15871         (enabled): Handle above.
15872         (define_cond_exec): Set predicated attribute to yes.
15874 2013-06-05  Mike Stump  <mikestump@comcast.net>
15876         * gdbinit.in (__FUNCTION__): Add.
15878 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15880         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15881         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15883 2013-06-05  Jan Hubicka  <jh@suse.cz>
15885         * varasm.c (mark_decl_referenced): Revert the removal until targets
15886         are fixed.
15888 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15890         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15891         instead of mark_decl_referenced.
15893 2013-06-05  Jan Hubicka  <jh@suse.cz>
15895         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15896         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15897         and symtab_used_from_object_file_p.
15898         (cgraph_make_node_local_1): Clear forced_by_abi.
15899         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15900         * cgraph.h (symtab_node_base): Add forced_by_abi.
15901         (decide_is_variable_needed): Remove.
15902         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15903         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15904         (decide_is_symbol_needed): ... this one; handle symbols in general;
15905         always analyze virtuals; honnor forced_by_abi.
15906         (cgraph_finalize_function): Update.
15907         (varpool_finalize_decl): Update.
15908         (symbol_defined_and_needed): Remove.
15909         (analyze_functions): Update.
15910         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15911         output_refs, input_overwrite_node): Handle forced_by_abi.
15912         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15913         (address_taken_from_non_vtable_p): ... this one.
15914         (comdat_can_be_unshared_p_1): New function.
15915         (cgraph_comdat_can_be_unshared_p): Rename to ...
15916         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15917         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15918         (function_and_variable_visibility): Clear forced_by_abi as needed.
15919         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15920         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15921         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15922         * varpool.c (decide_is_variable_needed): Remove.
15924 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15926         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15927         (arm_option_override): Override arm_restrict_it where appropriate.
15928         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15929         * config/arm/arm.opt (mrestrict-it): New command-line option.
15930         * doc/invoke.texi: Document -mrestrict-it.
15932 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15934         * tsan.c (tsan_atomic_table): Make const.
15936 2013-06-05  Richard Biener  <rguenther@suse.de>
15938         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15939         index associated with the tree we are supposed to replace.
15940         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15941         TREE_ASM_WRITTEN as zero for everything but SSA names.
15943 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15945         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15947 2013-06-04  Jan Hubicka  <jh@suse.cz>
15949         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15950         (input_node, input_varpool_node): Handle correctly external same
15951         body aliases.
15952         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15953         nodes at ltrans stage.
15955 2013-06-04  Jan Hubicka  <jh@suse.cz>
15957         * ipa-inline.c (update_caller_keys): Fix availability test.
15958         (update_callee_keys): Likewise.
15959         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15960         to follow ELF standard.
15962 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15964         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15965         (mips64r5900el-*-elf*): New configurations.
15966         * config/mips/mips-cpus.def (r5900): New processor.
15967         * config/mips/mips-tables.opt: Regenerate.
15968         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15969         (mips_issue_rate): Handle PROCESSOR_R5900.
15970         (mips_reorg_process_insns): Force reorder mode for the R5900.
15971         * config/mips/mips.h (TARGET_MIPS5900): Define.
15972         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15973         TARGET_MIPS5900.
15974         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15975         TARGET_MIPS5900.
15976         * config/mips/mips.md (processor): Add r5900.
15977         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15979 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15981         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15982         into function to generate MOVI instruction.
15983         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15984         (aarch64_preferred_simd_mode): Turn into wrapper.
15985         (aarch64_output_scalar_simd_mov_immediate): New function.
15986         * config/aarch64/aarch64-protos.h: Add prototype for above.
15988 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15990         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15991         element_char member.
15992         (sizetochar): Return signed char.
15993         (aarch64_simd_valid_immediate): Remove elchar and other
15994         unnecessary variables.
15995         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15996         Calculate element_char as required.
15997         * config/aarch64/aarch64-protos.h: Update and move prototype
15998         for aarch64_output_simd_mov_immediate.
15999         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16000         Update arguments.
16002 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
16004         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
16005         information completed by aarch64_simd_valid_immediate.
16006         (aarch64_legitimate_constant_p): Update arguments.
16007         (aarch64_simd_valid_immediate): Work with struct rather than many
16008         pointers.
16009         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
16010         (aarch64_simd_make_constant): Update arguments.
16011         (aarch64_output_simd_mov_immediate): Work with struct rather than
16012         many pointers.  Output immediate directly rather than as operand.
16013         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
16014         Update prototype.
16015         * config/aarch64/constraints.md (Dn): Update arguments.
16017 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
16019         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
16020         longer static.
16021         (aarch64_simd_immediate_valid_for_move): Remove.
16022         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
16023         (aarch64_simd_make_constant): Update call.
16024         (aarch64_output_simd_mov_immediate): Update call.
16025         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
16026         Add prototype.
16027         * config/aarch64/constraints.md (Dn): Update call.
16029 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
16031         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
16032         return type to bool for prototype.
16033         (aarch64_legitimate_constant_p): Check for true instead of not -1.
16034         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
16035         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
16037 2013-06-04  Catherine Moore  <clm@codesourcery.com>
16039         * config/mips/mips.opt (meva): New.
16040         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
16041         (ASM_SPEC): Handle -meva.
16042         * doc/invoke.texi (meva):  Document.
16044 2013-06-04  Alan Modra  <amodra@gmail.com>
16046         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
16047         constant output.
16049 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16051         * rtl.def: Add extra fourth optional field to define_cond_exec.
16052         * gensupport.c (process_one_cond_exec): Process attributes from
16053         define_cond_exec.
16054         * doc/md.texi: Document fourth field in define_cond_exec.
16056 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
16058         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
16059         out the processing order as in store_bit_field_1.
16061 2013-06-04  Jan Hubicka  <jh@suse.cz>
16063         PR middle-end/57500
16064         * cgraphunit.c (cgraph_process_same_body_aliases): Create
16065         non-VAR_DECL node if it does not exist yet.
16067 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
16069         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
16070         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
16071         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
16072         target_cpu_default setting.
16074 2013-06-03  Teresa Johnson  <tejohnson@google.com>
16076         * dumpfile.c (opt_info_switch_p): Change -fopt-info
16077         default to -fopt-info=optimized instead of all.
16078         * doc/invoke.texi: Ditto.
16079         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
16080         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
16081         (execute_vect_slp): Emit BB vectorization success under
16082         MSG_OPTIMIZED_LOCATIONS.
16083         * tree-vect-slp.c (vect_slp_transform_bb): Change
16084         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
16085         * tree-vect-loop.c (vect_transform_loop): Ditto.
16087 2013-06-03  Jason Merrill  <jason@redhat.com>
16089         PR c++/57415
16090         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16091         Use TARGET_EXPR for C++.
16093 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
16095         PR rtl-optimization/57268
16096         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
16097         if DEBUG_INSN_P (insn).
16099         Reapply
16100         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
16102         PR rtl-optimization/57268
16103         * sched-deps.c (sched_analyze_2): Flush dependence lists if
16104         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
16106 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
16108         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
16109         (ix86_avoid_lea_for_addr): Likewise.
16110         (exact_dependency_1): Likewise.
16111         (ix86_adjust_cost): Likewise.
16112         (swap_top_of_ready_list): Fix formatting and !reload_completed check
16113         removed.
16114         (do_reorder_for_imul): Fix typo, formatting and
16115         !reload_completed check removed.
16116         (ix86_sched_reorder): Fix typo and formatting.
16117         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
16118         list.
16120 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
16122         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
16124 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
16126         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
16127         <STRING_CST>: Likewise.
16128         <VECTOR_CST>: Likewise.
16130 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
16131             Mikael Morin  <mikael@gcc.gnu.org>
16133         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
16134         * config.in: Regenerated.
16135         * configure: Regenerated.
16137 2013-06-01  Jan Hubicka  <jh@suse.cz>
16139         PR middle-end/57366
16140         * cgraphunit.c (compile): When weakref is not supported,
16141         set up transparent aliases before final output pass.
16142         * varasm.c (assemble_alias): Do not try to do it here.
16144 2013-06-01  Jan Hubicka  <jh@suse.cz>
16146         PR middle-end/57467
16147         * passes.c (for_per_function): Skip unanalyzed functions.
16149 2013-06-01  Jan Hubicka  <jh@suse.cz>
16151         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
16152         (lto_symtab_merge_symbols_1): ... this one.
16153         (lto_symtab_merge_cgraph_nodes): Rename to ...
16154         (lto_symtab_merge_symbols): ... this one; simplify.
16155         * cgraph.c (same_body_aliases_done): Rename to ...
16156         (cpp_implicit_aliases_done): ... this one.
16157         (cgraph_create_function_alias): Update.
16158         (cgraph_same_body_alias): Update.
16159         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
16160         (verify_edge_corresponds_to_fndecl): Simplify.
16161         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
16162         (cgraph_node): Remove same_body_alias.
16163         (varpool_node): Remove alias_of and extra_name_alias.
16164         (same_body_aliases_done): Rename to ..
16165         (cpp_implicit_aliases_done): ... this one.
16166         (symtab_alias_ultimate_target): Add default parameter.
16167         (symtab_resolve_alias): New function.
16168         (fixup_same_cpp_alias_visibility): Declare.
16169         (cgraph_function_node): Add default parameter.
16170         (cgraph_node_asm_name): Likewise.
16171         (cgraph_function_or_thunk_node): Add default parameter; do
16172         not ICE when it is NULL.
16173         (varpool_variable_node): Likewise.
16174         * tree-emutls.c (create_emultls_var): Update.
16175         (ipa_lower_emutls): Update.
16176         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
16177         (cgraph_reset_node): Reset alias info.
16178         (cgraph_finalize_function): Update.
16179         (fixup_same_cpp_alias_visibility): Move to symtab.c.
16180         (analyze_function): Simplify.
16181         (cgraph_process_same_body_aliases): Simplify.
16182         (analyze_functions): Fixup same body aliases.
16183         (handle_alias_pairs): Simplify.
16184         (assemble_thunk): Update.
16185         (assemble_thunks_and_aliases): Update.
16186         (output_weakrefs): Rewrite.
16187         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
16188         (lto_output_varpool_node): Likewise.
16189         (compute_ltrans_boundary): Remve assert.
16190         (get_alias_symbol): New functoin.
16191         (input_node): Rewrite alias handling.
16192         (input_varpool_node): Likewise.
16193         * ipa-pure-const.c (propagate_pure_const): Fix formating.
16194         * ipa.c (process_references): Handle weakrefs correctly.
16195         (symtab_remove_unreachable_nodes): Likewise.
16196         * trans-mem.c (get_cg_data): Update.
16197         (ipa_tm_create_version_alias): Update.
16198         (ipa_tm_execute): Update.
16199         * symtab.c (dump_symtab_base): Dump aliases.
16200         (verify_symtab_base): Verify aliases.
16201         (symtab_node_availability): New function.
16202         (symtab_alias_ultimate_target): Simplify.
16203         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
16204         handle all the fixup cases.
16205         (symtab_resolve_alias): New function.
16206         * passes.c (ipa_write_summaries): Handle weakrefs.
16207         * varpool.c (varpool_analyze_node): Simplify.
16208         (assemble_aliases): Update.
16209         (varpool_create_variable_alias): Simplify.
16210         (varpool_extra_name_alias): Simplify.
16211         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
16212         (lto_symtab_merge_symbols): ... this one.
16214 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
16216         Revert
16217         PR rtl-optimization/57268
16218         * sched-deps.c (sched_analyze_2): Flush dependence lists if
16219         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
16221 2013-06-01  Tobias Burnus  <burnus@net-b.de>
16223         Partially reverted:
16224         2013-05-31  Tobias Burnus  <burnus@net-b.de>
16226         PR middle-end/57073
16227         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
16228         further up.
16230 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
16232         PR rtl-optimization/57268
16233         * sched-deps.c (sched_analyze_2): Flush dependence lists if
16234         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
16236 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
16238         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
16239         unordered comparison operators when -fno-trapping-math is in effect
16240         on the e500.
16241         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
16242         and implement unordered comparison operators properly on the e500.
16244 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
16246         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
16247         for constant scalar integers.
16248         (simplify_relational_operation_1): Likewise.
16250 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
16252         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
16253         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
16254         Fix comment.
16256 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
16257             Igor Zamyatin  <igor.zamyatin@intel.com>
16259         Silvermont (SLM) architecture performance tuning.
16260         * config/i386/i386.h (enum ix86_tune_indices): Add
16261         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
16262         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
16264         * config/i386/i386.c (initial_ix86_tune_features)
16265         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
16266         (ix86_lea_outperforms): Handle Silvermont tuning.
16267         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
16268         call.
16269         (ix86_use_lea_for_mov): Likewise.
16270         (ix86_avoid_lea_for_addr): Likewise.
16271         (ix86_lea_for_add_ok): Likewise.
16272         (exact_dependency_1): New function.
16273         (exact_store_load_dependency): Likewise.
16274         (ix86_adjust_cost): Handle Silvermont tuning.
16275         (do_reoder_for_imul): Likewise.
16276         (swap_top_of_ready_list): New function.
16277         (ix86_sched_reorder): Changed to handle Silvermont tuning.
16279         * config/i386/i386.md (peepholes that split memory operand in fp
16280         converts): New.
16282 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16284         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
16285         Remove un-necessary braces.
16287 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16289         * config/aarch64/aarch64.c (aarch64_classify_symbol):
16290         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
16292 2013-05-31  Tobias Burnus  <burnus@net-b.de>
16294         PR middle-end/57073
16295         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
16297 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16299         PR target/56315
16300         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
16301         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
16302         * config/arm/neon.md (iordi3_neon): Remove.
16303         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
16304         * config/arm/predicates.md (imm_for_neon_logic_operand):
16305         Move to earlier in the file.
16306         (neon_logic_op2): Likewise.
16307         (arm_iordi_operand_neon): New predicate.
16309 2013-05-31  Richard Biener  <rguenther@suse.de>
16311         PR tree-optimization/57478
16312         PR tree-optimization/57453
16313         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
16314         are life as well.
16316 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
16318         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
16319         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
16321 2013-05-30  Tobias Burnus  <burnus@net-b.de>
16322             Thomas Koenig  <tkoenig@gcc.gnu.org>
16324         PR middle-end/57073
16325         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
16326         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
16328 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
16330         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
16332 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
16334         * target.def (register_usage_leveling_p): New hook.
16335         * targhooks.c (default_register_usage_leveling_p): New.
16336         * targhooks.h (default_register_usage_leveling_p): New prototype.
16337         * lra-assigns.c (register_usage_leveling_p): Use the hook.
16338         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
16339         * doc/tm.texi: Update.
16340         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
16342 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
16344         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
16345         (*insv_reg<mode>): New define_insn.
16347 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
16349         PR rtl-optimization/57439
16350         * postreload.c (move2add_valid_value_p): Check that we have
16351         a zero subreg_regno_offset when accessing the register in
16352         the requested mode.
16354 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
16355             Igor Zamyatin  <igor.zamyatin@intel.com>
16357         Silvermont (SLM) architecture pipeline model, tuning and
16358         insn selection.
16359         * config.gcc: Add slm config options and target.
16361         * config/i386/slm.md: New.
16363         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
16365         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
16366         PROCESSOR_SLM.
16367         (ix86_target_macros_internal): Likewise.
16369         * config/i386/i386.c (slm_cost): New cost.
16370         (m_SLM): New macro flag.
16371         (initial_ix86_tune_features): Set m_SLM.
16372         (x86_accumulate_outgoing_args): Likewise.
16373         (x86_arch_always_fancy_math_387): Likewise.
16374         (processor_target_table): Add slm cost.
16375         (cpu_names): Add slm cpu name.
16376         (x86_option_override_internal): Set SLM ISA.
16377         (ix86_issue_rate): New case PROCESSOR_SLM.
16378         (ia32_multipass_dfa_lookahead): Likewise.
16379         (fold_builtin_cpu): Add slm.
16381         * config/i386/i386.h (TARGET_SLM): New target macro.
16382         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
16383         (processor_type): Add PROCESSOR_SLM.
16385         * config/i386/i386.md (cpu): Add new value "slm".
16386         (slm.md): Include slm.md.
16388 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
16389             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
16391         * config/arm/arm-protos.h: Add and update function protos.
16392         * config/arm/arm.c (use_simple_return_p): New added.
16393         (thumb2_expand_return): Check simple_return flag.
16394         * config/arm/arm.md: Add simple_return and conditional simple_return.
16395         * config/arm/iterators.md: Add iterator for return and simple_return.
16397 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
16399         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
16400         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
16401         (arm_emit_vfp_multi_reg_pop): Likewise.
16402         (thumb2_emit_ldrd_pop): Likewise.
16403         (arm_expand_epilogue): Add misc REG_CFA notes.
16404         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
16406 2013-05-29  Lawrence Crowl  <crowl@google.com>
16408         * config/arm/t-arm: Update for below.
16410         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
16411         Change type to hash_table.  Update dependent calls and types.
16413         * config/i386/t-cygming: Update for below.
16415         * config/i386/t-interix: Update for below.
16417         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
16418         Change type to hash_table.  Update dependent calls and types.
16419         (i386_find_on_wrapper_list::wrappers): Likewise.
16421         * config/ia64/t-ia64: Update for below.
16423         * config/ia64/ia64.c (bundle_state_table):
16424         Change type to hash_table.  Update dependent calls and types.
16426         * config/mips/mips.c (mips_reorg_process_insns::htab):
16427         Change type to hash_table.  Update dependent calls and types.
16429         * config/sol2.c (solaris_comdat_htab):
16430         Change type to hash_table.  Update dependent calls and types.
16432         * config/t-sol2: Update for above.
16434 2013-05-29  Teresa Johnson  <tejohnson@google.com>
16436         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
16437         functions are not yet marked as defined.
16439 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
16440             Pat Haugen  <pthaugen@us.ibm.com>
16441             Peter Bergner  <bergner@vnet.ibm.com>
16443         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
16444         instructions.
16445         (VEC_A): Likewise.
16446         (VEC_C): Likewise.
16447         (vrotl<mode>3): Likewise.
16448         (vashl<mode>3): Likewise.
16449         (vlshr<mode>3): Likewise.
16450         (vashr<mode>3): Likewise.
16452         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16453         support for power8 V2DI builtins.
16455         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
16456         power8 V2DI builtins.
16457         (vupkhsw): Likewise.
16458         (vupklsw): Likewise.
16459         (vaddudm): Likewise.
16460         (vminsd): Likewise.
16461         (vmaxsd): Likewise.
16462         (vminud): Likewise.
16463         (vmaxud): Likewise.
16464         (vpkudum): Likewise.
16465         (vpksdss): Likewise.
16466         (vpkudus): Likewise.
16467         (vpksdus): Likewise.
16468         (vrld): Likewise.
16469         (vsld): Likewise.
16470         (vsrd): Likewise.
16471         (vsrad): Likewise.
16472         (vsubudm): Likewise.
16473         (vcmpequd): Likewise.
16474         (vcmpgtsd): Likewise.
16475         (vcmpgtud): Likewise.
16476         (vcmpequd_p): Likewise.
16477         (vcmpgtsd_p): Likewise.
16478         (vcmpgtud_p): Likewise.
16479         (vupkhsw): Likewise.
16480         (vupklsw): Likewise.
16481         (vaddudm): Likewise.
16482         (vmaxsd): Likewise.
16483         (vmaxud): Likewise.
16484         (vminsd): Likewise.
16485         (vminud): Likewise.
16486         (vpksdss): Likewise.
16487         (vpksdus): Likewise.
16488         (vpkudum): Likewise.
16489         (vpkudus): Likewise.
16490         (vrld): Likewise.
16491         (vsld): Likewise.
16492         (vsrad): Likewise.
16493         (vsrd): Likewise.
16494         (vsubudm): Likewise.
16496         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
16497         support for power8 V2DI instructions.
16499         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
16500         power8 V2DI instructions.  Combine pack and unpack insns to use an
16501         iterator for each mode.  Check whether a particular mode supports
16502         Altivec instructions instead of just checking TARGET_ALTIVEC.
16503         (UNSPEC_VPKUWUM): Likewise.
16504         (UNSPEC_VPKSHSS): Likewise.
16505         (UNSPEC_VPKSWSS): Likewise.
16506         (UNSPEC_VPKUHUS): Likewise.
16507         (UNSPEC_VPKSHUS): Likewise.
16508         (UNSPEC_VPKUWUS): Likewise.
16509         (UNSPEC_VPKSWUS): Likewise.
16510         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
16511         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
16512         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
16513         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
16514         (UNSPEC_VUPKHSB): Likewise.
16515         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
16516         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
16517         (UNSPEC_VUPKHSH): Likewise.
16518         (UNSPEC_VUPKLSB): Likewise.
16519         (UNSPEC_VUPKLSH): Likewise.
16520         (VI2): Likewise.
16521         (VI_char): Likewise.
16522         (VI_scalar): Likewise.
16523         (VI_unit): Likewise.
16524         (VP): Likewise.
16525         (VP_small): Likewise.
16526         (VP_small_lc): Likewise.
16527         (VU_char): Likewise.
16528         (add<mode>3): Likewise.
16529         (altivec_vaddcuw): Likewise.
16530         (altivec_vaddu<VI_char>s): Likewise.
16531         (altivec_vadds<VI_char>s): Likewise.
16532         (sub<mode>3): Likewise.
16533         (altivec_vsubcuw): Likewise.
16534         (altivec_vsubu<VI_char>s): Likewise.
16535         (altivec_vsubs<VI_char>s): Likewise.
16536         (altivec_vavgs<VI_char>): Likewise.
16537         (altivec_vcmpbfp): Likewise.
16538         (altivec_eq<mode>): Likewise.
16539         (altivec_gt<mode>): Likewise.
16540         (altivec_gtu<mode>): Likewise.
16541         (umax<mode>3): Likewise.
16542         (smax<mode>3): Likewise.
16543         (umin<mode>3): Likewise.
16544         (smin<mode>3): Likewise.
16545         (altivec_vpkuhum): Likewise.
16546         (altivec_vpkuwum): Likewise.
16547         (altivec_vpkshss): Likewise.
16548         (altivec_vpkswss): Likewise.
16549         (altivec_vpkuhus): Likewise.
16550         (altivec_vpkshus): Likewise.
16551         (altivec_vpkuwus): Likewise.
16552         (altivec_vpkswus): Likewise.
16553         (altivec_vpks<VI_char>ss): Likewise.
16554         (altivec_vpks<VI_char>us): Likewise.
16555         (altivec_vpku<VI_char>us): Likewise.
16556         (altivec_vpku<VI_char>um): Likewise.
16557         (altivec_vrl<VI_char>): Likewise.
16558         (altivec_vsl<VI_char>): Likewise.
16559         (altivec_vsr<VI_char>): Likewise.
16560         (altivec_vsra<VI_char>): Likewise.
16561         (altivec_vsldoi_<mode>): Likewise.
16562         (altivec_vupkhsb): Likewise.
16563         (altivec_vupkhs<VU_char>): Likewise.
16564         (altivec_vupkls<VU_char>): Likewise.
16565         (altivec_vupkhsh): Likewise.
16566         (altivec_vupklsb): Likewise.
16567         (altivec_vupklsh): Likewise.
16568         (altivec_vcmpequ<VI_char>_p): Likewise.
16569         (altivec_vcmpgts<VI_char>_p): Likewise.
16570         (altivec_vcmpgtu<VI_char>_p): Likewise.
16571         (abs<mode>2): Likewise.
16572         (vec_unpacks_hi_v16qi): Likewise.
16573         (vec_unpacks_hi_v8hi): Likewise.
16574         (vec_unpacks_lo_v16qi): Likewise.
16575         (vec_unpacks_hi_<VP_small_lc>): Likewise.
16576         (vec_unpacks_lo_v8hi): Likewise.
16577         (vec_unpacks_lo_<VP_small_lc>): Likewise.
16578         (vec_pack_trunc_v8h): Likewise.
16579         (vec_pack_trunc_v4si): Likewise.
16580         (vec_pack_trunc_<mode>): Likewise.
16582         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
16583         V2DI builtins.
16584         (vec_vmaxsd): Likewise.
16585         (vec_vmaxud): Likewise.
16586         (vec_vminsd): Likewise.
16587         (vec_vminud): Likewise.
16588         (vec_vpksdss): Likewise.
16589         (vec_vpksdus): Likewise.
16590         (vec_vpkudum): Likewise.
16591         (vec_vpkudus): Likewise.
16592         (vec_vrld): Likewise.
16593         (vec_vsld): Likewise.
16594         (vec_vsrad): Likewise.
16595         (vec_vsrd): Likewise.
16596         (vec_vsubudm): Likewise.
16597         (vec_vupkhsw): Likewise.
16598         (vec_vupklsw): Likewise.
16600 2013-05-29  Jan Hubicka  <jh@suse.cz>
16602         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
16603         flags; reorder rest of fields in more consistent way.
16604         (varpool_node): Remove analyzed, finalized and alias.
16605         (cgraph_ndoe): Likewise.
16606         (symtab_alias_ultimate_target): New function.
16607         (cgraph_function_node): Move offline.
16608         (cgraph_reset_node): Declare.
16609         (cgraph_comdat_can_be_unshared_p): Remove.
16610         (varpool_remove_initializer): Declare.
16611         (varpool_first_defined_variable, varpool_next_defined_variable
16612         cgraph_first_defined_function, cgraph_next_defined_function): Update.
16613         (cgraph_function_with_gimple_body_p): Update.
16614         (varpool_all_refs_explicit_p): Update.
16615         (symtab_alias_target): New function.
16616         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
16617         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
16618         (cgraph_function_or_thunk_node): Simplify using
16619         symtab_alias_ultimate_target.
16620         (varpool_variable_node): Likewise.
16621         * cgraph.c (cgraph_create_function_alias): Update.
16622         (cgraph_add_thunk): Update.
16623         (cgraph_remove_node): Update.
16624         (dump_cgraph_node): Do not dump removed flags.
16625         (cgraph_function_body_availability): Update.
16626         (cgraph_propagate_frequency): Update.
16627         (verify_cgraph_node): Check sanity of local flag.
16628         (cgraph_function_node): Move here from cgraph.h; revamp for
16629         cgraph_function_or_thunk_node.
16630         * lto-symtab.c (lto_varpool_replace_node): Update.
16631         (lto_symtab_resolve_can_prevail_p): Update.
16632         (lto_symtab_merge_cgraph_nodes): Update.
16633         * ipa-cp.c (determine_versionability, initialize_node_lattices,
16634         propagate_constants_accross_call, devirtualization_time_bonus,
16635         ipcp_propagate_stage): Update.
16636         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16637         * ipa-inline-transform.c (clone_inlined_nodes,
16638         preserve_function_body_p): Update.
16639         * ipa-reference.c (propagate): Update.
16640         (write_node_summary_p): Update.
16641         * toplev.c (wrapup_global_declaration_2): Update.
16642         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16643         (analyze_function) ... this one.
16644         (cgraph_process_new_functions): Update.
16645         (cgraph_reset_node): Export.
16646         (cgraph_finalize_function): Update.
16647         (cgraph_add_new_function): Update.
16648         (process_function_and_variable_attributes): Update.
16649         (varpool_finalize_decl): Update.
16650         (symbol_finalized): Remove.
16651         (symbol_finalized_and_needed): Rename to ...
16652         (symbol_defined_and_needed): ... update.
16653         (cgraph_analyze_functions): Update.
16654         (handle_alias_pairs): Update.
16655         (mark_functions_to_output): Update.
16656         (assemble_thunk): Update.
16657         (output_in_order): Update.
16658         (output_weakrefs): Update.
16659         (finalize_compilation_unit): Update.
16660         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16661         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16662         input_node, input_varpool_node): Update.
16663         * dbxout.c (dbxout_expand_expr): Update.
16664         * cgraphclones.c (cgraph_clone_node): Update.
16665         (cgraph_copy_node_for_versioning): Update.
16666         (cgraph_materialize_clone): Update.
16667         (cgraph_materialize_all_clones): Update.
16668         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16669         propagate_pure_const, propagate_nothrow): Update.
16670         * lto-streamer-out.c (lto_output, write_symbol): Update.
16671         * ipa-utils.c (ipa_reverse_postorder): Update.
16672         * ipa-inline.c (can_inline_edge_p): Update.
16673         (update_caller_keys, ipa_inline): Update.
16674         * dwarf2out.c (reference_to_unused,
16675         premark_types_used_by_global_vars_helper): Update.
16676         * tree-eh.c (tree_could_trap_p): Update.
16677         * ipa-split.c (consider_split, execute_split_functions): Update.
16678         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16679         has_addr_references_p): Update; move ahead in file for better
16680         readability.
16681         (process_references): Simplify.
16682         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16683         bodies are removed.
16684         (cgraph_comdat_can_be_unshared_p): Make static.
16685         (cgraph_externally_visible_p): Update.
16686         (varpool_externally_visible_p): Update.
16687         (function_and_variable_visibility): Update.
16688         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16689         ipa_tm_mark_force_output_node): Update.
16690         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16691         estimate_edge_devirt_benefit, inline_generate_summary,
16692         inline_write_summary): Update.
16693         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16694         * ipa-prop.c (ipa_compute_jump_functions): Update.
16695         (ipa_print_node_params, ipa_prop_read_section,
16696         ipa_update_after_lto_read, read_replacements_section): Update.
16697         * varasm.c (mark_decl_referenced): Update.
16698         (assemble_alias, dump_tm_clone_pairs): Update.
16699         * tree-inline.c (copy_bb): Update.
16700         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16701         Update.
16702         * symtab.c (dump_symtab_base): Print new flags.
16703         (verify_symtab_base): Verify new flags.
16704         (symtab_alias_ultimate_target): New function.
16705         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16706         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16707         Update.
16708         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16709         Update.
16710         * i386.c (ix86_get_function_versions_dispatcher,
16711         ix86_generate_version_dispatcher_body): Update.
16712         (fold_builtin_cpu): Use varpool_add_new_variable.
16713         * varpool.c (varpool_remove_initializer): Break out from ...
16714         (varpool_remove_node): ... this one.
16715         (dump_varpool_node, varpool_node_for_asm,
16716         cgraph_variable_initializer_availability, varpool_analyze_node,
16717         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16718         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16719         Update.
16721 2013-05-29  Jan Hubicka  <jh@suse.cz>
16723         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16725 2013-05-29  Easwaran Raman  <eraman@google.com>
16727         PR tree-optimization/57442
16728         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16729         when control exits the main loop.
16731 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16733         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16734         and RX600.
16735         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16736         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16737         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16738         option.
16740 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16742         PR tree-optimization/57441
16743         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16744         Don't limit size of incr_vec to number of candidates.
16746 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16748         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16749         and mips16 directories.
16750         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16751         (MULTILIB_DIRNAMES): Ditto.
16752         (MULTILIB_EXCEPTIONS): Add new exceptions.
16753         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16754         (MULTILIB_DIRNAMES): Ditto.
16755         (MULTILIB_EXCEPTIONS): Add new exceptions.
16757 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16758             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16760         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16761         SYMBOL_TINY_ABSOLUTE.
16762         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16763         SYMBOL_TINY_ABSOLUTE.
16764         (aarch64_expand_mov_immediate): Likewise.
16765         (aarch64_classify_symbol): Likewise.
16766         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16767         Permit SYMBOL_TINY_ABSOLUTE.
16768         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16770 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16771             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16773         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16774         Refactor if/switch.  Replace gcc_assert with if.
16776 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16778         * config/i386/i386.c (initial_ix86_tune_features): Enable
16779         FP Reassociation for AMD bdver1 and bdver2.
16781 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16783         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16784         and IMAGPART_EXPR do not occur within other handled_components.
16786 2013-05-29  Richard Biener  <rguenther@suse.de>
16788         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16789         access on whether the use is in the BB we currently try to
16790         vectorize.
16791         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16792         vectorize to vect_bb_slp_scalar_cost.
16794 2013-05-29  Richard Biener  <rguenther@suse.de>
16796         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16797         computing scalar cost offsetted by stmts that are kept live
16798         by scalar uses.
16799         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16800         for computation of scalar cost.
16802 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16804         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16806 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16808         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16809         array notation built-in reduction functions.
16810         * doc/passes.texi (Passes): Added documentation about changes done
16811         for Cilk Plus.
16812         * doc/invoke.texi (C Dialect Options): Added documentation about
16813         the -fcilkplus flag.
16814         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16815         (BUILTINS_DEF): Depend on cilkplus.def.
16816         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16817         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16818         * cilkplus.def: New file.
16820 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16822         PR rtl-optimization/57439
16823         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16825 2013-05-28  Easwaran Raman  <eraman@google.com>
16827         PR tree-optimization/57337
16828         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16829         (find_insert_point): Correctly identify the insertion point
16830         when two statements with the same UID is compared.
16832 2013-05-28  Richard Biener  <rguenther@suse.de>
16834         PR tree-optimization/56787
16835         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16836         from the list of data references.
16837         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16838         clobbers.
16839         (vect_analyze_loop_operations): Likewise.
16840         (vect_transform_loop): Remove clobbers.
16842 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16844         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16845         and REALPART_EXPRs have scalar type.
16847 2013-05-28  Richard Biener  <rguenther@suse.de>
16849         PR tree-optimization/57411
16850         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16851         virtual operands.
16852         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16853         virtual operand propagation.
16855 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16857         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16858         destination register for bmasksi_vis.
16859         (vector_init_bshuffle): Likewise.
16860         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16862 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16864         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16865         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16866         mode if the instruction isn't available in the original mode.
16867         * config/sparc/sparc.opt (mfix-ut699): New option.
16868         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16869         (divdf3): Turn into expander.
16870         (divdf3_nofix): New insn.
16871         (divdf3_fix): Likewise.
16872         (divsf3): Disable if -mfix-ut699.
16873         (sqrtdf2): Turn into expander.
16874         (sqrtdf2_nofix): New insn.
16875         (sqrtdf2_fix): Likewise.
16876         (sqrtsf2): Disable if -mfix-ut699.
16878 2013-05-27  Richard Biener  <rguenther@suse.de>
16880         PR middle-end/57412
16881         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16882         block for the new loop.
16884 2013-05-27  Richard Biener  <rguenther@suse.de>
16886         PR tree-optimization/57343
16887         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16888         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16889         (number_of_iterations_cond): Do not build the folded tree.
16891 2013-05-27  Richard Biener  <rguenther@suse.de>
16893         Revert
16894         PR middle-end/57381
16895         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16896         OEP_CONSTANT_ADDRESS_OF retained.
16898         PR tree-optimization/57417
16899         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16900         for unchanged base.
16901         (set_ssa_val_to): Compare addresses using
16902         get_addr_base_and_unit_offset.
16904 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16906         PR rtl-optimization/56833
16907         * postreload.c (move2add_record_mode): New function.
16908         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16909         (move2add_use_add2_insn): Use move2add_record_sym_value.
16910         (move2add_use_add3_insn): Likewise.
16911         (reload_cse_move2add): Use move2add_valid_value_p and
16912         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16913         affected regs by setting reg_mode to VOIDmode.
16914         (move2add_note_store): Don't pretend the inside of a SUBREG is
16915         the actual destination.  Invalidate single/leading registers by
16916         setting reg_mode to VOIDmode.
16917         Use move2add_record_sym_value, move2add_valid_value_p and
16918         move2add_record_mode.
16920 2013-05-27  Richard Biener  <rguenther@suse.de>
16922         PR tree-optimization/57396
16923         * tree-affine.c (double_int_constant_multiple_p): Properly
16924         return false for val == 0 and div != 0.
16926 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16928         * config/mips/mips.h: Use #elif in preprocessor conditions.
16930 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16932         PR target/53916
16933         * config/mips/constraints.md (kl): New constraint.
16934         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16935         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16936         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16937         from LO for MIPS16.
16938         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16940 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16942         PR target/55777
16943         * config/mips/mips.c (mips_can_inline_p): New function.
16944         (TARGET_CAN_INLINE_P): Define.
16946 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16948         * sched-int.h (ds_t, dw_t): Make unsigned int.
16949         Fix documentation that describes how all the ds_t bits are used.
16950         Reserve the last bit for delayed-branch scheduling.
16951         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16952         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16953         (gen_dep_weak_1): Remove prototype.
16954         * sched-deps.c (get_dep_weak_1): Make static.
16955         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16956         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16957         * doc/tm.texi: Regenerate.
16958         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16960 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16962         PR debug/56950
16963         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16965 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16966             Sandra Loosemore  <sandra@codesourcery.com>
16968         * config.gcc (powerpc-*): Allow native for with-cpu.
16970 2013-05-24  Jeff Law  <law@redhat.com>
16972         PR tree-optimization/57124
16973         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16974         conversion feeding a condition if the range has an overflow
16975         if -fstrict-overflow.  Add warnings for when we do make the
16976         transformation.
16978 2013-05-24  Dehao Chen  <dehao@google.com>
16980         * tree-cfg.c (locus_discrim_map): Fix the typo.
16981         (locus_discrim_hasher): Likewise.
16982         (locus_discrim_hasher::hash): Likewise.
16983         (locus_discrim_hasher::equal): Likewise.
16985 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16987         PR tree-optimization/57294
16988         * cgraph.h (ipa_record_stmt_references): Declare.
16989         * cgraphbuild.c (ipa_record_stmt_references): New function.
16990         (build_cgraph_edges): Use ipa_record_stmt_references.
16991         (rebuild_cgraph_edges): Likewise.
16992         (cgraph_rebuild_references): Likewise.
16993         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16994         associated with the old statement and build references from the
16995         newly built statements.
16996         * ipa-ref.c (ipa_remove_stmt_references): New function.
16997         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16999 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
17001         * lra-constraints.c (emit_spill_move): Use smaller mode for
17002         mem-mem moves.
17003         (check_and_process_move): Consider mem-reg moves for secondary
17004         too.
17005         (curr_insn_transform): Don't lose insns emitted before for
17006         secondary memory moves.
17007         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
17008         reg set up in the current insn.
17010 2013-05-24  Dehao Chen  <dehao@google.com>
17012         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
17013         hash function.
17014         (locus_descrim_hasher::equal): Likewise.
17015         (build_gimple_cfg): New discriminator assignment algorithm.
17016         (make_edges): Likewise.
17017         (next_discriminator_for_locus): Likewise.
17018         (same_line_p): Likewise.
17019         (assign_discriminators): Likewise.
17020         (make_cond_expr_edges): Likewise.
17021         (make_gimple_switch_edges): Likewise.
17022         (make_goto_expr_edges): Likewise.
17023         (make_gimple_asm_edges): Likewise.
17025 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
17027         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
17028         X format specifier to only display bottom 16 bits.
17029         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
17030         immediate to match for operand 2, since it will be masked.
17032 2013-05-24  Richard Biener  <rguenther@suse.de>
17034         PR tree-optimization/57287
17035         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
17036         all SSA names that occur in abnormal PHIs.
17038 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17040         PR tree-ssa/57385
17041         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
17042         that index is not negative.
17044 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
17046         PR rtl-optimization/55177
17047         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
17048         (simplify_byte_swapping_operation): New.
17049         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
17050         (simplify_relational_operation_1): Deal with BSWAP.
17052 2013-05-23  Richard Henderson  <rth@redhat.com>
17054         PR target/56742
17055         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
17056         (ix86_reorg): Call it.
17058 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
17060         PR target/57379
17061         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
17062         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
17063         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
17065 2013-05-23  Christian Bruel  <christian.bruel@st.com>
17067         PR debug/57351
17068         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
17070 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
17071             Marcus Shawcroft  <marcus.shawcroft@arm.com>
17073         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
17074         * config/aarch64/constraints.md (Usa): Remove.
17075         * doc/md.texi (AArch64 Usa): Remove.
17077 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
17078             Marcus Shawcroft  <marcus.shawcroft@arm.com>
17080         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
17081         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
17082         * config/aarch64/predicates.md (aarch64_const_address): Remove.
17083         (aarch64_mov_operand): Use aarch64_mov_operand_p.
17085 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
17087         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
17088         instruction (AdvSIMD).
17089         * config/aarch64/aarch64-builtins.c
17090         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
17091         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
17093 2013-05-23  Martin Jambor  <mjambor@suse.cz>
17095         PR middle-end/57347
17096         * tree.h (contains_bitfld_component_ref_p): Declare.
17097         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
17098         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
17099         caller.
17100         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
17101         not access a bit-field.  Assert all final offsets are byte-aligned.
17103 2013-05-23  Richard Biener  <rguenther@suse.de>
17105         PR tree-optimization/57380
17106         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
17107         least one invariant or re-used load.
17108         * passes.c (init_optimization_passes): Move pass_phiprop before
17109         pass_forwprop.
17111 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
17113         * config/aarch64/aarch64-simd.md
17114         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
17116 2013-05-23  Richard Biener  <rguenther@suse.de>
17118         PR middle-end/57381
17119         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
17120         OEP_CONSTANT_ADDRESS_OF retained.
17122 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
17124         PR middle-end/57344
17125         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
17126         don't lower unit.  Handle unit not being always BITS_PER_WORD.
17128 2013-05-23  Richard Biener  <rguenther@suse.de>
17130         PR rtl-optimization/57341
17131         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
17132         instead of true_dependence.
17134 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
17136         * bb-reorder.c (branch_threshold): Make const.
17137         (exec_threshold): Ditto.
17139 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
17140             Pat Haugen  <pthaugen@us.ibm.com>
17141             Peter Bergner  <bergner@vnet.ibm.com>
17143         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
17144         documentation for the power8 crypto builtins.
17146         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
17148         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
17149         macros for defining power8 builtin functions.
17150         (BU_P8V_AV_2): Likewise.
17151         (BU_P8V_AV_P): Likewise.
17152         (BU_P8V_VSX_1): Likewise.
17153         (BU_P8V_OVERLOAD_1): Likewise.
17154         (BU_P8V_OVERLOAD_2): Likewise.
17155         (BU_CRYPTO_1): Likewise.
17156         (BU_CRYPTO_2): Likewise.
17157         (BU_CRYPTO_3): Likewise.
17158         (BU_CRYPTO_OVERLOAD_1): Likewise.
17159         (BU_CRYPTO_OVERLOAD_2): Likewise.
17160         (XSCVSPDP): Fix typo, point to the correct instruction.
17161         (VCIPHER): Add power8 crypto builtins.
17162         (VCIPHERLAST): Likewise.
17163         (VNCIPHER): Likewise.
17164         (VNCIPHERLAST): Likewise.
17165         (VPMSUMB): Likewise.
17166         (VPMSUMH): Likewise.
17167         (VPMSUMW): Likewise.
17168         (VPERMXOR_V2DI): Likewise.
17169         (VPERMXOR_V4SI: Likewise.
17170         (VPERMXOR_V8HI: Likewise.
17171         (VPERMXOR_V16QI: Likewise.
17172         (VSHASIGMAW): Likewise.
17173         (VSHASIGMAD): Likewise.
17174         (VPMSUM): Likewise.
17175         (VPERMXOR): Likewise.
17176         (VSHASIGMA): Likewise.
17178         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
17179         __CRYPTO__ if the crypto instructions are available.
17180         (altivec_overloaded_builtins): Add support for overloaded power8
17181         builtins.
17183         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
17184         support for power8 crypto builtins.
17185         (builtin_function_type): Likewise.
17186         (altivec_init_builtins): Add support for builtins that take vector
17187         long long (V2DI) arguments.
17189         * config/rs6000/crypto.md: New file, define power8 crypto
17190         instructions.
17192 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
17193             Pat Haugen  <pthaugen@us.ibm.com>
17194             Peter Bergner  <bergner@vnet.ibm.com>
17196         * doc/invoke.texi (Option Summary): Add power8 options.
17197         (RS/6000 and PowerPC Options): Likewise.
17199         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
17200         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
17201         wm, wn, wr documentation.
17203         * config/rs6000/constraints.md (wm): New constraint for VSX
17204         registers if direct move instructions are enabled.
17205         (wn): New constraint for no registers.
17206         (wq): New constraint for quad word even GPR registers.
17207         (wr): New constraint if 64-bit instructions are enabled.
17208         (wv): New constraint if power8 vector instructions are enabled.
17209         (wQ): New constraint for quad word memory locations.
17211         * config/rs6000/predicates.md (const_0_to_15_operand): New
17212         constraint for 0..15 for crypto instructions.
17213         (gpc_reg_operand): If VSX allow registers in VSX registers as well
17214         as GPR and floating point registers.
17215         (int_reg_operand): New predicate to match only GPR registers.
17216         (base_reg_operand): New predicate to match base registers.
17217         (quad_int_reg_operand): New predicate to match even GPR registers
17218         for quad memory operations.
17219         (vsx_reg_or_cint_operand): New predicate to allow vector logical
17220         operations in both GPR and VSX registers.
17221         (quad_memory_operand): New predicate for quad memory operations.
17222         (reg_or_indexed_operand): New predicate for direct move support.
17224         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
17225         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
17226         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
17227         (POWERPC_MASKS): Add power8 options.
17228         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
17229         various options.
17231         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
17232         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
17234         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
17235         (-mpower8-fusion): New power8 options.
17236         (-mpower8-fusion-sign): Likewise.
17237         (-mpower8-vector): Likewise.
17238         (-mcrypto): Likewise.
17239         (-mdirect-move): Likewise.
17240         (-mquad-memory): Likewise.
17242         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
17243         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
17244         registers.
17245         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
17246         (rs6000_debug_vector_unit): Add p8_vector.
17247         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
17248         definitions.  Also print fusion state.
17249         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
17250         (rs6000_builtin_mask_calculate): Add power8 builtin support.
17251         (rs6000_option_override_internal): Add support for power8.
17252         (rs6000_common_init_builtins): Add debugging for skipped builtins
17253         if -mdebug=builtin.
17254         (rs6000_adjust_cost): Add power8 support.
17255         (rs6000_issue_rate): Likewise.
17256         (insn_must_be_first_in_group): Likewise.
17257         (insn_must_be_last_in_group): Likewise.
17258         (force_new_group): Likewise.
17259         (rs6000_register_move_cost): Likewise.
17260         (rs6000_opt_masks): Likewise.
17262         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
17263         power8 capable assembler, default to power7 options.
17264         (TARGET_DIRECT_MOVE): Likewise.
17265         (TARGET_CRYPTO): Likewise.
17266         (TARGET_P8_VECTOR): Likewise.
