2013-09-26 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blobd16da9371787afb0b8bb32ee99fbd84ae5229df8
1 2013-09-26  Richard Biener  <rguenther@suse.de>
3         * alias.h (component_uses_parent_alias_set): Rename to ...
4         (component_uses_parent_alias_set_from): ... this.
5         * alias.c (component_uses_parent_alias_set): Rename to ...
6         (component_uses_parent_alias_set_from): ... this and return
7         the desired parent.
8         (reference_alias_ptr_type_1): Use the result from
9         component_uses_parent_alias_set_from instead of stripping
10         components one at a time.
11         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
15         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs): Move
16         prototypes to...
17         * tree-ssa-ter.h: New File.  Move prototypes here.
18         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
19         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
20         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
21         from is_replaceable_p. 
22         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
23         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
24         (process_replaceable): Use ter_is_replaceable_p.
25         (find_replaceable_in_bb): Use ter_is_replaceable_p.
26         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
27         newly refactored ssa_is_replaceable_p.
28         * cfgexpand.c: Include tree-outof-ssa.h.
29         * ssaexpand.h: Delete.
31 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
33         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
34         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
35         tree-ssa.c
36         (create_gimple_tmp): Delete.
37         (get_expr_type, build_assign, build_type_cast): Move to...
38         * gimple-builder.c: New File.
39         (get_expr_type): Relocate from gimple.c.
40         (build_assign, build_type_cast): Change to only create ssanames.
41         * gimple.h: Move prototypes to...
42         * gimple-builder.h: New File. Here.
43         * tree-ssa.h: And here.
44         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
45         count_uses_and_derefs): Relocate from gimple.c.
46         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
47         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
48         * tree-ssa-math-opts (execute_cse_reciprocals): Use
49         gimple_replace_ssa_lhs.
50         * asan.c: Include gimple-builder.h.
51         * Makefile.in: Add gimple-builder.o.
53 2013-09-26  Richard Biener  <rguenther@suse.de>
55         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
56         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
57         (loe_visit_block): Use gcc_checking_assert.
58         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
59         gimple_assign_ssa_name_copy_p.
60         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
61         change.
63 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
65         * config/rs6000/t-rs6000 (rs6000.o): Remove.
66         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
68 2013-09-26  Richard Biener  <rguenther@suse.de>
70         PR tree-optimization/58539
71         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
72         the fact that debug statements are not taking part in loop-closed
73         SSA construction.
75 2013