17267         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
17268         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
17269         (VECTOR_MEM_P8_VECTOR_P): Likewise.
17270         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
17271         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
17272         (TARGET_XSCVDPSPN): Likewise.
17273         (TARGET_XSCVSPDPN): Likewsie.
17274         (TARGET_SYNC_HI_QI): Likewise.
17275         (TARGET_SYNC_TI): Likewise.
17276         (MASK_CRYPTO): Likewise.
17277         (MASK_DIRECT_MOVE): Likewise.
17278         (MASK_P8_FUSION): Likewise.
17279         (MASK_P8_VECTOR): Likewise.
17280         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
17281         temporary used by some of the direct move instructions to get two FP
17282         temporary registers does not force creation of a stack frame.
17283         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
17284         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
17285         that any VSX registers are tieable, even if they are also an
17286         Altivec vector mode.
17287         (r6000_reg_class_enum): Add wm, wr, wv constraints.
17288         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
17289         (RS6000_BTM_CRYPTO): Likewise.
17290         (RS6000_BTM_COMMON): Likewise.
17292         * config/rs6000/rs6000.md (cpu attribute): Add power8.
17293         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
17294         (enum rs6000_vector): Add power8 vector support.
17296 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17298         PR target/19599
17299         PR target/57340
17300         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
17301         (any_sibcall_could_use_r3): this and handle indirect calls.
17302         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
17304 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17306         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
17308 2013-05-22  Richard Biener  <rguenther@suse.de>
17310         PR middle-end/57349
17311         * profile.c (branch_prob): Do not split blocks that are
17312         abnormally receiving from ECF_RETURNS_TWICE functions.
17314 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
17316         * recog.c (offsettable_address_addr_space_p): Fix calculation of
17317         address mode.  Move pointer mode initialization to the same place.
17319 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
17321         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
17322         while it has any effect.
17324 2013-05-21  Easwaran Raman  <eraman@google.com>
17326         PR tree-optimization/57322
17327         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
17328         UID of the statement added to the BB to be 1.
17330 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
17332         PR tree-optimization/57331
17333         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
17334         of conversion from pointer type to integral type with integer.
17336 2013-05-21  Martin Jambor  <mjambor@suse.cz>
17338         PR lto/57289
17339         * ipa-prop.c (ipa_read_node_info): Process param_used and
17340         controlled_uses in the same order as when writing.
17342 2013-05-21  Magnus Granberg  <baldrick@free.fr>
17344         PR plugins/56754
17345         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
17347 2013-05-21  Richard Biener  <rguenther@suse.de>
17349         PR tree-optimization/57318
17350         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
17351         estimate stmts with side-effects as likely eliminated.
17353 2013-05-21  Richard Biener  <rguenther@suse.de>
17355         PR tree-optimization/57330
17356         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
17357         preserve the call stmts fntype.
17359 2013-05-21  Richard Biener  <rguenther@suse.de>
17361         PR tree-optimization/57303
17362         * tree-ssa-sink.c (statement_sink_location): Improve killing
17363         stmt detection and properly handle self-assignments.
17365 2013-05-21  Christian Bruel  <christian.bruel@st.com>
17367         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
17368         spanning registers. LEAF_REG_REMAP is supported only for contiguous
17369         registers. Set register size out of the PARALLEL loop.
17371 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
17373         PR target/56547
17374         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
17375         (*fmasf4, *fmasf4_media): New insns.
17377 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
17379         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
17380         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
17381         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
17382         (mips_idiv_insns): Update the comments to say that the returned
17383         instruction counts are in units of BASE_INSN_LENGTH.
17384         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
17385         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
17386         using 2 rather than 4 as the length of indirect MIPS16 and
17387         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
17388         length of a NOP.  Don't divide MIPS16 lengths by 2.
17389         (mips16_split_long_branches): Assume a branch is long if the
17390         length is greater than 4 rather than 8.
17391         * config/mips/mips.md (length): Give MIPS16 lengths directly,
17392         rather than multiplying them by 2.  Multiply instruction counts
17393         by BASE_INSN_LENGTH rather than 4.
17394         (*jump_mips16, tls_get_tp_mips16_<mode>)
17395         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
17397 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
17399         * config/mips/mips.md (extended_mips16): Remove branch case.
17400         (length): Remove duplicated extended_mips16 test.
17402 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
17404         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
17406 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
17408         * recog.h (Recog_data): Rename to...
17409         (recog_data_d): ...this.
17410         (recog_data): Update accordingly.
17411         * recog.c (recog_data): Likewise.
17412         * reload.c (save_recog_data): Likewise.
17413         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
17414         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
17416 2013-05-17  Julian Brown  <julian@codesourcery.com>
17418         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
17419         found in a REG_EQUAL note, invalidate it.
17421 2013-05-17  Easwaran Raman  <eraman@google.com>
17423         * tree-ssa-reassoc.c (find_insert_point): New function.
17424         (insert_stmt_after): Likewise.
17425         (get_def_stmt): Likewise.
17426         (ensure_ops_are_available): Likewise.
17427         (not_dominated_by): Likewise.
17428         (rewrite_expr_tree): Do not move statements beyond what is
17429         necessary. Remove call to swap_ops_for_binary_stmt...
17430         (reassociate_bb): ... and move it here.
17431         (build_and_add_sum): Assign UIDs for new statements.
17432         (linearize_expr): Likewise.
17433         (do_reassoc): Renumber gimple statement UIDs.
17435 2013-05-17  Jan Hubicka  <jh@suse.cz>
17437         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
17438         weakrefs.
17439         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
17440         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
17441         weakrefs.
17442         (output_weakrefs): Update.
17444 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
17445             Martin Jambor  <mjambor@suse.cz>
17447         PR middle-end/57276
17448         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
17449         value that corresponds to the given aggval is found in values vector.
17451 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
17453         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
17454         sse, sse2, sse3, ssse3 and sse4a flags to options.
17456 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
17458         * gengtype-state.c: (s_expr_writer): New class, to handle
17459         prettifying of output layout of s-expressions.
17460         (state_writer): New class, to write out gtype.state.
17461         (state_written_type_count): Move this variable into member data of
17462         state_writer.
17463         (s_expr_writer::s_expr_writer): New code: constructor for new class
17464         (state_writer::state_writer(): ditto
17465         (s_expr_writer::write_new_line): New function
17466         (s_expr_writer::write_any_indent): ditto
17467         (s_expr_writer::begin_s_expr): ditto
17468         (s_expr_writer::end_s_expr): ditto
17469         (write_state_fileloc): convert to method of state_writer...
17470         (state_writer:: write_state_fileloc): ...and use methods of
17471         s_expr_writer to write indentation into the gtype.state output file
17472         to visually represent the hierarchical structure of the list
17473         structures
17474         (write_state_fields): ditto, renaming to...
17475         (state_writer::write_state_fields)
17476         (write_state_a_string): ditto, renaming to...
17477         (state_writer::write_state_a_string)
17478         (write_state_string_option): ditto, renaming to...
17479         (state_writer::write_state_string_option)
17480         (write_state_type_option): ditto, renaming to...
17481         (state_writer::write_state_type_option)
17482         (write_state_nested_option): ditto, renaming to...
17483         (state_writer::write_state_nested_option)
17484         (write_state_option): ditto, renaming to...
17485         (state_writer::write_state_option)
17486         (write_state_options): ditto, renaming to...
17487         (state_writer::write_state_options)
17488         (write_state_lang_bitmap): ditto, renaming to...
17489         (state_writer::write_state_lang_bitmap)
17490         (write_state_version): ditto, renaming to...
17491         (state_writer::write_state_version)
17492         (write_state_scalar_type): ditto, renaming to...
17493         (state_writer::write_state_scalar_type)
17494         (write_state_string_type): ditto, renaming to...
17495         (state_writer::write_state_string_type)
17496         (write_state_undefined_type): ditto, renaming to...
17497         (state_writer::write_state_undefined_type)
17498         (write_state_struct_union_type): ditto, renaming to...
17499         (state_writer::write_state_struct_union_type)
17500         (write_state_struct_type): ditto, renaming to...
17501         (state_writer::write_state_struct_type)
17502         (write_state_user_struct_type): ditto, renaming to...
17503         (state_writer::write_state_user_struct_type)
17504         (write_state_lang_struct_type): ditto, renaming to...
17505         (state_writer::write_state_lang_struct_type)
17506         (write_state_param_struct_type): ditto, renaming to...
17507         (state_writer::write_state_param_struct_type)
17508         (write_state_pointer_type): ditto, renaming to...
17509         (state_writer::write_state_pointer_type)
17510         (write_state_array_type): ditto, renaming to...
17511         (state_writer::write_state_array_type)
17512         (write_state_gc_used): ditto, renaming to...
17513         (state_writer::write_state_gc_used)
17514         (write_state_common_type_content): ditto, renaming to...
17515         (state_writer::write_state_common_type_content)
17516         (write_state_type): ditto, renaming to...
17517         (state_writer::write_state_type)
17518         (write_state_pair_list): ditto, renaming to...
17519         (state_writer::write_state_pair_list)
17520         (write_state_pair): ditto, renaming to...
17521         (state_writer::write_state_pair)
17522         (write_state_typedefs): ditto, renaming to...
17523         (state_writer::write_state_typedefs)
17524         (write_state_structures): ditto, renaming to...
17525         (state_writer::write_state_structures)
17526         (write_state_param_structs): ditto, renaming to...
17527         (state_writer::write_state_param_structs)
17528         (write_state_variables): ditto, renaming to...
17529         (state_writer::write_state_variables)
17530         (write_state_srcdir): ditto, renaming to...
17531         (state_writer::write_state_srcdir)
17532         (write_state_files_list): ditto, renaming to...
17533         (state_writer::write_state_files_list)
17534         (write_state_languages): ditto, renaming to...
17535         (state_writer::write_state_languages)
17536         (write_state): create a state_writer instance and use it when
17537         writing out the state file
17539 2013-05-17  Mike Stump  <mikestump@comcast.net>
17541         PR rtl-optimization/57304
17542         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
17543         accessing DF_REF_REAL_LOC.
17545 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17547         PR rtl-optimization/57281
17548         PR rtl-optimization/57300
17549         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
17550         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
17551         what the other splitter did if the registers are dead.
17553 2013-05-17  Richard Biener  <rguenther@suse.de>
17555         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
17556         MEM_REF offsets.
17558 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17560         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
17561         linking.
17563 2013-05-17  Marek Polacek  <polacek@redhat.com>
17565         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
17566         length when doing non-zero store of storing '\0' to '\0'.
17568 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
17570         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
17571         vect_external_def oprnd1 with loop_vinfo, try to emit
17572         optional cast, negation and and stmts on the loop preheader
17573         edge instead of into the pattern def seq.
17575         PR tree-optimization/57051
17576         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
17577         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
17579 2013-05-16  Nick Clifton  <nickc@redhat.com>
17581         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
17582         (rl78_is_naked_func): New function.
17583         (rl78_expand_prologue): Skip prologue generation for naked functions.
17584         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
17585         * doc/extend.texi (naked): Add RL78 to the list of processors
17586         that supports this attribute.
17588 2013-05-16  Jeff Law  <law@redhat.com>
17590         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
17592 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17594         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
17595         cache parameters using detect_caches_amd also for CYRIX,
17596         NSC and TM2 signatures.
17598 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
17599             Dzianis Kahanovich  <mahatma@eu.by>
17601         PR target/45359
17602         PR target/46396
17603         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17604         VIA/Centaur processors and determine their cache parameters
17605         using detect_caches_amd.
17607 2013-05-16  Teresa Johnson  <tejohnson@google.com>
17609         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
17610         (rtl_verify_edges): New function.
17611         (rtl_verify_bb_insns): Ditto.
17612         (rtl_verify_bb_pointers): Ditto.
17613         (rtl_verify_bb_insn_chain): Ditto.
17614         (rtl_verify_fallthru): Ditto.
17615         (rtl_verify_bb_layout): Ditto.
17616         (rtl_verify_flow_info_1): Outline checks into new functions.
17617         (rtl_verify_flow_info): Ditto.
17619 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
17621         * cfghooks.c (copy_bbs): Add update_dominance argument.
17622         * cfghooks.h (copy_bbs): Update prototype.
17623         * tree-cfg.c (gimple_duplicate_sese_region):
17624         Add update_dominance argument.
17625         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
17626         * tree-ssa-loop-ch.c (copy_loop_headers): Update
17627         gimple_duplicate_sese_region call.
17628         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
17629         Update copy_bbs call.
17630         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
17631         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
17633 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17635         * tree-vectorizer.h (NUM_PATTERNS): Increment.
17636         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17637         vect_recog_rotate_pattern.
17638         (vect_recog_rotate_pattern): New function.
17640 2013-05-16  Jason Merrill  <jason@redhat.com>
17642         * Makefile.in (LLINKER): New variable.
17643         (mostlyclean): Remove link mutex.
17644         * configure.ac: Handle --enable-link-mutex.
17645         * lock-and-run.sh: New script.
17647 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17649         PR target/19599
17650         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17651         for NULL decl.
17653 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17655         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17657 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17659         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17660         * config/arm/arm.c (next_consecutive_mem): New function.
17661         (gen_movmem_ldrd_strd): Likewise.
17662         * config/arm/arm.md (movmemqi): Update condition and code.
17663         (unaligned_loaddi, unaligned_storedi): New patterns.
17665 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17667         * config.gcc: Obsolete *-*-solaris2.9*.
17668         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17670 2013-05-16  Richard Biener  <rguenther@suse.de>
17672         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17673         earlier, after GRAPHITE transforms and IV canonicalization.
17675 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17677         * omp-low.c (extract_omp_for_data): For collapsed loops,
17678         if at least one of the loops is known at compile time to
17679         iterate zero times, set count to 0.
17680         (expand_omp_regimplify_p): New function.
17681         (expand_omp_for_generic): For collapsed loops, if at least
17682         one of the loops isn't known to iterate at least once,
17683         add runtime check with setting count to 0.
17684         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17685         For unsigned types if it isn't known at compile time that
17686         the loop will iterate at least once, add runtime check to bypass
17687         the whole loop if initial condition isn't true.
17689 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17691         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17693 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17695         PR middle-end/57286
17696         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17697         transformations to avoid an infinite loop.
17699 2013-05-16  Marek Polacek  <polacek@redhat.com>
17701         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17703 2013-05-15  Leif Ekblad  <leif@rdos.net>
17705         * config/i386/i386.c (ix86_decompose_address): Use
17706         DEFAULT_TLS_SEG_REG to access TLS segment register.
17707         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17708         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17709         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17711 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17713         PR target/57260
17714         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17715         sibling calls to functions that would normally be lazily bound,
17716         unless $gp is call-clobbered.
17718 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17720         * config/i386/i386.c (ix86_option_override_internal): Update
17721         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17722         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17723         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17724         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17725         of TARGET_3DNOW.
17726         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17728 2013-05-15  Andreas Schwab  <schwab@suse.de>
17730         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17731         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17732         third operand.
17734 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17736         * loop-unroll.c (report_unroll_peel): Check decision before
17737         emitting unroll/peel message.
17739 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17741         * function.h (has_bb_partition): New rtl_data flag.
17742         (bb_reorder_complete): Ditto.
17743         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17744         instead of flag_reorder_blocks_and_partition.
17745         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17746         with some enhancements.
17747         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17748         * bb-reorder.c (connect_traces): Check for has_bb_partition
17749         instead of flag_reorder_blocks_and_partition.
17750         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17751         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17752         verify_hot_cold_block_grouping.
17753         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17755 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17757         PR target/19599
17758         * config/arm/predicates.md (call_insn_operand): New predicate.
17759         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17760         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17761         if insn is not a tail call.
17762         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17763         registers.
17764         * config/arm/arm.h (enum reg_class): New caller save register class.
17765         (REG_CLASS_NAMES): Likewise.
17766         (REG_CLASS_CONTENTS): Likewise.
17767         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17768         without decls.
17770 2013-05-15  Richard Biener  <rguenther@suse.de>
17772         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17773         of MSG_OPTIMIZED_LOCATIONS.
17774         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17775         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17776         message.
17777         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17778         of MSG_OPTIMIZED_LOCATIONS.
17779         (execute_vect_slp): Likewise.
17780         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17781         (vect_create_cond_for_alias_checks): Likewise.
17782         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17783         (vect_recog_widen_mult_pattern): Likewise.
17784         (vect_recog_widen_sum_pattern): Likewise.
17785         (vect_recog_over_widening_pattern): Likewise.
17786         (vect_recog_widen_shift_pattern): Likewise.
17787         (vect_recog_vector_vector_shift_pattern): Likewise.
17788         (vect_recog_divmod_pattern): Likewise.
17789         (vect_recog_mixed_size_cond_pattern): Likewise.
17790         (vect_recog_bool_pattern): Likewise.
17791         (vect_pattern_recog_1): Likewise.
17793 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17795         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17796         non-functions to builtin_unreachable.
17797         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17798         correct when new direct edges were discovered.
17800 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17802         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17803         header, print symbol order instead of node uid, print more information
17804         about indirect edge targets.
17805         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17806         uids.
17807         (ipa_make_edge_direct_to_target): Likewise.
17808         (remove_described_reference): Likewise.
17809         (propagate_controlled_uses): Likewise.
17810         (ipa_print_node_params): Also print symbol order.
17811         (ipcp_transform_function): Print symbol order instead of node uids.
17812         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17813         (cgraph_get_create_real_symbol_node): Likewise.
17814         * ipa-cp.c (print_lattice): Likewise.
17815         (print_all_lattices): Likewise.
17816         (determine_versionability): Likewise.
17817         (initialize_node_lattices): Likewise.
17818         (estimate_local_effects): Likewise.
17819         (update_profiling_info): Likewise.
17820         (create_specialized_node): Likewise.
17821         (perhaps_add_new_callers): Likewise.
17822         (decide_about_value): Likewise.
17823         (decide_whether_version_node): Likewise.
17824         (identify_dead_nodes): Likewise.
17825         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17826         (dump_inline_summary): Likewise.
17827         (estimate_node_size_and_time): Likewise.
17828         (inline_analyze_function): Likewise.
17829         * ipa-inline.c (report_inline_failed_reason): Likewise.
17830         (want_early_inline_function_p): Likewise.
17831         (edge_badness): Likewise.
17832         (update_edge_key): Likewise.
17833         (inline_small_functions): Likewise.  Add dumping of order to two other
17834         dumps.
17835         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17836         instead of node uids.
17837         (propagate_pure_const): Likewise.
17838         (propagate_pure_const): Likewise.
17839         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17840         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17841         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17842         of node uids.
17843         * tree-pretty-print.c (dump_function_header): Likewise.
17844         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17845         Print symbol order instead of node uids.
17847 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17849         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17850         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17852 2013-05-15  Richard Biener  <rguenther@suse.de>
17854         PR tree-optimization/57275
17855         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17856         return value for fail to do runtime alias checks for gather loads.
17858 2013-05-15  Jan Hubicka  <jh@suse.cz>
17860         PR lto/57038
17861         PR lto/47375
17862         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17863         weakrefs are not external.
17864         (lto_symtab_merge_decls): Fix thinko when dealing with
17865         non-lto_symtab decls.
17866         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17867         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17868         * varpool.c (dump_varpool_node): Dump more flags.
17870 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17872         * config/i386/i386.c (processor_alias_table): Add instruction
17873         FSGSBASE for AMD bdver3 architecture.
17875 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17877         * tree.c (warn_deprecated_use): Print file:line using locus color.
17878         * diagnostic.c (diagnostic_report_current_module): Print file:line
17879         and file:line:column using locus color.
17881 2013-05-14  Mike Stump  <mikestump@comcast.net>
17883         * gdbinit.in: Add __null.
17885 2013-05-14  Mike Stump  <mikestump@comcast.net>
17887         * recog.h: Rename struct recog_data to Recog_data.
17888         * recog.c: Likewise.
17889         * reload.c (can_reload_into): Likewise.
17890         * config/picochip/picochip.c: Likewise.
17892 2013-05-14  Mike Stump  <mikestump@comcast.net>
17894         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17896 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17898         * resource.h (struct resources): Remove unch_memory member.
17899         (CLEAR_RESOURCE): Don't clear unch_memory.
17900         * resource.c (mark_referenced_resources): Don't set it.
17901         (mark_set_resources): Likewise.
17902         (mark_target_live_regs): Don't clear it.
17903         (init_resource_info): Likewise.
17904         * reorg.c (resource_conflicts_p): Don't compare it.
17905         (redundant_insn): Don't set it.
17907         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17908         Remove prototypes.
17909         * emit-rtl.c (next_label): Remove unused function.
17910         (skip_consecutive_labels, link_cc0_insns): Move to ...
17911         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17912         only place where these functions are used, and make them static.
17914 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17916         * fold-const.c (fold_negate_expr): Handle vectors.
17917         (fold_truth_not_expr): Make it static.
17918         (fold_invert_truthvalue): New static function.
17919         (invert_truthvalue_loc): Handle vectors. Do not call
17920         fold_truth_not_expr directly.
17921         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17922         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17923         (fold_comparison): Handle vector constants.
17924         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17925         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17926         * tree.h (fold_truth_not_expr): Remove declaration.
17928 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17930         * config/aarch64/aarch64-simd.md
17931         (aarch64_vcond_internal<mode>): Rename to...
17932         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17933         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17934         float modes. Clarify all iterator modes.
17935         (vcond<mode><mode>): Use new name for vcond expanders.
17936         (vcond<v_cmp_result><mode>): Likewise.
17937         (vcondu<mode><mode>: Likewise.
17938         * config/aarch64/iterators.md (VDQF_COND): New.
17940 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17942         PR bootstrap/57266
17943         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17944         variable for the shift amount. Check that we shift by non-negative
17945         amounts.
17947 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17949         PR target/42017
17950         * config/arm/arm.h (EPILOGUE_USES): Only return true
17951         for LR_REGNUM after epilogue_completed.
17953 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17955         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17956         is error_mark_node.
17958 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17960         PR target/57261
17961         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17962         and Solaris 11+/x86 with gld.
17963         * configure: Regenerate.
17965 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17967         * expmed.c (expand_shift_1): Canonicalize rotates by
17968         constant bitsize / 2 to bitsize - 1.
17969         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17970         case ROTATERT>: Likewise.
17972         Revert:
17973         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17975         * config/i386/i386.md (rotateinv): New code attr.
17976         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17977         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17978         roll $31, %eax, etc.
17980 2013-05-14  Richard Biener  <rguenther@suse.de>
17982         PR middle-end/57235
17983         * tree-eh.c (sink_clobbers): Give up for successors with
17984         multiple predecessors and no virtual uses.
17986 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17988         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17989         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17991 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17993         PR middle-end/57251
17994         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17995         the case when both op0 and op1 have VOIDmode.
17997 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17999         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
18000         in multiply-accumulate mode.
18002 2013-05-13  Guozhi Wei  <carrot@google.com>
18004         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
18006 2013-05-13  Kai Tietz  <ktietz@redhat.com>
18008         PR target/56975
18009         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
18010         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
18011         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
18012         * config/i386/i386.c (ix86_option_override_internal): Likewise.
18013         (ix86_expand_prologue): Likewise.
18014         (ix86_expand_split_stack_prologue): Likewise.
18015         (legitimate_pic_address_disp_p): Likewise.
18016         (legitimize_pic_address): Likewise.
18017         (legitimize_tls_address): Likewise.
18018         (legitimize_pe_coff_symbol): Likewise.
18019         (output_pic_addr_const): Likewise.
18020         (construct_plt_address): Likewise.
18021         (ix86_expand_call): Likewise.
18022         (x86_output_mi_thunk): Likewise.
18023         (x86_function_profiler): Likewise.
18025 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
18027         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
18028         similar switch cases.
18029         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
18030         (aarch64_simd_mov_to_<mode>low): Delete.
18031         (aarch64_simd_mov_to_<mode>high): Delete.
18032         (move_lo_quad_<mode>): Add w<-r alternative.
18033         (aarch64_simd_move_hi_quad_<mode>): Likewise.
18034         (aarch64_simd_mov_from_*): Update type attribute.
18035         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
18036         statement.
18038 2013-05-13  Jan Hubicka  <jh@suse.cz>
18040         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
18041         * config/i386/i386.c (ix86_compute_frame_layout,
18042         ix86_expand_epilogue, emit_i387_cw_initialization,
18043         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
18044         ix86_local_alignment): Fix use of size/speed predicates.
18046 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
18048         PR tree-optimization/45216
18049         PR tree-optimization/57157
18050         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
18051         the (-Y) & (B - 1) variant if OP is |.
18052         * expmed.c (expand_shift_1): For rotations by const0_rtx just
18053         return shifted.  Use (-op1) & (prec - 1) as other_amount
18054         instead of prec - op1.
18056 2013-05-13  Martin Jambor  <mjambor@suse.cz>
18058         PR middle-end/42371
18059         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
18060         (ipa_constant_data): New type.
18061         (ipa_jump_func): Use ipa_constant_data to hold information about
18062         constant jump functions.
18063         (ipa_get_jf_constant): Adjust to jump function type changes.
18064         (ipa_get_jf_constant_rdesc): New function.
18065         (ipa_param_descriptor): New field controlled_uses.
18066         (ipa_get_controlled_uses): New function.
18067         (ipa_set_controlled_uses): Likewise.
18068         * ipa-ref.h (ipa_find_reference): Declare.
18069         * ipa-prop.c (ipa_cst_ref_desc): New type.
18070         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
18071         changes.
18072         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
18073         New parameter cs.  Adjust all callers.
18074         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
18075         (remove_described_reference): New function.
18076         (jfunc_rdesc_usable): Likewise.
18077         (try_make_edge_direct_simple_call): Decrement controlled use count,
18078         attempt to remove reference if it hits zero.
18079         (combine_controlled_uses_counters): New function.
18080         (propagate_controlled_uses): Likewise.
18081         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
18082         (ipa_edge_duplication_hook): Duplicate reference descriptions.
18083         (ipa_print_node_params): Print described use counter.
18084         (ipa_write_jump_function): Adjust to jump function type changes.
18085         (ipa_read_jump_function): New parameter CS, pass it to
18086         ipa_set_jf_constant.  Adjust caller.
18087         (ipa_write_node_info): Stream controlled use count
18088         (ipa_read_node_info): Likewise.
18089         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
18090         asserting.
18091         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
18092         count.  Remove cloning-added reference if it reaches zero.
18093         * ipa-ref.c (ipa_find_reference): New function.
18095 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
18097         * config/i386/i386.c (processor_target_table): Modified default
18098         alignment values for AMD BD and BT architectures.
18100 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
18102         * tree-vect-generic.c (uniform_vector_p): Move ...
18103         * tree.c (uniform_vector_p): ... here.
18104         * tree.h (uniform_vector_p): Declare it.
18105         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
18106         into a scalar.
18108 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
18110         PR tree-optimization/57230
18111         * tree-ssa-strlen.c (handle_char_store): Record length for
18112         array store from STRING_CST.
18114         PR tree-optimization/57230
18115         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
18116         check.
18118 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
18120         * config/epiphany/epiphany.c (epiphany_init): Check size of
18121         NUM_MODES_FOR_MODE_SWITCHING.
18122         (epiphany_expand_prologue):
18123         Remove CONFIG_REGNUM initial value handling code.
18124         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
18125         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
18126         (emit_set_fp_mode, epiphany_mode_after): Likewise.
18127         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
18128         Don't return 1 for FP_MODE_NONE.
18129         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
18130         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
18131         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
18132         * config/epiphany/epiphany.md (save_config): New pattern.
18134 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
18136         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
18138 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
18140         * config/i386/i386.md (memory): Handle sseishft1.
18141         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
18142         (*vec_extractv2di_1): Ditto.
18144 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
18146         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
18147         saved registers.
18149 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
18151         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
18152         Add mthumb/march=armv7-a multilib.
18153         Add mthumb/march=armv7-r multilib.
18154         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
18156 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
18158         * config/v850/t-rtems: Add more multilibs.
18160 2013-05-10  Richard Biener  <rguenther@suse.de>
18162         PR tree-optimization/57214
18163         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
18164         not propagate from SSA names that occur in abnormal PHI nodes.
18166 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
18168         * stor-layout.c (element_precision): New function.
18169         * machmode.h (element_precision): Declare it.
18170         * tree.c (build_minus_one_cst): New function.
18171         (element_precision): Likewise.
18172         * tree.h (build_minus_one_cst): Declare new function.
18173         (element_precision): Likewise.
18174         * fold-const.c (operand_equal_p): Use element_precision.
18175         (fold_binary_loc): Handle vector types.
18176         * convert.c (convert_to_integer): Use element_precision.
18177         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
18178         separately.
18180 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
18182         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
18183         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
18184         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
18185         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
18186         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
18187         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
18188         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
18189         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
18190         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
18191         (Uuw8): New constraints.
18192         (Usb4): Move into alphabetical order.
18193         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
18194         (sd8_operand, ub8_operand, uw8_operand): New predicates.
18195         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
18196         previously unnamed patterns.
18197         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
18198         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
18199         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
18200         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
18201         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
18202         of specific lengths.
18204 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
18206         * config/i386/i386.md (rotateinv): New code attr.
18207         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
18208         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
18209         roll $31, %eax, etc.
18211         PR tree-optimization/45216
18212         PR tree-optimization/57157
18213         * tree-ssa-forwprop.c (simplify_rotate): New function.
18214         (ssa_forward_propagate_and_combine): Call it.
18216 2013-05-10  Richard Biener  <rguenther@suse.de>
18218         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
18219         disable peeling when we version for aliasing.
18220         (vector_alignment_reachable_p): Honor explicit user alignment.
18221         (vect_supportable_dr_alignment): Likewise.
18222         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
18223         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
18224         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
18225         then peeling to arrange for the cost-model check to come first.
18227 2013-05-10  Alan Modra  <amodra@gmail.com>
18229         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
18230         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
18231         * configure: Regenerate.
18233 2013-05-10  Alan Modra  <amodra@gmail.com>
18235         PR target/55033
18236         * varasm.c (default_elf_select_section): Move !DECL_P check..
18237         (get_named_section): ..to here before calling get_section_name.
18238         Adjust assertion.
18239         (default_section_type_flags): Add DECL_P check.
18240         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
18241         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
18243 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
18245         * config/epiphany/epiphany.c (epiphany_expand_prologue):
18246         When using gen_stack_adjust_str with a register offset, add a
18247         REG_FRAME_RELATED_EXPR note.
18249 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
18251         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
18252         (*vec_extractv4si_zext_mem): Ditto.
18253         (*vec_extractv2di): Add 0->x and x->x alternatives.
18254         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
18255         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
18257 2013-05-09  Jason Merrill  <jason@redhat.com>
18259         N3639 C++1y VLA support
18260         * gimplify.c (gimplify_vla_decl): Don't touch an existing
18261         DECL_VALUE_EXPR.
18263         * tree.c (build_constructor_va): New.
18264         * tree.h: Declare it.
18266 2013-05-09  Martin Jambor  <mjambor@suse.cz>
18268         PR lto/57084
18269         * gimple-fold.c (canonicalize_constructor_val): Call
18270         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
18272 2013-05-09  Jan Hubicka  <jh@suse.cz>
18273             Richard Biener  <rguenther@suse.de>
18275         PR lto/54095
18276         * symtab.c (symtab_make_decl_local): Do not add private names.
18278 2013-05-09  Jan Hubicka  <jh@suse.cz>
18280         PR lto/54095
18281         * symtab.c (insert_to_assembler_name_hash): Handle clones.
18282         (unlink_from_assembler_name_hash): Likewise.
18283         (symtab_prevail_in_asm_name_hash, symtab_register_node,
18284         symtab_unregister_node, symtab_initialize_asm_name_hash,
18285         change_decl_assembler_name): Update.
18287 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
18289         * config/aarch64/aarch64.md: New movtf split.
18290         (*movtf_aarch64): Update.
18291         (aarch64_movdi_tilow): Handle TF modes and rename to
18292         aarch64_movdi_<mode>low.
18293         (aarch64_movdi_tihigh): Handle TF modes and rename to
18294         aarch64_movdi_<mode>high
18295         (aarch64_movtihigh_di): Handle TF modes and rename to
18296         aarch64_mov<mode>high_di
18297         (aarch64_movtilow_di): Handle TF modes and rename to
18298         aarch64_mov<mode>low_di
18299         (aarch64_movtilow_tilow): Remove spurious whitespace.
18300         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
18301         splits.
18302         (aarch64_print_operand): Update.
18304 2013-05-09  Alan Modra  <amodra@gmail.com>
18306         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
18307         powerpc64le.
18308         * configure: Regenerate.
18310 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
18312         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
18313         splitter preparation statements.
18314         * config/i386/sse.md (*vec_extract* splitters): Ditto.
18315         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
18316         adjust_address_nv.
18318 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18320         * gimple-ssa-strength-reduction.c (count_candidates): Change
18321         return value to int.
18322         (analyze_candidates_and_replace): Change type of length to int.
18324 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
18326         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
18327         (*vec_extract<mode>): Use VI12_128 mode iterator.
18328         (*vec_extract<mode>_mem): Ditto.
18329         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
18330         attribute.
18332 2013-05-08  Diego Novillo  <dnovillo@google.com>
18334         PR bootstrap/54659
18336         Revert:
18337         2012-08-17  Diego Novillo  <dnovillo@google.com>
18339         PR bootstrap/54281
18340         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
18341         * config.in: Regenerate.
18342         * configure: Regenerate.
18343         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
18345 2013-05-08  Jan Hubicka  <jh@suse.cz>
18347         PR lto/54095
18348         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
18349         * cgraph.h (symtab_node_base): Add unique_name.
18350         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
18351         input_overwrite_node, input_varpool_node): Stream unique_name.
18352         * cgraphclones.c (cgraph_create_virtual_clone,
18353         cgraph_function_versioning): Set unique_name.
18354         * ipa.c (function_and_variable_visibility): Set unique_name.
18356 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18358         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
18359         (alloc_cand_and_find_basis): Restrict conditional candidate
18360         processing to CAND_MULTs.
18362 2013-05-08  Jan Hubicka  <jh@suse.cz>
18364         PR lto/54095
18365         lto-symtab.c (lto_symtab_symbol_p): New function.
18366         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
18367         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
18368         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
18369         Skip static symbols.
18371 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
18373         PR tree-optimization/57200
18374         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
18375         Only call inform if the preceding warning_at returns true.
18377 2013-05-07  Han Shen  <shenhan@google.com>
18379         * cfgexpand.c (record_or_union_type_has_array_p): New function.
18380         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
18381         * common.opt (fstack-protector-strong): New option.
18382         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
18383         * doc/invoke.texi (Optimization Options): Document
18384         "-fstack-protector-strong".
18385         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
18387 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18389         * config/mips/mips.c (mips_machine_reorg2): Return 0.
18391 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
18393         * ira.c (update_equiv_regs): Add insn having equiv memory even if
18394         it is not lhs of the insn.
18395         (setup_reg_equiv): Remove insn having equiv memory which it is not
18396         lhs of the insn.
18397         * lra-constraints.c (process_address): Try to improve generation
18398         code for address base + disp.
18399         (lra_constraints): Make correct the code for checking insn setting
18400         up backward equivalence.  Remove insn only if it is in the init
18401         insn list.
18402         * lra-eliminations.c (update_reg_eliminate): Change return value.
18403         (lra_eliminate): Use the result.
18405 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
18407         * config/i386/sse.md (ssescalarnummask): New mode attribute.
18408         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
18409         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
18410         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
18411         register target operands.
18412         (*vec_extractv8hi_sse2): New pattern.
18413         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
18414         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
18415         (*vec_extract<mode>_mem): New insn and split pattern.
18417 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
18419         * config/arm/arm.c (arm_asan_shadow_offset): New function.
18420         (TARGET_ASAN_SHADOW_OFFSET): Define.
18421         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
18422         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
18424 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18426         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
18427         (incr_vec_index): Return -1 if increment not found.
18428         (create_add_on_incoming_edge): Assert if increment not found.
18429         (record_increment): Limit number of increments recorded.
18430         (all_phi_incrs_profitable): Return false if an increment not found.
18431         (replace_profitable_candidates): Don't process increments that were
18432         not recorded.
18433         (analyze_candidates_and_replace): Limit size of incr_vec.
18435 2013-05-07  Richard Biener  <rguenther@suse.de>
18437         * calls.c (special_function_p): setjmp-like functions are leaf.
18438         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
18439         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
18441 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
18443         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
18444         (aarch64_simd_mov<mode>): New expander.
18445         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
18446         (aarch64_simd_mov_to_<mode>high): Likewise.
18447         (aarch64_simd_mov_from_<mode>low): Likewise.
18448         (aarch64_simd_mov_from_<mode>high): Likewise.
18449         (aarch64_dup_lane<mode>): Update.
18450         (aarch64_dup_lanedi): New instruction pattern.
18451         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
18452         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
18454 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18456         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
18457         (replace_mult_candidate): Remove unnecessary argument; remove
18458         unnecessary parameter from call to introduce_cast_before_cand.
18459         (replace_unconditional_candidate): Remove unnecessary parameter
18460         from call to replace_mult_candidate.
18461         (replace_conditional_candidate): Likewise.
18462         (insert_initializers): Use make_temp_ssa_name.
18463         (introduce_cast_before_cand): Remove unnecessary argument; use
18464         make_temp_ssa_name.
18465         (replace_one_candidate): Remove unnecessary argument; remove
18466         unnecessary parameter from calls to introduce_cast_before_cand.
18467         (replace_profitable_candidates): Remove unnecessary parameters
18468         from calls to replace_one_candidate.
18470 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18472         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
18473         phi def as possibly hiding a basis for a CAND_ADD whose operands
18474         have been commuted in the analysis.
18475         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
18477 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18479         * config/aarch64/aarch64.md
18480         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
18481         shift value between 0-4.
18483 2013-05-07  Richard Biener  <rguenther@suse.de>
18485         * double-int.h (rshift): New overload.
18486         * double-int.c (rshift): New function.
18487         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
18488         (create_reference_ops_from_ref): Remove.
18489         (vn_reference_insert): Use shared ops for constructing the
18490         reference and copy it.
18492 2013-05-07  Richard Biener  <rguenther@suse.de>
18494         PR middle-end/57190
18495         * tree-eh.c (sink_clobbers): Properly propagate
18496         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
18498 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
18500         PR tree-optimization/57149
18501         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
18502         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
18503         collect_phi_def_edges, execute_late_warn_uninitialized): Use
18504         uninit_undefined_value_p instead of ssa_undefined_value_p.
18506         PR debug/57184
18507         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
18508         for modifier == EXPAND_INITIALIZER.
18510 2013-05-07  Anton Blanchard  <anton@samba.org>
18512         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
18513         for powerpc64 little endian.
18514         * configure: Regenerate.
18516 2013-05-06  Graham Stott  <grahams@btinternet.com>
18518         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
18519         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
18520         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
18521         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
18523 2013-05-06  Graham Stott  <grahams@btinternet.com>
18525         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
18526         codes which allow non-lvalues.
18528 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
18530         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
18531         components are all 1s.
18532         (integer_minus_onep): New function.
18533         * tree.h (integer_minus_onep): Declare it.
18534         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
18535         integer_minus_onep instead of integer_all_onesp.
18537 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18539         PR target/52933
18540         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
18541         variations of these patterns.
18543 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18545         * config/i386/i386.md (isa): Add x64_sse4 member.
18546         (enabled): Handle x64_sse4.
18547         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
18548         instruction for 64bit SSE4_1 targets.  Update insn attributes.
18549         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
18550         instruction for SSE4_1 targets.  Update insn attributes.
18551         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18552         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
18553         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
18554         const_1 selector.
18555         (*vec_extractv4si): Rename from *sse4_1_pextrd.
18556         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
18557         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
18559 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18561         PR target/57108
18562         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
18564 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
18566         * final.c (do_assembler_dialects): Don't handle curly braces and
18567         vertical bar escaped by % as dialect delimiters.
18568         (output_asm_insn): Print curly braces and vertical bar if escaped
18569         by % and ASSEMBLER_DIALECT defined.
18570         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
18571         * doc/tm.texi: Regenerated.
18573 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
18575         * config/mips/mips.c: Include tree-pass.h.
18576         (mips_reorg): Split in pre- and post-dbr_schedule parts.
18577         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
18578         (pass_mips_machine_reorg2): New machine specific pass.
18579         (insert_pass_mips_machine_reorg2): New pass plugin definition.
18580         (mips_option_override): Register the new pass.
18581         * rtl.h (cleanup_barriers): Remove prototype.
18582         (dbr_schedule): Likewise.
18583         * jump.c (cleanup_barriers): Make static.
18584         * reorg.c (dbr_schedule): Likewise.
18586 2013-05-06  Richard Biener  <rguenther@suse.de>
18588         PR tree-optimization/57185
18589         * tree-parloops.c (add_field_for_reduction): Handle anonymous
18590         SSA names properly.
18592 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
18594         PR target/57106
18595         * config/i386/i386.c (add_parameter_dependencies): Add dependence
18596         between "first_arg" and "insn", not "last" and "insn".
18598 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18600         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
18601         (find_candidates_in_block): Re-enable slsr_process_phi.
18602         (create_phi_basis): Fix double counting of candidate adjustment.
18604 2013-05-06  Richard Biener  <rguenther@suse.de>
18606         PR middle-end/57147
18607         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
18608         the edge is also fallthru, preserve it and just clear the
18609         abnormal flag.
18610         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
18611         also complex, preserve that and just clear the fallthru flag.
18612         * tree-inline.c (update_ssa_across_abnormal_edges): Also
18613         update virtual operands.
18615 2013-05-06  Alan Modra  <amodra@gmail.com>
18617         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
18618         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
18619         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
18620         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
18621         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18622         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
18624 2013-05-06  Alan Modra  <amodra@gmail.com>
18626         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
18627         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
18628         (DEFAULT_ASM_ENDIAN): Define.
18629         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
18630         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
18631         Update -K PIC clause from sysv4.h.
18632         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
18633         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
18635 2013-05-06  Alan Modra  <amodra@gmail.com>
18637         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18638         twice for little-endian.
18639         (ashrdi3_no_power, ashrdi3): Support little-endian.
18641 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18643         PR target/55303
18644         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18645         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18646         related expanders.
18647         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18648         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18649         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18650         New predicates.
18652 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18653             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18655         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18656         * config/pa/pa.opt: Make mbig-switch a no-op.
18657         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18658         (CASE_VECTOR_MODE): Always return SImode.
18659         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18660         for the !TARGET_BIG_SWITCH case.
18661         * config/pa/pa-linux.h: Likewise.
18662         * config/pa/pa-openbsd.h: Likewise.
18663         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18664         * config/pa/pa.md (short_jump): Remove define_insn.
18665         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18666         (casesi0): Remove define_insn.
18667         (type): Remove btable_branch.
18668         (pa_combine_type): Likewise.
18669         (in_nullified_branch_delay): Likewise.
18670         (in_call_delay): Likewise.
18671         (define_delay): Likewise.
18672         (define_insn_reservation "Z3"): Likewise.
18673         (define_insn_reservation "Z4"): Likewise.
18674         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18675         (pa_adjust_insn_length): Remove adjustment for btable branches.
18676         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18677         and mno-big-switch
18679 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18681         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18682         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18683         Add m->r,x alternatives.
18684         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18685         splitters using SWI48x mode iterator.
18686         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18687         TARGET_64BIT.  Add m->x alternative.
18688         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18689         Add o->x alternative.  Enable for TARGET_SSE.
18690         (sse_storeq): Remove expander.
18691         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18692         with memory input operand.
18693         (*vec_extractv2di_1 splitter): New.
18694         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18695         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18697 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18699         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18700         (INT_LOWPART): Delete.
18701         (extract_MB): Adjust.
18702         (extract_ME): Adjust.
18703         (print_operand): Adjust.
18705 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18707         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18708         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18709         (reg_or_logical_cint_operand, easy_fp_constant,
18710         logical_const_operand): Delete "CONST_DOUBLE" case.
18711         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18712         "HOST_BITS_PER_WIDE_INT == 64" test.
18713         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18714         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18715         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18716         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18717         test.
18718         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18719         CONST_DOUBLE DImode/VOIDmode case.
18720         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18721         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18722         CONST_DOUBLE VOIDmode case.
18723         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18724         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18725         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18726         Delete CONST_DOUBLE case.
18727         (splitters for mov FMOVE64 const_double): Delete
18728         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18729         "HOST_BITS_PER_WIDE_INT >= 64" test.
18730         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18731         case.
18732         (mov DI const_double): Delete.
18734 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18736         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18737         on op shows all bits zero in mode of a lowpart subreg, return zero.
18739 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18741         PR target/57150
18742         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18743         to save TFmode registers and DImode to save TImode registers for
18744         caller save operations.
18745         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18746         mark being partially clobbered since they only use the first
18747         double word.
18749         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18750         and TDmode only use the upper 64-bits of each VSX register.
18752 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18754         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18755         (find_candidates_in_block): Disable slsr_process_phi.
18757 2013-05-03  Guozhi Wei  <carrot@google.com>
18759         * coverage.c (coverage_obj_init): Move the construction of gcov
18760         constructor to ...
18761         (build_init_ctor): ... here.
18763 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18765         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18766         (slsr_cand_d): Redefine def_phi.
18767         (stride_status, phi_adjust_status, count_phis_status): New enums.
18768         (find_phi_def): New.
18769         (find_basis_for_base_expr): New.
18770         (find_basis_for_candidate): Handle hidden bases.
18771         (alloc_cand_and_find_basis): Handle phi candidates.
18772         (slsr_process_phi): New.
18773         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18774         (create_mul_imm_cand): Likewise.
18775         (create_add_ssa_cand): Exclude phi base candidates.
18776         (create_add_imm_cand): Likewise.
18777         (slsr_process_cast): Likewise.
18778         (slsr_process_copy): Likewise.
18779         (find_candidates_in_block): Handle phi candidates.
18780         (dump_candidate): Likewise.
18781         (unconditional_cands): Delete.
18782         (unconditional_cands_with_known_stride_p): Delete.
18783         (phi_dependent_cand_p): New.
18784         (cand_increment): Handle phi-dependent candidates.
18785         (replace_dependent): Delete.
18786         (replace_mult_candidate): New.
18787         (replace_unconditional_candidate): New.
18788         (incr_vec_index): Move to avoid forward reference.
18789         (create_add_on_incoming_edge): New.
18790         (create_phi_basis): New.
18791         (replace_dependents): Delete.
18792         (replace_conditional_candidate): New.
18793         (phi_add_costs): New.
18794         (replace_uncond_cands_and_profitable_phis): New.
18795         (record_increment): Handle phi adjustments.
18796         (record_phi_increments): New.
18797         (record_increments): Handle phi adjustments.
18798         (phi_incr_cost): New.
18799         (lowest_cost_path): Handle phis.
18800         (total_savings): Likewise.
18801         (analyze_increments): Likewise.
18802         (ncd_with_phi): New.
18803         (ncd_of_cand_and_phis): New.
18804         (nearest_common_dominator_for_cands): Handle phi increments.
18805         (all_phi_incrs_profitable): New.
18806         (replace_profitable_candidates): Handle phi-dependent candidates.
18807         (analyze_candidates_and_replace): Likewise.
18809 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18811         PR bootstrap/57154
18812         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18813         do not exceed REG_BR_PROB_BASE.
18815 2013-05-03  Jeff Law  <law@redhat.com>
18817         PR tree-optimization/57144
18818         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18819         operand of the condition will bit into the new type when eliminating
18820         a cast feeding a condition.
18822 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18824         PR rtl-optimization/57130
18825         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18826         of COMPARE as in_code to the recursive call if needed.
18828 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18830         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18831         (enabled): Handle new members.
18832         * config/i386/sse.md (*vec_concatv2si): Merge from
18833         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18834         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18836 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18838         PR tree-optimization/57027
18839         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18840         for fnms opportunity, check we got the prerequisite kind
18841         of tree / gimple before using accessor functions.
18843 2013-05-03  Richard Biener  <rguenther@suse.de>
18845         * double-int.h (lshift): New overload without precision
18846         and arith argument.
18847         (operator *=, operator +=, operator -=): Move ...
18848         * double-int.c (operator *=, operator +=, operator -=): ... here
18849         and implement more efficiently.
18850         (mul_double_with_sign): Remove.
18851         (lshift_double): Adjust to take unsinged shift argument, push
18852         dispatching code to callers.
18853         (mul_double_wide_with_sign): Add early out for callers that
18854         are not interested in high parts or overflow.
18855         (lshift): New function.
18856         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18857         dispatch code here.
18858         (lrotate, rrotate): Use logical shifts.
18859         * expr.c (get_inner_reference): Use lshift.
18860         * fixed-value.c (do_fixed_divide): Likewise.
18861         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18862         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18863         (indirect_refs_may_alias_p): Likewise.
18864         (stmt_kills_ref_p_1): Likewise.
18866 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18868         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18870 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18872         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18873         scalar form of FABD instruction.
18875 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18877         * lra-constraints.c (process_alt_operands): Add checking alt
18878         number to choose the best alternative.
18880 2013-05-02  Richard Biener  <rguenther@suse.de>
18882         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18883         bitmap and its handling.
18884         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18886 2013-05-02  Richard Biener  <rguenther@suse.de>
18888         PR middle-end/57140
18889         * tree-inline.c (copy_loops): Properly handle removed loops.
18890         (copy_cfg_body): Mark destination loops for fixup if source
18891         loops needed fixup.
18893 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18895         PR target/56732
18896         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18897         generating simple_return for naked functions.
18899 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18901         PR middle-end/56988
18902         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18903         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18904         flags match.
18905         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18906         ipa_agg_replacement_value structures.
18907         (known_aggs_to_agg_replacement_list): Likewise.
18908         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18909         (read_agg_replacement_chain): Likewise.
18910         (ipcp_transform_function): Also check that by_ref flags match.
18912 2013-05-02  Richard Biener  <rguenther@suse.de>
18914         * graphds.h (struct graph): Add obstack member.
18915         * graphds.c (new_graph): Initialize obstack and allocate
18916         vertices from it.
18917         (add_edge): Allocate edge from the obstack.
18918         (free_graph): Free the obstack instead of all edges and vertices.
18920 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18922         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18923         divides.
18924         * cfg.c (update_bb_profile_for_threading): Ditto.
18925         * tree-inline.c (copy_bb): Ditto.
18926         (copy_edges_for_bb): Ditto.
18927         (initialize_cfun): Ditto.
18928         (copy_cfg_body): Ditto.
18929         (expand_call_inline): Ditto.
18930         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18931         (estimate_node_size_and_time): Ditto.
18932         (inline_merge_summary): Ditto.
18933         * cgraphclones.c (cgraph_clone_edge): Ditto.
18934         (cgraph_clone_node): Ditto.
18935         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18936         (compute_trg_info): Ditto.
18938 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18940         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18941         S reg when fp attribute set.
18942         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18944 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18946         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18947         New pattern.
18948         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18949         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18950         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18952 2013-05-02  Richard Biener  <rguenther@suse.de>
18954         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18955         (struct instantiate_cache_entry): New type.
18956         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18957         (struct instantiate_cache_type): New type.
18958         (set_instantiated_value, get_instantiated_value): Remove.
18959         (get_instantiated_value_entry): New function.
18960         (instantiate_scev_name): Use the new cache and adjust.
18961         (instantiate_scev_poly): Adjust.
18962         (instantiate_scev_binary): Likewise.
18963         (instantiate_array_ref): Likewise.
18964         (instantiate_scev_convert): Likewise.
18965         (instantiate_scev_not): Likewise.
18966         (instantiate_scev_3): Likewise.
18967         (instantiate_scev_2): Likewise.
18968         (instantiate_scev_r): Likewise.
18969         (instantiate_scev): Likewise.
18970         (resolve_mixers): Likewise.
18972 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18974         PR target/57091
18975         * lra-constraints.c (best_small_class_operands_num): Remove.
18976         (process_alt_operands): Remove small_class_operands_num.  Take
18977         small classes operands into losers and only if the operand is not
18978         matched.  Modify debugging output.
18979         (curr_insn_transform): Remove best_small_class_operands_num.
18980         Print insn name.
18982 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18984         * config/aarch64/aarch64-builtins.c
18985         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18986         * config/aarch64/aarch64-simd-builtins.def
18987         (reduc_splus_): Add new modes.
18988         (reduc_uplus_): New.
18989         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18990         (reduc_uplus_v4sf): Likewise.
18991         (reduc_splus_v4sf): Likewise.
18992         (aarch64_addv<mode>): Likewise.
18993         (reduc_uplus_<mode>): Likewise.
18994         (reduc_splus_<mode>): Likewise.
18995         (aarch64_addvv2di): Likewise.
18996         (reduc_uplus_v2di): Likewise.
18997         (reduc_splus_v2di): Likewise.
18998         (aarch64_addvv2si): Likewise.
18999         (reduc_uplus_v2si): Likewise.
19000         (reduc_splus_v2si): Likewise.
19001         (reduc_<sur>plus_<mode>): New.
19002         (reduc_<sur>plus_v2di): Likewise.
19003         (reduc_<sur>plus_v2si): Likewise.
19004         (reduc_<sur>plus_v4sf): Likewise.
19005         (aarch64_addpv4sf): Likewise.
19006         * config/aarch64/arm_neon.h
19007         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
19008         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
19009         add UNSPEC_SADDV, UNSPEC_UADDV.
19010         (SUADDV): New.
19011         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
19013 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19015         * config/aarch64/arm_neon.h
19016         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
19018 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19020         * config/aarch64/aarch64-builtins
19021         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
19023 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19025         * config/aarch64/aarch64-simd-builtins.def
19026         (reduc_smax_): New.
19027         (reduc_smin_): Likewise.
19028         (reduc_umax_): Likewise.
19029         (reduc_umin_): Likewise.
19030         (reduc_smax_nan_): Likewise.
19031         (reduc_smin_nan_): Likewise.
19032         (fmax): Remove.
19033         (fmin): Likewise.
19034         (smax): Update for V2SF, V4SF and V2DF modes.
19035         (smin): Likewise.
19036         (smax_nan): New.
19037         (smin_nan): Likewise.
19038         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
19039         (<su><maxmin><mode>3): ...This, refactor.
19040         (s<maxmin><mode>3): New.
19041         (<maxmin_uns><mode>3): Likewise.
19042         (reduc_<maxmin_uns>_<mode>): Refactor.
19043         (reduc_<maxmin_uns>_v4sf): Likewise.
19044         (reduc_<maxmin_uns>_v2si): Likewise.
19045         (aarch64_<fmaxmin><mode>: Remove.
19046         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
19047         new builtin names.
19048         (vmin<q>_f<32,64>): Likewise.
19049         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
19050         (FMAXMIN): New.
19051         (su): Add mappings for smax, smin, umax, umin.
19052         (maxmin): New.
19053         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
19054         (FMAXMIN): Rename as...
19055         (FMAXMIN_UNS): ...This.
19056         (maxminv): Remove.
19057         (fmaxminv): Likewise.
19058         (fmaxmin): Likewise.
19059         (maxmin_uns): New.
19060         (maxmin_uns_op): Likewise.
19062 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19064         * config/aarch64/arm_neon.h
19065         (vac<ge, gt><sd>_f<32, 64>): Rename to...
19066         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
19067         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
19069 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19071         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
19072         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
19074 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19076         * config/aarch64/aarch64-simd.md
19077         (vcond<mode>_internal): Handle special cases for constant masks.
19078         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
19079         (vcondu<mode><mode>): Likewise.
19080         (vcond<v_cmp_result><mode>): New.
19082 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19084         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
19085         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
19086         * config/aarch64/aarch64-simd-builtins.def
19087         (cmeq): Update to BUILTIN_VALLDI.
19088         (cmgt): Likewise.
19089         (cmge): Likewise.
19090         (cmle): Likewise.
19091         (cmlt): Likewise.
19092         * config/aarch64/arm_neon.h
19093         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
19094         to builtins or C as appropriate.
19096 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
19098         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
19099         (cmgeu): ...This.
19100         (cmhi): Rename to...
19101         (cmgtu): ...This.
19102         * config/aarch64/aarch64-simd.md
19103         (simd_mode): Add SF.
19104         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
19105         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
19106         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
19107         (cstore<mode>_neg): ...This.
19108         * config/aarch64/iterators.md
19109         (VALLF): new.
19110         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
19111         (COMPARISONS): New.
19112         (UCOMPARISONS): Likewise.
19113         (optab): Add missing comparisons.
19114         (n_optab): New.
19115         (cmp_1): Likewise.
19116         (cmp_2): Likewise.
19117         (CMP): Likewise.
19118         (cmp): Remove.
19119         (VCMP_S): Likewise.
19120         (VCMP_U): Likewise.
19121         (V_cmp_result): Add DF, SF modes.
19122         (v_cmp_result): Likewise.
19123         (v): Likewise.
19124         (vmtype): Likewise.
19125         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
19127 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
19129         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
19130         define_insn to define_insn_and_split.
19131         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
19132         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
19133         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
19134         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
19135         (thumb2_negscc): Likewise.
19137 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
19139         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
19141 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
19143         * config/arm/thumb2.md: Remove trailing whitespaces.
19145 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
19147         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
19148         Use gen_int_mode rather than GEN_INT.
19150 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
19152         * value-prof.c (stream_in_histogram_value): Remove the strayed
19153         debug_gimple_stmt.
19155 2013-04-30  Richard Biener  <rguenther@suse.de>
19157         PR middle-end/57122
19158         * cfghooks.c (split_edge): Properly check for the loop latch edge.
19160 2013-04-30  Richard Biener  <rguenther@suse.de>
19162         PR middle-end/57107
19163         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
19165 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
19167         PR rtl-optimization/56957
19168         PR rtl-optimization/57105
19169         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
19170         variable.  Use just INSN_UID for determining whether an insn
19171         should be only disconnected from the insn stream.
19172         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
19174 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
19176         PR tree-optimization/57104
19177         * tsan.c (instrument_expr): Don't instrument accesses to
19178         DECL_HARD_REGISTER VAR_DECLs.
19180 2013-04-30  Richard Biener  <rguenther@suse.de>
19182         * function.h (loops_for_fn): New inline function.
19183         (set_loops_for_fn): Likewise.
19184         * cfgloop.h (place_new_loop): Add struct function parameter.
19185         (get_loop): Likewise.
19186         (get_loops): Likewise.
19187         (number_of_loops): Likewise.
19188         (fel_next): Adjust.
19189         (fel_init): Likewise.
19190         * cfg.c (get_loop_copy): Adjust.
19191         * cfgloop.c (flow_loops_dump): Likewise.
19192         (record_loop_exits): Likewise.
19193         (verify_loop_structure): Likewise.
19194         * cfgloopanal.c (mark_irreducible_loops): Likewise.
19195         (estimate_reg_pressure_cost): Likewise.
19196         (mark_loop_exit_edges): Likewise.
19197         * cfgloopmanip.c (place_new_loop): Likewise.
19198         (add_loop): Likewise.
19199         (duplicate_loop): Likewise.
19200         * graph.c (draw_cfg_nodes): Likewise.
19201         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
19202         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
19203         (extract_affine_chrec): Likewise.
19204         (build_scop_iteration_domain): Likewise.
19205         * graphite.c (graphite_initialize): Likewise.
19206         * ira-build.c (create_loop_tree_nodes): Likewise.
19207         (more_one_region_p): Likewise.
19208         (rebuild_regno_allocno_maps): Likewise.
19209         (mark_loops_for_removal): Likewise.
19210         (mark_all_loops_for_removal): Likewise.
19211         (remove_unnecessary_regions): Likewise.
19212         (ira_build): Likewise.
19213         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19214         * loop-init.c (fix_loop_structure): Likewise.
19215         (gate_rtl_move_loop_invariants): Likewise.
19216         (gate_rtl_unswitch): Likewise.
19217         (gate_rtl_unroll_and_peel_loops): Likewise.
19218         (rtl_doloop): Likewise.
19219         * lto-streamer-in.c (input_cfg): Likewise.
19220         * lto-streamer-out.c (output_cfg): Likewise.
19221         * modulo-sched.c (sms_schedule): Likewise.
19222         * predict.c (tree_estimate_probability): Likewise.
19223         (tree_estimate_probability_driver): Likewise.
19224         (estimate_loops): Likewise.
19225         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
19226         (move_sese_region_to_fn): Likewise.
19227         (debug_loop_num): Likewise.
19228         * tree-chrec.c (chrec_evaluate): Likewise.
19229         (hide_evolution_in_other_loops_than_loop): Likewise.
19230         (chrec_component_in_loop_num): Likewise.
19231         (reset_evolution_in_loop): Likewise.
19232         (evolution_function_is_invariant_rec_p): Likewise.
19233         * tree-if-conv.c (main_tree_if_conversion): Likewise.
19234         * tree-inline.c (copy_loops): Likewise.
19235         (copy_cfg_body): Likewise.
19236         (tree_function_versioning): Likewise.
19237         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
19238         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
19239         Likewise.
19240         (add_to_evolution_1): Likewise.
19241         (scev_const_prop): Likewise.
19242         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
19243         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
19244         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
19245         (tree_ssa_lim_initialize): Likewise.
19246         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
19247         (verify_loop_closed_ssa): Likewise.
19248         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
19249         (tree_ssa_loop_im): Likewise.
19250         (tree_ssa_loop_unswitch): Likewise.
19251         (tree_vectorize): Likewise.
19252         (check_data_deps): Likewise.
19253         (tree_ssa_loop_ivcanon): Likewise.
19254         (tree_ssa_loop_bounds): Likewise.
19255         (tree_complete_unroll): Likewise.
19256         (tree_complete_unroll_inner): Likewise.
19257         (tree_parallelize_loops): Likewise.
19258         (tree_ssa_loop_prefetch): Likewise.
19259         (tree_ssa_loop_ivopts): Likewise.
19260         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
19261         * tree-vectorizer.c (vectorize_loops): Likewise.
19263 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
19265         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
19266         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
19267         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
19268         with EABI_LINK_SPEC.
19270 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
19272         PR target/44578
19273         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
19274         alternative.
19276 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
19278         PR target/57097
19279         * lra-constraints.c (process_alt_operands): Discourage a bit more
19280         using memory for pseudos.  Print cost dump for alternatives.
19281         Modify cost values for conflicts with early clobbers.
19282         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
19284 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
19286         PR target/57098
19287         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
19289 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
19291         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
19292         from/to S register.
19293         (movdi_aarch64): Support LDR/STR from/to D register.
19295 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
19297         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
19298         or higher by default.
19300 2013-04-29  Richard Biener  <rguenther@suse.de>
19302         PR middle-end/57075
19303         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
19304         even if not adding abnormal edges for calls that can make
19305         abnormal gotos.
19307 2013-04-29  Richard Biener  <rguenther@suse.de>
19309         PR middle-end/57103
19310         * tree-cfg.c (move_stmt_op): Fix condition under which to update
19311         TREE_BLOCK.
19312         (move_stmt_r): Remove redundant checking.
19314 2013-04-29  Teresa Johnson  <tejohnson@google.com>
19316         PR bootstrap/57077
19317         * basic-block.h (apply_scale): New function.
19318         (apply_probability): Use apply_scale.
19319         * gimple-streamer-in.c (input_bb): Ditto.
19320         * lto-streamer-in.c (input_cfg): Ditto.
19321         * lto-cgraph.c (merge_profile_summaries): Ditto.
19322         * tree-optimize.c (execute_fixup_cfg): Ditto.
19323         * tree-inline.c (copy_bb): Update comment to use apply_scale.
19324         (copy_edges_for_bb): Ditto.
19325         (copy_cfg_body): Ditto.
19327 2013-04-29  Tom de Vries  <tom@codesourcery.com>
19329         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
19330         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
19331         (tail_merge_optimize): Handle current_loops == NULL.
19333 2013-04-26  Jeff Law  <law@redhat.com>
19335         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
19336         (simplify_cond_using_ranges): Generalize code to simplify
19337         COND_EXPRs where one argument is a constant and the other
19338         is an SSA_NAME created by an integral type conversion.
19340 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19342         * config/arm/arm.md (store_minmaxsi): Use only when
19343         optimize_insn_for_size_p.
19345 2013-04-29  Christian Bruel  <christian.bruel@st.com>
19347         PR target/57108
19348         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
19350 2013-04-29  Richard Biener  <rguenther@suse.de>
19352         PR middle-end/57089
19353         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
19354         loop tree make sure to schedule a fixup for the child as well.
19355         (expand_omp_for_generic): Properly add loops.
19356         (expand_omp_for_static_nochunk): Likewise.
19357         (expand_omp_for_static_chunk): Likewise.
19358         (expand_omp_for): For the degenerate case fixup loops.
19359         (expand_omp_sections): Fix default bb placement in loops.
19360         (expand_omp_atomic_pipeline): Properly add loops.
19362 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19364         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
19366 2013-04-29  Tom de Vries  <tom@codesourcery.com>
19368         * tree-ssa-tail-merge.c: Update header comment.
19370 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19372         * config/aarch64/arm_neon.h
19373         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
19374         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
19375         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
19376         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
19377         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
19378         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
19379         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
19380         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
19382 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19384         * config/aarch64/aarch64-simd.md
19385         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
19386         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
19387         fix_trunc, fixuns_trunc.
19388         (ftrunc<VDQF:mode>2): New.
19389         * config/aarch64/iterators.md (optab): Add fix, fixuns.
19390         (fix_trunc_optab): New.
19392 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19394         * config/aarch64/aarch64-builtins.c
19395         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
19396         iceilf, lround, iroundf.
19398 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
19400         PR target/54349
19401         * config/i386/i386.h (enum ix86_tune_indices)
19402         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
19403         New, split from X86_TUNE_INTER_UNIT_MOVES.
19404         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
19405         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
19406         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
19407         (TARGET_INTER_UNIT_MOVES): Remove.
19408         * config/i386/i386.c (initial_ix86_tune_features): Update.
19409         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
19410         (ix86_expand_convert_uns_didf_sse): Use
19411         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
19412         (ix86_expand_vector_init_one_nonzero): Ditto.
19413         (ix86_expand_vector_init_interleave): Ditto.
19414         (inline_secondary_memory_needed): Return true for moves from SSE class
19415         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
19416         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19417         * config/i386/constraints.md (Yi, Ym): Depend on
19418         TARGET_INTER_UNIT_MOVES_TO_VEC.
19419         (Yj, Yn): New constraints.
19420         * config/i386/i386.md (*movdi_internal): Change constraints of
19421         operand 1 from Yi to Yj and from Ym to Yn.
19422         (*movsi_internal): Ditto.
19423         (*movdf_internal): Ditto.
19424         (*movsf_internal): Ditto.
19425         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
19426         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
19427         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
19428         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
19429         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
19430         * config/i386/sse.md (movdi_to_sse): Ditto.
19431         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
19432         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
19433         TARGET_INTER_UNIT_MOVES.
19434         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
19435         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
19436         instead of TARGET_INTER_UNIT_MOVES.
19437         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
19438         operand 1 from Yi to Yj and from Ym to Yn.
19440 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19442         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
19443         (float_truncate_hi_): Likewise.
19444         (float_extend_lo_): Likewise.
19445         (float_truncate_lo_): Likewise.
19446         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
19447         (aarch64_float_extend_lo_v2df): Likewise.
19448         (vec_unpacks_hi_v4sf): Likewise.
19449         (aarch64_float_truncate_lo_v2sf): Likewise.
19450         (aarch64_float_truncate_hi_v4sf): Likewise.
19451         (vec_pack_trunc_v2df): Likewise.
19452         (vec_pack_trunc_df): Likewise.
19454 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19456         * config/aarch64/aarch64-builtins.c
19457         (aarch64_fold_builtin): Fold float conversions.
19458         * config/aarch64/aarch64-simd-builtins.def
19459         (floatv2si, floatv4si, floatv2di): New.
19460         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
19461         * config/aarch64/aarch64-simd.md
19462         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
19463         * config/aarch64/iterators.md (FLOATUORS): New.
19464         (optab): Add float, floatuns.
19465         (su_optab): Likewise.
19467 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19469         * config/aarch64/aarch64-builtins.c
19470         (aarch64_builtin_vectorized_function): Use new names for
19471         fcvt builtins.
19472         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
19473         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
19474         (fcvtzu): Split as...
19475         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
19476         (fcvtas): Split as...
19477         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
19478         (fcvtau): Split as...
19479         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
19480         (fcvtps): Split as...
19481         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
19482         (fcvtpu): Split as...
19483         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
19484         (fcvtms): Split as...
19485         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
19486         (fcvtmu): Split as...
19487         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
19488         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
19489         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
19490         (lfrintnusf, lfrintnudf): Likewise.
19491         * config/aarch64/aarch64-simd.md
19492         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
19493         define_insn.
19494         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
19495         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
19496         (fcvt_pattern): Likewise.
19498 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19500         * config/aarch64/aarch64-simd.md
19501         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
19502         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
19504 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19506         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
19507         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
19508         (vrnd<a,m,n,p>_f32): Implement using builtins.
19509         (vrnd<i,x><q>_f<32, 64>): New.
19511 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
19513         * config/aarch64/aarch64-builtins.c
19514         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
19515         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
19516         (frintz): Rename to...
19517         (btrunc): ...this.
19518         (frintp): Rename to...
19519         (ceil): ...this.
19520         (frintm): Rename to...
19521         (floor): ...this.
19522         (frinti): Rename to...
19523         (nearbyint): ...this.
19524         (frintx): Rename to...
19525         (rint): ...this.
19526         (frinta): Rename to...
19527         (round): ...this.
19528         * config/aarch64/aarch64-simd.md
19529         (aarch64_frint<frint_suffix><mode>): Delete.
19530         (<frint_pattern><mode>2): Convert to insn.
19531         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
19532         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
19533         (frint_pattern): Likewise.
19534         (frint_suffix): Likewise.
19536 2013-04-29  Richard Biener  <rguenther@suse.de>
19538         PR tree-optimization/57081
19539         * loop-init.c: Include tree-flow.h.
19540         (loop_optimizer_finalize): Free number of iteration estimates.
19541         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
19543 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
19545         PR tree-optimization/57083
19546         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
19547         non-singleton shift count range, zero extend low_bound for uns case.
19549         * config/i386/predicates.md (general_vector_operand): New predicate.
19550         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
19551         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
19552         if they aren't nonimmediate operands.  If their original values
19553         satisfy const_vector_equal_evenodd_p, don't shift them.
19554         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
19555         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
19556         if not nonimmediate_operand.
19557         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
19558         instead of register_operand.
19559         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
19561 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
19563         * stor-layout.c (finalize_size_functions): Allocate a structure and
19564         reset cfun before dumping the functions.
19566 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
19568         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
19570         PR target/56866
19571         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
19572         use xop_pmacsdqh if uns_p.
19573         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
19574         the immediate rotate count.
19576 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
19578         * rtl.h (struct rtx_def): Add comment for field jump.
19579         (LRA_SUBREG_P): New macro.
19580         * recog.c (register_operand): Check LRA_SUBREG_P.
19581         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19582         stack frame.
19583         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19584         (lra_final_code_change): Skip subreg change for operators.
19585         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19586         if there are no operand changes.
19587         * lra-constraints.c (curr_insn_set): New.
19588         (match_reload): Set LRA_SUBREG_P.
19589         (emit_spill_move): Ditto.
19590         (check_and_process_move): Use curr_insn_set. Process only single
19591         set insns.  Don't initialize sec_mem_p and change_p.
19592         (simplify_operand_subreg): Use LRA_SUBREG_P.
19593         (reg_in_class_p): New function.
19594         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19595         of #ifdef.  Add code to remove cycling.
19596         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19597         non-null disp.  Reload inner instead of disp when base and index
19598         are null.  Try to put lo_sum into register.
19599         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19600         (check_and_process_move): Move code for move cost check to
19601         simple_move_p.  Remove equiv_substitution.
19602         (simple_move_p): New function.
19603         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19604         curr_insn_set.  Call check_and_process_move only for single set
19605         insns.  Use the new function.  Move call of check_and_process_move
19606         after operand equiv substitution and address process.
19608 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19610         PR go/57045
19611         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
19612         with nonlocal goto receivers or returns twice calls, ignore
19613         unininitialized values from abnormal edges to nl goto receiver
19614         or returns twice call.
19616 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19618         PR tree-optimization/57051
19619         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
19620         and VEC_RSHIFT_EXPR if shift count is a multiple of element
19621         bitsize.
19623 2013-04-26  Richard Biener  <rguenther@suse.de>
19625         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
19626         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
19627         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
19628         (fixup_loop_arrays_after_move): New function.
19629         (move_sese_region_to_fn): Properly outline the loop tree parts
19630         of the SESE region.
19632 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
19634         * config/i386/i386.md (type, unit): Fix long lines.
19636 2013-04-26  Richard Biener  <rguenther@suse.de>
19638         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19639         (lto-streamer-out.o): Likewise.
19640         * cfgloop.c (init_loops_structure): Export, add struct function
19641         argument and adjust.
19642         (flow_loops_find): Adjust.
19643         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19644         (init_loops_structure): Declare.
19645         * lto-streamer-in.c: Include cfgloop.h.
19646         (input_cfg): Input the loop tree.
19647         * lto-streamer-out.c: Include cfgloop.h.
19648         (output_cfg): Output the loop tree.
19649         (output_struct_function_base): Do not drop PROP_loops.
19651 2013-03-26  Richard Biener  <rguenther@suse.de>
19653         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19654         (pass_build_cfg): Provide PROP_loops.
19655         (move_sese_region_to_fn): Remove loops that are outlined into fn
19656         for now.
19657         * tree-inline.c: Include cfgloop.h.
19658         (initialize_cfun): Do not drop PROP_loops.
19659         (copy_loops): New function.
19660         (copy_cfg_body): Copy loop structure.
19661         (tree_function_versioning): Initialize destination loop tree.
19662         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19663         (pass_parallelize_loops): Do IL verification.
19664         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19665         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19666         the CFG make sure we fixup loops as well.
19667         * tree-ssa-tail-merge.c: Include cfgloop.h.
19668         (replace_block_by): When merging loop latches mark loops for fixup.
19669         * lto-streamer-out.c (output_struct_function_base): Drop
19670         PROP_loops for now.
19671         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19672         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19673         * ipa-split.c: Include cfgloop.h.
19674         (split_function): Add the new return block to the loop tree root.
19675         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19676         whether we have removed the forwarder block.
19677         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19678         * cfgloop.h (place_new_loop): Declare.
19679         * cfgloopmanip.c (place_new_loop): Export.
19680         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19681         (tree-switch-conversion.o): Likewise.
19682         (tree-complex.o): Likewise.
19683         (tree-inline.o): Likewise.
19684         (tree-ssa-tailmerge.o): Likewise.
19685         (ipa-split.o): Likewise.
19686         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19687         (tree-ssa-copy.o): Likewise.
19688         * tree-switch-conversion.c: Include cfgloop.h
19689         (process_switch): If we emit a bit-test cascade, schedule loops
19690         for fixup.
19691         * tree-complex.c: Include cfgloop.h.
19692         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19693         * asan.c: Include cfgloop.h.
19694         (create_cond_insert_point): Properly add new basic-blocks to
19695         loops, schedule loop fixup.
19696         * cfgloop.c (verify_loop_structure): Check that looks are not
19697         marked for fixup.
19698         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19699         to loops.
19700         (expand_omp_for_generic): Likewise.
19701         (expand_omp_sections): Likewise.
19702         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19703         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19704         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19705         is initialized, not when loops are present.
19706         * tree-parloops.c (parallelize_loops): Remove checking here.
19707         * passes.c (init_optimization_passes): Schedule a copy-propagation
19708         pass before complete unrolling of inner loops.
19710 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19712         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19713         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19714         (colorize_init): Add argument to _WIN32 version.
19715         * toplev.c: Include diagnostic-color.h.
19716         (process_options): Default to -fdiagnostics-color=auto if
19717         GCC_COLORS env var is in the environment.
19718         * common.opt (fdiagnostics-color=): Add Var and Init.
19719         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19720         env var is in the environment, the default is auto rather than never.
19722         * diagnostic.h (file_name_as_prefix): Add context argument.
19723         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19724         the string as locus.
19725         * langhooks.c (lhd_print_error_function): Adjust caller.
19727 2013-04-25  Lawrence Crowl  <crowl@google.com>
19729         * var-tracking.c (shared_hash_def::htab):
19730         Change type to hash_table.  Update dependent calls and types.
19732 2013-04-25  Lawrence Crowl  <crowl@google.com>
19734         * Makefile.in: Update as needed below.
19736         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19737         Move declaration to after the type's method definitons.
19739         * attribs.c (htab_t scoped_attributes::attribute_hash):
19740         Change type to hash_table.  Update dependent calls and types.
19742         * bitmap.c (htab_t bitmap_desc_hash):
19743         Change type to hash_table.  Update dependent calls and types.
19745         * cselib.c (htab_t cselib_hash_table):
19746         Change type to hash_table.  Update dependent calls and types.
19748         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19749         (hash_string_slot_node): Move implementation into lto-streamer.h
19750         struct string_slot_hasher.
19751         (eq_string_slot_node): Likewise.
19753         * data-streamer-out.c: Update output_block::string_hash_table
19754         dependent calls and types.
19756         * dwarf2cfi.c (htab_t trace_index):
19757         Change type to hash_table.  Update dependent calls and types.
19759         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19760         Change type to hash_table.  Update dependent calls and types.
19761         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19762         (htab_t optimize_external_refs::map): Likewise.
19763         (htab_t output_comp_unit::extern_map): Likewise.
19764         (htab_t output_comdat_type_unit::extern_map): Likewise.
19765         (htab_t output_macinfo::macinfo_htab): Likewise.
19766         (htab_t optimize_location_lists::htab): Likewise.
19767         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19769         * except.c (htab_t ehspec_hash_type):
19770         Change type to hash_table.  Update dependent calls and types.
19771         (assign_filter_values::ttypes): Likewise.
19772         (assign_filter_values::ehspec): Likewise.
19773         (sjlj_assign_call_site_values::ar_hash): Likewise.
19774         (convert_to_eh_region_ranges::ar_hash): Likewise.
19776         * gcse.c (htab_t pre_ldst_table):
19777         Change type to hash_table.  Update dependent calls and types.
19779         * ggc-common.c (htab_t saving_htab):
19780         Change type to hash_table.  Update dependent calls and types.
19781         (htab_t loc_hash): Likewise.
19782         (htab_t ptr_hash): Likewise.
19783         (call_count): Rename ggc_call_count.
19784         (call_alloc): Rename ggc_call_alloc.
19785         (loc_descriptor): Rename make_loc_descriptor.
19786         (add_statistics): Rename ggc_add_statistics.
19788         * ggc-common.c (saving_htab):
19789         Change type to hash_table.  Update dependent calls and types.
19791         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19792         (push_gimplify_context): Likewise.
19793         (pop_gimplify_context): Likewise.
19794         (struct gimple_temp_hash_elt): Added.
19795         (struct gimplify_hasher): Likewise.
19796         (struct gimplify_ctx.temp_htab):
19797         Change type to hash_table.  Update dependent calls and types.
19799         * gimple-fold.c: Include gimplify-ctx.h.
19801         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19802         Change type to hash_table.  Update dependent calls and types.
19803         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19804         avoid potential global name collision.
19806         * gimplify.c: Include gimplify-ctx.h.
19807         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19808         (htab_t gimplify_ctx::temp_htab):
19809         Update dependent calls and types for new type hash_table.
19810         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19811         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19813         * gimplify-ctx.h: New.
19814         (struct gimple_temp_hash_elt): Move from gimplify.c.
19815         (class gimplify_hasher): New.
19816         (struct gimplify_ctx): Move from gimple.h.
19817         (htab_t gimplify_ctx::temp_htab):
19818         Change type to hash_table.  Update dependent calls and types.
19820         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19821         (htab_t ivs_params::newivs_index):
19822         Change type to hash_table.  Update dependent calls and types.
19823         (htab_t ivs_params::params_index): Likewise.
19824         (htab_t print_generated_program::params_index): Likewise.
19825         (htab_t gloog::newivs_index): Likewise.
19826         (htab_t gloog::params_index): Likewise.
19828         * graphite.c: Include graphite-htab.h.
19829         4htab_t graphite_transform_loops::bb_pbb_mapping):
19830         Change type to hash_table.  Update dependent calls and types.
19832         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19833         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19834         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19836         * graphite-dependences.c: Include graphite-htab.h.
19837         (loop_is_parallel_p): Change hash table type of parameter.
19839         * graphite-htab.h: New.
19840         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19841         (extern find_pbb_via_hash): Move from graphite-poly.h.
19842         (extern loop_is_parallel_p): Move from graphite-poly.h.
19843         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19845         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19846         (extern loop_is_parallel_p): Move to graphite-htab.h.
19847         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19849         * haifa-sched.c (htab_t delay_htab):
19850         Change type to hash_table.  Update dependent calls and types.
19851         (htab_t delay_htab_i2): Likewise.
19853         * ira-color.c (htab_t allocno_hard_regs_htab):
19854         Change type to hash_table.  Update dependent calls and types.
19856         * ira-costs.c (htab_t cost_classes_htab):
19857         Change type to hash_table.  Update dependent calls and types.
19859         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19860         Change type to hash_table.  Update dependent calls and types.
19862         * loop-iv.c (htab_t bivs):
19863         Change type to hash_table.  Update dependent calls and types.
19865         * loop-unroll.c (htab_t opt_info::insns_to_split):
19866         Change type to hash_table.  Update dependent calls and types.
19867         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19869         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19870         (struct string_slot_hasher): New.
19871         (htab_t output_block::string_hash_table):
19872         Change type to hash_table.  Update dependent calls and types.
19874         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19875         (htab_t file_name_hash_table):
19876         Change type to hash_table.  Update dependent calls and types.
19878         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19879         calls and types.
19881         * lto-streamer.c (htab_t tree_htab):
19882         Change type to hash_table.  Update dependent calls and types.
19884         * omp-low.c: Include gimplify-ctx.h.
19886         * passes.c (htab_t name_to_pass_map):
19887         Change type to hash_table.  Update dependent calls and types.
19888         (pass_traverse): Rename to passes_pass_traverse.
19890         * plugin.c (htab_t event_tab):
19891         Change type to hash_table.  Update dependent calls and types.
19893         * postreload-gcse.c (htab_t expr_table):
19894         Change type to hash_table.  Update dependent calls and types.
19895         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19897         * sese.c (debug_rename_map_1): Make extern.
19898         (htab_t copy_bb_and_scalar_dependences::rename_map):
19899         Change type to hash_table.  Update dependent calls and types.
19901         * sese.h (extern debug_rename_map): Move to .c file.
19903         * store-motion.c (htab_t store_motion_mems_table):
19904         Change type to hash_table.  Update dependent calls and types.
19906         * trans-mem.c (htab_t tm_new_mem_hash):
19907         Change type to hash_table.  Update dependent calls and types.
19909         * tree-browser.c (htab_t TB_up_ht):
19910         Change type to hash_table.  Update dependent calls and types.
19912         * tree-cfg.c (htab_t discriminator_per_locus):
19913         Change type to hash_table.  Update dependent calls and types.
19915         * tree-complex.c: Include tree-hasher.h
19916         (htab_t complex_variable_components):
19917         Change type to hash_table.  Update dependent calls and types.
19919         * tree-eh.c (htab_t finally_tree):
19920         Change type to hash_table.  Update dependent calls and types.
19922         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19923         struct int_tree_hasher.
19924         (extern int_tree_map_eq): Likewise.
19925         (uid_decl_map_hash): Removed.
19926         (extern decl_tree_map_eq): Likewise.
19928         * tree-hasher.h: New.
19929         (struct int_tree_hasher): New.
19930         (typedef int_tree_htab_type): New.
19932         * tree-inline.c: Include gimplify-ctx.h.
19934         * tree-mudflap.c: Include gimplify-ctx.h.
19936         * tree-parloops.c: Include tree-hasher.h.
19937         (htab_t eliminate_local_variables_stmt::decl_address):
19938         Change type to hash_table.  Update dependent calls and types.
19939         (htab_t separate_decls_in_region::decl_copies): Likewise.
19941         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19942         Change type to hash_table.  Update dependent calls and types.
19944         * tree-sra.c (candidates):
19945         Change type to hash_table.  Update dependent calls and types.
19947         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19948         in tree-flow.h.
19949         (int_tree_map_hash): Likewise.
19951         * tree-ssa-dom.c (htab_t avail_exprs):
19952         Change type to hash_table.  Update dependent calls and types.
19954         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19955         Change type to hash_table.  Update dependent calls and types.
19957         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19958         Change type to hash_table.  Update dependent calls and types.
19960         * tree-ssa-phiopt.c (seen_ssa_names):
19961         Change type to hash_table.  Update dependent calls and types.
19963         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19964         Change type to hash_table.  Update dependent calls and types.
19966         * tree-ssa-uncprop.c (equiv):
19967         Change type to hash_table.  Update dependent calls and types.
19969 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19971         PR rtl-optimization/57003
19972         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19973         call note_stores with kill_clobbered_value callback again after
19974         killing regs_invalidated_by_call.
19976 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19978         * config/aarch64/aarch64-simd.md
19979         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19980         (aarch64_simd_bsl<mode>): Likewise.
19981         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19983 2013-04-25  Marek Polacek  <polacek@redhat.com>
19985         PR tree-optimization/57066
19986         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19988 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19990         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19992 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19994         * config/aarch64/aarch64-builtins.c
19995         (aarch64_fold_builtin): New.
19996         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19997         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19998         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19999         * config/aarch64/arm_neon.h
20000         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
20002 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
20003             Tejas Belagod  <tejas.belagod@arm.com>
20005         * config/aarch64/aarch64-builtins.c
20006         (aarch64_gimple_fold_builtin): New.
20007         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
20008         * config/aarch64/aarch64-simd-builtins.def (addv): New.
20009         * config/aarch64/aarch64-simd.md (addpv4sf): New.
20010         (addvv4sf): Update.
20011         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
20013 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20015         * config/aarch64/aarch64.md
20016         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
20018 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20020         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
20021         (*ngcsi_uxtw): New pattern.
20023 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20024             Julian Brown  <julian@codesourcery.com>
20026         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
20027         (TB_DREG): Add T_V4HF.
20028         (v4hf_UP): New macro.
20029         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
20030         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
20031         Handle initialisation of V4HF. Adjust initialisation of reinterpret
20032         built-ins.
20033         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
20034         (arm_vector_mode_supported_p): Handle V4HF.
20035         (arm_mangle_map): Handle V4HFmode.
20036         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
20037         * config/arm/arm_neon_builtins.def: Add entries for
20038         vcvtv4hfv4sf, vcvtv4sfv4hf.
20039         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
20040         (neon_vcvtv4hfv4sf): Likewise.
20041         * config/arm/neon-gen.ml: Handle half-precision floating point
20042         features.
20043         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
20044         * config/arm/arm_neon.h: Regenerate.
20045         * config/arm/neon.ml (type elts): Add F16.
20046         (type vectype): Add T_float16x4, T_floatHF.
20047         (type vecmode): Add V4HF.
20048         (type features): Add Requires_FP_bit feature.
20049         (elt_width): Handle F16.
20050         (elt_class): Likewise.
20051         (elt_of_class_width): Likewise.
20052         (mode_of_elt): Refactor.
20053         (type_for_elt): Handle F16, fix error messages.
20054         (vectype_size): Handle T_float16x4.
20055         (vcvt_sh): New function.
20056         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
20057         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
20058         (string_of_mode): Handle V4HF.
20059         * doc/arm-neon-intrinsics.texi: Regenerate.
20061 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
20063         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
20064         format specifier in 'X' case.
20066 2013-04-25  Alan Modra  <amodra@gmail.com>
20068         PR target/57052
20069         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
20070         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
20071         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
20072         Repeat for many other rotate/shift and mask patterns using subregs.
20073         Name lshiftrt insns.
20074         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
20075         on WORDS_BIG_ENDIAN.
20077 2013-04-25  Alan Modra  <amodra@gmail.com>
20079         * config.gcc: Support little-endian powerpc-linux targets.
20080         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
20081         (LINK_OS_LINUX_SPEC): Define.
20082         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
20083         Preserve MASK_LITTLE_ENDIAN.
20084         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
20085         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
20086         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
20087         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
20088         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
20089         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
20090         Correct fp word order for little-endian.  Don't shift toc entries
20091         smaller than a word for little-endian.
20092         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
20093         (bswapdi2 splits): Correct low-part subreg for little-endian.
20094         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
20095         low/high where such is correct only for be.
20096         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
20097         little-endian for -mcall-aixdesc.
20099 2013-04-25  Alan Modra  <amodra@gmail.com>
20101         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
20102         replace_equiv_address_nv.
20104 2013-04-25  Alan Modra  <amodra@gmail.com>
20106         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
20108 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
20110         Revert:
20111         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
20112         * rtl.h (struct rtx_def): ...
20114 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
20116         PR rtl-optimizations/57046
20117         * lra-constraints (split_reg): Set up lra_risky_transformations_p
20118         for multi-reg splits.
20120 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
20122         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
20124 2013-04-24  Sterling Augustine  <saugustine@google.com>
20126         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
20127         (comp_dir_string, debug_str_dwo_section): New.
20128         (DEBUG_STR_DWO_SECTION): Rename to ...
20129         (DEBUG_DWO_STR_SECTION): ... this.
20130         (DEBUG_NORM_STR_SECTION): Delete.
20131         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
20132         (DEBUG_STR_DWO_SECTION_FLAGS): New.
20133         (find_AT_string): Move most logic to ...
20134         (find_AT_string_in_table): ... here.  New.
20135         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
20136         add_skeleton_AT_string.  Delete logic.
20137         (output_skeleton_debug_sections): Remove call to
20138         add_top_level_skeleton_die_attrs.
20139         (add_comp_dir_attribute): Move logic to comp_dir_string.
20140         (dwarf2out_init): Initialize debug_str_dwo_section.
20141         (output_indirect_string): Call find_string_form.
20142         (output_indirect_strings): Rewrite.
20143         (prune_unused_types): Empty skeleton_debug_str_hash.
20144         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
20145         (dwarf2out_finish):  Call output_indirect_strings.
20147 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
20149         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
20151 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
20153         * rtl.h (struct rtx_def): Add comment for field jump.
20154         (LRA_SUBREG_P): New macro.
20155         * recog.c (register_operand): Check LRA_SUBREG_P.
20156         * lra.c (lra): Add note at the end of RTL code. Align non-empty
20157         stack frame.
20158         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
20159         (lra_final_code_change): Skip subreg change for operators.
20160         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
20161         if there are no operand changes.
20162         * lra-constraints.c (curr_insn_set): New.
20163         (match_reload): Set LRA_SUBREG_P.
20164         (emit_spill_move): Ditto.
20165         (check_and_process_move): Use curr_insn_set. Process only single
20166         set insns.  Don't initialize sec_mem_p and change_p.
20167         (simplify_operand_subreg): Use LRA_SUBREG_P.
20168         (reg_in_class_p): New function.
20169         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
20170         of #ifdef.  Add code to remove cycling.
20171         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
20172         non-null disp.  Reload inner instead of disp when base and index
20173         are null.  Try to put lo_sum into register.
20174         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
20175         (check_and_process_move): Move code for move cost check to
20176         simple_move_p.  Remove equiv_substitution.
20177         (simple_move_p): New function.
20178         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
20179         curr_insn_set.  Call check_and_process_move only for single set
20180         insns.  Use the new function.  Move call of check_and_process_move
20181         after operand equiv substitution and address process.
20183 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
20185         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
20186         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
20187         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
20189 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
20191         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
20193 2013-04-24  Marek Polacek  <polacek@redhat.com>
20195         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
20196         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
20197         (select_loops_exit_conditions): Likewise.
20198         (number_of_iterations_for_all_loops): Likewise.
20199         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
20200         (scev_analysis): Likewise.
20202 2013-04-02  Catherine Moore  <clm@codesourcery.com>
20203             Chao-ying Fu  <fu@mips.com>
20205         * config/mips/micromips.md (jraddiusp): New pattern.
20206         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
20207         instruction if possible.
20209 2013-04-24  Alan Modra  <amodra@gmail.com>
20211         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
20213 2013-04-24  Julian Brown  <julian@codesourcery.com>
20214             Chung-Lin Tang  <cltang@codesourcery.com>
20216         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
20217         dependency behavior in enumeration type DIE generation. Add TODO note
20218         to comments about future DW_FORM_sdata/udata re-work of related code.
20220 2013-04-23  Lawrence Crowl  <crowl@google.com>
20222         * Makefile.in: Update as needed below.
20224         * hash-table.h (class hash_table):
20225         Correct many methods with parameter types compare_type to the correct
20226         value_type.  (Correct code was unlikely to notice the change.)
20227         (hash_table::elements_with_deleted) New.
20228         (class hashtable::iterator): New.
20229         (hashtable::begin()): New.
20230         (hashtable::end()): New.
20231         (FOR_EACH_HASH_TABLE_ELEMENT): New.
20233         * statistics.c (statistics_hashes):
20234         Change type to hash_table.  Update dependent calls and types.
20236         * tree-into-ssa.c (var_infos):
20237         Change type to hash_table.  Update dependent calls and types.
20239         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
20240         Change type to hash_table.  Update dependent calls and types.
20242         * tree-ssa-loop-im.c (struct mem_ref.refs):
20243         Change type to hash_table.  Update dependent calls and types.
20245         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
20246         Change type to hash_table.  Update dependent calls and types.
20248         * tree-ssa-sccvn.c (vn_tables_s::nary):
20249         Change type to hash_table.  Update dependent calls and types.
20250         (vn_tables_s::phis): Likewise.
20251         (vn_tables_s::references): Likewise.
20253         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
20254         (vn_reference_eq): Update parameter and return types.
20256         * tree-ssa-structalias.c (pointer_equiv_class_table):
20257         Change type to hash_table.  Update dependent calls and types.
20258         (location_equiv_class_table): Likewise.
20260         * tree-vect-data-refs.c: Consequential changes for making
20261         peeling a hash_table.
20263         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
20264         (destroy_loop_vec_info): Dependent hash_table update.
20266         * tree-vectorizer.h (peeling_htab):
20267         Change type to hash_table.  Update dependent calls and types.
20269 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
20271         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
20272         to check the register content is equal or not.
20273         * lra-constraints.c (match_reload): Use lra_assign_reg_val
20274         to assign register content record.
20275         * lra-eliminations.c (update_reg_eliminate): Use
20276         lra_update_reg_val_offset to update register content offset.
20277         * lra-int.h (struct lra_reg): Add offset member.
20278         (lra_reg_val_equal_p): New static inline function.
20279         (lra_update_reg_val_offset): New static inline function.
20280         (lra_assign_reg_val): New static inline function.
20281         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
20282         to assign register content record.
20283         (initialize_lra_reg_info_element): Initial offset to zero.
20285 2013-04-23  Catherine Moore  <clm@codesourcery.com>
20287         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
20288         operands.  Record compression.
20290 2013-04-23  Xinliang David Li  <davidxl@google.com>
20292         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
20294 2013-04-23  Richard Biener  <rguenther@suse.de>
20296         PR middle-end/57036
20297         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
20298         parameter, only add abnormal goto edges from the copied body
20299         if the call could perform abnormal gotos.
20300         (copy_cfg_body): Adjust.
20302 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
20304         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
20306 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
20308         * coretypes.h (gimple_stmt_iterator): Add struct to make
20309         compatible with C.
20311 2013-04-23  Richard Biener  <rguenther@suse.de>
20313         PR tree-optimization/57026
20314         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
20315         from SSA names occuring in abnormal PHI nodes.
20317 2013-04-22  Andi Kleen  <ak@linux.intel.com>
20319         * lto/lto.c (print_lto_report_1): Fix LTO report names.
20321 2013-04-22  Andi Kleen  <ak@linux.intel.com>
20323         * lto/lto.c (print_lto_report_1): Declare early.
20324         (read_cgraph_and_symbols): Call print_lto_report_1 early.
20326 2013-04-22  Andi Kleen  <ak@linux.intel.com>
20328         * common.opt (-flto-report-wpa): Add.
20329         * doc/invoke.texi (-flto-report-wpa): Add.
20330         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
20331         (lto_main): dito.
20333 2013-04-22  Xinliang David Li  <davidxl@google.com>
20335         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
20336         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
20337         * Makefile.in: New dependency
20339             David Daney <ddaney.cavm@gmail.com>
20341         * configure.ac (gcc_cv_as_micromips_support):  Use the
20342         --fatal-warnings option.
20343         * configure: Regenerate.
20345 2013-04-22  Marek Polacek  <polacek@redhat.com>
20347         PR sanitizer/56990
20348         * tsan.c (instrument_expr): Don't instrument expression
20349         in case its size is zero.
20351 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
20353         PR target/57032
20354         Revert:
20355         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
20357         * config/alpha/alpha.c (TARGET_LRA_P): New define.
20359 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
20361         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
20362         (gimple_stmt_iterator): New typedef.
20363         * gimple.h (gimple_stmt_iterator): Rename to...
20364         (gimple_stmt_iterator_d): ... This.
20365         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
20366         trees be valid for GIMPLE and GENERIC.
20367         (TARGET_GIMPLE_FOLD_BUILTIN): New.
20368         * gimple-fold.c (gimple_fold_call): Call target hook
20369         gimple_fold_builtin.
20370         * hooks.c (hook_bool_gsiptr_false): New.
20371         * hooks.h (hook_bool_gsiptr_false): New.
20372         * target.def (fold_stmt): New.
20373         * doc/tm.texi: Regenerate.
20375 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
20377         PR target/57018
20378         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
20379         a set sp if no stack realignment.
20381 2013-04-22  Nick Clifton  <nickc@redhat.com>
20383         * config.gcc (tilegx-linux): Extend extra_objs rather than
20384         overwriting it.
20385         (tilepro-linux): Likewise.
20387 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
20389         * config/aarch64/aarch64-builtins.c
20390         (CF): Remove.
20391         (CF0, CF1, CF2, CF3, CF4, CF10): New.
20392         (VAR<1-12>): Add MAP parameter.
20393         (BUILTIN_*): Likewise.
20394         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
20395         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
20396         (aarch64_ushl_n<mode>): Likewise.
20397         (aarch64_sshr_n<mode>): Likewise.
20398         (aarch64_ushr_n<mode>): Likewise.
20399         (aarch64_<maxmin><mode>): Likewise.
20400         (aarch64_sqrt<mode>): Likewise.
20401         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
20402         (vshr<q>_n_*): Likewise.
20404 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
20406         * config/aarch64/aarch64-builtins.c
20407         (aarch64_simd_builtin_type_mode): Handle SF types.
20408         (sf_UP): Define.
20409         (BUILTIN_GPF): Define.
20410         (aarch64_init_simd_builtins): Handle SF types.
20411         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
20412         (frecps): Likewise.
20413         (frecpx): Likewise.
20414         * config/aarch64/aarch64-simd.md
20415         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
20416         (aarch64_frecpe<mode>): New.
20417         (aarch64_frecps<mode>): Likewise.
20418         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
20419         (v8type): Add frecp<esx>.
20420         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
20421         (aarch64_frecps<mode>): Likewise.
20422         * config/aarch64/iterators.md (FRECP): New.
20423         (frecp_suffix): Likewise.
20424         * config/aarch64/arm_neon.h
20425         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
20427 2013-04-22  Christian Bruel  <christian.bruel@st.com>
20429         PR target/56995
20430         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
20431         (REG_CLASS_NAMES): Idem.
20432         (REG_CLASS_CONTENTS): Idem.
20433         (REGCLASS_HAS_FP_REG): Idem.
20434         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
20435         (sh_conditional_register_usage): Idem.
20437 2013-04-21  Jeff Law  <law@redhat.com>
20439         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
20440         (ssa_forward_propagate_and_combine): Use it.
20442 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
20444         * lra.c: Update the flow chart diagram.
20446 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
20448         PR rtl-optimization/56847
20449         * lra-constraints.c (process_alt_operands): Discourage alternative
20450         with non-matche doffsettable memory constraint fro memory with
20451         known offset.
20453 2013-04-19  Richard Biener  <rguenther@suse.de>
20455         PR tree-optimization/56982
20456         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
20457         function.
20458         * gimplify.c (gimplify_call_expr): Notice special calls.
20459         (gimplify_modify_expr): Likewise.
20460         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
20461         abnormal control flow receivers.
20462         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
20463         in the same way as cfun->has_nonlocal_labels.
20464         (gimple_purge_dead_abnormal_call_edges): Likewise.
20465         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
20466         receivers start a basic-block.
20468 2013-04-19  Richard Biener  <rguenther@suse.de>
20470         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
20471         member ...
20472         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
20473         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
20474         (SLP_TREE_LOAD_PERMUTATION): Add.
20475         (vect_transform_slp_perm_load): Adjust prototype.
20476         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
20477         (vect_free_slp_instance): Likewise.
20478         (vect_create_new_slp_node): Likewise.
20479         (vect_supported_slp_permutation_p): Remove.
20480         (vect_slp_rearrange_stmts): Adjust.
20481         (vect_supported_load_permutation_p): Likewise.  Inline
20482         vect_supported_slp_permutation_p here.
20483         (vect_analyze_slp_instance): Compute load permutations per
20484         slp node instead of per instance.
20485         (vect_get_slp_defs): Adjust.
20486         (vect_transform_slp_perm_load): Likewise.
20487         (vect_schedule_slp_instance): Remove redundant code.
20488         (vect_schedule_slp): Remove hack for PR56270, add it ...
20489         * tree-vect-stmts.c (vectorizable_load): ... here, do not
20490         CSE loads for SLP.  Adjust.
20492 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
20494         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
20495         spelling in two comments.
20497 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
20499         PR target/56797
20500         * config/arm/arm.c (load_multiple_sequence): Require SP
20501         as base register for loads if SP is in the register list.
20503 2013-04-19  Martin Jambor  <mjambor@suse.cz>
20505         PR tree-optimization/56718
20506         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
20507         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
20508         and made public.  Adjusted all callers.
20509         (ipa_intraprocedural_devirtualization): New function.
20510         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
20511         (ipa_intraprocedural_devirtualization): Likewise.
20512         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
20514 2013-04-19  Richard Biener  <rguenther@suse.de>
20516         PR tree-optimization/57000
20517         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
20519 2013-04-19  Terry Guo  <terry.guo@arm.com>
20521         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
20522         Replace with ...
20523         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
20524         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
20525         (cortex_m4_fmacs): Use new reservations.
20526         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
20528 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
20530         PR rtl-optimization/56999
20531         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
20532         related code.
20533         (lra_coalesce): Remove split_origin_bitmap and related code.
20534         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
20535         ranges if necessary.
20537 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
20539         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
20540         New array.
20541         (ix86_expand_call): Remove clobbered_registers array and use
20542         x86_64_ms_sysv_extra_clobbered_registers instead.
20543         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
20544         Declare here.
20545         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
20546         predicate.
20547         * config/i386/i386.md (*call_rex64_ms_sysv): Use
20548         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
20549         (*call_value_rex64_ms_sysv): Ditto.
20551 2013-04-18  Cary Coutant  <ccoutant@google.com>
20553         * dwarf2out.c (output_pubnames): Check die_perennial_p of
20554         parent instead of die_mark.
20556 2013-04-18  Diego Novillo  <dnovillo@google.com>
20558         * gimple.c (create_gimple_tmp): New.
20559         (get_expr_type): New.
20560         (build_assign): New.
20561         (build_type_cast): New.
20562         * gimple.h (enum ssa_mode): Define.
20563         (gimple_seq_set_location): New.
20564         * asan.c (build_check_stmt): Change some gimple_build_* calls
20565         to use build_assign and build_type_cast.
20567 2013-04-18  Richard Biener  <rguenther@suse.de>
20569         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
20570         handle negative step.  Remove redundant checks.
20571         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
20572         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
20573         for negative step and grouped loads fail to vectorize.
20575 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20577         * emit-rtl.c (reset_insn_used_flags): New function.
20578         (reset_all_used_flags): Use it.
20579         (verify_insn_sharing): New function.
20580         (verify_rtl_sharing): Fix verification for SEQUENCEs.
20582 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
20584         PR tree-optimization/56984
20585         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
20586         and (x >> M) >= N don't register any assertion if N << M is the
20587         minimum value.
20589 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
20591         * lower-subreg.c (resolve_simple_move): If called self-recursive,
20592         do not delete_insn insns that have not yet been emitted, only
20593         unlink them with remove_insn.
20594         * df-scan.c (df_insn_delete): Revert r197492.
20596 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
20598         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
20599         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
20601 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20603         * config/arm/arm.md (movsicc_insn): Convert define_insn into
20604         define_insn_and_split.
20605         (and_scc,ior_scc,negscc): Likewise.
20606         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
20608 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
20610         * config/arm/arm.c (use_return_insn): Return 0 for targets that
20611         can benefit from using a sequence of LDRD instructions in epilogue
20612         instead of a single LDM instruction.
20614 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20616         PR 45688
20617         * doc/extend.texi: Fix typo.
20619 2013-04-17  Richard Biener  <rguenther@suse.de>
20621         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
20622         (vect_build_slp_tree): ... here.
20623         (vect_build_slp_tree_1): Compute which stmts of the SLP group
20624         match.  Remove special-casing of mismatched complex loads.
20625         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
20626         re-try the match with swapped commutative operands.
20627         (vect_supported_load_permutation_p): Remove special-casing of
20628         mismatched complex loads.
20629         (vect_analyze_slp_instance): Adjust.
20631 2013-04-17  Richard Biener  <rguenther@suse.de>
20633         PR rtl-optimization/56921
20634         * cfgloop.h (struct loop): Add simple_loop_desc member.
20635         (struct niter_desc): Mark with GTY(()).
20636         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20637         * loop-iv.c (get_simple_loop_desc): Likewise.
20638         (free_simple_loop_desc): Likewise.
20640         Revert
20641         2013-04-16  Richard Biener  <rguenther@suse.de>
20643         PR rtl-optimization/56921
20644         * loop-init.c (pass_rtl_move_loop_invariants): Add
20645         TODO_do_not_ggc_collect to todo_flags_finish.
20646         (pass_rtl_unswitch): Same.
20647         (pass_rtl_unroll_and_peel_loops): Same.
20648         (pass_rtl_doloop): Same.
20650 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20652         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20653         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20654         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20655         references.
20656         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20657         * tree-streamer.c (record_common_node): Adjust reference in comment.
20659 2013-04-17  Terry Guo  <terry.guo@arm.com>
20661         * config/arm/cortex-m4.md: Add a new bypass.
20663 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20665         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20666         New pattern.
20667         (*subs_<optab><mode>_multp2): New pattern.
20668         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20669         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20671 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20673         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20674         (*subs_mul_imm_<mode>): New pattern.
20676 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20678         PR target/56948
20679         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20680         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20681         (vsx_movti_32bit): Change j->wa to O->wa.
20683 2013-04-16  Richard Biener  <rguenther@suse.de>
20685         PR rtl-optimization/56921
20686         * loop-init.c (pass_rtl_move_loop_invariants): Add
20687         TODO_do_not_ggc_collect to todo_flags_finish.
20688         (pass_rtl_unswitch): Same.
20689         (pass_rtl_unroll_and_peel_loops): Same.
20690         (pass_rtl_doloop): Same.
20692 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20694         * config/arm/arm.c (emit_multi_reg_push): New declaration
20695         for an existing function.
20696         (arm_emit_strd_push): New function.
20697         (arm_expand_prologue): Used here.
20698         (arm_emit_ldrd_pop): New function.
20699         (arm_expand_epilogue): Used here.
20700         (arm_get_frame_offsets): Update condition.
20701         (arm_emit_multi_reg_pop): Add a special case for load of a single
20702         register with writeback.
20704 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20706         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20707         description.
20709 2013-04-16  Richard Biener  <rguenther@suse.de>
20711         PR tree-optimization/56756
20712         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20713         (first_mem_ref_loc): New.
20714         (execute_sm): Place the load temporarily before a previous
20715         access instead of in the latch edge to ensure its SSA dependencies
20716         are defined at points dominating the load.
20718 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20720         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20721         correct fix by moving header and footer insn to the footer of
20722         the merged basic block.  Clear BB_END of the merged-away block.
20724         PR middle-end/43631
20725         * emit-rtl.c (make_note_raw): New function.
20726         (link_insn_into_chain): New static inline function.
20727         (add_insn): Use it.
20728         (add_insn_before, add_insn_after): Factor insn chain linking code...
20729         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20730         using link_insn_into_chain.
20731         (note_outside_basic_block_p): New helper function for emit_note_after
20732         and emit_note_before.
20733         (emit_note_after): Use nobb variant of add_insn_after if the note
20734         should not be contained in a basic block.
20735         (emit_note_before): Use nobb variant of add_insn_before if the note
20736         should not be contained in a basic block.
20737         (emit_note_copy): Use make_note_raw.
20738         (emit_note): Likewise.
20739         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20740         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20741         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20742         the moved barrier the tail of the basic block it follows.
20743         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20745 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20747         PR tree-optimization/56962
20748         * gimple-ssa-strength-reduction.c (record_increment): Only set
20749         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20750         either rhs1 or rhs2 is equal to c->base_expr.
20752 2013-04-15  Richard Biener  <rguenther@suse.de>
20754         PR tree-optimization/56933
20755         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20756         member.
20757         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20758         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20759         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20760         dependence check ...
20761         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20762         ... here.
20763         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20764         GROUP_READ_WRITE_DEPENDENCE.
20766 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20768         * emit-rtl.c (reset_all_used_flags): New function.
20769         (verify_rtl_sharing): Call reset_all_used_flags before and after
20770         performing the checks.
20772 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20774         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20775         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20776         * config/arm/constraints.md (De): New constraint.
20777         * config/arm/neon.md (anddi3_neon): Delete.
20778         (neon_vand<mode>): Expand to standard anddi3 pattern.
20779         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20780         Move earlier in the file.
20781         (neon_inv_logic_op2): Likewise.
20782         (arm_anddi_operand_neon): New predicate.
20784 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20786         * configure.ac (gcc_cv_ld_as_needed): Set
20787         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20788         Use -z ignore, -z record on *-*-solaris2*.
20789         (HAVE_LD_AS_NEEDED): Update comment.
20790         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20791         * configure: Regenerate.
20792         * config.in: Regenerate.
20793         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20794         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20795         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20796         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20797         equivalents.  Fix markup.
20798         * doc/tm.texi: Regenerate.
20800 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20802         * config/i386/i386.opt: New option mstack-protector-guard=.
20803         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20804         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20805         TARGET_SSP_TLS_GUARD.
20806         * config/i386/i386.c (ix86_option_override_internal): Set
20807         ix86_stack_protector_guard.
20808         * config/i386/i386.md (stack_protect_set): Enable for
20809         TARGET_SSP_TLS_GUARD only.
20810         (stack_protect_set_<mode>): Ditto.
20811         (stack_protect_test): Ditto.
20812         (stack_protect_test_<mode>): Ditto.
20813         * doc/invoke.texi (i386 Option): Document.
20815 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20817         PR target/56890
20818         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20819         (S_MODES): Set H_MODE bit.
20820         (SF_MODES): Set only S_MODE and SF_MODE bits.
20821         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20822         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20823         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20824         <MODE_FLOAT>: Likewise.
20826 2013-04-15  Joey Ye  <joey.ye@arm.com>
20828         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20830 2013-04-15  Joey Ye  <joey.ye@arm.com>
20832         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20833         for real far jump.
20834         (thumb_far_jump_used_p): Count instruction size and set
20835         far_jump_used.
20837 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20839         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20840         * resource.c (mark_target_live_regs): Reformat conditional block.
20842 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20844         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20845         notes, they are emitted only just before final.
20846         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20848 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20850         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20851         * cfgrtl.c (delete_insn): Call it here instead.
20852         * lra-spills.c (lra_final_code_change): Use delete_insn.
20853         * haifa-sched.c (sched_remove_insn): Likewise.
20854         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20855         returning to the nop pool.
20856         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20857         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20859 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20861         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20862         * doc/tm.texi: Regenerated.
20864 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20866         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20867         QImode checks.
20869 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20871         * df-core.c (df_find_def): Compare register numbers.
20872         (df_find_use): Likewise.
20874 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20876         PR target/56903
20877         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20878         lra_in_progress for return.
20880 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20882         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20883         define_insn into define_insn_and_split and emit movsicc patterns.
20885 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20887         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20889 2013-04-12  Richard Biener  <rguenther@suse.de>
20891         * tree-pass.h (TODO_do_not_ggc_collect): New.
20892         * passes.c (execute_one_ipa_transform_pass): Honor
20893         TODO_do_not_ggc_collect.
20894         (execute_one_pass): Likewise.
20896         Revert
20897         2013-04-10  Richard Biener  <rguenther@suse.de>
20899         * passes.c (init_optimization_passes): Remove reload pass.
20900         * ira.c (do_reload): Merge into ...
20901         (ira): ... this.
20902         (rest_of_handle_reload): Remove.
20903         (pass_reload): Likewise.
20904         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20905         of reload for vzeroupper pass placement.
20907 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20909         PR tree-optimization/56918
20910         PR tree-optimization/56920
20911         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20912         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20913         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20914         use wide_mul_with_sign method.
20916 2013-04-12  Richard Biener  <rguenther@suse.de>
20918         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20919         not be considered a gimple constant.
20921 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20923         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20924         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20925         operations.
20927 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20928             Jakub Jelinek  <jakub@redhat.com>
20930         * opts.c: Include diagnostic-color.h.
20931         (common_handle_option): Handle OPT_fdiagnostics_color_.
20932         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20933         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20934         (diagnostic-color.o): New.
20935         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20936         (diagnostic_color_rule): New enum.
20937         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20938         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20939         the location string.
20940         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20941         either NULL, or color kind.
20942         * diagnostic-color.c: New file.
20943         * diagnostic-color.h: New file.
20944         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20945         arguments.
20946         * doc/invoke.texi (-fdiagnostics-color): Document.
20947         * pretty-print.h (pp_show_color): Define.
20948         (struct pretty_print_info): Add show_color field.
20949         * diagnostic.c: Include diagnostic-color.h.
20950         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20951         macros.  Colorize error:, warning: etc. strings and also the location
20952         string.
20953         (diagnostic_show_locus): Colorize the caret line.
20954         * pretty-print.c: Include diagnostic-color.h.
20955         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20956         inside of %< %> quotes or quoted through q format modifier.
20958 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20960         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20962 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20964         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20965         code in CC_NZ mode.
20966         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20967         pattern.
20969 2013-04-11  Marek Polacek  <polacek@redhat.com>
20971         PR tree-optimization/48184
20972         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20974 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20976         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20977         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20978         (skip_simple_arithmetic): Tidy up.
20979         * tree.h (skip_simple_constant_arithmetic): Declare.
20981 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20983         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20985 2013-04-11  Richard Biener  <rguenther@suse.de>
20987         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20988         generate vector constants.
20990 2013-04-11  Richard Biener  <rguenther@suse.de>
20992         PR tree-optimization/56878
20993         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20994         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20995         New function.
20996         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20997         Prefer to align the DR with the most invariant base address.
20999 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
21001         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
21002         comment.
21004 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
21006         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
21007         floating-point vector comparisons against 0.
21009 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
21011         PR tree-optimization/56899
21012         * fold-const.c (extract_muldiv_1): Apply distributive law
21013         only if TYPE_OVERFLOW_WRAPS (ctype).
21015 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
21017         PR target/56124
21018         * ira-costs.c (scan_one_insn): Check whether the source rtx of
21019         loading has side effect.
21021 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
21023         * config/sparc/sparc.c: Include tree-pass.h.
21024         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
21025         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
21026         head of file.  Change return type.  Split off gate function.
21027         (sparc_gate_work_around_errata): New function.
21028         (pass_work_around_errata): New pass definition.
21029         (insert_pass_work_around_errata) New pass insert definition to
21030         insert pass_work_around_errata just after delayed-branch scheduling.
21031         (sparc_option_override): Insert the pass.
21032         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
21034 2013-04-10  David S. Miller  <davem@davemloft.net>
21036         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
21037         or -mcpu=hypersparc.
21039         * target.def (cstore_mode): New hook.
21040         * target.h: Include insn-codes.h
21041         * targhooks.c: Likewise.
21042         (default_cstore_mode): New function.
21043         * targhooks.h: Declare it.
21044         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
21045         * doc/tm.texi: Rebuild.
21046         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
21047         target hook, rather than inspecting the insn_data.
21048         * config/sparc/sparc.c (sparc_cstore_mode): New function.
21049         (TARGET_CSTORE_MODE): Redefine.
21050         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
21051         result patterns.
21052         * config/sparc/predicates.md (cstore_result_operand): New special
21053         predicate.
21054         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
21055         Use it for operand 0.
21056         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
21057         (*snesi_special): Likewise.
21058         (*snesi_zero): Likewise.
21059         (*seqsi_zero): Likewise.
21060         (*sltu_insn): Likewise.
21061         (*sgeu_insn): Likewise.
21062         (*seqdi_special): Make operand 0 and comparison operation be of
21063         DImode.
21064         (*snedi_special): Likewise.
21065         (*snedi_special_vis3): Likewise.
21066         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
21067         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
21068         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
21069         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
21070         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
21071         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
21072         (*sltu_extend_sp64): Likewise.
21073         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
21074         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
21075         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
21076         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
21077         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
21079 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
21081         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
21082         (aarch64_start_file): Use the new function.
21084 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
21085             Jason Merrill  <jason@redhat.com>
21087         * common.opt: Add -gdwarf.
21088         * opts.c (common_handle_option): Handle it.
21089         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
21091 2013-04-10  Richard Biener  <rguenther@suse.de>
21093         * passes.c (execute_todo): Do not call ggc_collect conditional here.
21094         (execute_one_ipa_transform_pass): But unconditionally here.
21095         (execute_one_pass): And here.
21096         (init_optimization_passes): Remove reload pass.
21097         * tree-pass.h (TODO_ggc_collect): Remove.
21098         (pass_reload): Likewise.
21099         * ira.c (do_reload): Merge into ...
21100         (ira): ... this.
21101         (rest_of_handle_reload): Remove.
21102         (pass_reload): Likewise.
21103         * config/i386/i386.c (ix86_option_override): Refer to ira instead
21104         of reload for vzeroupper pass placement.
21105         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
21106         and todo_flags_finish of all passes.
21108 2013-04-10  Richard Biener  <rguenther@suse.de>
21110         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
21111         first_const_oprnd field, rename first_def_type to first_op_type.
21112         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
21113         (vect_get_and_check_slp_defs): Always use the type of the
21114         operand.  Allow mixed vect_external_def, vect_constant_def types.
21115         (vect_get_constant_vectors): Handle mixed vect_external_def,
21116         vect_constant_def types.
21118 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
21120         PR tree-optimization/55524
21121         * tree-ssa-math-opts.c
21122         (convert_mult_to_fma): Don't use an fms construct
21123         when we don't have an fms operation, but fnma, and it looks
21124         likely that we'll be able to use the latter.
21126 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
21128         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
21129         function.
21130         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
21131         inline fail caused by overwritable functions.
21133 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
21135         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
21136         unnecessary bits in the constant power of two case.
21138 2013-04-10  Richard Biener  <rguenther@suse.de>
21140         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
21141         broken code swapping operands.
21142         (vect_build_slp_tree): Do not compute load permutations here.
21143         (vect_analyze_slp_instance): Compute load permutations here,
21144         after building the SLP tree.
21146 2013-04-09  Christian Bruel  <christian.bruel@st.com>
21148         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
21149         of next/prev_real_insn.
21151 2013-04-09  Jan Hubicka  <jh@suse.cz>
21153         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
21154         Drop aliased parameter.
21155         (function_and_variable_visibility): Do not handle alias pairs.
21156         * cgraph.c (varpool_externally_visible_p): Update prototype.
21157         * varpool.c (varpool_add_new_variable): Update.
21159 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21161         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
21163 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
21165         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
21167         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
21169 2013-04-09  Marek Polacek  <polacek@redhat.com>
21171         PR tree-optimization/48762
21172         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
21174 2013-04-09  Richard Biener  <rguenther@suse.de>
21176         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
21177         dealing with cost.
21178         (vect_build_slp_tree): Likewise.
21179         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
21180         calculating the cost of a SLP instance.
21181         (vect_analyze_slp_instance): Use it from here, after building
21182         the SLP tree.
21184 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
21186         PR middle-end/56883
21187         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
21188         expand_omp_for_static_chunk): Use simple_p = true in
21189         force_gimple_operand_gsi calls when assigning to addressable decls.
21191 2013-04-09  Jeff Law  <law@redhat.com>
21193         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
21194         when the boolean was created by converting a wider object which
21195         had a boolean range.
21197 2013-04-09  Richard Biener  <rguenther@suse.de>
21199         * tree-vectorizer.h (slp_void_p): Remove.
21200         (slp_tree): Typedef before _slp_tree declaration.
21201         (struct _slp_tree): Use a vector of slp_tree as children.
21202         (vect_get_place_in_interleaving_chain): Remove.
21203         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
21204         Move ...
21205         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
21206         and make static.
21207         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
21208         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
21209         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
21210         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
21211         Use slp_node instead of slp_void_p and adjust.
21213 2013-04-09  Richard Biener  <rguenther@suse.de>
21215         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
21216         work that is not necessary.
21218 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
21220         PR tree-optimization/56854
21221         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
21222         forward into clobber stmts if it would change MEM_REF lhs into
21223         non-MEM_REF.
21225 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
21227         * tree.c (type_hash_lookup, type_hash_add): Make static.
21228         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
21230 2013-04-09  Richard Biener  <rguenther@suse.de>
21232         * tree.h (unsave_expr_now): Remove.
21233         * tree-inline.c (mark_local_for_remap_r): Remove.
21234         (unsave_expr_1): Likewise.
21235         (unsave_r): Likewise.
21236         (unsave_expr_now): Likewise.
21237         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
21238         (propagate_tree_value): Likewise.
21240 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
21242         * doc/rtl.texi (sequence): Rewrite documentation to match the
21243         current use of SEQUENCE rtl objects.
21244         * rtl.def (SEQUENCE): Likewise.
21246         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
21247         Update documentation.
21248         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
21249         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
21251         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
21253 2013-04-08  Teresa Johnson  <tejohnson@google.com>
21255         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
21256         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
21257         methods.
21258         (estimate_edge_size_and_time): Add comment to suggest using rounding
21259         methods.
21260         (estimate_node_size_and_time): Ditto.
21261         (remap_edge_change_prob): Use helper rounding divide methods.
21262         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
21263         (gimple_mod_pow2_value_transform): Ditto.
21264         (gimple_mod_subtract_transform): Ditto.
21265         (gimple_ic_transform): Ditto.
21266         (gimple_stringops_transform): Ditto.
21267         * stmt.c (conditional_probability): Ditto.
21268         (emit_case_dispatch_table): Ditto.
21269         * lto-cgraph.c (merge_profile_summaries): Ditto.
21270         * tree-optimize.c (execute_fixup_cfg): Ditto.
21271         * cfgcleanup.c (try_forward_edges): Ditto.
21272         * cfgloopmanip.c (scale_loop_profile): Ditto.
21273         (loopify): Ditto.
21274         (duplicate_loop_to_header_edge): Ditto.
21275         (lv_adjust_loop_entry_edge): Ditto.
21276         * tree-vect-loop.c (vect_transform_loop): Ditto.
21277         * profile.c (compute_branch_probabilities): Ditto.
21278         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
21279         * lto-streamer-in.c (input_cfg): Ditto.
21280         * gimple-streamer-in.c (input_bb): Ditto.
21281         * ipa-cp.c (update_profiling_info): Ditto.
21282         (update_specialized_profile): Ditto.
21283         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
21284         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
21285         rounding methods.
21286         * sched-rgn.c (compute_dom_prob_ps): Ditto.
21287         (compute_trg_info): Ditto.
21288         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
21289         (purge_dead_edges): Ditto.
21290         * loop-unswitch.c (unswitch_loop): Ditto.
21291         * cgraphclones.c (cgraph_clone_edge): Ditto.
21292         (cgraph_clone_node): Ditto.
21293         * tree-inline.c (copy_bb): Ditto.
21294         (copy_edges_for_bb): Ditto.
21295         (initialize_cfun): Ditto.
21296         (copy_cfg_body): Ditto.
21297         (expand_call_inline): Ditto.
21299 2013-04-08  Kai Tietz  <ktietz@redhat.com>
21301         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
21302         TARGET_CYGWIN64 by TARGET_64BIT.
21304 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
21306         * config/epiphany/epiphany.md (GPR_1): New constant.
21307         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
21308         * config/epiphany/epiphany.c (gen_compare_reg):
21309         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
21310         is already in place.
21311         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
21312         Don't require being called during rtl expansion; If y operlaps r0,
21313         return 0.
21314         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
21315         (epiphany_expand_epilogue): Likewise.
21317         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
21318         Don't use CC_FPmode for ORDERED / UNORDERED.
21319         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
21321         * config/epiphany/constraints.md (CnL): New constraint.
21322         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
21323         * config/epiphany/predicates.md (add_operand): Allow 1024.
21325         * config/epiphany/epiphany.md (logical_op): New code iterator.
21326         (op_mnc): New code attribute.
21327         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
21328         (mov_f+1, mov_f+2): New peephole2 patterns.
21330         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
21331         (cstoresi4): Also allow re-use of zero result when doing a NE
21332         comparison to a non-zero operand.
21333         Use (clobber (scratch)) for first insn if the gpr output is not needed.
21335         * config/epiphany/epiphany.md (<insn_opname>v2si3):
21336         Use gen_addsi3_i / gen_subsi3_i.
21338 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
21340         PR c++/34949
21341         PR c++/50243
21342         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
21343         contain anything but clobbers, at most one __builtin_stack_restore,
21344         optionally debug stmts and final resx, and if it has at least one
21345         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
21346         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
21347         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
21348         which isn't defaut definition, remove them.
21349         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
21350         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
21351         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
21352         with MEM_REF LHS with SSA_NAME address.
21354 2013-04-08  Jeff Law  <law@redhat.com>
21356         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
21358 2013-04-08  Richard Biener  <rguenther@suse.de>
21360         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
21361         extra newline.
21362         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
21363         determined vector type.
21364         (vect_analyze_data_refs): Likewise.
21365         (vect_get_new_vect_var): Adjust.
21366         (vect_create_destination_var): Preserve SSA name versions.
21367         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
21368         not dump anything here.
21370 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
21372         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
21373         Add member lr_slot_known.
21374         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
21375         if necessary.
21376         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
21377         Remove code that sets lr_slot_offset according to what a previous
21378         version of epiphany_emit_save_restore used to do.
21379         (epiphany_emit_save_restore): When doing an lr save or restore,
21380         set/verify lr_slot_known and lr_slot_offset.
21382 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
21384         PR target/54338
21385         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
21386         in ALL_REGS.
21388 2013-04-08  Richard Biener  <rguenther@suse.de>
21390         * alias.c (find_base_term): Fix thinko in previous change.
21392 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
21394         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
21395         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
21396         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
21397         if possible to compute val.
21398         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
21399         For QImode integers don't require anything about precision.  Use
21400         const_with_all_bytes_same to find out if the constant doesn't have
21401         repeated bytes in it.
21403 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21405         * config/s390/s390.c (s390_expand_insv): Only accept insertions
21406         within mode size.
21408 2013-04-08  Marek Polacek  <polacek@redhat.com>
21410         PR rtl-optimization/48182
21411         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
21412         value to 1.
21414 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21416         PR target/55487
21417         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
21418         nuses, make sure we have a label.
21420 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
21422         PR target/56843
21423         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
21424         (rs6000_emit_swdiv_low_precision): Remove.
21425         (rs6000_emit_swdiv): Rewrite to handle between one and four
21426         iterations of Newton-Raphson generally; modify required number of
21427         iterations for some cases.
21428         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
21430 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
21432         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
21433         set-but-unused variable.
21435         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
21436         basic blocks of released function bodies garbage-collectable.
21438         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
21439         (struct rtl_opt_pass): Add TODO_df_finish.
21441         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
21443 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21445         * config/arm/constraints.md (q): New constraint.
21446         * config/arm/ldrdstrd.md: New file.
21447         * config/arm/arm.md (ldrdstrd.md) New include.
21448         (arm_movdi): Use "q" instead of "r" constraint
21449         for double-word memory access.
21450         (movdf_soft_insn): Likewise.
21451         * config/arm/vfp.md (movdi_vfp): Likewise.
21452         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
21453         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
21454         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
21455         (mem_ok_for_ldrd_strd): Likewise.
21456         (output_move_double): Update assertion.
21458 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21460         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
21462 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21464         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
21465         define_insn_and_split.
21466         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
21468 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21470         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
21471         define_insn_and_split.
21472         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
21473         (shiftsi3_compare): New pattern.
21474         (rrx): New pattern.
21475         * config/arm/unspecs.md (UNSPEC_RRX): New.
21477 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21479         * config/arm/arm.md (negdi_extendsidi): New pattern.
21480         (negdi_zero_extendsidi): Likewise.
21482 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21484         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
21485         define_insn_and_split.
21486         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
21487         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
21489 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21491         * config/arm/arm.md (arm_subdi3): Convert define_insn into
21492         define_insn_and_split.
21493         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
21494         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
21496 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21498         * config/arm/arm.md (subsi3_carryin): New pattern.
21499         (subsi3_carryin_const): Likewise.
21500         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
21501         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
21503 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21505         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
21507 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
21509         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
21510         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
21512 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21514         * config/arm/arm.c (arm_expand_builtin): Change fcode
21515         type to unsigned int.
21517 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21519         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
21521 2013-04-04  Ian Lance Taylor  <iant@google.com>
21523         * doc/standards.texi (Standards): The Go frontend supports the Go 1
21524         language standard.
21526 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
21528         PR middle-end/56729
21529         * df-scan.c (df_insn_delete): Disable failing assert.
21531 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21533         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
21534         New function prototype.
21535         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
21536         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
21537         (arm_builtin_vectorized_function): New function.
21539 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21541         * config/arm/arm_neon_builtins.def: New file.
21542         * config/arm/arm.c (neon_builtin_data): Move contents to
21543         arm_neon_builtins.def.
21544         (enum arm_builtins): Include neon builtin definitions.
21545         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
21546         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
21548 2013-04-04  Marek Polacek  <polacek@redhat.com>
21550         PR tree-optimization/48186
21551         * predict.c (maybe_hot_frequency_p): Return false if
21552         HOT_BB_FREQUENCY_FRACTION is 0.
21553         (cgraph_maybe_hot_edge_p): Likewise.
21555 2013-04-04  Richard Biener  <rguenther@suse.de>
21557         PR tree-optimization/56826
21558         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
21559         more accurately.
21561 2013-04-04  Richard Biener  <rguenther@suse.de>
21563         PR tree-optimization/56213
21564         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
21565         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
21567 2013-04-04  Richard Biener  <rguenther@suse.de>
21569         PR tree-optimization/56837
21570         * tree-loop-distribution.c (classify_partition): For non-zero
21571         values require that the value has the same precision as its
21572         mode to be useful as memset value.
21574 2013-04-03  Nick Clifton  <nickc@redhat.com>
21576         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
21577         (fmssf4): Use fmsf.s on E3V5 architectures.
21578         (fnmasf4): Use fnmaf.s on E3V5 architectures.
21579         (fnmssf4): Use fnmsf.s on E3V5 architectures.
21581 2013-04-03  Jeff Law  <law@redhat.com>
21583         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
21584         (lra-eliminations.o): Likewise.
21586 2013-04-03  Teresa Johnson  <tejohnson@google.com>
21588         * gcov-io.c (compute_working_sets): Moved most of body of old
21589         compute_working_sets here from profile.c.
21590         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
21591         (gcov_working_set_t): Moved typedef here from basic-block.h
21592         (compute_working_set): Declare.
21593         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
21594         (get_working_sets): Renamed from compute_working_set,
21595         replace most of body with call to new compute_working_sets.
21596         (get_exec_counts): Replace call to compute_working_sets
21597         to get_working_sets.
21598         * profile.h (get_working_sets): Renamed from compute_working_set.
21599         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
21600         to get_working_sets.
21601         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
21602         * gcov-dump.c (dump_working_sets): New function.
21604 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
21606         * hwint.c (sext_hwi, zext_hwi): New functions.
21607         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
21608         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
21609         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
21610         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
21611         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
21612         (sext_hwi, zext_hwi): New functions.
21614 2013-04-03  Jeff Law  <law@redhat.com>
21616         PR tree-optimization/56799
21617         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
21618         back test for widening conversion erroneously dropped in prior change.
21620 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21622         PR target/56809
21623         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
21624         instead of next_real_insn.
21626 2013-04-03  Marek Polacek  <polacek@redhat.com>
21628         PR sanitizer/55702
21629         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
21631 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21633         PR target/56809
21634         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
21635         next_real_insn.
21636         (thumb1_output_casesi): Likewise.
21637         (thumb2_output_casesi): Likewise.
21639 2013-04-03  Richard Biener  <rguenther@suse.de>
21641         PR tree-optimization/56817
21642         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21643         Split out ...
21644         (tree_unroll_loops_completely_1): ... new function to manually
21645         walk the loop tree, properly defering outer loops of unrolled
21646         loops to later iterations.
21648 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21650         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21651         (vectorizable_load): Likewise.
21652         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21653         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21655 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21657         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21658         BIT_FIELD_REF.
21660 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21662         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21664 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21666         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21668 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21670         PR tree-optimization/56790
21671         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21672         folding.
21674 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21676         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21677         Handle VEC_MERGE.
21678         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21679         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21680         equal arguments.
21682 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21684         PR c/19449
21685         * tree.h (force_folding_builtin_constant_p): New decl.
21686         * builtins.c (force_folding_builtin_constant_p): New variable.
21687         (fold_builtin_constant_p): Fold immediately also if
21688         force_folding_builtin_constant_p.
21690 2013-04-03  Richard Biener  <rguenther@suse.de>
21692         PR tree-optimization/56812
21693         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21694         DRs of the same interleaving chain are independent.
21696 2013-04-02  Jason Merrill  <jason@redhat.com>
21698         * gdbinit.in (pbb): Use debug fn.
21700 2013-04-02  Lawrence Crowl  <crowl@google.com>
21702         * sese.h (struct ivtype_map_elt_s): Remove unused.
21703         (extern debug_ivtype_map): Remove unused.
21704         (extern eq_ivtype_map_elts): Remove unused.
21705         * sese.c (debug_ivtype_map): Removed unused.
21706         (debug_ivtype_map_1): Removed unused.
21707         (debug_ivtype_elt): Remove unused.
21708         (eq_ivtype_map_elts): Remove unused.
21711 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21713         PR target/52790
21714         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21715         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21716         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21717         function.
21718         (legitimize_pe_coff_symbol): Likewise.
21719         (is_imported_p): New helper-function.
21720         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21721         for Windows x64 targets.
21722         (ix86_expand_prologue): Optimize for pe-coff targets.
21723         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21724         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21725         medium/large code-model.
21726         (legitimize_pic_address): Likewise.
21727         (legitimize_tls_address): Likewise.
21728         (ix86_expand_call): Likewise.
21729         (x86_output_mi_thunk): Likewise.
21730         (get_dllimport_decl): Add new beimport argument.
21731         (construct_plt_address): Don't assert for x64 pe-coff targets.
21732         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21733         targets.
21734         (SYMBOL_FLAG_STUBVAR): New macro.
21735         (SYMBOL_REF_STUBVAR_P): Likewise.
21736         * config/i386/winnt.c (stub_list): New structure.
21737         (stub_head): New local variable.
21738         (i386_pe_record_stub): New function.
21739         (i386_pe_file_end): Emit refptr-stubs.
21741 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21743         PR rtl-optimization/56745
21744         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21745         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21747         PR c++/34949
21748         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21749         and both of them are MEM_REFs, just compare first argument for
21750         equality and attempt to deal even with differing offsets.
21752         PR c++/34949
21753         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21754         of gimple_clobber_p to be MEM_REF.
21755         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21756         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21757         after gimplification.
21758         * asan.c (get_mem_ref_of_assignment): Don't instrument
21759         gimple_clobber_p stmts.
21760         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21761         gimple_clobber_p stmt if they have MEM_REF lhs and
21762         are dead because of another gimple_clobber_p stmt.
21763         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21764         gimple_clobber_p stmts like debug stmts.
21765         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21766         that refer to unused VAR_DECLs or uninitialized values.
21767         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21768         gimple_clobber_p stmts if they refer to removed parameters.
21769         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21770         formatting.
21772 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21774         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21775         using SWI48 mode attribute.
21777 2013-04-02  Wei Mi  <wmi@google.com>
21779         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21780         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21781         *<rotate_insn><mode>3_mask in i386.md.
21783 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21785         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21787 2013-04-02  Richard Biener  <rguenther@suse.de>
21789         PR tree-optimization/56778
21790         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21791         Runtime alias tests are not supported for gather loads.
21792         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21793         stmts referenced from SSA operands before updating SSA form.
21795 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21796             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21798         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21799         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21800         * config/arm/cortex-a53.md: New file.
21801         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21802         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21803         * config/arm/arm.c (arm_issue_rate): Likewise.
21804         * config/arm/arm-tune.md: Regenerate
21805         * config/arm/arm-tables.opt: Regenerate.
21806         * config/arm/arm-cores.def: Add cortex-a53.
21808 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21810         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21811         non-static link.
21813 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21815         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21816         scalar load/store operations using B/H registers.
21817         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21819 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21821         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21822         scalar move.
21823         * config/aarch64/aarch64.c
21824         (aarch64_simd_scalar_immediate_valid_for_move): New.
21825         * config/aarch64/aarch64-protos.h
21826         (aarch64_simd_scalar_immediate_valid_for_move): New.
21827         * config/aarch64/constraints.md (Dh, Dq): New.
21828         * config/aarch64/iterators.md (hq): New.
21830 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21832         * reorg.c (get_branch_condition): Deal with conditional returns.
21833         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21835 2013-04-01  Wei Mi  <wmi@google.com>
21837         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21838         Truncate operand 2 using %b asm operand modifier.
21839         (*<shift_insn><mode>3_mask): Ditto.
21840         (*<rotate_insn><mode>3_mask): Ditto.
21842 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21844         PR middle-end/56798
21845         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21847 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21849         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21850         of next_real_insn.
21851         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21853 2013-03-30  Lawrence Crowl  <crowl@google.com>
21855         * dse.c (clear_alias_sets): Remove never set.
21856         (disqualified_clear_alias_sets): Remove never set.
21857         (clear_alias_mode_pool): Remove never set.
21858         (dse_step0): Remove condition that is never true.
21859         (canon_address): Remove condition that is never true.
21860         (dse_step7): Remove condition that is never true.
21861         (rest_of_handle_dse): Remove condition that is never true.
21862         (rest_of_handle_dse::did_global): Remove never read from above.
21863         (dse_step2_spill): Remove never called from above.
21864         (dse_step5_spill): Remove never called from above.
21866 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21868         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21869         JUMP_TABLE_DATA changes.
21870         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21871         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21872         (Insns) <jump_table_data>: New entry.
21873         * doc/tm.texi: Regenerate.
21875         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21877         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21878         for table jump at the end of a basic block using tablejump_p.
21879         * targhooks.c (default_invalid_within_doloop): Likewise.
21880         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21881         target hook implementation that is identical to the default hook.
21882         (rs6000_invalid_within_doloop): Remove.
21884         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21885         unused variable from tablejump_p call.
21887         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21888         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21889         (INSN_DELETED_P): Likewise.
21890         (emit_jump_table_data): New prototype.
21891         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21892         after 4th as unused.
21893         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21894         * sched-vis.c (print_insn): Likewise.
21895         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21896         insn for compatibility with back ends that use next_active_insn to
21897         identify jump table data.
21898         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21899         (remove_insn): Likewise.
21900         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21901         to be emitted.
21902         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21903         (emit_jump_table_data): New function.
21905         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21906         basic block, a JUMP_TABLE_DATA never is.
21907         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21908         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21909         off from code handling real insns.
21910         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21911         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21912         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21913         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21914         is not a NONDEBUG_INSN_P.
21915         * ira-costs.c (scan_one_insn): Likewise.
21916         * jump.c (mark_all_labels): Likewise.
21917         (mark_jump_label_1): Likewise.
21918         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21919         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21920         a basic block.
21921         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21922         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21923         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21924         JUMP_TABLE_DATA_P insns.
21925         (calculate_elim_costs_all_insns): Likewise.
21926         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21927         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21928         (delete_output_reload): Code style fixups.
21929         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21930         insn flags on this non-insn.
21931         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21932         as scheduling barriers, for pre-change compatibility.
21933         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21934         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21936         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21937         redundant JUMP_TABLE_DATA_P test.
21938         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21939         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21940         (frv_for_each_packet): Likewise.
21941         * config/i386/i386.c (min_insn_size): Likewise.
21942         (ix86_avoid_jump_mispredicts): Likewise.
21943         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21944         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21945         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21946         (mips16_insn_length): Robustify.
21947         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21948         (mips16_split_long_branches): Likewise.
21949         * config/pa/pa.c (pa_combine_instructions): Likewise.
21950         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21951         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21952         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21953         as contributing to pool range lengths.
21954         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21955         Remove redundant JUMP_TABLE_DATA_P test.
21956         (sh_loop_align): Likewise.
21957         (split_branches): Likewise.
21958         (sh_insn_length_adjustment): Likewise.
21959         * config/spu/spu.c (get_branch_target): Likewise.
21961 2013-03-29  Jan Hubicka  <jh@suse.cz>
21963         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21964         gcov streaming; stream hot bb threshold to ltrans.
21965         * predict.c (get_hot_bb_threshold): Break out from ....
21966         (maybe_hot_count_p): ... here.
21967         (set_hot_bb_threshold): New function.
21968         * lto-section-in.c (lto_section_name): Add profile.
21969         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21970         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21971         and data-streamer.h
21972         (histogram_entry): New structure.
21973         (histogram, histogram_pool): New global vars.
21974         (histogram_hash): New structure.
21975         (histogram_hash::hash): New method.
21976         (histogram_hash::equal): Likewise.
21977         (account_time_size): New function.
21978         (cmp_counts): New function.
21979         (dump_histogram): New function.
21980         (ipa_profile_generate_summary): New function.
21981         (ipa_profile_write_summary): New function.
21982         (ipa_profile_read_summary): New function.
21983         (ipa_profile): Decide on threshold.
21984         (pass_ipa_profile): Add ipa_profile_write_summary and
21985         ipa_profile_read_summary.
21986         * Makefile.in (ipa.o): Update dependencies.
21987         * lto-streamer.h (LTO_section_ipa_profile): New section.
21989 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21991         * tree.h (VAR_P): New.
21993 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21995         PR lto/56777
21996         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21998 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
22000         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
22001         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
22002         (control_flow_insn_p): Likewise.
22003         * cfgrtl.c (duplicate_insn_chain): Likewise.
22004         * final.c (get_attr_length_1): Likewise.
22005         (shorten_branches): Likewise.
22006         (final_scan_insn): Likewise.
22007         * function.c (instantiate_virtual_regs): Likewise.
22008         * gcse.c (insert_insn_end_basic_block): Likewise.
22009         * ira-costs.c (scan_one_insn): Likewise.
22010         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
22011         * lra.c (check_rtl): Likewise.
22012         * reload1.c (elimination_costs_in_insn): Likewise.
22013         * reorg.c (follow_jumps): Likewise.
22015         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
22016         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
22017         (thumb_far_jump_used_p): Likewise.
22018         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
22019         (workaround_speculation): Likewise.
22020         (add_sched_insns_for_speculation): Likewise.
22021         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
22022         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
22023         (frv_for_each_packet): Likewise.
22024         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
22025         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
22026         (final_emit_insn_group_barriers): Likewise.
22027         * config/m32r/m32r.c (m32r_is_insn): Likewise.
22028         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
22029         (mips16_insn_length): Likewise.
22030         * config/pa/pa.c (pa_reorg): Likewise.
22031         (pa_combine_instructions): Likewise.
22032         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
22033         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
22034         (sh_reorg): Likewise.
22035         (split_branches): Likewise.
22036         * config/spu/spu.c (get_branch_target): Likewise.
22038         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
22039         JUMP_TABLE_DATA_P.
22041 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
22043         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
22044         Fix declaration name.
22046 2013-03-28  Lawrence Crowl  <crowl@google.com>
22048         * graphds.h (struct graph.indicies): Remove unused.
22049         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
22050         (SCOP_ORIGINAL_PDDRS): Remove unused.
22051         * sese.h (extern insert_loop_close_phis): Removed unused.
22052         (extern insert_guard_phis): Removed unused.
22053         (extern ivtype_map_elt_info): Removed unused.
22054         (new_ivtype_map_elt): Removed unused.
22055         * sese.c (ivtype_map_elt_info): Removed unused.
22057 2013-03-28  Lawrence Crowl  <crowl@google.com>
22059         * Makefile.in: Add several missing include dependences.
22060         (DUMPFILE_H): New.
22061         (test-dump.o): New.  This object is not added to any executable,
22062         but is present for ad-hoc testing.
22063         * bitmap.c
22064         (debug (const bitmap_head_def &)): New.
22065         (debug (const bitmap_head_def *)): New.
22066         * bitmap.h
22067         (extern debug (const bitmap_head_def &)): New.
22068         (extern debug (const bitmap_head_def *)): New.
22069         * cfg.c
22070         (debug (edge_def &)): New.
22071         (debug (edge_def *)): New.
22072         * cfghooks.c
22073         (debug (basic_block_def &)): New.
22074         (debug (basic_block_def *)): New.
22075         * dumpfile.h
22076         (dump_node (const_tree, int, FILE *)): Correct source file.
22077         * dwarf2out.c
22078         (debug (die_struct &)): New.
22079         (debug (die_struct *)): New.
22080         * dwarf2out.h
22081         (extern debug (die_struct &)): New.
22082         (extern debug (die_struct *)): New.
22083         * gimple-pretty-print.c
22084         (debug (gimple_statement_d &)): New.
22085         (debug (gimple_statement_d *)): New.
22086         * gimple-pretty-print.h
22087         (extern debug (gimple_statement_d &)): New.
22088         (extern debug (gimple_statement_d *)): New.
22089         * ira-build.c
22090         (debug (ira_allocno_copy &)): New.
22091         (debug (ira_allocno_copy *)): New.
22092         (debug (ira_allocno &)): New.
22093         (debug (ira_allocno *)): New.
22094         * ira-int.h
22095         (extern debug (ira_allocno_copy &)): New.
22096         (extern debug (ira_allocno_copy *)): New.
22097         (extern debug (ira_allocno &)): New.
22098         (extern debug (ira_allocno *)): New.
22099         * ira-lives.c
22100         (debug (live_range &)): New.
22101         (debug (live_range *)): New.
22102         * lra-int.h
22103         (debug (lra_live_range &)): New.
22104         (debug (lra_live_range *)): New.
22105         * lra-lives.c
22106         (debug (lra_live_range &)): New.
22107         (debug (lra_live_range *)): New.
22108         * omega.c
22109         (debug (omega_pb_d &)): New.
22110         (debug (omega_pb_d *)): New.
22111         * omega.h
22112         (extern debug (omega_pb_d &)): New.
22113         (extern debug (omega_pb_d *)): New.
22114         * print-rtl.c
22115         (debug (const rtx_def &)): New.
22116         (debug (const rtx_def *)): New.
22117         * print-tree.c
22118         (debug_tree (tree): Move within file.
22119         (debug_raw (const tree_node &)): New.
22120         (debug_raw (const tree_node *)): New.
22121         (dump_tree_via_hooks (const tree_node *, int)): New.
22122         (debug (const tree_node &)): New.
22123         (debug (const tree_node *)): New.
22124         (debug_verbose (const tree_node &)): New.
22125         (debug_verbose (const tree_node *)): New.
22126         (debug_head (const tree_node &)): New.
22127         (debug_head (const tree_node *)): New.
22128         (debug_body (const tree_node &)): New.
22129         (debug_body (const tree_node *)): New.
22130         (debug_vec_tree (tree): Move and reimplement in terms of dump.
22131         (debug (vec<tree, va_gc> &)): New.
22132         (debug (vec<tree, va_gc> *)): New.
22133         * rtl.h
22134         (extern debug (const rtx_def &)): New.
22135         (extern debug (const rtx_def *)): New.
22136         * sbitmap.c
22137         (debug_raw (simple_bitmap_def &)): New.
22138         (debug_raw (simple_bitmap_def *)): New.
22139         (debug (simple_bitmap_def &)): New.
22140         (debug (simple_bitmap_def *)): New.
22141         * sbitmap.h
22142         (extern debug (simple_bitmap_def &)): New.
22143         (extern debug (simple_bitmap_def *)): New.
22144         (extern debug_raw (simple_bitmap_def &)): New.
22145         (extern debug_raw (simple_bitmap_def *)): New.
22146         * sel-sched-dump.c
22147         (debug (vinsn_def &)): New.
22148         (debug (vinsn_def *)): New.
22149         (debug_verbose (vinsn_def &)): New.
22150         (debug_verbose (vinsn_def *)): New.
22151         (debug (expr_def &)): New.
22152         (debug (expr_def *)): New.
22153         (debug_verbose (expr_def &)): New.
22154         (debug_verbose (expr_def *)): New.
22155         (debug (vec<rtx> &)): New.
22156         (debug (vec<rtx> *)): New.
22157         * sel-sched-dump.h
22158         (extern debug (vinsn_def &)): New.
22159         (extern debug (vinsn_def *)): New.
22160         (extern debug_verbose (vinsn_def &)): New.
22161         (extern debug_verbose (vinsn_def *)): New.
22162         (extern debug (expr_def &)): New.
22163         (extern debug (expr_def *)): New.
22164         (extern debug_verbose (expr_def &)): New.
22165         (extern debug_verbose (expr_def *)): New.
22166         (extern debug (vec<rtx> &)): New.
22167         (extern debug (vec<rtx> *)): New.
22168         * sel-sched-ir.h
22169         (_list_iter_cond_expr): Make inline instead of static.
22170         * sreal.c
22171         (debug (sreal &)): New.
22172         (debug (sreal *)): New.
22173         * sreal.h
22174         (extern debug (sreal &)): New.
22175         (extern debug (sreal *)): New.
22176         * tree.h
22177         (extern debug_raw (const tree_node &)): New.
22178         (extern debug_raw (const tree_node *)): New.
22179         (extern debug (const tree_node &)): New.
22180         (extern debug (const tree_node *)): New.
22181         (extern debug_verbose (const tree_node &)): New.
22182         (extern debug_verbose (const tree_node *)): New.
22183         (extern debug_head (const tree_node &)): New.
22184         (extern debug_head (const tree_node *)): New.
22185         (extern debug_body (const tree_node &)): New.
22186         (extern debug_body (const tree_node *)): New.
22187         (extern debug (vec<tree, va_gc> &)): New.
22188         (extern debug (vec<tree, va_gc> *)): New.
22189         * tree-cfg.c
22190         (debug (struct loop &)): New.
22191         (debug (struct loop *)): New.
22192         (debug_verbose (struct loop &)): New.
22193         (debug_verbose (struct loop *)): New.
22194         * tree-dump.c: Add header dependence.
22195         * tree-flow.h
22196         (extern debug (struct loop &)): New.
22197         (extern debug (struct loop *)): New.
22198         (extern debug_verbose (struct loop &)): New.
22199         (extern debug_verbose (struct loop *)): New.
22200         * tree-data-ref.c
22201         (debug (data_reference &)): New.
22202         (debug (data_reference *)): New.
22203         (debug (vec<data_reference_p> &)): New.
22204         (debug (vec<data_reference_p> *)): New.
22205         (debug (vec<ddr_p> &)): New.
22206         (debug (vec<ddr_p> *)): New.
22207         * tree-data-ref.h
22208         (extern debug (data_reference &)): New.
22209         (extern debug (data_reference *)): New.
22210         (extern debug (vec<data_reference_p> &)): New.
22211         (extern debug (vec<data_reference_p> *)): New.
22212         (extern debug (vec<ddr_p> &)): New.
22213         (extern debug (vec<ddr_p> *)): New.
22214         * tree-ssa-alias.c
22215         (debug (pt_solution &)): New.
22216         (debug (pt_solution *)): New.
22217         * tree-ssa-alias.h
22218         (extern debug (pt_solution &)): New.
22219         (extern debug (pt_solution *)): New.
22220         * tree-ssa-alias.c
22221         (debug (_var_map &)): New.
22222         (debug (_var_map *)): New.
22223         (debug (tree_live_info_d &)): New.
22224         (debug (tree_live_info_d *)): New.
22225         * tree-ssa-alias.h
22226         (extern debug (_var_map &)): New.
22227         (extern debug (_var_map *)): New.
22228         (extern debug (tree_live_info_d &)): New.
22229         (extern debug (tree_live_info_d *)): New.
22231 2013-03-28  Jan Hubicka  <jh@suse.cz>
22233         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
22235 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
22237         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
22238         record only when desired or required.
22240 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
22242         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
22243         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
22245 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22247         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
22248         (*andsi3_compare0_uxtw): New pattern.
22249         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
22250         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
22252 2013-03-28  Jan Hubicka  <jh@suse.cz>
22254         * data-streamer-in.c (streamer_read_gcov_count): New function.
22255         * gimple-streamer-out.c: Include value-prof.h.
22256         (output_gimple_stmt): Output histogram.
22257         (output_bb): Use streamer_write_gcov_count.
22258         * value-prof.c: Include data-streamer.h
22259         (dump_histogram_value): Add HIST_TYPE_MAX.
22260         (stream_out_histogram_value): New function.
22261         (stream_in_histogram_value): New function.
22262         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
22263         (stream_out_histogram_value, stream_in_histogram_value): Declare.
22264         * data-streamer-out.c (streamer_write_gcov_count): New function.
22265         (streamer_write_gcov_count_stream): New function.
22266         * lto-cgraph.c (lto_output_edge): Update counter streaming.
22267         (lto_output_node): Likewise.
22268         (input_node, input_edge): Likewise.
22269         * lto-streamer-out.c (output_cfg): Update streaming.
22270         * lto-streamer-in.c (input_cfg): Likewise.
22271         * data-streamer.h (streamer_write_gcov_count,
22272         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
22273         * gimple-streamer-in.c: Include value-prof.h
22274         (input_gimple_stmt): Input histograms.
22275         (input_bb): Update profile streaming.
22277 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
22279         * genmodes.c (emit_max_int): New function.
22280         (emit_insn_modes_h): Added call to emit_max_function.
22281         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
22282         Added doc.
22283         * machmode.def: Fixed comment.
22285 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
22287         * combine.c (try_combine): Removed useless assert.
22288         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
22290 2013-03-28  Marek Polacek  <polacek@redhat.com>
22291             Richard Biener  <rguenther@suse.de>
22293         PR tree-optimization/56695
22294         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
22295         build signed result of a vector comparison.
22296         * tree-cfg.c (verify_gimple_comparison): Check that a result
22297         of a vector comparison has signed type.
22299 2013-03-28  Richard Biener  <rguenther@suse.de>
22301         PR tree-optimization/37021
22302         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
22303         do not restrict gaps between groups.
22304         * tree-vect-stmts.c (vectorizable_load): Properly account for
22305         a gap between groups.
22307 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
22309         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
22310         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
22311         is not enabled.
22313 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
22315         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
22316         * doc/extend.texi (Named Address Spaces): Ditto.
22317         (Variable Attributes): Ditto.
22319 2013-03-27  Kai Tietz  <ktietz@redhat.com>
22321         * config.build: Add support for cygwin x64 target.
22322         * config.gcc: Likewise.
22323         * config.host: Likewise.
22324         * configure.ac: Likewise
22325         * configure: Regenerated.
22327 2013-03-27  Kai Tietz  <ktietz@redhat.com>
22329         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
22330         * config/i386/t-cygwin-w64: New file.
22331         * config/i386/cygwin-w64.h: New file.
22332         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
22333         and add support for x64-cygwin target.
22334         (CPP_SPEC): Likewise.
22335         (CXX_WRAP_SPEC_LIST): Undefine before define.
22336         (LIBGCJ_SONAME): Use 15 as version.
22338 2013-03-27  Richard Biener  <rguenther@suse.de>
22340         PR tree-optimization/56716
22341         * tree-ssa-structalias.c (perform_var_substitution): Adjust
22342         dumping for ref nodes.
22344 2013-03-27  Martin Jambor  <mjambor@suse.cz>
22346         PR tree-optimization/55334
22347         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
22348         restricted pointers to arrays.
22350 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
22352         * Makefile.in (.SUFFIXES): Add .cc.
22353         (.c.o): Apply same recipe for implicit rule .cc.o.
22355 2013-03-27  Richard Biener  <rguenther@suse.de>
22357         PR tree-optimization/37021
22358         * tree-vect-data-refs.c (vect_check_strided_load): Allow
22359         REALPART/IMAGPART_EXPRs around the supported refs.
22360         * tree-ssa-structalias.c (find_func_aliases): Assume that
22361         floating-point values are not used to transfer pointers.
22363 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
22365         * target.def (TARGET_HAS_IFUNC_P): New target hook.
22366         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
22367         * doc/tm.texi: Regenerate.
22368         * targhooks.h (default_has_ifunc_p): New.
22369         * targhooks.c (default_has_ifunc_p): Ditto.
22370         * config/linux-protos.h: New file.
22371         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
22372         hook for linux which disables support of indirect functions in android.
22373         * config/linux-android.c: New file.
22374         * config/t-linux-android.c: Ditto.
22375         * config.gcc: Added new object file linux-android.o.
22376         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
22377         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
22378         * varasm.c (do_assemble_alias): Likewise.
22379         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
22380         doesn't support indirect functions.
22381         * configure: Regenerate.
22383 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
22385         PR target/56102
22386         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
22387         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
22388         mult-word mode.
22390 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22392         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
22394 2013-03-27  Terry Guo  <terry.guo@arm.com>
22396         * config/arm/arm-cores.def: Added core cortex-r7.
22397         * config/arm/arm-tune.md: Regenerated.
22398         * config/arm/arm-tables.opt: Regenerated.
22399         * doc/invoke.texi: Added entry for core cortex-r7.
22401 2013-03-27  Walter Lee  <walt@tilera.com>
22403         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
22404         double-decrement of next_scratch_regno.
22406 2013-03-27  Walter Lee  <walt@tilera.com>
22408         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
22409         input operands.
22410         (insn_v1mulus): Ditto.
22411         (insn_v2muls): Ditto.
22413 2013-03-27  Walter Lee  <walt@tilera.com>
22415         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
22416         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
22418 2013-03-27  Walter Lee  <walt@tilera.com>
22420         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
22421         (*sibcall_value): Ditto.
22423 2013-03-27  Walter Lee  <walt@tilera.com>
22425         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
22426         (insn_mnz_v8qi): ... this ...
22427         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
22428         vector equivalent.
22429         (insn_v<n>mnz): Replaced by ...
22430         (insn_v1mnz): ... this ...
22431         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
22432         equivalent.
22433         (insn_mz_<mode>): Replaced by ...
22434         (insn_mz_v8qi): ... this ...
22435         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
22436         vector equivalent.
22437         (insn_v<n>mz): Replaced by ...
22438         (insn_v1mz): ... this ...
22439         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
22440         equivalent.
22442 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
22444         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
22446 2013-03-26  Roland McGrath  <mcgrathr@google.com>
22448         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
22449         than fprintf with a non-constant, non-format string.
22451 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
22453         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
22454         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
22455         operand 0 predicate.
22456         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
22457         attribute.  Use general_x64nomem_operand as operand 1 predicate.
22458         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
22459         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
22460         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
22461         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
22462         (mov<mode>_insv_1): Remove expander.  Merge insn with
22463         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
22464         Use general_x64nomem_operand as operand 1 predicate.
22465         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
22466         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
22467         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
22468         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
22469         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
22470         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
22471         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
22472         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
22473         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
22474         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
22475         (general_x64nomem_operand): Ditto.
22477 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
22479         * config/rtems.opt: Add -pthread option.
22481 2013-03-26  Richard Biener  <rguenther@suse.de>
22483         * alias.c (find_base_term): Avoid redundant and not used recursion.
22484         (base_alias_check): Get the initial base term from the caller.
22485         (true_dependence_1): Compute and pass base terms to base_alias_check.
22486         (write_dependence_p): Likewise.
22487         (may_alias_p): Likewise.
22489 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
22491         * config/aarch64/aarch64.c (aarch64_classify_address): Support
22492         PC-relative load in SI modes and above only.
22494 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
22496         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
22497         * config/arm/iwmmxt.md (WCGR0): Update.
22498         (WCGR1, WCGR2, WCGR3): Likewise.
22500 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
22502         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
22503         Use x64 and nox64 isa attributes.
22505 2013-03-26  Richard Biener  <rguenther@suse.de>
22507         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
22508         alignment computations and rely on get_object_alignment_1
22509         for the !TYPE_P case.
22510         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
22512 2013-03-26  Walter Lee  <walt@tilera.com>
22514         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
22515         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
22517 2013-03-25  Jeff Law  <law@redhat.com>
22519         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
22520         check for INTEGRAL_TYPE_P that was missing due to checking in
22521         wrong version of prior patch.
22523 2013-03-25  Walter Lee  <walt@tilera.com>
22525         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
22526         TILEGX_INSN_SHUFFLEBYTES1.
22527         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
22528         shufflebytes1.
22529         (tilegx_builtins): Ditto.
22530         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
22532 2013-03-25  Walter Lee  <walt@tilera.com>
22534         * config/tilegx/tilegx.md (floatsisf2): New pattern.
22535         (floatunssisf2): New pattern.
22536         (floatsidf2): New pattern.
22537         (floatunssidf2): New pattern.
22539 2013-03-25  Walter Lee  <walt@tilera.com>
22541         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
22542         tests for constraint J, K, N, P.
22544 2013-03-25  Walter Lee  <walt@tilera.com>
22546         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
22547         Use indirect/pcrel encoding.
22548         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
22549         Ditto.
22551 2013-03-25  Steve Ellcey  <sellcey@mips.com>
22553         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
22554         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
22555         * config/mips/mips.c (mips_option_override): Set IMADD default.
22556         * config/mips/mips.h (PTF_AVOID_IMADD): New.
22557         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
22558         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
22559         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
22560         * doc/invoke.texi (-mimadd/-mno-imadd): New.
22562 2013-03-25  Jeff Law  <law@redhat.com>
22564         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
22565         slightly to avoid creating and folding useless trees.  Simplify
22566         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
22568 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
22570         * config/i386/i386.md (*zero_extendsidi2): Merge with
22571         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
22572         * config/i386/predicates.md (x86_64_zext_operand): Rename from
22573         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
22574         targets.  Clarify comment.
22576 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22578         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
22579         pass-through jump functions differently.
22580         (ipa_read_jump_function): Likewise.  Also use setter functions to set
22581         up jump functions.
22583 2013-03-25  Martin Jambor  <mjambor@suse.cz>
22585         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
22586         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
22587         process it.
22588         (ipa_get_indirect_edge_target): New function.
22589         (devirtualization_time_bonus): New parameter known_aggs, pass it to
22590         ipa_get_indirect_edge_target.  Update all callers.
22591         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
22592         ipa_get_indirect_edge_target_1 instead of calling
22593         ipa_get_indirect_edge_target.
22594         (create_specialized_node): Pass aggvlas to
22595         ipcp_discover_new_direct_edges.
22597 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22599         * config/arm/arm.md (f_sels, f_seld): New types.
22600         (*cmov<mode>): New pattern.
22601         * config/arm/predicates.md (arm_vsel_comparison_operator): New
22602         predicate.
22604 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22606         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
22607         POSIX-printf for mingw-hosted  builds.
22609 2013-03-25  Richard Biener  <rguenther@suse.de>
22611         PR middle-end/56694
22612         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
22613         must-not-throw stmt location.
22615 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22617         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
22618         Emit load-acquire versions when acq is true.
22619         (arm_emit_store_exclusive): Add rel parameter.
22620         Emit store-release versions when rel is true.
22621         (arm_split_compare_and_swap): Use acquire-release instructions
22622         instead.
22623         of barriers when appropriate.
22624         (arm_split_atomic_op): Likewise.
22625         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
22626         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
22627         (VUNSPEC_SLX): Likewise.
22628         (VUNSPEC_LDA): Likewise.
22629         (VUNSPEC_STL): Likewise.
22630         * config/arm/sync.md (atomic_load<mode>): New pattern.
22631         (atomic_store<mode>): Likewise.
22632         (arm_load_acquire_exclusive<mode>): Likewise.
22633         (arm_load_acquire_exclusivesi): Likewise.
22634         (arm_load_acquire_exclusivedi): Likewise.
22635         (arm_store_release_exclusive<mode>): Likewise.
22637 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22639         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22640         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22641         * config/mip/predicates.md (lwsp_swsp_operand,
22642         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22643         sb16_operand, db4_operand, db7_operand, ib3_operand,
22644         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22645         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22646         andi16_operand): New predicates.
22647         * config/mips/mips.md (compression): New attribute.
22648         (enabled): New attribute.
22649         (length): Consider compression in computing length.
22650         (shift_compression): New code attribute.
22651         (*add<mode>3): New operands. Record compression.
22652         (sub<mode>3): Likewise.
22653         (one_cmpl<mode>2): Likewise.
22654         (*and<mode>3): Likewise.
22655         (*ior<mode>3): Likewise.
22656         (unnamed pattern for xor): Likewise.
22657         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22658         (*<optab><mode>3): Likewise.
22659         (*mov<mode>_internal: Likewise.
22660         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22661         (mips_unsigned_immediate_p): New.
22662         (umips_lwsp_swsp_address_p): New.
22663         (m16_based_address_p): New.
22664         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22665         (mips_unsigned_immediate_p): New prototype.
22666         (lwsp_swsp_address_p): New prototype.
22667         (m16_based_address_p): New prototype.
22668         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22669         (mips_signed_immediate_p): New function.
22670         (m16_based_address_p): New function.
22671         (lwsp_swsp_address_p): New function.
22672         (mips_print_operand_punctuation): Recognize short delay slot insns
22673         for microMIPS.add<mode>3"
22675 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22677         PR target/56720
22678         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22679         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22681 2013-03-25  Richard Biener  <rguenther@suse.de>
22683         PR tree-optimization/56689
22684         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22685         any edge.
22687 2013-03-25  Richard Biener  <rguenther@suse.de>
22689         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22690         of bitmap.
22691         (memory_references): Likewise.
22692         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22693         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22694         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22695         (gather_mem_refs_in_loops): Fold into ...
22696         (analyze_memory_references): ... this.  Move initialization
22697         to tree_ssa_lim_initialize.
22698         (fill_always_executed_in): Rename to ...
22699         (fill_always_executed_in_1): ... this.
22700         (fill_always_executed_in): Move contains_call computation to
22701         this new function from ...
22702         (tree_ssa_lim_initialize): ... here.
22703         (tree_ssa_lim): Call fill_always_executed_in.
22705 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22707         * postreload.c (reload_combine): Fix code detecting returns.
22709 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22711         * function.c (emit_use_return_register_into_block): On cc0 targets,
22712         do not emit the sequence between cc0 setter and user.
22714 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22716         * config/i386/predicates.md (local_symbolic_operand): Interpret
22717         dll-imported symbols as none-local.
22719 2013-03-25  Richard Biener  <rguenther@suse.de>
22721         * tree-ssa-loop-im.c (struct depend): Remove.
22722         (struct lim_aux_data): Make depends a vec of gimples.
22723         (free_lim_aux_data): Adjust.
22724         (add_dependency): Likewise.
22725         (set_level): Likewise.
22727 2013-03-25  Richard Biener  <rguenther@suse.de>
22729         PR middle-end/56434
22730         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22731         the pointer returned by calls with ECF_MALLOC set.
22733 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22735         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22737 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22739         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22740         using MMXMODE mode iterator.
22741         (*move<mode>_internal): Merge with *movv2sf_internal and
22742         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22744 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22746         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22747         (record_last_mem_set_info): Likewise.
22749         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22750         of XNEWVEC followed by memset.
22751         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22753 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22755         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22756         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22757         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22758         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22759         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22760         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22761         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22762         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22763         BARRIER_P instead of GET_CODE.
22765 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22767         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22768         inaccuracy in the probing code.
22770         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22771         (ctrapdi4): Likewise.
22773 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22775         * calls.c (expand_call): Add missing guard to code handling return
22776         of non-BLKmode structures in MSB.
22777         * function.c (expand_function_end): Likewise.
22779 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22781         * combine.c (try_combine): Adjust comment.  Do not add the set of
22782         insn #0 if the destination indirectly is set or dies in insn #2.
22783         Tidy up code to distribute a new note.
22785 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22787         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22788         also for alternatives 16 and 17.
22790 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22792         * config/i386/sse.md (*mov<mode>_internal): Merge with
22793         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22794         Emit insn template depending on type attribute.  Use
22795         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22796         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22797         attribute calculation.  Remove unit attribute calculation.
22798         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22799         Set prefix_data16 attribute for DImode ssemov types.
22800         Use Ym instead of y for SSE-MMX conversion alternatives.
22801         Reorder operand constraints.
22803 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22805         * df.h (df_insn_delete): Adjust prototype.
22806         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22807         and let it decide whether mark the basic block dirty.
22808         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22809         * df-scan.c (df_insn_info_delete): New helper function, split
22810         off from df_insn_delete.
22811         (df_scan_free_bb_info): Use it.
22812         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22813         Likewise.
22814         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22815         that the insn is actually an insn and it has a non-NULL basic block.
22816         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22818 2013-03-22  Richard Biener  <rguenther@suse.de>
22820         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22821         dep_ref members.
22822         (mem_ref_alloc): Do not allocate them.
22823         (refs_independent_p): Do not query or maintain a cache.
22825 2013-03-22  Richard Biener  <rguenther@suse.de>
22827         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22828         (gather_mem_refs_in_loops): Do not compute it.
22829         (analyze_memory_references): Do not allocate it.
22830         (tree_ssa_lim_finalize): Do not free it.
22831         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22833 2013-03-22  Richard Biener  <rguenther@suse.de>
22835         * is-a.h (as_a): Use gcc_checking_assert.
22837 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22839         * config/aarch64/aarch64.c (aarch64_print_operand): New
22840         format specifier for printing a constant in hex.
22841         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22842         format specifier for printing second operand.
22844 2013-03-22  Richard Biener  <rguenther@suse.de>
22846         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22847         bitmaps.
22848         (gather_mem_refs_in_loops): Perform store accumulation here.
22849         (create_vop_ref_mapping_loop): Remove.
22850         (create_vop_ref_mapping): Likewise.
22851         (analyze_memory_references): Initialize refs_stored_in_loop.
22852         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22853         (record_indep_loop): Remove.
22854         (record_dep_loop): New function.
22855         (ref_indep_loop_p_1): Adjust to only walk over references
22856         in the loop, not its subloops.
22857         (ref_indep_loop_p): Rename to ...
22858         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22859         maintaining a more fine-grained cache.
22860         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22861         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22863 2013-03-22  Richard Biener  <rguenther@suse.de>
22865         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22866         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22867         aggregate mem_ref_loc.
22868         (free_mem_ref_locs): Inline into ...
22869         (memref_free): ... this and adjust.
22870         (mem_ref_alloc): Adjust.
22871         (mem_ref_locs_alloc): Remove.
22872         (record_mem_ref_loc): Adjust.
22873         (get_all_locs_in_loop): Rewrite into ...
22874         (for_all_locs_in_loop): ... this iterator.
22875         (rewrite_mem_ref_loc): New functor.
22876         (rewrite_mem_refs): Use for_all_locs_in_loop.
22877         (sm_set_flag_if_changed): New functor.
22878         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22879         (ref_always_accessed): New functor.
22880         (ref_always_accessed_p): Use for_all_locs_in_loop.
22882 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22884         * tree-pass.h (PROP_gimple_lvec): New.
22885         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22886         (init_optimization_passes): Move pass_lower_vector.
22887         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22888         PROP_gimple_lvec.
22889         (pass_lower_vector): Provide PROP_gimple_lvec.
22890         (pass_lower_vector_ssa): Likewise.
22891         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22893 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22895         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22897 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22899         * config/i386/i386.md (*movdi_internal): Disparage slightly
22900         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22901         conversion alternatives.
22903 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22905         PR middle-end/48087
22906         * diagnostic.def (DK_WERROR): New kind.
22907         * diagnostic.h (werrorcount): Define.
22908         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22909         promoted to DK_ERROR, increment DK_WERROR counter instead of
22910         DK_ERROR counter.
22911         * toplev.c (toplev_main): Call print_ignored_options even if
22912         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22913         even if just werrorcount is non-zero.
22915         PR debug/55608
22916         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22917         on failure.
22918         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22919         (string_cst_pool_decl): New function.
22920         (optimize_one_addr_into_implicit_ptr): New function.
22921         (resolve_addr_in_expr): Optimize DWARF location expression
22922         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22923         which doesn't live in memory, but has DW_AT_location or
22924         DW_AT_const_value, or refers to a string literal, into
22925         DW_OP_GNU_implicit_pointer.
22926         (optimize_location_into_implicit_ptr): New function.
22927         (resolve_addr): If removing DW_AT_location of a variable because
22928         it was DW_OP_addr of address of the variable, but the variable doesn't
22929         live in memory, try to emit const value attribute for the initializer.
22931 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22933         * tree.h (VECTOR_TYPE_P): New macro.
22934         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22935         TYPE_MODE): Use it.
22936         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22937         VEC_COND_EXPR cannot be lvalues.
22938         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22940 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22942         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22943         Restrict the transformation to equal modes.
22945 2013-03-21  Richard Biener  <rguenther@suse.de>
22947         PR tree-optimization/39326
22948         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22949         (MEM_ANALYZABLE): Adjust.
22950         (record_mem_ref_loc): Move bitmap ops ...
22951         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22952         unanalyzable refs, do not record locations for it.
22953         (analyze_memory_references): Allocate ref zero as shared
22954         unanalyzable ref.
22955         (refs_independent_p): Do not test for unanalyzed mems here.
22956         (ref_indep_loop_p_1): Special-case disambiguation against
22957         the unanalyzed ref.
22958         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22960 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22962         * config/arm/arm-protos.h (tune_params): Add
22963         prefer_neon_for_64bits field.
22964         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22965         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22966         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22967         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22968         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22969         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22970         (arm_option_override): Handle -mneon-for-64bits new option.
22971         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22972         (prefer_neon_for_64bits): Declare new variable.
22973         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22974         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22975         (arch_enabled): Handle new arch types. Remove support for onlya8
22976         and nota8.
22977         (one_cmpldi2): Use new arch names.
22978         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22979         * config/arm/arm.opt (mneon-for-64bits): Add option.
22980         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22981         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22982         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22983         of onlya8.
22984         * doc/invoke.texi (-mneon-for-64bits): Document.
22986 2013-03-21  Richard Biener  <rguenther@suse.de>
22988         PR tree-optimization/39326
22989         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22990         (sort_bbs_in_loop_postorder_cmp): New function.
22991         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22993 2013-03-21  Richard Biener  <rguenther@suse.de>
22995         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22996         (vect_insert_into_interleaving_chain): Likewise.
22997         (vect_drs_dependent_in_basic_block): Inline ...
22998         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22999         split out from ...
23000         (vect_analyze_data_ref_dependence): ... here.  Simplify.
23001         (vect_check_interleaving): Simplify.
23002         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
23003         (vect_slp_analyze_data_ref_dependences): ... this new function.
23004         (dr_group_sort_cmp): New function.
23005         (vect_analyze_data_ref_accesses): Compute data-reference groups
23006         here instead of in vect_analyze_data_ref_dependence.  Use
23007         a more efficient algorithm.
23008         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
23009         vect_slp_analyze_data_ref_dependences.  Call
23010         vect_analyze_data_ref_accesses earlier.
23011         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
23012         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
23013         (vect_slp_analyze_data_ref_dependences): New prototype.
23015 2013-03-21  Richard Biener  <rguenther@suse.de>
23017         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
23018         ref is stored in the loop.
23019         (find_refs_for_sm): Walk only over all stores.
23020         (store_motion_loop): Allocate from lim_bitmap_obstack.
23021         (store_motion): Likewise.
23023 2013-03-21  Richard Biener  <rguenther@suse.de>
23025         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
23026         Update virtual SSA form.
23028 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
23030         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
23031         * configure: Regenerate.
23032         * config.in: Regenerate.
23033         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
23034         if !HAVE_LD_EH_FRAME_CIEV3.
23036 2013-03-21  Richard Biener  <rguenther@suse.de>
23038         * tree-cfg.c (verify_expr_no_block): New function.
23039         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
23040         nor DECL_VALUE_EXPR have locations with associated blocks.
23041         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
23042         (clear_unused_block_pointer): Remove code dealing with
23043         blocks in DECL_DEBUG_EXPR locations.
23045 2013-03-21  Richard Biener  <rguenther@suse.de>
23047         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
23048         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
23049         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
23050         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
23051         instead of DECL_DEBUG_EXPR_IS_FROM.
23052         * gimplify.c (gimplify_modify_expr): Likewise.
23053         * tree-cfg.c (verify_expr_location_1): Likewise.
23054         * tree-complex.c (create_one_component_var): Likewise.
23055         * tree-sra.c (create_access_replacement): Likewise.
23056         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
23057         (clear_unused_block_pointer): Likewise.
23058         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
23059         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
23060         * var-tracking.c (var_debug_decl): Likewise.
23061         (track_expr_p): Likewise.
23062         * tree-inline.c (add_local_variables): Likewise.  Set
23063         DECL_HAS_DEBUG_EXPR_P after copying it.
23064         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
23065         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
23067 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
23069         PR bootstrap/56656
23070         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
23071         * configure: Regenerate.
23072         * config.in: Regenerate.
23073         * config/i386/i386.md (*movdf_internal): Use
23074         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
23075         movd instead of movq mnemonic for interunit moves.
23076         (*movdi_internal): Ditto.
23078 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23080         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
23081         (abd<mode>_3): New pattern.
23082         (aba<mode>_3): New pattern.
23083         (fabd<mode>_3): New pattern.
23085 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23087         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
23088         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
23089         occurrence of REGISTER_PREFIX as its empty string.
23091 2013-03-20  Jeff Law  <law@redhat.com>
23093         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
23094         addititional equivalences for equality comparisons between an SSA_NAME
23095         and a constant where the SSA_NAME was set from a widening conversion.
23097 2013-03-20  Walter Lee  <walt@tilera.com>
23099         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
23101 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23103         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
23104         depending on type attribute.
23105         (*movti_internal): Ditto.
23106         (*movtf_internal): Ditto.
23107         (*movxf_internal): Ditto.
23108         (*movdf_internal): Ditto.
23109         (*movsf_internal): Ditto.
23111 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23113         * config/i386/i386.md (*movti_internal): Set prefix attribute to
23114         maybe_vex for sselog1 and ssemov types.
23115         (*movdi_internal): Reorder operand constraints.
23116         (*movsi_internal): Ditto.  Set prefix attribute to
23117         maybe_vex for sselog1 and ssemov types.
23118         (*movtf_internal): Set prefix attribute to maybe_vex
23119         for sselog1 and ssemov types.
23120         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
23121         DImode ssemov types.  Reorder operand constraints.
23122         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
23123         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
23124         attribute for SImode ssemov types.  Reorder operand constraints.
23126 2013-03-20  Martin Jambor  <mjambor@suse.cz>
23128         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
23129         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
23131 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
23133         * config/rs6000/predicates.md (indexed_address, update_address_mem
23134         update_indexed_address_mem): New predicates.
23135         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
23136         attribute for load/store instructions.
23137         * config/rs6000/dfp.md (movsd_store): Likewise.
23138         (movsd_load): Likewise.
23139         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
23140         (unnamed HI->DI extend define_insn): Likewise.
23141         (unnamed SI->DI extend define_insn): Likewise.
23142         (unnamed QI->SI extend define_insn): Likewise.
23143         (unnamed QI->HI extend define_insn): Likewise.
23144         (unnamed HI->SI extend define_insn): Likewise.
23145         (unnamed HI->SI extend define_insn): Likewise.
23146         (extendsfdf2_fpr): Likewise.
23147         (movsi_internal1): Likewise.
23148         (movsi_internal1_single): Likewise.
23149         (movhi_internal): Likewise.
23150         (movqi_internal): Likewise.
23151         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
23152         attribute for load/store instructions.
23153         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
23154         instructions.
23155         (mov<mode>_softfloat): Likewise.
23156         (mov<mode>_hardfloat32): Likewise.
23157         (mov<mode>_hardfloat64): Likewise.
23158         (mov<mode>_softfloat64): Likewise.
23159         (movdi_internal32): Likewise.
23160         (movdi_internal64): Likewise.
23161         (probe_stack_<mode>): Likewise.
23163 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
23165         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
23166         floating point, and decimal floating point to reload iterator.
23168         * config/rs6000/constraints.md (wl constraint): New constraints to
23169         return FLOAT_REGS if certain options are used to reduce the number
23170         of separate patterns that exist in the file.
23171         (wx constraint): Likewise.
23172         (wz constraint): Likewise.
23174         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
23175         -mdebug=reg, print wg, wl, wx, and wz constraints.
23176         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
23177         Initialize the reload functions for 64-bit binary/decimal floating
23178         point types.
23179         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
23180         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
23181         create the buffer on the stack to overcome not having a 32-bit
23182         load and store.
23183         (rs6000_emit_move): Likewise.
23184         (rs6000_secondary_memory_needed_rtx): Likewise.
23185         (rs6000_alloc_sdmode_stack_slot): Likewise.
23186         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
23187         via xxlxor, just like DFmode 0.0.
23189         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
23190         define as 1 if we are running on a power7 or newer.
23191         (enum r6000_reg_class_enum): Add new constraints.
23193         * config/rs6000/dfp.md (movsd): Delete, combine with binary
23194         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
23195         with other moves by using conditional constraits (wg).  Use LFIWZX
23196         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
23197         (movsd splitter): Likewise.
23198         (movsd_hardfloat): Likewise.
23199         (movsd_softfloat): Likewise.
23201         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
23202         binary and decimal floating point moves.
23203         (fmove_ok): New attributes to combine binary and decimal floating
23204         point moves, and to combine power6x (mfpgpr) moves along normal
23205         floating moves.
23206         (real_value_to_target): Likewise.
23207         (f32_lr): Likewise.
23208         (f32_lm): Likewise.
23209         (f32_li): Likewise.
23210         (f32_sr): Likewise.
23211         (f32_sm): Likewise.
23212         (f32_si): Likewise.
23213         (movsf): Combine binary and decimal floating point moves.  Combine
23214         power6x (mfpgpr) moves with other moves by using conditional
23215         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
23216         (mov<mode> for SFmode/SDmode); Likewise.
23217         (SFmode/SDmode splitters): Likewise.
23218         (movsf_hardfloat): Likewise.
23219         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
23220         (movsf_softfloat): Likewise.
23221         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
23223         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
23224         wx and wz constraints.
23226         * config/rs6000/constraints.md (wg constraint): New constraint to
23227         return FLOAT_REGS if -mmfpgpr (power6x) was used.
23229         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
23230         constraint.
23232         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
23233         -mdebug=reg, print wg, wl, wx, and wz constraints.
23234         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
23235         Initialize the reload functions for 64-bit binary/decimal floating
23236         point types.
23237         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
23238         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
23239         create the buffer on the stack to overcome not having a 32-bit
23240         load and store.
23241         (rs6000_emit_move): Likewise.
23242         (rs6000_secondary_memory_needed_rtx): Likewise.
23243         (rs6000_alloc_sdmode_stack_slot): Likewise.
23244         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
23245         via xxlxor, just like DFmode 0.0.
23247         * config/rs6000/dfp.md (movdd): Delete, combine with binary
23248         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
23249         with other moves by using conditional constraits (wg).  Use LFIWZX
23250         and STFIWX for loading SDmode on power7.
23251         (movdd splitters): Likewise.
23252         (movdd_hardfloat32): Likewise.
23253         (movdd_softfloat32): Likewise.
23254         (movdd_hardfloat64_mfpgpr): Likewise.
23255         (movdd_hardfloat64): Likewise.
23256         (movdd_softfloat64): Likewise.
23258         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
23259         64-bit binary and decimal floating point moves.
23260         (FMOVE64X): Likewise.
23261         (movdf): Combine 64-bit binary and decimal floating point moves.
23262         Combine power6x (mfpgpr) moves with other moves by using
23263         conditional constraits (wg).
23264         (mov<mode> for DFmode/DDmode): Likewise.
23265         (DFmode/DDmode splitters): Likewise.
23266         (movdf_hardfloat32): Likewise.
23267         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
23268         (movdf_softfloat32): Likewise.
23269         (movdf_hardfloat64_mfpgpr): Likewise.
23270         (movdf_hardfloat64): Likewise.
23271         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
23272         (movdf_softfloat64): Likewise.
23273         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
23274         (reload_<mode>_load): Move to later in the file so they aren't in
23275         the middle of the floating point move insns.
23276         (reload_<mode>_store): Likewise.
23278         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
23279         constraint.
23281         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
23282         constraint if -mdebug=reg.
23283         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
23284         Enable using dd reload support if needed.
23286         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
23287         binary and decimal floating point moves in rs6000.md.
23288         (movtd_internal): Likewise.
23290         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
23291         decimal floating point moves.
23292         (movtf): Likewise.
23293         (movtf_internal): Likewise.
23294         (mov<mode>_internal, TDmode/TFmode): Likewise.
23295         (movtf_softfloat): Likewise.
23296         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
23298         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
23299         movdi_internal64, using wg constraint for move direct operations.
23300         (movdi_internal64): Likewise.
23302         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
23303         MODES_TIEABLE_P for selected modes.  Print the numerical value of
23304         the various virtual registers. Use GPR/FPR first/last values,
23305         instead of hard coding the register numbers.  Print which modes
23306         have reload functions registered.
23307         (rs6000_option_override_internal): If -mdebug=reg, trace the options
23308         settings before/after setting cpu, target and subtarget settings.
23309         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
23310         and for secondary reload failures in rs6000_secondary_reload_inner.
23311         (rs6000_secondary_reload_fail): Likewise.
23312         (rs6000_secondary_reload_inner): Likewise.
23314         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
23315         macros for first/last GPR and FPR registers.
23316         (LAST_GPR_REGNO): Likewise.
23317         (FIRST_FPR_REGNO): Likewise.
23318         (LAST_FPR_REGNO): Likewise.
23320         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
23321         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
23322         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
23323         (vcond<mode><mode>): Likewise.
23324         (vcondu<mode><mode>): Likewise.
23325         (vector_gtu<mode>): Likewise.
23326         (vector_gte<mode>): Likewise.
23327         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
23328         to prevent the compiler from converting DImode operations to TImode.
23329         (ior<mode>3): Likewise.
23330         (and<mode>3): Likewise.
23331         (one_cmpl<mode>2): Likewise.
23332         (nor<mode>3): Likewise.
23333         (andc<mode>3): Likewise.
23335         * config/rs6000/constraints.md (wt constraint): New constraint
23336         that returns VSX_REGS if TImode is allowed in VSX registers.
23338         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
23339         constant under VSX.
23341         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
23342         similar to TImode, but it is restricted to being in the GPRs.
23344         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
23345         TImode to occupy a single VSX register.
23347         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
23348         -mvsx-timode for power7/power8.
23349         (power7 cpu): Likewise.
23350         (power8 cpu): Likewise.
23352         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
23353         sure that TFmode/TDmode take up two registers if they are ever
23354         allowed in the upper VSX registers.
23355         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
23356         registers.
23357         (rs6000_init_hard_regno_mode_ok): Likewise.
23358         (rs6000_debug_reg_global): Add debugging for PTImode and wt
23359         constraint.  Print if LRA is turned on.
23360         (rs6000_option_override_internal): Give an error if -mvsx-timode
23361         and VSX is not enabled.
23362         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
23363         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
23364         to reg+offset addressing.  Use PTImode when checking offset
23365         addresses for validity.
23366         (reg_offset_addressing_ok_p): Likewise.
23367         (rs6000_legitimate_offset_address_p): Likewise.
23368         (rs6000_legitimize_address): Likewise.
23369         (rs6000_legitimize_reload_address): Likewise.
23370         (rs6000_legitimate_address_p): Likewise.
23371         (rs6000_eliminate_indexed_memrefs): Likewise.
23372         (rs6000_emit_move): Likewise.
23373         (rs6000_secondary_reload): Likewise.
23374         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
23375         reloads to fpr registers to continue to use reg+offset addressing,
23376         but 64-bit reloads to altivec registers need reg+reg addressing.
23377         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
23378         it.  Treat LO_SUM like a PLUS operation.
23379         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
23380         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
23381         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
23382         registers to share a register with a smaller sized type, since VSX
23383         puts scalars in the upper 64-bits.
23384         (print_operand): Add support for PTImode.
23385         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
23386         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
23387         registers, but don't have arithmetic support.
23388         (rs6000_memory_move_cost): Add test for VSX.
23389         (rs6000_opt_masks): Add -mvsx-timode.
23391         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
23392         for TImode.
23393         (VSs): Likewise.
23394         (VSr): Use wt constraint for TImode.
23395         (VSv): Drop TImode support.
23396         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
23397         (vsx_movti_64bit): Likewise.
23398         (vsx_movti_32bit): Likewise.
23399         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
23400         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
23401         one '?' on the appropriate output constraint.  Do not allow TImode
23402         logical operations on 32-bit systems.
23403         (vsx_ior<mode>3): Likewise.
23404         (vsx_xor<mode>3): Likewise.
23405         (vsx_one_cmpl<mode>2): Likewise.
23406         (vsx_nor<mode>3): Likewise.
23407         (vsx_andc<mode>3): Likewise.
23408         (vsx_concat_<mode>): Likewise.
23409         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
23411         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
23412         OPTION_MASK_VSX_TIMODE.
23413         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
23414         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
23416         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
23417         (TI2 iterator): New iterator for TImode, PTImode.
23418         (wd mode attribute): Add values for vector types.
23419         (movti_string): Replace TI move operations with operations for TImode
23420         and PTImode.  Add support for TImode being allowed in VSX registers.
23421         (mov<mode>_string, TImode/PTImode): Likewise.
23422         (movti_ppc64): Likewise.
23423         (mov<mode>_ppc64, TImode/PTImode): Likewise.
23424         (TI mode splitters): Likewise.
23426         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
23427         constraint.
23429 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
23431         PR tree-optimization/56355
23432         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
23433         Also handle integers with undefined overflow.
23435 2013-03-20  Catherine Moore  <clm@codesourcery.com>
23436             Maciej W. Rozycki  <macro@codesourcery.com>
23437             Tom de Vries  <tom@codesourcery.com>
23438             Nathan Sidwell  <nathan@codesourcery.com>
23439             Iain Sandoe  <iain@codesourcery.com>
23440             Nathan Froyd  <froydnj@codesourcery.com>
23441             Chao-ying Fu  <fu@mips.com>
23443         * doc/extend.texi: (micromips, nomicromips, nocompression):
23444         Document new function attributes.
23445         * doc/invoke.texi (minterlink-compressed, mmicromips,
23446         m14k, m14ke, m14kec): Document new options.
23447         (minterlink-mips16): Update documentation.
23448         * doc/md.texi (ZC, ZD): Document new constraints.
23449         * configure.ac (gcc_cv_as_micromips): Check if linker
23450         supports the .set micromips directive.
23451         * configure: Regenerate.
23452         * config.in: Regenerate.
23453         * config/mips/mips-tables.opt: Regenerate.
23454         * config/mips/micromips.md: New file.
23455         * constraints.md (ZC, ZD): New constraints.
23456         * config/mips/predicates.md (movep_src_register): New predicate.
23457         (movep_src_operand): New predicate.
23458         (non_volatile_mem_operand): New predicate.
23459         * config/mips/mips.md (multimem): New type.
23460         (length): Differentiate between 17-bit and 18-bit branch offsets.
23461         (MOVEP1, MOVEP2): New mode iterator.
23462         (mov_<load>l): Use ZC constraint.
23463         (mov_<load>r): Likewise.
23464         (mov_<store>l): Likewise.
23465         (mov_<store>r): Likewise.
23466         (*branch_equality<mode>_inverted): Add microMIPS support.
23467         (*branch_equality<mode>): Likewise.
23468         (*jump_absolute): Likewise.
23469         (indirect_jump_<mode>): Likewise.
23470         (tablejump_<mode>): Likewise.
23471         (<optab>_internal): Likewise.
23472         (sibcall_internal): Likewise.
23473         (sibcall_value_internal): Likewise.
23474         (prefetch): Use constraint ZD.
23475         * config/mips/mips.opt (minterlink-compressed): New option.
23476         (minterlink-mips16): Now an alias for minterlink-compressed.
23477         (mmicromips): New option.
23478         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
23479         (compare_and_swap_12): Likewise.
23480         (sync_add<mode>): Likewise.
23481         (sync_<optab>_12): Likewise.
23482         (sync_old_<optab>_12): Likewise.
23483         (sync_new_<optab>_12): Likewise.
23484         (sync_nand_12): Likewise.
23485         (sync_old_nand_12): Likewise.
23486         (sync_new_nand_12): Likewise.
23487         (sync_sub<mode>): Likewise.
23488         (sync_old_add<mode>): Likewise.
23489         (sync_old_sub<mode>): Likewise.
23490         (sync_new_add<mode>): Likewise.
23491         (sync_new_sub<mode>): Likewise.
23492         (sync_<optab><mode>): Likewise.
23493         (sync_old_<optab><mode>): Likewise.
23494         (sync_new_<optab><mode>): Likewise.
23495         (sync_nand<mode>): Likewise.
23496         (sync_old_nand<mode>): Likewise.
23497         (sync_new_nand<mode>): Likewise.
23498         (sync_lock_test_and_set<mode>): Likewise.
23499         (test_and_set_12): Likewise.
23500         (atomic_compare_and_swap<mode>): Likewise.
23501         (atomic_exchange<mode>_llsc): Likewise.
23502         (atomic_fetch_add<mode>_llsc): Likewise.
23503         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
23504         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
23505         (umips_save_restore_pattern_p): Likewise.
23506         (umips_load_store_pair_p): Likewise.
23507         (umips_output_load_store_pair): Likewise.
23508         (umips_movep_target_p): Likewise.
23509         (umips_12bit_offset_address_p): Likewise.
23510         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
23511         (mips_base_mips16): Rename this...
23512         (mips_base_compression_flags): ...to this. Update all uses.
23513         (mips_attribute_table): Add micromips, nomicromips and nocompression.
23514         (mips_mips16_decl_p): Delete.
23515         (mips_nomips16_decl_p): Delete.
23516         (mips_get_compress_on_flags): New function.
23517         (mips_get_compress_off_flags): New function.
23518         (mips_get_compress_mode): New function.
23519         (mips_get_compress_on_name): New function.
23520         (mips_get_compress_off_name): New function.
23521         (mips_insert_attributes): Support multiple compression types.
23522         (mips_merge_decl_attributes): Likewise.
23523         (umips_12bit_offset_address_p): New function.
23524         (mips_start_function_definition): Emit .set micromips directive.
23525         (mips_call_may_need_jalx_p): New function.
23526         (mips_function_ok_for_sibcall): Add microMIPS support.
23527         (mips_print_operand_punctuation): Support short delay slots and
23528         compact jumps.
23529         (umips_swm_mask, umips_swm_encoding): New.
23530         (umips_build_save_restore): New function.
23531         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
23532         (was_mips16_p): Remove.
23533         (old_compression_mode): New.
23534         (mips_set_compression_mode): New function.
23535         (mips_set_current_function): Add microMIPS support.
23536         (mips_option_override): Likewise.
23537         (umips_save_restore_pattern_p): New function.
23538         (umips_output_save_restore): New function.
23539         (umips_load_store_pair_p_1): New function.
23540         (umips_load_store_pair_p): New function.
23541         (umips_output_load_store_pair_1): New function.
23542         (umips_output_load_store_pair): New function.
23543         (umips_movep_target_p) New function.
23544         (mips_prepare_pch_save): Add microMIPS support.
23545         * config/mips/mips.h (TARGET_COMPRESSION): New.
23546         (TARGET_CPU_CPP_BUILTINS): Update macro
23547         to use new compression flags and to support microMIPS.
23548         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
23549         (MIPS_ARCH_FLOAT_SPEC): Likewise.
23550         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
23551         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
23552         (ASM_SPEC): Support mmicromips and mno-micromips.
23553         (M16STORE_REG_P): New macro.
23554         (MIPS_CALL): Support TARGET_MICROMIPS.
23555         (MICROMIPS_J): New macro.
23556         (mips_base_mips16): Rename this...
23557         (mips_base_compression_flags): ...to this.
23558         (UMIPS_12BIT_OFFSET_P): New macro.
23559         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
23560         (MULTILIB_DIRNAMES): Likewise.
23561 2013-03-20  Richard Biener  <rguenther@suse.de>
23563         PR tree-optimization/56661
23564         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
23565         the result does not have to be distinct.
23567 2013-03-20  Richard Biener  <rguenther@suse.de>
23569         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
23570         remap_gimple_op_r.
23572 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23573             Steven Bosscher  <steven@gcc.gnu.org>
23575         PR rtl-optimization/56605
23576         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
23578 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
23580         PR bootstrap/56656
23581         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
23582         that require movd instead of movq.
23584 2013-03-20  Richard Biener  <rguenther@suse.de>
23586         * tree-ssa-structalias.c (struct variable_info): Add pointer
23587         to the first field of an aggregate with sub-vars.  Make
23588         this and the pointer to the next subfield its ID.
23589         (vi_next): New function.
23590         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
23591         storedanything_id, integer_id): Increment by one.
23592         (new_var_info, get_call_vi, lookup_call_clobber_vi,
23593         get_call_clobber_vi): Adjust.
23594         (solution_set_expand): Simplify and speedup.
23595         (solution_set_add): Inline into ...
23596         (set_union_with_increment): ... this.  Adjust accordingly.
23597         (do_sd_constraint): Likewise.
23598         (do_ds_constraint): Likewise.
23599         (do_complex_constraint): Simplify.
23600         (build_pred_graph): Adjust.
23601         (solve_graph): Likewise.  Simplify and speedup.
23602         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
23603         get_constraint_for_component_ref, get_constraint_for_1,
23604         first_vi_for_offset, first_or_preceding_vi_for_offset,
23605         create_function_info_for, create_variable_info_for_1,
23606         create_variable_info_for, intra_create_variable_infos): Adjust.
23607         (init_base_vars): Push NULL for ID zero.
23608         (compute_points_to_sets): Adjust.
23610 2013-03-20  Richard Biener  <rguenther@suse.de>
23612         * cfgloop.c (verify_loop_structure): Streamline and avoid
23613         ICEing on corrupt loop tree.
23614         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
23615         loop tree.
23617 2013-03-20  Richard Biener  <rguenther@suse.de>
23619         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
23620         check whether an SSA update is needed.
23622 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
23624         * config/mips/constraints.md (T): Rename to...
23625         (Yf): ...this.
23626         (U): Rename to...
23627         (Yd): ...this.
23628         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
23629         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
23631 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23633         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
23634         (*subsi3_carryin_uxtw): Likewise.
23636 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23638         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23639         (*rorsi3_insn_uxtw): Likewise.
23641 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23643         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23644         (*extrsi5_insn_uxtw): Likewise.
23646 2013-03-19  Richard Biener  <rguenther@suse.de>
23648         PR tree-optimization/56273
23649         * passes.c (init_optimization_passes): Move second VRP after DOM.
23651 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23653         * config/i386/i386.md (*movti_internal): Merge from
23654         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23655         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23656         nox64 isa attributes.
23658 2013-03-18  Richard Biener  <rguenther@suse.de>
23660         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23661         (unite): Likewise.
23662         (merge_node_constraints): Likewise.
23663         (build_succ_graph): Likewise.
23664         (valid_graph_edge): Inline into single caller.
23665         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23666         and cache varinfo.
23667         (scc_visit): Fix formatting and variable use.
23668         (do_sd_constraint): Use gcc_checking_assert.
23669         (do_ds_constraint): Likewise.
23670         (do_complex_constraint): Likewise.
23671         (condense_visit): Likewise.  Cleanup.
23672         (dump_pred_graph): New function.
23673         (perform_var_substitution): Dump the pred-graph before
23674         variable substitution.
23675         (find_equivalent_node): Use gcc_checking_assert.
23676         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23678 2013-03-18  Richard Biener  <rguenther@suse.de>
23680         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23681         Remove cond_expr_stmt_list argument and do not gimplify the
23682         built expression.
23683         (vect_loop_versioning): Adjust.
23684         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23685         Cleanup to use less temporaries.
23686         (vect_create_data_ref_ptr): Cleanup.
23688 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23690         PR tree-optimization/56635
23691         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23692         require types_compatible_p types.
23694 2013-03-18  Nick Clifton  <nickc@redhat.com>
23696         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23697         spurious backslash.
23699         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23700         Add missing line to comment describing function.
23702 2013-03-18  Richard Biener  <rguenther@suse.de>
23704         PR tree-optimization/56210
23705         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23706         Handle string / character search functions.
23707         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23709 2013-03-18  Richard Biener  <rguenther@suse.de>
23711         PR middle-end/56483
23712         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23713         and implement properly.
23714         * gimple.h (gimple_cond_single_var_p): Remove.
23716 2013-03-18  Richard Biener  <rguenther@suse.de>
23718         * tree-data-ref.h (find_data_references_in_loop): Declare.
23719         * tree-data-ref.c (get_references_in_stmt): Use a stack
23720         vector pre-allocated in the callers.
23721         (find_data_references_in_stmt): Adjust.
23722         (graphite_find_data_references_in_stmt): Likewise.
23723         (create_rdg_vertices): Likewise.
23724         (find_data_references_in_loop): Export.
23725         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23726         Compute dependences here...
23727         (vect_analyze_data_refs): ...not here.  When we encounter
23728         a non-vectorizable data reference in basic-block vectorization
23729         truncate the data reference vector.  Do not bother to
23730         fixup data-dependence information for gather loads.
23731         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23732         of data references, as reported.
23734 2013-03-18  Richard Biener  <rguenther@suse.de>
23736         PR tree-optimization/3713
23737         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23738         has_constants and expr.
23739         (stmt_has_constants): Properly valueize SSA names when deciding
23740         whether the stmt has constants.
23742 2013-03-18  Richard Biener  <rguenther@suse.de>
23744         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23745         whole function when there is nothing to do.
23746         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23747         * tree-vectorizer.c (vectorize_loops): Update virtual and
23748         loop-closed SSA once.
23749         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23751 2013-03-18  Richard Biener  <rguenther@suse.de>
23753         PR middle-end/56113
23754         * domwalk.c (bb_postorder): New global static.
23755         (cmp_bb_postorder): New function.
23756         (walk_dominator_tree): Replace scheme imposing an order for
23757         visiting dominator sons by one sorting them at the time they
23758         are pushed on the stack.
23760 2013-03-18  Richard Biener  <rguenther@suse.de>
23762         PR tree-optimization/39326
23763         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23764         (struct mem_ref): Replace mem member with ao_ref typed member.
23765         (MEM_ANALYZABLE): Adjust.
23766         (memref_eq): Likewise.
23767         (mem_ref_alloc): Likewise.
23768         (gather_mem_refs_stmt): Likewise.
23769         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23770         (execute_sm_if_changed_flag_set): Adjust.
23771         (execute_sm): Likewise.
23772         (ref_always_accessed_p): Likewise.
23773         (refs_independent_p): Likewise.
23774         (can_sm_ref_p): Likewise.
23776 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23778         PR c/56566
23779         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23780         return 1 even for !unsignedp.
23782 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23784         * config/i386/i386.md (isa): Add x64 and nox64.
23785         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23786         (*pushtf): Enable *roF alternative for x64 isa only.
23787         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23788         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23789         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23790         (*movtf_internal): Merge from *movtf_internal_rex64 and
23791         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23792         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23793         nox64 isa attributes.
23794         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23795         nox64 isa attributes.
23796         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23798 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23800         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23802 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23804         PR target/56640
23805         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23806         class names.  Remove trailing comma after "ALL_REGS".
23808 2013-03-16  Jan Hubicka  <jh@suse.cz>
23810         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23811         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23812         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23813         of cgraph_get_create_node.
23814         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23816 2013-03-16  Jason Merrill  <jason@redhat.com>
23818         PR debug/49090
23819         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23820         with DW_AT_default_value.
23822 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23824         * BASE-VER: Set to 4.9.0.
23826 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23828         PR target/56619
23829         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23830         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23831         Document _x* TSX intrinsics.
23833 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23834             David Holsgrove  <david.holsgrove@xilinx.com>
23836         * configure.ac: Add MicroBlaze TLS support detection.
23837         * configure: Regenerate.
23838         * config/microblaze/microblaze-protos.h
23839         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23840         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23841         * config/microblaze/microblaze.c (microblaze_address_type): Add
23842         ADDRESS_TLS and tls_reloc address types.
23843         (microblaze_address_info): Add tls_reloc.
23844         (TARGET_HAVE_TLS): Define.
23845         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23846          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23847          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23848         load_tls_operand,  microblaze_call_tls_get_addr,
23849         microblaze_legitimize_tls_address): New functions.
23850         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23851         (get_base_reg): Use microblaze_tls_symbol_p.
23852         (microblaze_classify_address): Handle TLS.
23853         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23854         label_mentioned_p and microblaze_tls_referenced_p.
23855         (microblaze_legitimize_address): Handle TLS.
23856         (microblaze_address_insns): Handle ADDRESS_TLS.
23857         (pic_address_needs_scratch): Handle TLS.
23858         (print_operand_address): Handle TLS.
23859         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23860         (microblaze_expand_move): Handle TLS.
23861         (microblaze_legitimate_constant_p): Check
23862         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23863         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23864         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23865         (PIC_OFFSET_TABLE_REGNUM): Set.
23866         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23867         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23868         (addsi3, movsi_internal2, movdf_internal): Update constraints
23869         * config/microblaze/predicates.md (arith_plus_operand): Define
23870         (move_operand): Redefine as move_src_operand,
23871         check microblaze_tls_referenced_p.
23873 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23875         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23876         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23878 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23880         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23881         CC mode for AND.
23883 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23885         PR tree-optimization/53265
23886         * common.opt (Waggressive-loop-optimizations): New option.
23887         * tree-ssa-loop-niter.c: Include tree-pass.h.
23888         (do_warn_aggressive_loop_optimizations): New function.
23889         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23890         if number_of_latch_executions returned constant.
23891         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23892         early.  If number_of_latch_executions returned constant, set
23893         nb_iterations_upper_bound back to it.
23894         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23895         field.
23896         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23897         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23899         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23900         (MULTILIB_OSDIRNAMES): Set.
23901         * genmultilib: If defaultosdirname doesn't start with :: , set
23902         defaultosdirname2 instead, clear it and emit two . multilib_raw
23903         entries instead of just one.
23905 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23907         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23908         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23909         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23910         (SUBTARGET_OVERRIDE_OPTIONS): New.
23912 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23914         PR target/49880
23915         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23916         (musermode): Convert to Var(TARGET_USERMODE).
23917         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23918         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23919         * config/sh/sh.c (sh_option_override): Use
23920         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23921         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23922         condition.
23923         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23924         TARGET_SH4.
23925         (udivsi3_i4_single, divsi3_i4_single): Use
23926         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23928 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
23930         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23931         default setting.
23933 2013-03-13  Richard Biener  <rguenther@suse.de>
23935         PR tree-optimization/56608
23936         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23937         calls when vectorizing basic-blocks.
23939 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23941         PR plugins/45078
23942         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23943         tm_file.
23945 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23947         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23949 2013-03-11  Jan Hubicka  <jh@suse.cz>
23951         PR lto/56557
23952         * lto-streamer-out.c (output_symbol_p): Skip references from
23953         constructors of external variables.
23955 2013-03-11  Jan Hubicka  <jh@suse.cz>
23957         PR middle-end/56571
23958         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23959         from pseudos.
23960         * emit-rtl.c (verify_rtx_sharing): Likewise.
23961         (copy_insn_1): Likewise.
23962         * rtl.c (copy_rtx): Likewise.
23964 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23966         PR target/56591
23967         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23968         output_operand_lossage message.
23970 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23972         PR target/56470
23973         * arm.c (shift_op): Validate RTL pattern on the fly.
23974         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23975         the RTL.
23977 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23979         PR target/56347
23980         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23981         new call patterns that clobber %fr12.
23982         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23983         split and postreload patterns.
23984         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23985         registers %fr12 and %fr12R as call used.
23987 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23989         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23990         (canon_address, record_store, replace_read, check_mem_read_rtx,
23991         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23992         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23993         rest_of_handle_dse): Likewise.
23995 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23997         PR middle-end/56524
23998         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23999         Add base_optabs.
24000         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
24001         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
24002         (save_optabs_if_changed): Replace with...
24003         (init_tree_optimization_optabs): ...this.
24004         * optabs.c (save_optabs_if_changed): Rename to...
24005         (init_tree_optimization_optabs): ...this.  Take the optimization node
24006         as argument.  Do nothing if the base optabs are already correct.
24007         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
24008         to recompute optabs.
24009         * function.h (function): Remove optabs field.
24010         * function.c (invoke_set_current_function_hook): Call
24011         init_tree_optimization_optabs.  Use the result to initialize
24012         this_fn_optabs.
24014 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
24016         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
24017         if GTMA_HAS_NO_INSTRUMENTATION.
24018         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
24019         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
24020         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
24021         * gimple-pretty-print.c (dump_gimple_transaction): Handle
24022         GTMA_HAS_NO_INSTRUMENTATION.
24024 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
24026         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
24027         libasan_preinit.o.
24029 2013-03-08  Marek Polacek  <polacek@redhat.com>
24030             Jakub Jelinek  <jakub@redhat.com>
24032         PR tree-optimization/56478
24033         * predict.c (is_comparison_with_loop_invariant_p): Change the
24034         type of loop_step to tree.
24035         (predict_loops): Adjust.
24036         (predict_iv_comparison): Perform the computations on double_ints.
24038 2013-03-08  Richard Biener  <rguenther@suse.de>
24040         PR tree-optimization/56570
24041         * tree-cfg.c (verify_expr_location_1): Verify locations for
24042         DECL_DEBUG_EXPR.
24043         * tree-sra.c (create_access_replacement): Strip locations
24044         from DECL_DEBUG_EXPRs.
24046 2013-03-08  Richard Biener  <rguenther@suse.de>
24048         * tree-inline.c (expand_call_inline): Do not associate
24049         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
24050         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
24052 2013-03-08  Richard Biener  <rguenther@suse.de>
24054         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
24055         or block changes with -Og.  Fix for location / block encoding
24056         changes and PHI arguments with locations.
24058 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
24060         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
24061         for all counters.
24062         (struct output_info): Likewise.
24063         (register_overhead): Remove bad gcc_assert.
24064         (bitmap_find_bit): If there is only a single bitmap element, do not
24065         count a miss as a search.
24066         (print_statistics): Update for counter type changes.
24067         (dump_bitmap_statistics): Likewise.  Print headers such that they
24068         are properly lined up with the printed counters.
24070 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
24072         PR tree-optimization/56559
24073         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
24074         check that it has only a single use.
24076 2013-03-07  Richard Biener  <rguenther@suse.de>
24078         * doc/invoke.texi (fwhole-program): Discourage use in combination
24079         with -flto.
24081 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
24083         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
24085         PR tree-optimization/56539
24086         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
24087         instead of GSI_CONTINUE_LINKING as last argument to
24088         force_gimple_operand_gsi.  Adjust function comment.
24090         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
24091         aarch64-cores.def.
24093         PR middle-end/56548
24094         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
24095         promoted mode, convert the result back to the original mode.
24097 2013-03-06  Richard Biener  <rguenther@suse.de>
24099         PR middle-end/56294
24100         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
24101         (insert_updated_phi_nodes_compare_uids): New function.
24102         (update_ssa): Sort symbols_to_rename after UID before
24103         traversing it to insert PHI nodes.
24105 2013-03-06  Richard Biener  <rguenther@suse.de>
24107         PR middle-end/50494
24108         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
24109         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
24111         Revert
24112         2013-02-13  Richard Biener  <rguenther@suse.de>
24114         PR lto/50494
24115         * varasm.c (output_constant_def_1): Get the decl representing
24116         the constant as argument.
24117         (output_constant_def): Wrap output_constant_def_1.
24118         (make_decl_rtl): Use output_constant_def_1 with the decl
24119         representing the constant.
24120         (build_constant_desc): Optionally re-use a decl already
24121         representing the constant.
24122         (tree_output_constant_def): Adjust.
24124 2013-03-06  Joey Ye  <joey.ye@arm.com>
24126         PR lto/50293
24127         * gcc.c (convert_white_space): New function.
24128         (main): Handles white space in function name.
24130 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
24132         PR target/56529
24133         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
24134         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
24135         to SH_DIV_CALL_TABLE for TARGET_SH2.
24136         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
24137         list.
24138         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
24139         call-table options.
24141 2013-03-05  Sterling Augustine  <saugustine@google.com>
24142             Cary Coutant  <ccoutant@google.com>
24144         PR debug/55364
24145         * dwarf2out.c (resolve_addr): Don't call
24146         remove_loc_list_addr_table_entries a second time for the same
24147         expression.
24149 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24151         PR debug/56510
24152         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
24153         (avoid_complex_debug_insns): New function.
24154         (expand_debug_locations): Call it.
24156         PR rtl-optimization/56484
24157         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
24158         lifetimes of hard registers on small register class machines.
24160 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
24162         * config/microblaze/microblaze-protos.h: Rename
24163         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
24164         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
24165         fast_interrupt.
24166         (microblaze_fast_interrupt_function_p): New function.
24167         (microblaze_is_interrupt_handler): Rename to
24168         microblaze_is_interrupt_variant and add fast_interrupt check.
24169         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
24170         (save_restore_insns): Likewise.
24171         (compute_frame_size): Likewise.
24172         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
24173         (microblaze_globalize_label): Likewise.
24174         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
24175         * config/microblaze/microblaze.md: Use wrapper
24176         microblaze_is_interrupt_variant.
24178 2013-03-05  Kai Tietz  <ktietz@redhat.com>
24180         * sdbout.c (sdbout_one_type): Switch to current function's section
24181         supporting cold/hot.
24183 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
24185         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
24186         -mxl-reorder.
24188 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24190         PR middle-end/56461
24191         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
24192         if VALGRIND_GET_VBITS is defined, temporarily make object
24193         memory all defined, and restore previous valgrind addressability
24194         and definability afterwards.  Free this_object at the end.
24196         PR middle-end/56461
24197         * lra.c (lra): Call lra_clear_live_ranges if live_p,
24198         right before calling lra_create_live_ranges, also call it
24199         when clearing live_p.  Only call lra_clear_live_ranges
24200         at the end if live_p.
24202         PR middle-end/56461
24203         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
24205 2013-03-05  Richard Biener  <rguenther@suse.de>
24207         PR tree-optimization/56521
24208         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
24209         value-id.
24211 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
24213         PR c++/55135
24214         * except.h (remove_unreachable_eh_regions): New prototype.
24215         * except.c (remove_eh_handler_splicer): New function, split out
24216         of remove_eh_handler.
24217         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
24218         warning about running it on many EH regions one at a time.
24219         (remove_unreachable_eh_regions_worker): New function, walk the
24220         EH tree in depth-first order and remove non-marked regions.
24221         (remove_unreachable_eh_regions): New function.
24222         * tree-eh.c (mark_reachable_handlers): New function, split out
24223         from remove_unreachable_handlers.
24224         (remove_unreachable_handlers): Use mark_reachable_handlers and
24225         remove_unreachable_eh_regions.
24226         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
24227         and remove_unreachable_eh_regions.
24229 2013-03-05  Richard Biener  <rguenther@suse.de>
24231         PR middle-end/56525
24232         * loop-init.c (fix_loop_structure): Remove loops in two stages,
24233         not freeing them until the end.
24235 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
24237         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
24239 2013-03-05  Richard Biener  <rguenther@suse.de>
24241         PR tree-optimization/56270
24242         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
24243         of loads after scheduling an SLP instance.
24245 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24247         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
24248         tic6x.exp.
24249         (check_gcc_parallelize): Run guality.exp as a separate job from
24250         vect.exp with unsorted.exp and $(dg_target_exps) separately from
24251         struct-layout-1.exp with stackalign.exp.
24253         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
24255         PR middle-end/56461
24256         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
24257         load_index sbitmap even if some bit in it isn't set.
24259         PR middle-end/56461
24260         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
24261         (discover_iteration_bound_by_body_walk): Change queues to
24262         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
24263         spelling in comment.  Call safe_push on queues[bound_index] directly.
24264         Release queues[queue_index] in every iteration unconditionally.
24265         Release bounds vector.
24267         PR middle-end/56461
24268         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
24269         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
24270         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
24271         inner_phis vector.
24273 2013-03-05  Richard Biener  <rguenther@suse.de>
24275         PR lto/56515
24276         * tree-inline.c (remap_blocks_to_null): New function.
24277         (expand_call_inline): When expanding a call stmt without
24278         an associated block inline remap all callee blocks to NULL.
24280 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
24282         PR rtl-optimization/56494
24283         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
24284         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
24285         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
24287         PR middle-end/56461
24288         * sel-sched-ir.c (free_sched_pools): Release
24289         succs_info_pool.stack[succs_info_pool.max_top] vectors too
24290         if succs_info_pool.max_top isn't -1.
24292         PR bootstrap/56509
24293         * opts.c (opts_obstack, opts_concat): Moved to...
24294         * opts-common.c (opts_obstack, opts_concat): ... here.
24296 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
24298         PR middle-end/56461
24299         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
24301 2013-03-04  Martin Jambor  <mjambor@suse.cz>
24303         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
24304         all appropriate places.
24306 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
24308         PR tree-optimization/56424
24309         * ipa-split.c (split_function): Do not set the RSO flag if result is
24310         not by reference and its type is a register type.
24312 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
24314         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
24315         (microblaze_legitimate_pic_operand): Likewise
24316         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
24317         new function microblaze_legitimate_pic_operand
24318         * config/microblaze/microblaze-protos.h
24319         (microblaze_legitimate_pic_operand): Declare.
24321 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24323         * config/microblaze/predicates.md (call_insn_simple_operand):
24324         New predicate for supported rtx code types.
24325         * config/microblaze/microblaze.md (call_internal1): Use
24326         call_insn_simple_operand predicate.
24328 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
24330         PR middle-end/56461
24331         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
24332         partitions.ordered_remove.
24334         PR middle-end/56461
24335         * tree-vect-stmts.c (vectorizable_conversion): Don't call
24336         vec_oprnds0.create (1) for modifier == NONE.
24338         PR middle-end/56461
24339         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
24340         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
24341         vec_oprnds1 right before pushing anything to it for
24342         scalar_shift_arg.
24344         PR middle-end/56461
24345         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
24346         set nbbs to 0 instead of having separate code path.
24347         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
24348         instead of false as last argument if returning NULL.
24350 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
24352         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
24353         the attribute is now called "target" instead of "option".
24354         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
24355         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
24356         attribute/pragma name for TARGET_OPTION_VALID_P and
24357         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
24358         * doc/tm.texi: Regenerated.
24360 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
24362         * config/microblaze/microblaze.c:
24363         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
24364         * config/microblaze/microblaze.h: Add -mxl-reorder to
24365         DRIVER_SELF_SPECS.
24366         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
24367         instructions emitted if TARGET_REORDER.
24368         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
24369         or 0 for -m/-mno case, but initialises as 2 to detect default use case
24370         separately.
24372 2013-03-01  Xinliang David Li  <davidxl@google.com>
24374         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
24375         walk length.
24377 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
24379         PR middle-end/56461
24380         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
24381         vector even when returning true.  Fix up function comment formatting.
24383         PR middle-end/56461
24384         * ira-build.c (ira_loop_nodes_count): New variable.
24385         (create_loop_tree_nodes): Initialize it.
24386         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
24388         PR middle-end/56461
24389         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
24390         method on dr_chain and result_chain.
24391         * tree-vect-stmts.c (vectorizable_store): Only call
24392         result_chain.create if j == 0.
24394         PR middle-end/56461
24395         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
24396         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
24397         before overwriting it.
24399 2013-03-01  Tobias Burnus  <burnus@net-b.de>
24401         * doc/extended.texi (C Extensions): Change order in @menu
24402         to match @node.
24403         (Other MIPS Built-in Functions): Move last MIPS entry before
24404         "picoChip Built-in Functions".
24405         (SH Built-in Functions): Move after RX Built-in Functions.
24406         * doc/gcc.texi (Introduction): Change order in @menu
24407         to match @node.
24408         * doc/md.texi (Constraints): Ditto.
24409         * gty.texi (Type Information): Ditto.
24410         (User-provided marking routines for template types): Make
24411         subsection.
24412         * doc/invoke.texi (AArch64 Options): Move before
24413         "Adapteva Epiphany Options".
24415 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
24416             Jakub Jelinek  <jakub@redhat.com>
24418         PR sanitizer/56454
24419         * asan.c (gate_asan): Lookup no_sanitize_address instead of
24420         no_address_safety_analysis attribute.
24421         * doc/extend.texi (no_address_safety_attribute): Rename to
24422         no_sanitize_address attribute, mention no_address_safety_analysis
24423         attribute as deprecated alias.
24425 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
24427         PR middle-end/56461
24428         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
24429         type to vec<vec<tree> > *.
24430         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
24431         to be vec<tree> instead of vec<tree> *, set vec_defs
24432         to vNULL and call vec_defs.create (number_of_vects), adjust other
24433         uses of vec_defs.
24434         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
24435         vectorizable_condition): Adjust vect_get_slp_defs callers.
24437 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24439         * config/aarch64/aarch64.c
24440         (aarch64_float_const_representable): Remove unused variable.
24442 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24444         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
24446 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24448         * config/aarch64/aarch64-builtins.c
24449         (aarch64_init_simd_builtins): Make static.
24451 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
24453         * config/aarch64/aarch64.c
24454         (aarch64_simd_make_constant): Make static.
24456 2013-02-28  Martin Jambor  <mjambor@suse.cz>
24458         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
24459         with no initialization to the RHS of debug statements.
24461 2013-02-28  Martin Jambor  <mjambor@suse.cz>
24463         PR tree-optimization/56294
24464         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
24465         Adjust dumping.
24466         (get_access_replacement): Do not call create_access_replacement.
24467         Assert a replacement exists.
24468         (get_repl_default_def_ssa_name): Create the replacement declaration
24469         itself.
24471 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
24473         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
24474         final_end_function.
24476 2013-02-28  Marek Polacek  <polacek@redhat.com>
24478         PR rtl-optimization/56466
24479         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
24480         if we're changing a loop.
24481         (peel_loops_completely): Likewise.
24483 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
24485         PR c++/55813
24486         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
24488 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24490         PR target/56445
24491         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
24492         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
24493         INTX_FTYPE_FX, FX_FTYPE_INTX.
24494         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
24496 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24498         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
24499         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
24500         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
24501         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
24502         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
24503         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
24504         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
24505         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
24506         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
24507         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
24508         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
24509         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
24510         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
24511         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
24512         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
24513         (avrxmega6): Increase max flash segments from 5 to 6.
24514         * config/avr/t-multilib: Regenerate.
24515         * config/avr/avr-tables.opt: Regenerate.
24516         * doc/avr-mmcu.texi: Regenerate.
24518 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
24520         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
24521         (avr_device_to_arch): Rename to avr_device_to_ld.
24522         (avr_device_to_as): New prototype.
24523         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
24524         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
24525         * config/avr/driver-avr.c (avr_device_to_as): New.
24526         (avr_device_to_arch): Rename to avr_device_to_ld.
24528 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24530         PR middle-end/56461
24531         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
24532         method on dr_chain and result_chain.
24534         PR middle-end/56461
24535         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
24536         pointer_set_destroy on not_executed_last_iteration.
24538         PR middle-end/56461
24539         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
24541         PR middle-end/56461
24542         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
24543         FOR_EACH_DEFINED_FUNCTION when freeing state.
24545         PR middle-end/56461
24546         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
24547         pool_free.
24548         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
24549         overwriting it.
24551         PR middle-end/56461
24552         * ipa-cp.c (decide_whether_version_node): Call vec_free on
24553         known_aggs[i].items and release known_aggs vector.
24555         PR middle-end/56461
24556         * ipa-reference.c (propagate): Free node_info even for alias nodes.
24558 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24560         * config/microblaze/microblaze.c (microblaze_emit_compare):
24561         Use xor for EQ/NE comparisions.
24562         * config/microblaze/microblaze.md (cstoresf4): Add constraints
24563         (cbranchsf4): Adjust operator to comparison_operator.
24565 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24567         PR middle-end/56461
24568         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
24569         vector.
24570         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
24571         vec_safe_push, always update *slot.
24572         (redirect_edge_var_map_clear): Use vec_free.
24573         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
24574         (free_var_map_entry): Use vec_free.
24575         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
24576         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
24578 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
24580         PR middle-end/45472
24581         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
24582         when the may_trap_p bit of the exprs being merged differs.
24583         Reorder tests for speculativeness in the logical and operator.
24585 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
24587         * incpath.c (add_standard_paths): Use reconcat instead of concat
24588         where appropriate and avoid leaking memory.
24590         * opts.h: Include obstack.h.
24591         (opts_concat): New prototype.
24592         (opts_obstack): New declaration.
24593         * opts.c (opts_concat): New function.
24594         (opts_obstack): New variable.
24595         (init_options_struct): Call gcc_init_obstack on opts_obstack.
24596         (finish_options): Use opts_concat instead of concat
24597         and XOBNEWVEC instead of XNEWVEC.
24598         * opts-common.c (generate_canonical_option, decode_cmdline_option,
24599         generate_option): Likewise.
24600         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
24601         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
24603         PR target/56455
24604         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
24605         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
24607 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24609         PR middle-end/56461
24610         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
24612 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24614         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
24615         (arm_block_move_unaligned_straight): Likewise.
24616         (arm_adjust_block_mem): Likewise.
24618 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
24620         PR target/48901
24621         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
24622         temp, cond and label.
24623         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
24625         PR target/52500
24626         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
24627         * config/c6x/c6x.h (dbx_register_map): Update declaration.
24629         PR target/52501
24630         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
24631         of prologue/epilogue functions.
24633         PR target/52550
24634         * config/tilegx/tilegx.c (tilegx_expand_prologue):
24635         Remove unused variable cfa_offset.
24636         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24638         PR target/54639
24639         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24640         type promotion to unsigned.
24642         PR target/54640
24643         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24644         for HOST_WIDE_INT of 32 bit / same size as int.
24645         (arm_block_move_unaligned_straight): Likewise.
24646         (arm_adjust_block_mem): Likewise.
24648         PR target/54662
24649         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24650         ALL_CFLAGS.
24652 2013-02-26  Marek Polacek  <polacek@redhat.com>
24654         PR tree-optimization/56426
24655         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24657 2013-02-26  Richard Biener  <rguenther@suse.de>
24659         PR target/56444
24660         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24661         unused variable loops.
24663 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24665         PR tree-optimization/56448
24666         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24667         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24668         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24669         later operands of the references, or even first operand for
24670         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24672         PR tree-optimization/56443
24673         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24674         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24675         to type_for_mode langhook.
24677 2013-02-25  Matt Turner  <mattst88@gmail.com>
24679         * doc/invoke.texi: Document r4700.
24681 2013-02-25  Richard Biener  <rguenther@suse.de>
24683         PR tree-optimization/56175
24684         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24685         split out from ...
24686         (simplify_bitwise_binary): ... here.  Also guard the conversion
24687         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24689 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24691         Revert:
24692         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24693             Maciej W. Rozycki  <macro@codesourcery.com>
24694             Tom de Vries  <tom@codesourcery.com>
24695             Nathan Sidwell  <nathan@codesourcery.com>
24696             Iain Sandoe  <iain@codesourcery.com>
24697             Nathan Froyd  <froydnj@codesourcery.com>
24698             Chao-ying Fu  <fu@mips.com>
24700         * doc/extend.texi: (micromips, nomicromips, nocompression):
24701         Document new function attributes.
24702         * doc/invoke.texi (minterlink-compressed, mmicromips,
24703         m14k, m14ke, m14kec): Document new options.
24704         (minterlink-mips16): Update documentation.
24705         * doc/md.texi (ZC, ZD): Document new constraints.
24706         * configure.ac (gcc_cv_as_micromips): Check if linker
24707         supports the .set micromips directive.
24708         * configure: Regenerate.
24709         * config.in: Regenerate.
24710         * config/mips/mips-tables.opt: Regenerate.
24711         * config/mips/micromips.md: New file.
24712         * constraints.md (ZC, AD): New constraints.
24713         * config/mips/predicates.md (movep_src_register): New predicate.
24714         (movep_src_operand): New predicate.
24715         (non_volatile_mem_operand): New predicate.
24716         * config/mips/mips.md (multimem): New type.
24717         (length): Differentiate between 17-bit and 18-bit branch offsets.
24718         (MOVEP1, MOVEP2): New mode iterator.
24719         (mov_<load>l): Use ZC constraint.
24720         (mov_<load>r): Likewise.
24721         (mov_<store>l): Likewise.
24722         (mov_<store>r): Likewise.
24723         (*branch_equality<mode>_inverted): Add microMIPS support.
24724         (*branch_equality<mode>): Likewise.
24725         (*jump_absolute): Likewise.
24726         (indirect_jump_<mode>): Likewise.
24727         (tablejump_<mode>): Likewise.
24728         (<optab>_internal): Likewise.
24729         (sibcall_internal): Likewise.
24730         (sibcall_value_internal): Likewise.
24731         (prefetch): Use constraint ZD.
24732         * config/mips/mips.opt (minterlink-compressed): New option.
24733         (minterlink-mips16): Now an alias for minterlink-compressed.
24734         (mmicromips): New option.
24735         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24736         (compare_and_swap_12): Likewise.
24737         (sync_add<mode>): Likewise.
24738         (sync_<optab>_12): Likewise.
24739         (sync_old_<optab>_12): Likewise.
24740         (sync_new_<optab>_12): Likewise.
24741         (sync_nand_12): Likewise.
24742         (sync_old_nand_12): Likewise.
24743         (sync_new_nand_12): Likewise.
24744         (sync_sub<mode>): Likewise.
24745         (sync_old_add<mode>): Likewise.
24746         (sync_old_sub<mode>): Likewise.
24747         (sync_new_add<mode>): Likewise.
24748         (sync_new_sub<mode>): Likewise.
24749         (sync_<optab><mode>): Likewise.
24750         (sync_old_<optab><mode>): Likewise.
24751         (sync_new_<optab><mode>): Likewise.
24752         (sync_nand<mode>): Likewise.
24753         (sync_old_nand<mode>): Likewise.
24754         (sync_new_nand<mode>): Likewise.
24755         (sync_lock_test_and_set<mode>): Likewise.
24756         (test_and_set_12): Likewise.
24757         (atomic_compare_and_swap<mode>): Likewise.
24758         (atomic_exchange<mode>_llsc): Likewise.
24759         (atomic_fetch_add<mode>_llsc): Likewise.
24760         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24761         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24762         (umips_save_restore_pattern_p): Likewise.
24763         (umips_load_store_pair_p): Likewise.
24764         (umips_output_load_store_pair): Likewise.
24765         (umips_movep_target_p): Likewise.
24766         (umips_12bit_offset_address_p): Likewise.
24767         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24768         (mips_base_mips16): Rename this...
24769         (mips_base_compression_flags): ...to this. Update all uses.
24770         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24771         (mips_mips16_decl_p): Delete.
24772         (mips_nomips16_decl_p): Delete.
24773         (mips_get_compress_on_flags): New function.
24774         (mips_get_compress_off_flags): New function.
24775         (mips_get_compress_mode): New function.
24776         (mips_get_compress_on_name): New function.
24777         (mips_get_compress_off_name): New function.
24778         (mips_insert_attributes): Support multiple compression types.
24779         (mips_merge_decl_attributes): Likewise.
24780         (umips_12bit_offset_address_p): New function.
24781         (mips_start_function_definition): Emit .set micromips directive.
24782         (mips_call_may_need_jalx_p): New function.
24783         (mips_function_ok_for_sibcall): Add microMIPS support.
24784         (mips_print_operand_punctuation): Support short delay slots and
24785         compact jumps.
24786         (umips_swm_mask, umips_swm_encoding): New.
24787         (umips_build_save_restore): New function.
24788         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24789         (was_mips16_p): Remove.
24790         (old_compression_mode): New.
24791         (mips_set_compression_mode): New function.
24792         (mips_set_current_function): Add microMIPS support.
24793         (mips_option_override): Likewise.
24794         (umips_save_restore_pattern_p): New function.
24795         (umips_output_save_restore): New function.
24796         (umips_load_store_pair_p_1): New function.
24797         (umips_load_store_pair_p): New function.
24798         (umips_output_load_store_pair_1): New function.
24799         (umips_output_load_store_pair): New function.
24800         (umips_movep_target_p) New function.
24801         (mips_prepare_pch_save): Add microMIPS support.
24802         * config/mips/mips.h (TARGET_COMPRESSION): New.
24803         (TARGET_CPU_CPP_BUILTINS): Update macro
24804         to use new compression flags and to support microMIPS.
24805         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24806         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24807         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24808         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24809         (ASM_SPEC): Support mmicromips and mno-micromips.
24810         (M16STORE_REG_P): New macro.
24811         (MIPS_CALL): Support TARGET_MICROMIPS.
24812         (MICROMIPS_J): New macro.
24813         (mips_base_mips16): Rename this...
24814         (mips_base_compression_flags): ...to this.
24815         (UMIPS_12BIT_OFFSET_P): New macro.
24816         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24817         (MULTILIB_DIRNAMES): Likewise.
24819 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24821         PR rtl-optimization/56131
24822         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24823         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24824         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24826 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24828         * doc/invoke.texi (-fsanitize=): Move from optimization
24829         to debugging options.
24831 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24833         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24835 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24836             Alexander Monakov  <amonakov@ispras.ru>
24838         PR middle-end/56077
24839         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24840         flush pending lists also on non-jumps.  Adjust comment.
24842 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24843             Maciej W. Rozycki  <macro@codesourcery.com>
24844             Tom de Vries  <tom@codesourcery.com>
24845             Nathan Sidwell  <nathan@codesourcery.com>
24846             Iain Sandoe  <iain@codesourcery.com>
24847             Nathan Froyd  <froydnj@codesourcery.com>
24848             Chao-ying Fu  <fu@mips.com>
24850         * doc/extend.texi: (micromips, nomicromips, nocompression):
24851         Document new function attributes.
24852         * doc/invoke.texi (minterlink-compressed, mmicromips,
24853         m14k, m14ke, m14kec): Document new options.
24854         (minterlink-mips16): Update documentation.
24855         * doc/md.texi (ZC, ZD): Document new constraints.
24856         * configure.ac (gcc_cv_as_micromips): Check if linker
24857         supports the .set micromips directive.
24858         * configure: Regenerate.
24859         * config.in: Regenerate.
24860         * config/mips/mips-tables.opt: Regenerate.
24861         * config/mips/micromips.md: New file.
24862         * constraints.md (ZC, AD): New constraints.
24863         * config/mips/predicates.md (movep_src_register): New predicate.
24864         (movep_src_operand): New predicate.
24865         (non_volatile_mem_operand): New predicate.
24866         * config/mips/mips.md (multimem): New type.
24867         (length): Differentiate between 17-bit and 18-bit branch offsets.
24868         (MOVEP1, MOVEP2): New mode iterator.
24869         (mov_<load>l): Use ZC constraint.
24870         (mov_<load>r): Likewise.
24871         (mov_<store>l): Likewise.
24872         (mov_<store>r): Likewise.
24873         (*branch_equality<mode>_inverted): Add microMIPS support.
24874         (*branch_equality<mode>): Likewise.
24875         (*jump_absolute): Likewise.
24876         (indirect_jump_<mode>): Likewise.
24877         (tablejump_<mode>): Likewise.
24878         (<optab>_internal): Likewise.
24879         (sibcall_internal): Likewise.
24880         (sibcall_value_internal): Likewise.
24881         (prefetch): Use constraint ZD.
24882         * config/mips/mips.opt (minterlink-compressed): New option.
24883         (minterlink-mips16): Now an alias for minterlink-compressed.
24884         (mmicromips): New option.
24885         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24886         (compare_and_swap_12): Likewise.
24887         (sync_add<mode>): Likewise.
24888         (sync_<optab>_12): Likewise.
24889         (sync_old_<optab>_12): Likewise.
24890         (sync_new_<optab>_12): Likewise.
24891         (sync_nand_12): Likewise.
24892         (sync_old_nand_12): Likewise.
24893         (sync_new_nand_12): Likewise.
24894         (sync_sub<mode>): Likewise.
24895         (sync_old_add<mode>): Likewise.
24896         (sync_old_sub<mode>): Likewise.
24897         (sync_new_add<mode>): Likewise.
24898         (sync_new_sub<mode>): Likewise.
24899         (sync_<optab><mode>): Likewise.
24900         (sync_old_<optab><mode>): Likewise.
24901         (sync_new_<optab><mode>): Likewise.
24902         (sync_nand<mode>): Likewise.
24903         (sync_old_nand<mode>): Likewise.
24904         (sync_new_nand<mode>): Likewise.
24905         (sync_lock_test_and_set<mode>): Likewise.
24906         (test_and_set_12): Likewise.
24907         (atomic_compare_and_swap<mode>): Likewise.
24908         (atomic_exchange<mode>_llsc): Likewise.
24909         (atomic_fetch_add<mode>_llsc): Likewise.
24910         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24911         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24912         (umips_save_restore_pattern_p): Likewise.
24913         (umips_load_store_pair_p): Likewise.
24914         (umips_output_load_store_pair): Likewise.
24915         (umips_movep_target_p): Likewise.
24916         (umips_12bit_offset_address_p): Likewise.
24917         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24918         (mips_base_mips16): Rename this...
24919         (mips_base_compression_flags): ...to this. Update all uses.
24920         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24921         (mips_mips16_decl_p): Delete.
24922         (mips_nomips16_decl_p): Delete.
24923         (mips_get_compress_on_flags): New function.
24924         (mips_get_compress_off_flags): New function.
24925         (mips_get_compress_mode): New function.
24926         (mips_get_compress_on_name): New function.
24927         (mips_get_compress_off_name): New function.
24928         (mips_insert_attributes): Support multiple compression types.
24929         (mips_merge_decl_attributes): Likewise.
24930         (umips_12bit_offset_address_p): New function.
24931         (mips_start_function_definition): Emit .set micromips directive.
24932         (mips_call_may_need_jalx_p): New function.
24933         (mips_function_ok_for_sibcall): Add microMIPS support.
24934         (mips_print_operand_punctuation): Support short delay slots and
24935         compact jumps.
24936         (umips_swm_mask, umips_swm_encoding): New.
24937         (umips_build_save_restore): New function.
24938         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24939         (was_mips16_p): Remove.
24940         (old_compression_mode): New.
24941         (mips_set_compression_mode): New function.
24942         (mips_set_current_function): Add microMIPS support.
24943         (mips_option_override): Likewise.
24944         (umips_save_restore_pattern_p): New function.
24945         (umips_output_save_restore): New function.
24946         (umips_load_store_pair_p_1): New function.
24947         (umips_load_store_pair_p): New function.
24948         (umips_output_load_store_pair_1): New function.
24949         (umips_output_load_store_pair): New function.
24950         (umips_movep_target_p) New function.
24951         (mips_prepare_pch_save): Add microMIPS support.
24952         * config/mips/mips.h (TARGET_COMPRESSION): New.
24953         (TARGET_CPU_CPP_BUILTINS): Update macro
24954         to use new compression flags and to support microMIPS.
24955         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24956         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24957         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24958         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24959         (ASM_SPEC): Support mmicromips and mno-micromips.
24960         (M16STORE_REG_P): New macro.
24961         (MIPS_CALL): Support TARGET_MICROMIPS.
24962         (MICROMIPS_J): New macro.
24963         (mips_base_mips16): Rename this...
24964         (mips_base_compression_flags): ...to this.
24965         (UMIPS_12BIT_OFFSET_P): New macro.
24966         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24967         (MULTILIB_DIRNAMES): Likewise.
24969 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24971         PR target/52555
24972         * target-globals.c (save_target_globals): For init_reg_sets and
24973         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24975 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
24977         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24978         * config/aarch64/t-aarch64
24979         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24981 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24983         PR inline-asm/56148
24984         * lra-constraints.c (process_alt_operands): Reload operand
24985         conflicting with earlier clobber only if no more other conflicting
24986         operands.
24988 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24990         PR sanitizer/56393
24991         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24992         if not linking a shared library.
24994 2013-02-22  Seth LaForge  <sethml@google.com>
24996         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24998 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
25000         * config/arm/arm.md (split for extendsidi): Update condition.
25001         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
25002         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
25003         (qhs_zextenddi_cstr): Likewise.
25005 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
25007         PR middle-end/56420
25008         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
25009         avoid signed wrapping.
25010         (expand_mult): Handle properly multiplication by
25011         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
25012         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
25013         in the compiler if coeff is HOST_WIDE_INT_MIN.
25014         (expand_divmod): Don't make ext_op1 static, change it's type to
25015         uhwi.  Avoid undefined behavior in -INTVAL (op1).
25017         PR rtl-optimization/50339
25018         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
25019         field.
25020         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
25021         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
25022         into splitting_ashiftrt field.
25023         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
25024         ASHIFTRT.
25025         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
25026         choices.
25028 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
25030         PR middle-end/56108
25031         * trans-mem.c (execute_tm_mark): Do not expand transactions that
25032         are sure to go irrevocable.
25034 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
25036         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
25037         scalars are valid operands.
25039 2013-02-21  Martin Jambor  <mjambor@suse.cz>
25041         PR tree-optimization/56310
25042         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
25043         only matching indices and non-negative final offsets.
25044         (intersect_aggregates_with_edge): Pass src_idx to
25045         agg_replacements_to_vector.  Pass src_idx insstead of index to
25046         intersect_with_agg_replacements.
25048 2013-02-21  Martin Jambor  <mjambor@suse.cz>
25050         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
25051         instead of hard-wired defaults.
25053 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
25055         * doc/invoke.texi (MIPS Options): Update documentation of the
25056         floating-point multiply-accumulate instruction restrictions.
25058 2013-02-21  Kostya Serebryany  <kcc@google.com>
25060         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
25061         asan_shadow_offset on x86_64 linux.
25063 2013-02-21  Richard Biener  <rguenther@suse.de>
25065         PR tree-optimization/56415
25066         Revert
25067         2013-02-11  Richard Biener  <rguenther@suse.de>
25069         PR tree-optimization/56273
25070         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
25071         first VRP run.
25073 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
25075         PR bootstrap/56258
25076         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
25077         instead of @itemx.
25079         PR inline-asm/56405
25080         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
25081         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
25083 2013-02-20  Jan Hubicka  <jh@suse.cz>
25085         PR tree-optimization/56265
25086         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
25087         when target is referenced for first time.
25089 2013-02-20  Richard Biener  <rguenther@suse.de>
25091         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
25092         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
25093         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
25094         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
25095         not return anything.
25096         (rename_ssa_copies): Do not remove unused locals.
25097         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
25098         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
25099         * passes.c (execute_function_todo): Do not schedule unused locals
25100         removal if cleanup_tree_cfg did something.
25101         * tree-ssa-live.c (remove_unused_locals): Dump statistics
25102         about the number of removed locals.
25104 2013-02-20  Richard Biener  <rguenther@suse.de>
25106         PR tree-optimization/56398
25107         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
25109 2013-02-20  Martin Jambor  <mjambor@suse.cz>
25111         PR tree-optimization/55334
25112         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
25113         restricted pointers to arrays.
25115 2013-02-20  Richard Biener  <rguenther@suse.de>
25116             Jakub Jelinek  <jakub@redhat.com>
25118         PR tree-optimization/56396
25119         * tree-ssa-ccp.c (n_const_val): New static variable.
25120         (get_value): Return NULL for SSA names we don't have a lattice
25121         entry for.
25122         (ccp_initialize): Initialize n_const_val.
25123         * tree-ssa-copy.c (n_copy_of): New static variable.
25124         (init_copy_prop): Initialize n_copy_of.
25125         (get_value): Return NULL_TREE for SSA names we don't have a
25126         lattice entry for.
25128 2013-02-20  Martin Jambor  <mjambor@suse.cz>
25130         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
25132 2013-02-20  Richard Biener  <rguenther@suse.de>
25134         * genpreds.c (write_lookup_constraint): Do not compare first
25135         letter of the constraint again.
25137 2013-02-20  Richard Biener  <rguenther@suse.de>
25139         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
25140         and ceil_log2.
25141         (get_use_iv_cost): Terminate hashtable walk when coming across
25142         an empty entry.
25144 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
25146         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
25147         reassociation for avx2 targets.
25149 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25151         * config/microblaze/microblaze.c: microblaze_has_clz = 0
25152         Add version check for v8.10.a to enable microblaze_has_clz
25153         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
25154         version and TARGET_PATTERN_COMPARE check
25155         * config/microblaze/microblaze.md: New clzsi2 instruction
25157 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25159         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
25160         function before branching.
25162 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
25164         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
25165         DUMP_INSN_RTX_UID.
25166         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
25168 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
25170         PR middle-end/55889
25171         * sel-sched.c: Include ira.h.
25172         (implicit_clobber_conflict_p): New function.
25173         (moveup_expr): Use it.
25174         * Makefile.in (sel-sched.o): Depend on ira.h.
25176 2013-02-19  Richard Biener  <rguenther@suse.de>
25178         PR tree-optimization/56384
25179         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
25180         (vn_hash_type): Split out from ...
25181         (vn_hash_constant_with_type): ... here.
25182         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
25183         (vn_phi_eq): Compare types from vn_phi_s structure.
25184         (vn_phi_lookup): Populate vn_phi_s type.
25185         (vn_phi_insert): Likewise.
25187 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
25189         PR tree-optimization/56350
25190         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
25191         if haven't found reduction or nested cycle operand, rather than
25192         asserting we must find it.
25194         PR tree-optimization/56381
25195         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
25196         to fold_build3.
25198 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
25199             Jakub Jelinek  <jakub@redhat.com>
25201         PR target/52555
25202         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
25203         (swap_optab_enable): Same.
25204         (init_all_optabs): Use argument instead of global.
25205         * tree.h (struct tree_optimization_option): New field target_optabs.
25206         * expr.h (init_all_optabs): Add argument to prototype.
25207         (TREE_OPTIMIZATION_OPTABS): New.
25208         (save_optabs_if_changed): Protoize.
25209         * optabs.h: Declare this_fn_optabs.
25210         * optabs.c (save_optabs_if_changed): New.
25211         Declare this_fn_optabs.
25212         (init_optabs): Add argument to init_all_optabs() call.
25213         * function.c (invoke_set_current_function_hook): Handle per
25214         function optabs.
25215         * function.h (struct function): New field optabs.
25216         * config/mips/mips.c (mips_set_mips16_mode): Handle when
25217         optimization_current_node has changed.
25218         * target-globals.h (save_target_globals_default_opts): Protoize.
25219         * target-globals.c (save_target_globals_default_opts): New.
25221 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25223         PR target/56347
25224         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
25225         registers %fr12 and %fr12R as call used.
25227         PR target/56214
25228         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
25229         and HImode, require all displacements to be an integer multiple of
25230         their mode size.
25231         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
25232         only allow QImode and HImode when reload is in progress and strict is
25233         true.  Likewise for symbolic addresses.  Use base14_operand to check
25234         displacements in REG+BASE addresses.
25236 2013-02-18  Richard Biener  <rguenther@suse.de>
25238         PR tree-optimization/56366
25239         * tree-vect-loop.c (get_initial_def_for_induction): Properly
25240         handle sign-conversion of outer-loop initial induction value.
25242 2013-02-18  Richard Biener  <rguenther@suse.de>
25244         PR middle-end/56349
25245         * cfghooks.c (merge_blocks): If we merge a latch into another
25246         block adjust references to it.
25247         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
25248         (verify_loop_structure): Verify that a recorded latch is in fact
25249         a latch.
25251 2013-02-18  Richard Biener  <rguenther@suse.de>
25253         PR tree-optimization/56321
25254         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
25255         order SSA name release and virtual operand unlinking.
25257 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25259         * config/microblaze/microblaze.md (save_stack_block): Define.
25260         (restore_stack_block): Likewise.
25262 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25264         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
25265         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
25266         * config/microblaze/microblaze.c (microblaze_option_override):
25267         Bail out early for PIC modes when target does not support PIC.
25269 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25271         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
25272         Replace with a microblaze version.
25273         (microblaze_trampoline_init): Adapt for microblaze.
25274         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
25275         microblaze.
25277 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
25278             Dodji Seketeli  <dodji@redhat.com>
25280         PR asan/56330
25281         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
25282         (instrument_mem_region_access): Do not forget to always put
25283         instrumentation of the of 'base' and 'base + len' in a "if (len !=
25284         0) statement, even for cases where either 'base' or 'base + len'
25285         are not instrumented -- because they have been previously
25286         instrumented.  Simplify the logic by putting all the statements
25287         instrument 'base + len' inside a sequence, and then insert that
25288         sequence right before the current insertion point.  Then, to
25289         instrument 'base + len', just get an iterator on that statement.
25290         And do not forget to update the pointer to iterator the function
25291         received as argument.
25293 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
25295         PR rtl-optimization/56348
25296         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
25298 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
25300         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
25301         (clean_graph_dump_file): Pass base to start_graph_dump.
25303 2013-02-14  Richard Henderson  <rth@redhat.com>
25305         PR target/55941
25306         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
25308 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
25310         * collect2-aix.h: Define F_LOADONLY.
25312 2013-02-14  Richard Biener  <rguenther@suse.de>
25314         PR lto/50494
25315         * varasm.c (output_constant_def_1): Get the decl representing
25316         the constant as argument.
25317         (output_constant_def): Wrap output_constant_def_1.
25318         (make_decl_rtl): Use output_constant_def_1 with the decl
25319         representing the constant.
25320         (build_constant_desc): Optionally re-use a decl already
25321         representing the constant.
25322         (tree_output_constant_def): Adjust.
25324 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
25326         Fix an asan crash
25327         * asan.c (instrument_builtin_call):  Really put the length of the
25328         second source argument into src1_len.
25330 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
25332         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
25333         argument.  If it is false, don't create edge from then_bb to
25334         fallthru_bb.
25335         (insert_if_then_before_iter): Pass true to it.
25336         (build_check_stmt): Pass false to it.
25337         (transform_statements): Flush hash table only on extended basic
25338         block boundaries, rather than at the beginning of every bb.
25339         Don't flush hash table on nonfreeing_call_p calls.
25340         * tree-flow.h (nonfreeing_call_p): New prototype.
25341         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
25343 2013-02-13  David S. Miller  <davem@davemloft.net>
25345         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
25347 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
25349         PR target/56184
25350         * ira.c (max_regno_before_ira): Move from ...
25351         (ira): ... here.
25352         (fix_reg_equiv_init): Use max_regno_before_ira instead of
25353         vec_safe_length.
25355 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
25357         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
25359 2013-02-13  Richard Biener  <rguenther@suse.de>
25361         PR lto/56295
25362         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
25363         globals in MEM_REFs.
25365 2013-02-13  Richard Biener  <rguenther@suse.de>
25367         * loop-init.c (loop_optimizer_init): Clear loop state when
25368         re-initializing preserved loops.
25369         * loop-unswitch.c (unswitch_single_loop): Return whether
25370         we unswitched the loop.  Do not verify loop state here.
25371         (unswitch_loops): When we unswitched a loop discover new loops.
25373 2013-02-13  Kostya Serebryany  <kcc@google.com>
25375         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
25376         on x86_64 linux.
25377         * sanitizer.def: Rename __asan_init to __asan_init_v1.
25379 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
25381         Avoid instrumenting duplicated memory access in the same basic block
25382         * Makefile.in (asan.o): Add new dependency on hash-table.h
25383         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
25384         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
25385         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
25386         (free_mem_ref_resources, has_mem_ref_been_instrumented)
25387         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
25388         (get_mem_ref_of_assignment): New functions.
25389         (get_mem_refs_of_builtin_call): Extract from
25390         instrument_builtin_call and tweak a little bit to make it fit with
25391         the new signature.
25392         (instrument_builtin_call): Use the new
25393         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
25394         of is_gimple_builtin_call.
25395         (instrument_derefs, instrument_mem_region_access): Insert the
25396         instrumented memory reference into the hash table.
25397         (maybe_instrument_assignment): Renamed instrument_assignment into
25398         this, and change it to advance the iterator when instrumentation
25399         actually happened and return true in that case.  This makes it
25400         homogeneous with maybe_instrument_assignment, and thus give a
25401         chance to callers to be more 'regular'.
25402         (transform_statements): Clear the memory reference hash table
25403         whenever we enter a new BB, when we cross a function call, or when
25404         we are done transforming statements.  Use
25405         maybe_instrument_assignment instead of instrumentation.  No more
25406         need to special case maybe_instrument_assignment and advance the
25407         iterator after calling it; it's now handled just like
25408         maybe_instrument_call.  Update comment.
25410 2013-02-13  Richard Biener  <rguenther@suse.de>
25412         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
25413         Fix loop discovery code.
25415 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
25417         PR inline-asm/56148
25418         * lra-constraints.c (process_alt_operands): Match early clobber
25419         operand with itself.  Check conflicts with earlyclobber only if
25420         the operand is not reloaded.  Prefer to reload conflicting operand
25421         if earlyclobber and matching operands are the same.
25423 2013-02-12  Richard Biener  <rguenther@suse.de>
25425         PR lto/56297
25426         * lto-streamer-out.c (write_symbol): Do not output symbols
25427         for hard register variables.
25429 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
25431         PR target/54222
25432         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
25433         (umulsidi3_insn, mulsidi3_insn): New insns.
25435 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
25437         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
25438         (struct tune_params): Add vec_costs field.
25439         * config/arm/arm.c (arm_builtin_vectorization_cost)
25440         (arm_add_stmt_cost): New functions.
25441         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
25442         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
25443         (arm_default_vec_cost): New struct of type cpu_vec_costs.
25444         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
25445         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
25446         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
25447         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
25449 2013-02-12  Richard Biener  <rguenther@suse.de>
25451         PR lto/56295
25452         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
25453         decls again if possible.
25455 2013-02-12  Richard Biener  <rguenther@suse.de>
25457         PR middle-end/56288
25458         * tree-ssa.c (verify_ssa_name): Fix check, move
25459         SSA_NAME_IN_FREE_LIST check up.
25461 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
25462             Steven Bosscher   <steven@gcc.gnu.org>
25464         PR rtl-optimization/56151
25465         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
25466         equal to op0 or op1, and last_insn pattern is CODE operation
25467         with MEM dest and one of the operands matches that MEM.
25469 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
25471         * doc/extend.texi: Document Function Multiversioning and "default"
25472         parameter string to target attribute.
25473         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
25474         target attribute parameter is "default".
25475         (ix86_compare_version_priority): Remove checks for target attribute.
25476         (ix86_mangle_function_version_assembler_name): Change error to sorry.
25477         Remove check for target attribute equal to NULL. Add assert.
25478         (ix86_generate_version_dispatcher_body): Change error to sorry.
25480 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
25481             Jack Howarth  <howarth@bromo.med.uc.edu>
25482             Patrick Marlier  <patrick.marlier@gmail.com>
25484         PR libitm/55693
25485         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
25486         define ENDFILE_SPEC as TM_DESTRUCTOR.
25487         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
25489 2013-02-11  Alexander Potapenko  <glider@google.com>
25490             Jack Howarth  <howarth@bromo.med.uc.edu>
25491             Jakub Jelinek  <jakub@redhat.com>
25493         PR sanitizer/55617
25494         * config/darwin.c (cdtor_record): Rename ctor_record.
25495         (sort_cdtor_records): Rename sort_ctor_records.
25496         (finalize_dtors): New routine to sort destructors by
25497         priority before use in assemble_integer.
25498         (machopic_asm_out_destructor): Use finalize_dtors if needed.
25500 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
25502         PR rtl-optimization/56275
25503         * simplify-rtx.c (avoid_constant_pool_reference): Check that
25504         offset is non-negative and less than cmode size before
25505         calling simplify_subreg.
25507 2013-02-11  Richard Biener  <rguenther@suse.de>
25509         PR tree-optimization/56264
25510         * cfgloop.h (fix_loop_structure): Adjust prototype.
25511         * loop-init.c (fix_loop_structure): Return the number of
25512         newly discovered loops.
25513         * tree-cfgcleanup.c (repair_loop_structures): When new loops
25514         are discovered, do a full loop-closed SSA rewrite.
25516 2013-02-11  Richard Biener  <rguenther@suse.de>
25518         PR tree-optimization/56273
25519         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
25520         first VRP run.
25521         (check_array_ref): Fix missing newline in dumps.
25522         (search_for_addr_array): Likewise.
25524 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
25526         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
25528 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
25530         PR target/56256
25531         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
25533 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
25535         PR rtl-optimization/56246
25536         * lra-constraints.c (simplify_operand_subreg): Try to reuse
25537         reload pseudo.
25538         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
25539         constraints are satisfied.
25541 2013-02-08  Jeff Law  <law@redhat.com>
25543         PR debug/53948
25544         * emit-rtl.c (reg_is_parm_p): New function.
25545         * regs.h (reg_is_parm_p): New prototype.
25546         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
25547         callee-clobbered registers.
25549 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
25551         PR target/56043
25552         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
25553         If there is no implicit builtin declaration, just return NULL.
25555 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25557         * config/i386/sse.md (FMAMODEM): New mode iterator.
25558         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
25559         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
25561 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25563         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
25564         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
25565         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
25567 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25569         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
25570         (microblaze*-*-elf): Likewise.
25571         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
25572         LINK_SPEC.
25573         * config/microblaze/microblaze-c.c: Add builtin defines for
25574         _LITTLE_ENDIAN and _BIG_ENDIAN.
25575         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
25576         add to TARGET_DEFAULT flags.
25577         Expand ASM_SPEC and LINK_SPEC.
25578         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
25579         * config/microblaze/microblaze.md: Update extendsidi2 and
25580         movdi_internal instructions to use low-order / high-order reg
25581         print_operands.
25582         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
25583         options and inversemask / mask of LITTLE_ENDIAN.
25584         * config/microblaze/t-microblaze: Expand multilib options to
25585         include mlittle-endian (le) and update exceptions patterns.
25587 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
25589         PR rtl-optimization/56195
25590         * lra-constraints.c (get_reload_reg): Don't reuse regs
25591         if they have smaller mode than requested, if they have
25592         wider mode than requested, try to return a SUBREG.
25594         PR tree-optimization/56250
25595         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
25596         if type is unsigned and code isn't MULT_EXPR.
25598 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25600         PR tree-optimization/56064
25601         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
25602         bits according to mode.
25603         * fixed-value.h (fixed_from_double_int)
25604         (const_fixed_from_double_int): Adjust comments.
25606 2013-02-08  Richard Biener  <rguenther@suse.de>
25608         PR lto/56231
25609         * lto-streamer.h (struct data_in): Remove current_file, current_line
25610         and current_col members.
25611         * lto-streamer-out.c (lto_output_location): Stream changed bits
25612         en-block for efficiency.
25613         * lto-streamer-in.c (clear_line_info): Remove.
25614         (lto_input_location): Cache current file, line and column
25615         globally via local statics.  Read changed bits en-block.
25616         (input_function): Do not call clear_line_info.
25617         (lto_read_body): Likewise.
25618         (lto_input_toplevel_asms): Likewise.
25620 2013-02-08  Michael Matz  <matz@suse.de>
25622         PR tree-optimization/52448
25623         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
25624         (nt_call_phase): New static.
25625         (add_or_mark_expr): Only mark accesses with newer phase than any
25626         call seen.
25627         (nonfreeing_call_p): New.
25628         (nt_init_block): Update nt_call_phase, mark blocks as visited.
25629         (nt_fini_block): Keep blocks marked as visited.
25630         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
25632 2013-02-08  Richard Biener  <rguenther@suse.de>
25634         * ira.c (ira): Free broken dominator information.
25636 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25638         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25640 2013-02-08  Marek Polacek  <polacek@redhat.com>
25642         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25644 2013-02-08  Richard Biener  <rguenther@suse.de>
25646         PR middle-end/56181
25647         * cfgloop.h (flow_loops_find): Adjust.
25648         (bb_loop_header_p): Declare.
25649         * cfgloop.c (bb_loop_header_p): New function split out from ...
25650         (flow_loops_find): ... here.  Adjust function signature,
25651         support incremental loop structure update.
25652         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25653         * cfgloopmanip.c (fix_loop_structure): Move ...
25654         * loop-init.c (fix_loop_structure): ... here.
25655         (apply_loop_flags): Split out from ...
25656         (loop_optimizer_init): ... here.
25657         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25658         in incremental mode, only remove dead loops here.
25660 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25662         PR target/54222
25663         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25664         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25665         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25666         (*round<mode>3.libgcc): New insns for fixed-modes.
25667         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25668         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25669         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25670         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25671         implementations.  Define to __builtin_avr_absFX,
25672         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25673         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25674         __builtin_avr_countlsFX, respectively.
25675         * config/avr/avr-c.c (target.h): Include it.
25676         (enum avr_builtin_id): New enum.
25677         (avr_resolve_overloaded_builtin): New static function.
25678         (avr_register_target_pragmas): Use it to set
25679         targetm.resolve_overloaded_builtin.
25680         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25681         tree nodes used by DEF_BUILTIN.
25682         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25683         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25684         <AVR_BUILTIN_xxBITS>: Same.
25686 2013-02-08  Richard Biener  <rguenther@suse.de>
25688         * cfgloop.c (verify_loop_structure): Properly handle
25689         a loop exiting to another loop header.
25690         * ira-int.h (ira_loops): Remove.
25691         * ira.c (ira_loops): Remove.
25692         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25693         (do_reload): Use loop_optimizer_finalize.
25694         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25695         number_of_loops to access the loop tree.
25696         (more_one_region_p): Likewise.
25697         (finish_loop_tree_nodes): Likewise.
25698         (rebuild_regno_allocno_maps): Likewise.
25699         (mark_loops_for_removal): Likewise.
25700         (mark_all_loops_for_removal): Likewise.
25701         (remove_unnecessary_regions): Likewise.
25702         (ira_build): Likewise.
25703         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25705 2013-02-08  Richard Biener  <rguenther@suse.de>
25707         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25708         * ipa-pure-const.c (analyze_function): Avoid calling
25709         mark_irreducible_loops twice.
25710         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25712 2013-02-07  David S. Miller  <davem@davemloft.net>
25714         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25715         on 'reg'.
25716         * var-tracking.c (vt_add_function_parameter): Test the presence of
25717         HAVE_window_save properly and do not remap argument registers when
25718         we have a leaf function.
25720 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25722         PR bootstrap/56227
25723         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25724         instead of "ll".
25725         * config/i386/i386.c (ix86_print_operand): Ditto.
25727 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25729         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25731 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25733         PR rtl-optimization/56225
25734         * lra-constraints.c (process_alt_operands): Check that reload hard
25735         reg can hold value for strict_low_part.
25737 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25739         PR debug/56154
25740         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25741         dwarf2out_end_function.
25742         (in_first_function_p, maybe_at_text_label_p,
25743         first_loclabel_num_not_at_text_label): New variables.
25744         (dwarf2out_var_location): In the first function find out
25745         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25746         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25747         functions.
25749 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25751         PR rtl-optimization/56178
25752         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25753         SUBREG of a register.  Tidy up related block of code.
25754         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25755         note if the source is a register or a SUBREG of a register.
25757 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25759         PR target/56228
25760         * config/rs6000/rs6000.md (ptrm): New mode attr.
25761         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25762         call_value_indirect_aix<pttrsize>,
25763         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25764         m in constraints.
25766 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25768         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25769         if -bnortl. Convert to strcmp and strncmp.
25771 2013-02-07  Alan Modra  <amodra@gmail.com>
25773         PR target/54009
25774         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25775         addresses won't wrap when offsetting.
25776         (rs6000_secondary_reload): Provide secondary reloads needed for
25777         wrapping LO_SUM addresses.
25779 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25781         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25782         MACH, just __MACH__.
25784 2013-02-06  Richard Biener  <rguenther@suse.de>
25786         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25787         instead of calling fix_loop_structure.
25789 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25791         PR middle-end/56217
25792         * omp-low.c (use_pointer_for_field): Return false if
25793         lower_send_shared_vars doesn't generate any copy-out code.
25795 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25797         PR rtl-optimization/56131
25798         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25799         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25800         of the label is NULL.  Add comment.
25802 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25804         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25806         PR sanitizer/55374
25807         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25808         (STATIC_LIBTSAN_LIBS): Likewise.
25809         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25810         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25811         is defined, don't add anything else beyond that.
25812         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25813         (LINK_COMMAND_SPEC): Use them.
25815         PR tree-optimization/56205
25816         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25817         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25818         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25820 2013-02-05  Richard Biener  <rguenther@suse.de>
25822         PR tree-optimization/53342
25823         PR tree-optimization/53185
25824         * tree-vectorizer.h (vect_check_strided_load): Remove.
25825         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25826         not disallow peeling for vectorized strided loads.
25827         (vect_check_strided_load): Make static and simplify.
25828         (vect_analyze_data_refs): Adjust.
25829         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25830         correctly when vectorizing strided loads.
25832 2013-02-05  Richard Biener  <rguenther@suse.de>
25834         * doc/install.texi: Refer to ISL, not PPL.
25836 2013-02-05  Jan Hubicka  <jh@suse.cz>
25838         PR tree-optimization/55789
25839         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25841 2013-02-05  Jan Hubicka  <jh@suse.cz>
25843         PR tree-optimization/55789
25844         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25845         the dead call anyway.
25847 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25849         PR sanitizer/55374
25850         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25852 2013-02-04  Alexander Potapenko  <glider@google.com>
25853             Jack Howarth  <howarth@bromo.med.uc.edu>
25854             Jakub Jelinek  <jakub@redhat.com>
25856         PR sanitizer/55617
25857         * config/darwin.c (sort_ctor_records): Stabilized qsort
25858         on constructor priority by using original position.
25859         (finalize_ctors): New routine to sort constructors by
25860         priority before use in assemble_integer.
25861         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25863 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25865         PR libstdc++/54314
25866         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25867         about visibility on artificial decls.
25868         * config/sol2.c (solaris_assemble_visibility): Likewise.
25870 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25872         PR target/56186
25873         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25874         argument and improve checking of return-argument types for 16-byte
25875         modes.
25876         (ix86_function_value_1): Add additional valtype argument on call
25877         of function_value_64.
25878         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25879         handling infunction_value_64 function.
25881 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25883         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25885 2013-02-04  Richard Biener  <rguenther@suse.de>
25887         PR tree-optimization/56188
25888         * tree-ssa-structalias.c (label_visit): Consider case with
25889         initially non-empty points-to set.
25890         (perform_var_substitution): Dump node mapping and clean up.
25892 2013-02-04  Richard Guenther  <rguenther@suse.de>
25894         PR lto/56168
25895         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25896         node prevail as last resort.
25897         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25898         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25900 2013-02-04  Richard Biener  <rguenther@suse.de>
25902         PR tree-optimization/56113
25903         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25904         Merge into ...
25905         (equiv_class_lookup_or_add): ... this.
25906         (label_visit): Adjust and fix error in previous patch.
25907         (perform_var_substitution): Adjust.
25909 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25911         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25912         * config/sh/sh4-300.md: Likewise.
25913         * config/sh/sh4a.md: Likewise.
25914         * config/sh/constraints.md: Likewise.
25915         * config/sh/sh.md: Likewise.
25916         * config/sh/netbsd-elf.h: Likewise.
25917         * config/sh/predicates.md: Likewise.
25918         * config/sh/sh-protos.h: Likewise.
25919         * config/sh/ushmedia.h: Likewise.
25920         * config/sh/linux.h: Likewise.
25921         * config/sh/sh.c: Likewise.
25922         * config/sh/superh.h: Likewise.
25923         * config/sh/elf.h: Likewise.
25924         * config/sh/sh4.md: Likewise.
25925         * config/sh/sh.h: Likewise.
25927 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25929         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25930         constraint to match_test floating_point_store_memory_operand().
25931         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25932         (base14_operand): New.
25933         (floating_point_store_memory_operand): New.
25934         (integer_store_memory_operand): Revise to use base14_operand and
25935         reg_plus_base_memory_operand.
25936         (move_dest_operand): Allow symbolic_memory_operands.
25937         (symbolic_memory_operand): Check for LO_SOM.
25938         (symbolic_operand): Change default case to break.
25939         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25940         CONST_DOUBLE values to be reloaded by putting them into memory when
25941         the destination is a floating point register.
25942         (movdf): Remove code to handle CONST_DOUBLE.
25943         (movsf): Likewise.
25944         (reload_indf_r1): New.
25945         (reload_insf_r1): New.
25946         Consistently use "Q" and "T" constraints with integer and floating
25947         point move instructions, respectively.
25948         (movdi): Remove FAIL.
25949         Change predicate for source operand unamed DImode move from
25950         general_operand to move_src_operand.
25951         (umulsidi3): Change predicate for destination operand to
25952         register_operand.
25953         Likewise for similar unamed patterns.
25954         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25955         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25956         (hppa_legitimize_address): Simplify mask calculation.
25957         (pa_emit_move_sequence): Revised handling of secondary reloads from
25958         REG+D addresses for floating point loads and stores.  Directly handle
25959         loading CONST0_RTX (mode) to a floating point register.
25960         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25961         to floating point registers.  Don't restrict secondary reloads to
25962         floating point registers to integer modes.  Revise some comments and
25963         cleanup some code.
25964         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25965         (pa_legitimate_address_p): New.
25966         (pa_legitimize_reload_address): New.
25967         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25968         (STRICT_REG_OK_FOR_BASE_P): New.
25969         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25970         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25972 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25973             Andrew Dixie  <andrewd@gentrack.com>
25975         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25976         flag set.
25978 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25980         * expmed.c (extract_bit_field_1): Pass the full width of the
25981         structure to get_best_reg_extraction_insn.
25983 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25985         PR target/54601
25986         * configure.ac (use_cxa_atexit): Add AIX.
25987         * configure: Regenerate.
25989         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25991 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25993         PR debug/54793
25994         * final.c (need_profile_function): New variable.
25995         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25996         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25997         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25998         notes, targetm.asm_out.function_prologue doesn't emit anything,
25999         HAVE_prologue and profiler should be emitted before prologue,
26000         set need_profile_function instead of emitting it.
26001         (final_scan_insn): If need_profile_function, emit
26002         profile_function on the first NOTE_INSN_BASIC_BLOCK or
26003         NOTE_INSN_FUNCTION_BEG note.
26005 2013-02-01  Richard Henderson  <rth@redhat.com>
26007         * config/rs6000/rs6000.md (smulditi3): New.
26008         (umulditi3): New.
26010         * config/alpha/alpha.md (umulditi3): New.
26012 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
26014         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
26015         (ASM_OUTPUT_ALIGNED_LOCAL): New.
26017 2013-02-01  Richard Biener  <rguenther@suse.de>
26019         PR tree-optimization/56113
26020         * tree-ssa-structalias.c (label_visit): Reduce work for
26021         single-predecessor nodes.
26023 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
26025         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
26026         range isn't testing for zero.
26028 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
26030         PR middle-end/56113
26031         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
26033 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
26034             Nick Clifton  <nickc@redhat.com>
26036         * config/v850/constraints.md (Q): Define as a memory constraint.
26037         * config/v850/predicates.md (label_ref_operand): New predicate.
26038         (e3v5_shift_operand): New predicate.
26039         (ior_operator): New predicate.
26040         * config/v850/t-v850: Add e3v5 multilib.
26041         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
26042         (v850_gen_movdi): Prototype.
26043         * config/v850/v850.c: Add support for e3v5 architecture.
26044         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
26045         TARGET_V850E_UP.
26046         (construct_save_jarl): Add e3v5 long JARL support.
26047         (v850_adjust_insn_length): New function.  Adjust length of call
26048         insns when using e3v5 instructions.
26049         (v850_gen_movdi): New function: Generate instructions to move a
26050         DImode value.
26051         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
26052         (CPP_SPEC): Define __v850e3v5__ as appropriate.
26053         (TARGET_USE_FPU): Enable for e3v5.
26054         (CONST_OK_FOR_W): New macro.
26055         (ADJUST_INSN_LENGTH): Define.
26056         * config/v850/v850.md (UNSPEC_LOOP): Define.
26057         (attr cpu): Add v850e3v5.
26058         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
26059         (movdi): New pattern.
26060         (movdi_internal): New pattern.
26061         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
26062         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
26063         (cstoresf4): Likewise.
26064         (cstoredf4): Likewise.
26065         (insv): New pattern.
26066         (rotlso3_a): New pattern.
26067         (rotlsi3_b): New pattern
26068         (rotlsi3_v850e3v5): New pattern.
26069         (doloop_begin): New pattern.
26070         (fix_loop_counter): New pattern.
26071         (doloop_end): New pattern.
26072         (branch_normal): Add e3v5 long branch support.
26073         (branch_invert): Likewise.
26074         (branch_z_normal): Likewise.
26075         (branch_z_invert): Likewise.
26076         (branch_nz_normal): Likewise.
26077         (branch_nz_invert): Likewise.
26078         (call_internal_short): Add e3v5 register-indirect JARL support.
26079         (call_internal_long): Likewise.
26080         (call_value_internal_short): Likewise.
26081         (call_value_internal_long): Likewise.
26082         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
26083         (mloop): New option.
26084         * config.gcc: Add support for configuring v840e3v5 target.
26085         * doc/invoke.texi: Document new v850 specific command line options.
26087 2013-01-31  Paul Koning  <ni1d@arrl.net>
26089         PR debug/55059
26090         PR debug/54508
26091         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
26092         children if parent is a class.
26093         (prune_unused_types_prune): Don't add DW_AT_declaration.
26095 2013-01-31  Richard Biener  <rguenther@suse.de>
26097         PR tree-optimization/56157
26098         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
26099         match up operand with SLP child.
26101 2013-01-31  Jason Merrill  <jason@redhat.com>
26103         PR debug/54410
26104         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
26105         parameters the first time.
26106         (gen_scheduled_generic_parms_dies): Check completeness here.
26108 2013-01-31  Richard Biener  <rguenther@suse.de>
26110         PR middle-end/53073
26111         * common.opt (faggressive-loop-optimizations): New flag,
26112         enabled by default.
26113         * doc/invoke.texi (faggressive-loop-optimizations): Document.
26114         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
26115         infer_loop_bounds_from_undefined by it.
26117 2013-01-31  Richard Biener  <rguenther@suse.de>
26119         PR tree-optimization/56150
26120         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
26121         visit virtual operands.
26122         (find_uses_to_rename_bb): Likewise.
26124 2013-01-31  Richard Biener  <rguenther@suse.de>
26126         PR tree-optimization/56150
26127         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
26128         mixed store non-store stmts.
26130 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
26132         PR sanitizer/55374
26133         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
26134         LIBASAN_EARLY_SPEC is defined.
26135         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
26136         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
26137         before %o.
26138         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
26140         PR c++/55742
26141         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
26142         invalid args instead of ICEing on it.
26143         (ix86_valid_target_attribute_tree): Return error_mark_node if
26144         ix86_valid_target_attribute_inner_p failed.
26145         (ix86_valid_target_attribute_p): Return false only if
26146         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
26147         target("default") attribute.
26148         (sorted_attr_string): Change argument from const char * to tree,
26149         merge in all target attribute arguments rather than just one.
26150         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
26151         instead of free.  Avoid using strcat.
26152         (ix86_mangle_function_version_assembler_name): Mangle
26153         target("default") as if no target attribute is present.  Adjust
26154         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
26155         instead of xmalloc and XDELETEVEC instead of free.
26156         (ix86_function_versions): Don't return true if one of the decls
26157         doesn't have target attribute.  If they don't and one of the decls
26158         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
26159         sorted_attr_string caller.  Use XDELETEVEC instead of free.
26160         (ix86_supports_function_versions): Remove.
26161         (make_name): Fix up formatting.
26162         (make_dispatcher_decl): Remove resolver_name and its initialization.
26163         Avoid leaking memory.
26164         (is_function_default_version): Return true if there is
26165         target("default") attribute rather than no target attribute at all.
26166         (make_resolver_func): Avoid leaking memory.
26167         (ix86_generate_version_dispatcher_body): Likewise.
26168         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
26169         * target.def (supports_function_versions): Remove.
26170         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
26171         * doc/tm.texi: Regenerated.
26173 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
26175         PR rtl-optimization/56144
26176         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
26177         for values with side effects.
26179 2013-01-30  Richard Biener  <rguenther@suse.de>
26181         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
26182         (sparseset_pop): Likewise.
26183         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
26184         to be able to use quick_push in the worker loop.
26186 2013-01-30  Marek Polacek  <polacek@redhat.com>
26188         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
26190 2013-01-30  Richard Biener  <rguenther@suse.de>
26192         PR lto/56147
26193         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
26195 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
26197         PR tree-optimization/56064
26198         * fixed-value.c (fixed_from_double_int): New function.
26199         * fixed-value.h (fixed_from_double_int): New prototype.
26200         (const_fixed_from_double_int): New static inline function.
26201         * fold-const.c (native_interpret_fixed): New static function.
26202         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
26203         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
26204         (native_encode_fixed): New static function.
26205         (native_encode_expr) <FIXED_CST>: Use it.
26206         (native_interpret_int): Move double_int worker code to...
26207         * double-int.c (double_int::from_buffer): ...this new static method.
26208         * double-int.h (double_int::from_buffer): Prototype it.
26210 2013-01-30  Richard Biener  <rguenther@suse.de>
26212         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
26213         New pointer-map and obstack.
26214         (init_alias_vars): Allocate pointer-map and obstack.
26215         (delete_points_to_sets): Free them.
26216         (find_what_var_points_to): Cache result.
26217         (find_what_p_points_to): Adjust for changed interface of
26218         find_what_var_points_to.
26219         (compute_points_to_sets): Likewise.
26220         (ipa_pta_execute): Likewise.
26222 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
26224         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
26225         * configure: Regenerate.
26226         * config.in: Regenerate.
26227         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
26228         #nobits/#progbits if supported.
26230 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
26232         PR target/56121
26233         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
26234         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
26235         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
26237 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26239         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
26240         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
26242 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26244         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
26245         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
26247 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26249         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
26250         declaration.
26251         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
26252         * config/arm/cortex-a7.md: New bypasses using
26253         arm_mac_accumulator_is_result.
26255 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26257         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
26258         (cortex_a7_neon_mla): Likewise.
26259         (cortex_a7_fpfmad): New reservation.
26260         (cortex_a7_fpmacs): Use ffmas and update required units.
26261         (cortex_a7_fpmuld): Update required units and latency.
26262         (cortex_a7_fpmacd): Likewise.
26263         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
26264         (cortex_a7_neon). Likewise.
26265         (bypass) Update participating units.
26267 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
26269         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
26270         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
26271         from fmac to ffma.
26272         * config/arm/vfp11.md (vfp_farith): Use ffmas.
26273         (vfp_fmul): Use ffmad.
26274         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
26275         (cortex_r4_fmacd): Use ffmad.
26276         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
26277         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
26278         (cortex_a9_fmacd): Use ffmad.
26279         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
26280         (cortex_a8_vfp_macd): Use ffmad.
26281         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
26282         (cortex_a5_fpmacd): Use ffmad.
26283         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
26284         (cortex_a15_vfp_macd): Use ffmad.
26285         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
26287 2013-01-29  Jason Merrill  <jason@redhat.com>
26289         PR libstdc++/54314
26290         * varasm.c (default_assemble_visibility): Don't warn about
26291         visibility on artificial decls.
26293 2013-01-29  Richard Biener  <rguenther@suse.de>
26295         PR tree-optimization/56113
26296         * tree-ssa-structalias.c (equiv_class_lookup): Also return
26297         the bitmap leader.
26298         (label_visit): Free duplicate bitmaps and record the leader instead.
26299         (perform_var_substitution): Adjust.
26301 2013-01-29  Richard Biener  <rguenther@suse.de>
26303         PR tree-optimization/55270
26304         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
26305         the CFG, schedule loops for fixup.
26307 2013-01-29  Nick Clifton  <nickc@redhat.com>
26309         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
26310         SP_REG.
26312 2013-01-28  Leif Ekblad  <leif@rdos.net>
26314         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
26315         * config/i386/i386.h (TARGET_RDOS): New macro.
26316         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
26317         * config/i386/i386.c (ix86_option_override_internal): For 64bit
26318         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
26319         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
26320         DEFAULT_LARGE_SECTION_THRESHOLD.
26321         * config/i386/i386.md (R14_REG, R15_REG): New constants.
26322         * config/i386/rdos.h: New file.
26323         * config/i386/rdos64.h: New file.
26325 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
26327         PR other/54814
26328         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
26329         TEST_HARD_REG_BIT.
26331 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
26333         PR rtl-optimization/56117
26334         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
26335         call cselib_lookup_from_insn on the MEM before calling
26336         add_insn_mem_dependence.
26338 2013-01-28  Richard Biener  <rguenther@suse.de>
26340         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
26341         to a stmt that didn't have one.
26342         (copy_phis_for_bb): Likewise for PHI arguments.
26343         (copy_debug_stmt): Likewise for debug stmts.
26345 2013-01-28  Richard Biener  <rguenther@suse.de>
26347         PR tree-optimization/56034
26348         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
26349         (partition_builtin_p): Adjust.
26350         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
26351         it is the last partition.
26352         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
26353         up the vertex for the definition.
26354         (classify_partition): Classify whether a partition is a
26355         PKIND_REDUCTION, thus has uses outside of the loop.
26356         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
26357         Merge all PKIND_REDUCTION partitions into the last partition.
26358         (tree_loop_distribution): Seed partitions from reductions as well.
26360 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
26362         PR tree-optimization/56125
26363         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
26364         pow(x,c) into sqrt(x) * powi(x, n/2) or
26365         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
26366         optimizing for size.
26367         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
26368         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
26369         integer.
26371         PR tree-optimization/56094
26372         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
26373         to UNKNOWN_LOCATION while gimplifying expr.
26375 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
26377         PR target/56114
26378         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
26379         operand 0 in movabs insn template for -masm=intel asm alternative.
26380         (*movabs<mode>_2): Ditto for operand 1.
26382 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
26384         PR target/54663
26385         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
26386         of microblaze-c.o
26388 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26390         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
26391         tm_file.
26393 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26395         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
26396         Undef to avoid warning.
26398 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
26400         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
26401         * configure: Regenerate.
26403 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
26405         PR tree-optimization/56098
26406         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
26407         for stmts with volatile ops.
26408         (cond_store_replacement): Don't optimize if assign has volatile ops.
26409         (cond_if_else_store_replacement_1): Don't optimize if either
26410         then_assign or else_assign have volatile ops.
26411         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
26412         volatile ops.
26414 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
26416         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
26418 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
26420         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
26421         missing ':' in asm example.
26423 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
26425         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
26426         entries into lane and laneq entries.
26427         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
26428         Remove AdvSIMD scalar modes.
26429         (aarch64_sq<r>dmulh_laneq<mode>): New.
26430         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
26431         modes.
26432         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
26433         builtin implementations to relfect changes in RTL in aarch64-simd.md.
26434         * config/aarch64/iterators.md (VCOND): New.
26435         (VCONQ): New.
26437 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
26439         PR target/54222
26440         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
26441         Add NULL LIBNAME argument to existing definitions.
26442         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
26443         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
26444         * config/avr/avr.c (DEF_BUILTIN): Same.
26445         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
26446         (avr_expand_builtin): Expand to a vanilla call if a libgcc
26447         implementation is available (DECL_ASSEMBLER_NAME is set).
26448         (avr_fold_absfx): New static function.
26449         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
26450         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
26451         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
26452         AVR_BUILTIN_ABSLLK.
26453         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
26454         (abshk, absk, abslk, absllk): Provide as static inline functions.
26456 2013-01-25  Marek Polacek  <polacek@redhat.com>
26458         PR tree-optimization/56035
26459         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
26461 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
26463         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
26464         (*movtf_internal_rex64): Add (!o,C) alternative
26465         (*movxf_internal_rex64): Ditto.
26466         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
26468 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
26470         * doc/invoke.texi: fix typo.
26471         * doc/objc.texi: fix typo.
26473 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
26475         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
26476         for the first two alternatives.
26478 2013-01-24  Diego Novillo  <dnovillo@google.com>
26480         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
26481         (ggc-zone.o): Remove.
26482         * configure.ac: Remove option --with-gc.
26483         * configure: Re-generate.
26484         * doc/install.texi: Remove documentation for --with-gc.
26485         * gengtype.c (write_enum_defn): Remove.  Update all users.
26486         (write_Types_process_field): Remove generation of gt_e_* argument.
26487         (output_type_enum): Remove.  Update all users.
26488         (write_enum_defn): Remove.  Update all users.
26489         (enum alloc_zone): Remove.  Update all users.
26490         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
26491         * ggc-common.c (ggc_splay_alloc): Remove first argument.
26492         Update all callers.
26493         (struct ptr_data): Remove field TYPE.  Update all users.
26494         (gt_pch_note_object): Remove argument TYPE.  Update all users.
26495         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
26496         Update all users.
26497         * ggc-none.c (ggc_alloc_typed_stat): Remove.
26498         (struct alloc_zone): Remove.
26499         (ggc_internal_alloc_zone_stat): Remove.
26500         (ggc_internal_cleared_alloc_zone_stat): Remove.
26501         * ggc-page.c (ggc_alloc_typed_stat): Remove.
26502         (ggc_pch_count_object): Remove last argument.  Update all users.
26503         (ggc_pch_alloc_object): Remove last argument.  Update all users.
26504         (struct alloc_zone): Remove.
26505         * ggc-zone.c: Remove.
26506         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
26507         (struct alloc_zone): Remove.
26508         (ggc_alloc_typed_stat): Remove.
26509         (ggc_alloc_typed): Remove.
26510         (ggc_splay_alloc): Remove first argument.
26511         (rtl_zone): Remove.  Update all users.
26512         (tree_zone): Remove.  Update all users.
26513         (tree_id_zone): Remove.  Update all users.
26514         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
26515         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
26516         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
26517         * tree-ssanames.c: Remove references to zone allocator in comments.
26519 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
26521         * config/avr/avr.c (avr_out_fract): Make register numbers that
26522         might be outside of source operand signed.
26524 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
26526         * config/i386/constraints.md (Yf): New constraint.
26527         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
26528         of f constraint to conditionaly disable x87 register preferences.
26529         (*movdf_internal): Ditto.
26530         (*movsf_internal): Ditto.
26532 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
26534         PR inline-asm/55934
26535         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
26536         that have operands with impossible constraints.
26537         Add a FIXME for a speed-up opportunity.
26538         * lra-constraints.c (process_alt_operands): Verify that a class
26539         selected from constraints on asms is valid for the operand mode.
26540         (curr_insn_transform): Remove incorrect comment.
26542 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
26544         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
26545         TOC operand is a valid symbol ref in the constant pool.
26547 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26549         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
26551 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
26553         PR target/54222
26554         * config/avr/stdfix.h: New file.
26555         * t-avr (stdfix-gcc.h): New rule to build it.
26556         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
26558 2013-01-23  Kostya Serebryany  <kcc@google.com>
26560         * config/darwin.h: remove dependency on
26561         CoreFoundation (asan on Mac OS).
26563 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
26565         PR target/49069
26566         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
26567         instead of cmpdi_operand for first comparison operand.
26568         Don't assert that comparison operands aren't both constants.
26570 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26572         * doc/install.texi (Downloading the Source): Update references to
26573         downloading separate components.
26575 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
26577         * doc/extend.texi (__int128): Improve grammar.
26579 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
26581         PR target/56028
26582         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
26583         alternative to (o,r).
26584         (*movdi_internal_rex64): Remove (!o,n) alternative.
26585         (DImode immediate->memory splitter): Remove.
26586         (DImode immediate->memory peephole2): Remove.
26587         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
26588         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
26589         alternative to (!o,*r).
26590         (*movtf_internal_sse): New pattern.
26591         (*movxf_internal_rex64): New pattern.
26592         (*movxf_internal): Disable for TARGET_64BIT.
26593         (*movdf_internal_rex64): Remove (!o,F) alternative.
26595 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
26597         PR middle-end/56074
26598         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
26599         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
26600         * tree-vect-loop-manip.c (find_loop_location): Also ignore
26601         stmt locations where LOCATION_LOCUS of the stmt location is
26602         UNKNOWN_LOCATION or BUILTINS_LOCATION.
26604         PR target/55686
26605         * config/i386/i386.md (UNSPEC_STOS): New.
26606         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
26607         *strsetqi_1): Add UNSPEC_STOS.
26609 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
26611         PR c++/56067
26612         * doc/invoke.texi: Remove left over -Wsynth example.
26614 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26616         PR tree-optimization/56051
26617         * fold-const.c (fold_binary_loc): Don't fold
26618         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
26619         a narrowing conversion, or widening conversion from signed
26620         to unsigned.
26622 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26624         PR rtl-optimization/56023
26625         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
26626         dependent on debug instruction.
26628 2013-01-21  Martin Jambor  <mjambor@suse.cz>
26630         PR middle-end/56022
26631         * function.c (allocate_struct_function): Call
26632         invoke_set_current_function_hook earlier.
26634 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
26636         * reload1.c (init_reload): Only initialize reload_obstack
26637         during the first call.
26639 2013-01-21  Marek Polacek  <polacek@redhat.com>
26641         * cfgloop.c (verify_loop_structure): Fix up grammar.
26643 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26645         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26646         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26648 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26650         PR target/56058
26651         * config/arm/marvell-pj4.md: Update copyright year.
26652         Fix up use of alu to alu_reg and simple_alu_imm.
26654 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26656         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26658 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26660         PR target/55433
26661         * lra-constraints.c (curr_insn_transform): Don't reuse original
26662         insn for secondary memory move when memory mode should be different.
26664 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26666         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26667         atomic_storedi_1): New patterns.
26669 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26671         btver2 pipeline descriptions.
26672         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26673         descriptions.
26674         * config/i386/i386.md (btver2_decode): New type attributes.
26675         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26676         type attributes.
26677         * config/i386/btver2.md: New file describing btver2 pipelines.
26679 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26681         PR tree-optimization/52631
26682         * tree-ssa-sccvn (visit_use): Before looking up the original
26683         statement, try looking up the simplified expression.
26685 2013-01-19  Anthony Green  <green@moxielogic.com>
26687         * config/moxie/moxie.c (moxie_expand_prologue): Set
26688         current_function_static_stack_size.
26690 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26692         PR tree-optimization/56029
26693         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26694         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26696 2013-01-18  Sharad Singhai  <singhai@google.com>
26698         PR tree-optimization/55995
26699         * dumpfile.c (dump_loc): Print location only if available.
26700         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26702 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26704         PR target/55433
26705         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26706         secondary memory move.
26707         (inherit_reload_reg): Use rclass instead of cl for
26708         check_secondary_memory_needed_p.
26710 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26712         PR middle-end/56015
26713         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26714         the case where writing real complex part of target modifies op1.
26716 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26718         * config/aarch64/aarch64-simd.md
26719         (aarch64_vcond_internal<mode>): Handle unordered cases.
26720         * config/aarch64/iterators.md (v_cmp_result): New.
26722 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26723             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26725         * config/arm/marvell-pj4.md: New file.
26726         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26727         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26728         (generic_vfp): Likewise.
26729         * config/arm/arm-cores.def: Add marvell-pj4.
26730         * config/arm/arm-tune.md: Regenerate.
26731         * config/arm/arm-tables.opt: Regenerate.
26732         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26733         * doc/invoke.texi: Document marvell-pj4.
26735 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26737         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26739 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26741         PR debug/54114
26742         PR debug/54402
26743         PR debug/49888
26744         * var-tracking.c (negative_power_of_two_p): New.
26745         (global_get_addr_cache, local_get_addr_cache): New.
26746         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26747         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26748         heading comment.
26749         (vt_stack_offset_p): Remove.
26750         (vt_canon_true_dep): Always canonicalize loc's address.
26751         (clobber_overlapping_mems): Make sure we have a MEM.
26752         (local_get_addr_clear_given_value): New.
26753         (val_reset): Clear local cached entries.
26754         (compute_bb_dataflow): Create and release the local cache.
26755         Disable duplicate MEMs clobbering.
26756         (emit_notes_in_bb): Clobber MEMs likewise.
26757         (vt_emit_notes): Create and release the local cache.
26758         (vt_initialize, vt_finalize): Create and release the global
26759         cache, respectively.
26760         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26762 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26764         PR libmudflap/53359
26765         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26766         not found in the symtab.
26768 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26770         PR debug/56006
26771         PR rtl-optimization/55547
26772         PR rtl-optimization/53827
26773         PR debug/53671
26774         PR debug/49888
26775         * alias.c (offset_overlap_p): New, factored out of...
26776         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26777         the conservative special case for symbolic constants.  Don't
26778         adjust zero sizes on alignment.
26780 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26782         PR rtl-optimization/52573
26783         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26784         REG_UNUSED for the same register.
26786 2013-01-17  Richard Biener  <rguenther@suse.de>
26787             Marek Polacek  <polacek@redhat.com>
26789         PR rtl-optimization/55833
26790         * loop-unswitch.c (unswitch_loops): Move loop verification...
26791         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26792         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26793         Set it to true when we're removing a loop from hierarchy tree in
26794         an irreducible region.
26795         (fix_bb_placements): Adjust caller.
26796         (fix_loop_placements): Likewise.
26798 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26800         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26801         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26802         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26803         Remove ID.  Adjust comments.
26804         * config/avr/avr-c.c (avr_builtin_name): Remove.
26805         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26806         * config/avr/avr.c (avr_tolower): New static function.
26807         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26808         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26809         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26810         default expansion.
26812 2013-01-17  Jan Hubicka  <jh@suse.cz>
26814         PR tree-optimization/55273
26815         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26817 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26819         PR target/55981
26820         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26821         store through atomic_store<mode>_1.
26822         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26824 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26826         PR tree-optimizations/55264
26827         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26828         for virtual methods.
26829         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26830         virtual methods before inlining is over.
26831         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26832         virtual functions.
26833         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26834         non-virtual.
26836 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26838         PR rtl-optimization/56005
26839         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26840         pending reads for prefetch.
26842 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26844         * config/aarch64/aarch64.md
26845         (*cstoresi_neg_uxtw): New pattern.
26846         (*cmovsi_insn_uxtw): New pattern.
26847         (*<optab>si3_uxtw): New pattern.
26848         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26849         (*<optab>si3_insn_uxtw): New pattern.
26850         (*bswapsi2_uxtw): New pattern.
26852 2013-01-16  Richard Biener  <rguenther@suse.de>
26854         * tree-inline.c (tree_function_versioning): Remove set but
26855         never used variable.
26857 2013-01-16  Richard Biener  <rguenther@suse.de>
26859         PR tree-optimization/55964
26860         * tree-flow.h (rename_variables_in_loop): Remove.
26861         (rename_variables_in_bb): Likewise.
26862         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26863         (copy_loop_before): Adjust and delete update-ssa status.
26864         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26865         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26866         (rename_variables_in_loop): Remove.
26867         (slpeel_update_phis_for_duplicate_loop): Likewise.
26868         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26869         use available cfg machinery instead of duplicating it.
26870         Update PHI nodes and perform poor-mans SSA update here.
26871         (slpeel_tree_peel_loop_to_edge): Adjust.
26873 2013-01-16  Richard Biener  <rguenther@suse.de>
26875         PR tree-optimization/54767
26876         PR tree-optimization/53465
26877         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26878         (vrp_visit_phi_node): For PHI arguments coming via backedges
26879         drop all symbolical range information.
26880         (execute_vrp): Compute backedges.
26882 2013-01-16  Richard Biener  <rguenther@suse.de>
26884         * doc/install.texi: Update CLooG and ISL requirements to
26885         0.18.0 and 0.11.1.
26887 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26889         PR target/55301
26890         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26891         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26892         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26894 2013-01-16  DJ Delorie  <dj@redhat.com>
26896         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26897         (UNSPECV_SP_SWITCH_E): New.
26898         (sp_switch_1): Change to an unspec.
26899         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26900         replace $r15.
26902 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26904         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26905         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26906         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26907         (expand_mem_thread_fence): Ditto.
26908         (expand_mem_signal_fence): Ditto.
26909         (expand_atomic_load): Ditto.
26910         (expand_atomic_store): Ditto.
26912 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26914         PR rtl-optimization/55547
26915         PR rtl-optimization/53827
26916         PR debug/53671
26917         PR debug/49888
26918         * alias.c (memrefs_conflict_p): Set sizes to negative after
26919         AND adjustments.
26921 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26923         PR target/55940
26924         * function.c (thread_prologue_and_epilogue_insns): Always
26925         add crtl->drap_reg to set_up_by_prologue.set, even if
26926         stack_realign_drap is false.
26928 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26930         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26931         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26932         *call): Fix indention.
26934 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26936         PR target/55876
26937         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26938         Update comment.
26940 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26942         PR rtl-optimization/55153
26943         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26945 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26947         PR tree-optimization/55920
26948         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26949         accesses as grp_to_be_debug_replaced.
26951 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26953         PR tree-optimization/55920
26954         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26955         there is non-useless type conversion needed from debug rhs to lhs,
26956         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26958 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26959             Mikael Pettersson  <mikpe@it.uu.se>
26961         PR target/43961
26962         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26963         Thumb.
26964         (ASM_OUTPUT_CASE_LABEL): Remove.
26965         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26966         * final.c (shorten_branches): Update alignment of labels before
26967         jump tables if CASE_VECTOR_SHORTEN_MODE.
26969 2013-01-15  Richard Biener  <rguenther@suse.de>
26971         PR bootstrap/55961
26972         * system.h: Do not include gmp.h for building host tools.
26974 2013-01-15  Richard Biener  <rguenther@suse.de>
26976         PR middle-end/55882
26977         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26978         account for bitpos when computing alignment.
26980 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26982         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26983         (ix86_target_macros_internal): Likewise.
26985         * config/i386/i386.c (m_CORE2I7): Removed.
26986         (m_CORE_HASWELL): New macro.
26987         (m_CORE_ALL): Likewise.
26988         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26989         (initial_ix86_arch_features): Likewise.
26990         (processor_target_table): Initializations for Core avx2.
26991         (cpu_names): New names "core-avx2".
26992         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26993         PROCESSOR_CORE_HASWELL.
26994         (ix86_issue_rate): New case.
26995         (ia32_multipass_dfa_lookahead): Likewise.
26996         (ix86_sched_init_global): Likewise.
26998         * config/i386/i386.h (TARGET_HASWELL): New macro.
26999         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
27000         (processor_type): New PROCESSOR_HASWELL.
27002 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
27004         PR tree-optimization/55955
27005         * tree-vect-loop.c (vectorizable_reduction): Give up early on
27006         *SHIFT_EXPR and *ROTATE_EXPR codes.
27008         PR tree-optimization/48766
27009         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
27010         -ftrapv disable -fwrapv.
27012 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
27014         PR target/55974
27015         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
27016         etc. to 1 and not to __flash.
27017         Use LL suffix for __INT24_MAX__ with -mint8.
27018         Use ULL suffix for __UINT24_MAX__ with -mint8.
27020 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
27022         * config/avr/avr-arch.h
27023         (struct base_arch_s): Use typedef avr_arch_t instead.
27024         (struct arch_info_s): Use typedef avr_arch_info_t instead.
27025         (struct mcu_type_s): Use typedef avr_mcu_t instead.
27026         * config/avr/avr.c: Same.
27027         * config/avr/avr-devices.c: Same.
27028         * config/avr/driver-avr.c: Same.
27029         * config/avr/gen-avr-mmcu-texi.c: Same.
27030         * config/avr/avr-mcus.def: Adjust comment.
27032 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
27034         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
27035         * config/aarch64/iterators.md (VALLDI): New.
27037 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
27038             Andi Kleen  <ak@linux.intel.com>
27040         PR target/55948
27041         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
27042         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
27043         memmodel flag.
27045 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
27047         * config/avr/avr-stdint.h: Remove trailing blanks.
27048         * config/avr/avr-log.h: Same.
27049         * config/avr/avr-arch.h: Same.
27050         * config/avr/avr-devices.c: Same.
27051         * config/avr/avr-dimode.md: Same.
27052         * config/avr/predicates.md: Same.
27053         * config/avr/avr-c.c: Same.  And fix typo.
27055         * config/avr/avr-protos.h: Same.  And:
27056         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
27057         (init_cumulative_args): Rename to avr_init_cumulative_args.
27058         (expand_prologue): Rename to avr_expand_prologue.
27059         (expand_epilogue): Rename to avr_expand_epilogue.
27060         (adjust_insn_length): Rename to avr_adjust_insn_length.
27061         (notice_update_cc): Rename to avr_notice_update_cc.
27062         (final_prescan_insn): Rename to avr_final_prescan_insn.
27063         * config/avr/avr.c: Same.
27064         * config/avr/avr.h: Same.
27065         * config/avr/avr.md: Remove trailing blanks.
27066         (prologue): Use avr_expand_prologue.
27067         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
27069 2013-01-14  Richard Biener  <rguenther@suse.de>
27071         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
27072         verify_location, collect_subblocks): New functions.
27073         (verify_gimple_in_cfg): Verify that locations only reference
27074         BLOCKs in the functions BLOCK tree.
27076 2013-01-14  Richard Biener  <rguenther@suse.de>
27078         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
27079         PHI argument.
27080         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
27081         unshare reference.
27082         (insert_out_of_ssa_copy_on_edge): Likewise.
27083         (rewrite_close_phi_out_of_ssa): Likewise.
27084         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
27085         debug expressions.
27086         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
27087         propagated constants.
27088         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
27089         can not be shared.
27091 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
27093         * config/avr/avr-modes.def: Add GPL copyright notice.
27095 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
27097         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
27098         MEMMODEL_MASK to determine memory model.
27099         (atomic_store<mode>): Ditto from operands[2].
27100         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
27102 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
27104         PR fortran/55935
27105         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
27106         (fold_gimple_assign): Don't call unshare_expr here.
27107         (fold_ctor_reference): Call unshare_expr.
27109 2013-01-13  Terry Guo  <terry.guo@arm.com>
27111         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
27112         * doc/fragments.texi: Document MULTILIB_REUSE.
27113         * gcc.c (multilib_reuse): New internal spec.
27114         (set_multilib_dir): Also search multilib from multilib_reuse.
27115         * genmultilib (tmpmultilib3): Refactor code.
27116         (tmpmultilib4): Ditto.
27117         (multilib_reuse): New multilib argument.
27119 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
27121         * Makefile.in: Update copyright.
27123 2013-01-12  Tom de Vries  <tom@codesourcery.com>
27125         PR middle-end/55890
27126         * calls.c (expand_call): Check if arg_nr is valid.
27128 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
27130         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
27131         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
27132         documentation.  Add missing '__' in front of
27133         __builtin_ia32_packssdw256.
27135 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
27137         PR target/55719
27138         * config/s390/s390.c (s390_preferred_reload_class): Do not return
27139         NO_REGS for larl operands.
27140         (s390_reload_larl_operand): Use s390_load_address instead of
27141         emit_move_insn.
27143 2013-01-11  Richard Biener  <rguenther@suse.de>
27145         * tree-cfg.c (verify_node_sharing_1): Split out from ...
27146         (verify_node_sharing): ... here.
27147         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
27149 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
27151         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
27152         Substitute TREECHECKING.
27153         * configure: Regenerate.
27154         * Makefile.in (TREECHECKING): New.
27156 2013-01-11  Richard Guenther  <rguenther@suse.de>
27158         PR tree-optimization/44061
27159         * tree-vrp.c (extract_range_basic): Compute zero as
27160         value-range for __builtin_constant_p of function parameters.
27162 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
27164         Update copyright years.
27166 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
27168         PR rtl-optimization/55672
27169         * lra-eliminations.c (mark_not_eliminable): Permit addition with
27170         const to be eliminable.
27172 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
27174         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
27175         * configure: Regenerate.
27177 2013-01-10  Richard Biener  <rguenther@suse.de>
27179         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
27181 2013-01-10  Richard Biener  <rguenther@suse.de>
27183         PR bootstrap/55792
27184         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
27185         locations for virtual PHI arguments.
27186         (rewrite_update_phi_arguments): Likewise.
27188 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
27190         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
27191         on to assembler.
27193 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
27195         PR tree-optimization/55921
27196         * tree-complex.c (expand_complex_asm): New function.
27197         (expand_complex_operations_1): Call it for GIMPLE_ASM.
27199 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
27201         PR target/55718
27202         * config/s390/s390.c (s390_symref_operand_p)
27203         (s390_loadrelative_operand_p): Merge the two functions.
27204         (s390_check_qrst_address, print_operand_address): Add parameters
27205         to s390_loadrelative_operand_p invokation.
27206         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
27207         (s390_reload_larl_operand, s390_secondary_reload): Use
27208         s390_loadrelative_operand_p instead of s390_symref_operand_p.
27209         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
27211 2013-01-09  Mike Stump  <mikestump@comcast.net>
27213         * dse.c (record_store): Remove unnecessary assert.
27215 2013-01-09  Jan Hubicka  <jh@suse.cz>
27217         PR tree-optimization/55569
27218         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
27219         * cfgloop.h (scale_loop_profile): Likewise.
27221 2013-01-09  Jan Hubicka  <jh@suse.cz>
27223         PR lto/45375
27224         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
27225         functions.
27226         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
27228 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
27230         PR middle-end/55114
27231         * expr.h (maybe_emit_group_store): Declare.
27232         * expr.c (maybe_emit_group_store): New function.
27233         * builtins.c (expand_builtin_int_roundingfn): Call it.
27234         (expand_builtin_int_roundingfn_2): Likewise.
27236 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
27238         PR rtl-optimization/55829
27239         * lra-constraints.c (match_reload): Add code for absent output.
27240         (curr_insn_transform): Add code for reloads of matched inputs
27241         without output.
27243 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
27245         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
27246         attribute of movddup insn to DF.
27247         (*vec_interleave_lowv2df): Ditto.
27248         (vec_dupv2df): Ditto.
27250 2013-01-09  Jan Hubicka  <jh@suse.cz>
27252         PR tree-optimiation/55875
27253         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
27254         EVERY_ITERATION parameter.
27255         (number_of_iterations_exit): Check if exit is executed every iteration.
27256         (idx_infer_loop_bounds): Similarly here.
27257         (n_of_executions_at_most): Simplify
27258         to only test for cases where statement is dominated by the
27259         particular bound; handle correctly the "postdominance" test.
27260         (scev_probably_wraps_p): Use max loop iterations info
27261         as a global bound first.
27263 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
27264             Nick Clifton  <nickc@redhat.com>
27266         * config/v850/v850.md (cbranchsf4): New pattern.
27267         (cstoresf4): New pattern.
27268         (cbranchdf4): New pattern.
27269         (cstoredf4): New pattern.
27270         (movsicc): Disallow floating point comparisons.
27271         (cmpsf_le_insn): Fix order of operators.
27272         (cmpsf_lt_insn): Likewise.
27273         (cmpsf_eq_insn): Likewise.
27274         (cmpdf_le_insn): Likewise.
27275         (cmpdf_lt_insn): Likewise.
27276         (cmpdf_eq_insn): Likewise.
27277         (cmpsf_ge_insn): Use LE comparison.
27278         (cmpdf_ge_insn): Likewise.
27279         (cmpsf_gt_insn): Use LT comparison.
27280         (cmpdf_gt_insn): Likewise.
27281         (cmpsf_ne_insn): Delete pattern.
27282         (cmpdf_ne_insn): Delete pattern.
27283         * config/v850/v850.c (v850_gen_float_compare): Use
27284         gen_cmpdf_eq_insn for NE comparison.
27285         (v850_float_z_comparison_operator)
27286         (v850_float_nz_comparison_operator): Move from here ...
27287         * config/v850/predicates.md: ... to here.  Move GT and GE
27288         comparisons into v850_float_z_comparison_operator.
27289         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
27290         Delete prototype.
27291         (v850_float_nz_comparison_operator): Likewise.
27293 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27295         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
27296         with calls to gen_insvsi/gen_insvdi.
27298 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
27300         * config/i386/i386.c (initial_ix86_tune_features): Set up
27301         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
27303 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
27304             Jakub Jelinek  <jakub@redhat.com>
27306         PR tree-optimization/48189
27307         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
27308         If nitercst is 0, don't predict the exit edge.
27310 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
27312         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
27313         in asm_fprintf with reg_names.
27314         (aarch64_print_operand_address): Likewise.
27315         (aarch64_return_addr): Likewise.
27316         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
27318 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27320         * config/pa/pa.h (VAL_U6_BITS_P): Define.
27321         (INT_U6_BITS): Likewise.
27322         * config/pa/predicates.md (uint6_operand): New predicate.
27323         (shift5_operand, shift6_operand): Likewise.
27324         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
27325         arith32_operand.
27326         (lshrdi3): Use shift6_operand.
27327         (shrpsi4, shrpdi4): New insn patterns.
27328         (extzv): Delete expander.
27329         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
27330         predicates in unamed zero extract patterns.  Tighten common constraint.
27331         (extv): Delete expander.
27332         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
27333         predicates in unamed sign extract patterns.  Tighten common constraint.
27334         (insv): Delete expander.
27335         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
27336         predicates in unamed insert patterns.  Tighten common constraint.
27337         Change uint32_operand predicate to uint6_operand predicate in unamed
27338         DImode pattern to insert constant values of type 1...1xxxx.
27340 2013-01-04  Jan Hubicka  <jh@suse.cz>
27342         PR tree-optimization/55823
27343         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
27344         issue.
27346 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27347             Uros Bizjak  <ubizjak@gmail.com>
27349         PR rtl-optimization/55845
27350         * df-problems.c (can_move_insns_across): Stop scanning at
27351         volatile_insn_p source instruction or give up if
27352         across_from .. across_to range contains any volatile_insn_p
27353         instructions.
27355 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
27357         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
27358         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
27359         Declare.
27360         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
27361         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
27363 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27365         PR fortran/55341
27366         * asan.c (asan_clear_shadow): New function.
27367         (asan_emit_stack_protection): Use it.
27369 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
27371         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
27372         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
27373         with tab instead of space.
27375 2013-01-08  Nick Clifton  <nickc@redhat.com>
27377         * config/rl78/rl78.c (rl78_expand_prologue): Always select
27378         register bank 0 at the start of an interrupt handler.
27379         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
27380         MDBH registers.
27382 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
27384         * config/aarch64/aarch64-simd.md
27385         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
27386         (aarch64_simd_bsl): Likewise.
27387         (aarch64_vcond_internal<mode>): Likewise.
27388         (vcond<mode><mode>): Likewise.
27389         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
27390         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
27392 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
27394         * config/aarch64/aarch64-builtins.c
27395         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
27397 2013-01-08  Martin Jambor  <mjambor@suse.cz>
27399         PR debug/55579
27400         * tree-sra.c (analyze_access_subtree): Return true also after
27401         potentially creating a debug-only replacement.
27403 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27405         PR middle-end/55890
27406         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
27408         PR tree-optimization/54120
27409         * tree-vrp.c (range_fits_type_p): Don't allow
27410         src_precision < precision from signed vr to unsigned_p
27411         if vr->min or vr->max is negative.
27412         (simplify_float_conversion_using_ranges): Test can_float_p
27413         against CODE_FOR_nothing.
27415 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
27416             Richard Biener  <rguenther@suse.de>
27418         PR middle-end/55851
27419         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
27420         types instead of just INTEGER_TYPE types.
27422 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
27424         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
27425         TF_SIZE): Define.
27427 2013-01-07  Steve Ellcey  <sellcey@mips.com>
27429         PR target/42661
27430         * config/mips/mips.opt: Change mad to mmad to match documentation.
27432 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27434         PR target/55897
27435         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
27436         .progmemx.data now.
27438 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27440         PR target/55897
27441         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
27442         (avr_addrspace_t): Add .section_name field.
27443         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
27444         array size.
27445         (avr_addrspace): Same.  Initialize .section_name.  Remove last
27446         NULL entry.  Put __memx into .progmemx.data.
27447         (progmem_section_prefix): Remove.
27448         (avr_asm_init_sections): No need to initialize progmem_section.
27449         (avr_asm_named_section): Use avr_addrspace[].section_name to get
27450         section name prefix.
27451         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
27452         retrieve the progmem section.
27453         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
27454         boundary to run over avr_addrspace[].
27455         (avr_register_target_pragmas): Ditto.
27457 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
27459         * varasm.c (output_constant_def_contents): For asan_protect_global
27460         protected strings, adjust DECL_ALIGN if needed, before testing for
27461         anchored symbols.
27462         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
27463         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
27464         normal decls.
27465         (output_object_block): For asan protected decls, emit asan padding
27466         after their contents.
27467         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
27468         (asan_finish_file): Test it here instead.
27470 2013-01-07  Nick Clifton  <nickc@redhat.com>
27471             Matthias Klose  <doko@debian.org>
27472             Doug Kwan  <dougkwan@google.com>
27473             H.J. Lu  <hongjiu.lu@intel.com>
27475         PR driver/55470
27476         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
27478         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
27480         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
27482         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
27484         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
27486 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27488         PR target/54461
27489         * doc/install.texi (Cross-Compiler-Specific Options): Document
27490         --with-avrlibc.
27492 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
27494         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
27495         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
27496         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
27497         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
27498         vqmovun_high_s64): Fix source operand number and update copyright.
27500 2013-01-07  Richard Biener  <rguenther@suse.de>
27502         PR middle-end/55890
27503         * gimple.h (gimple_call_builtin_p): New overload.
27504         * gimple.c (validate_call): New function.
27505         (gimple_call_builtin_p): Likewise.
27506         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27507         Use gimple_call_builtin_p.
27508         (find_func_clobbers): Likewise.
27509         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
27510         (strlen_optimize_stmt): Likewise.
27512 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27514         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
27515         (vld1q_dup_*): Likewise.
27516         (vld1_*): Likewise.
27517         (vld1q_*): Likewise.
27518         (vld1_lane_*): Likewise.
27519         (vld1q_lane_*): Likewise.
27521 2013-01-07  Richard Biener  <rguenther@suse.de>
27523         * lto-streamer.h (LTO_minor_version): Bump to 2.
27525 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
27527         * config/aarch64/aarch64-protos.h
27528         (aarch64_const_double_zero_rtx_p): Rename to...
27529         (aarch64_float_const_zero_rtx_p): ...this.
27530         (aarch64_float_const_representable_p): New.
27531         (aarch64_output_simd_mov_immediate): Likewise.
27532         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
27533         move immediate case.
27534         * config/aarch64/aarch64.c
27535         (aarch64_const_double_zero_rtx_p): Rename to...
27536         (aarch64_float_const_zero_rtx_p): ...this.
27537         (aarch64_print_operand): Allow printing of new constants.
27538         (aarch64_valid_floating_const): New.
27539         (aarch64_legitimate_constant_p): Check for valid floating-point
27540         constants.
27541         (aarch64_simd_valid_immediate): Likewise.
27542         (aarch64_vect_float_const_representable_p): New.
27543         (aarch64_float_const_representable_p): Likewise.
27544         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
27545         (aarch64_output_simd_mov_immediate): New.
27546         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
27547         (*movdf_aarch64): Likewise.
27548         * config/aarch64/constraints.md (Ufc): New.
27549         (Y): call aarch64_float_const_zero_rtx.
27550         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
27552 2013-01-07  Richard Biener  <rguenther@suse.de>
27554         PR tree-optimization/55888
27555         PR tree-optimization/55862
27556         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
27557         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
27558         not if it is contained therein.
27560 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27562         * config/avr/t-avr: Typo.
27564 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
27566         PR55243
27567         * config/avr/t-avr: Don't automatically rebuild
27568         $(srcdir)/config/avr/t-multilib
27569         $(srcdir)/config/avr/avr-tables.opt
27570         $(srcdir)/doc/avr-mmcu.texi
27571         (avr-mcus): New phony target to build them on request.
27572         (s-avr-mlib, s-avr-mmcu-texi): Remove.
27573         * avr/avr-mcus.def: Adjust comments.
27575 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
27577         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
27579 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27581         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
27583 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
27585         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
27587 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
27589         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
27590         to generate profiling.
27591         * config/rs6000/aix64.h (LIB_SPEC): Same.
27593 2013-01-04  Andrew Pinski  <apinski@cavium.com>
27595         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
27596         New function.
27597         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
27599 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
27601         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
27602         unconditionally.
27603         (ix86_expand_move): Ditto.
27604         (ix86_zero_extend_to_Pmode): Ditto.
27605         (ix86_expand_call): Ditto.
27606         (ix86_expand_special_args_builtin): Ditto.
27607         (ix86_expand_builtin): Ditto.
27609 2013-01-04  Richard Biener  <rguenther@suse.de>
27611         PR tree-optimization/55862
27612         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
27613         translating them through PHI nodes.
27615 2013-01-04  Martin Jambor  <mjambor@suse.cz>
27617         PR tree-optimization/55755
27618         * tree-sra.c (sra_modify_assign): Do not check that an access has no
27619         children when trying to avoid producing a VIEW_CONVERT_EXPR.
27621 2013-01-04  Marek Polacek  <polacek@redhat.com>
27623         PR middle-end/55859
27624         * opts.c (default_options_optimization): Clarify error message.
27626 2013-01-04  Richard Biener  <rguenther@suse.de>
27628         PR middle-end/55863
27629         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
27630         reassociation.
27632 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27634         PR target/53789
27635         * config/pa/pa.md (movsi): Revert previous change.
27636         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27637         references.
27639 2013-01-03  Richard Henderson  <rth@redhat.com>
27641         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27642         after eliminating TLS symbols.
27644 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27646         PR bootstrap/50167
27647         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27648         * graphite-poly.c (debug_gmp_value): Likewise.
27650 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27652         PR target/55712
27653         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27654         selected code model, define __code_mode_small__, __code_model_medium__,
27655         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27656         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27657         xchg temporary register with %k.  Declare temporary register as
27658         early clobbered.
27659         [__x86_64__]: For medium and large code models, preserve %rbx register.
27661 2013-01-03  Richard Biener  <rguenther@suse.de>
27663         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27664         (dump_subscript): Adjust.
27665         (finalize_ddr_dependent): Do not dump redundant info.
27666         (analyze_siv_subscript): Adjust.
27667         (subscript_dependence_tester): Likewise.
27668         (compute_affine_dependence): Likewise.
27670 2013-01-03  Richard Biener  <rguenther@suse.de>
27672         Revert
27673         2013-01-03  Richard Biener  <rguenther@suse.de>
27675         PR tree-optimization/55857
27676         * tree-vect-stmts.c (vectorizable_load): Do not setup
27677         re-alignment for invariant loads.
27679         2013-01-02  Richard Biener  <rguenther@suse.de>
27681         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27682         invariant load do not generate a vector load from the scalar location.
27684 2013-01-03  Richard Biener  <rguenther@suse.de>
27686         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27687         for not vectorizing.
27688         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27689         not build INDIRECT_REFs, call get_name once only.
27690         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27691         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27693 2013-01-03  Richard Biener  <rguenther@suse.de>
27695         PR tree-optimization/55857
27696         * tree-vect-stmts.c (vectorizable_load): Do not setup
27697         re-alignment for invariant loads.
27699 2013-01-03  Richard Biener  <rguenther@suse.de>
27701         PR lto/55848
27702         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27703         prefer a built-in decl.
27705 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27707         * gcc.c (process_command): Update copyright notice dates.
27708         * gcov.c (print_version): Likewise.
27709         * gcov-dump.c (print_version): Likewise.
27711         PR rtl-optimization/55838
27712         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27713         iv0.step, iv1.step and step.
27715 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27716             Marc Glisse  <marc.glisse@inria.fr>
27718         PR tree-optimization/55832
27719         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27720         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27721         integer_{one,zero}_node.
27723 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27725         PR debug/54402
27726         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27727         * var-tracking.c (reverse_op): Don't add reverse ops to
27728         VALUEs that have already
27729         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27731 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27733         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27735 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27737         * dumpfile.c (dump_loc): Print filename with location.
27738         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27739         new location_t parameter to emit complete unroll message with
27740         new dump framework.
27741         (canonicalize_loop_induction_variables): Compute loops location
27742         and pass to try_unroll_loop_completely.
27743         * loop-unroll.c (report_unroll_peel): New function.
27744         (peel_loops_completely): Use new dump format with location
27745         for main dumpfile message, and invoke report_unroll_peel on success.
27746         (decide_unrolling_and_peeling): Ditto.
27747         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27748         by report_unroll_peel.
27749         (decide_peel_completely): Ditto.
27750         (decide_unroll_constant_iterations): Ditto.
27751         (decide_unroll_runtime_iterations): Ditto.
27752         (decide_peel_simple): Ditto.
27753         (decide_unroll_stupid): Ditto.
27754         * cfgloop.c (get_loop_location): New function.
27755         * cfgloop.h (get_loop_location): Declare.
27757 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27759         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27760         NULL.
27762 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27764         PR middle-end/55198
27765         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27766         BLKmode objects when EXPAND_MEMORY is specified.
27768 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27770         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27771         in loop predicate.
27772         (fold_builtin_cpu): Do not share cpu model decls across statements.
27774 2013-01-02  Jason Merrill  <jason@redhat.com>
27776         PR c++/55804
27777         * tree.c (build_array_type_1): Revert earlier change.
27779 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27781         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27782         "cortex-a57".
27783         * config/aarch64/aarch64-tune.md: Re-generate.
27785 2013-01-02  Richard Biener  <rguenther@suse.de>
27787         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27788         invariant load do not generate a vector load from the scalar location.
27790 2013-01-02  Richard Biener  <rguenther@suse.de>
27792         PR bootstrap/55784
27793         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27794         * configure: Regenerate.
27796 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27798         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27799         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27800         (expand_builtin_int_roundingfn_2): Keep the original target around
27801         for the fallback case.
27803 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27805         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27806         to be clear for sign changes.
27808 2013-01-01  Jan Hubicka  <jh@suse.cz>
27810         * ipa-inline-analysis.c: Fix formatting.
27812 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27814         PR tree-optimization/55831
27815         * tree-vect-loop.c (get_initial_def_for_induction): Use
27816         gsi_after_labels instead of gsi_start_bb.
27818 Copyright (C) 2013 Free Software Foundation, Inc.
27820 Copying and distribution of this file, with or without modification,
27821 are permitted in any medium without royalty provided the copyright
27822 notice and this notice are preserved.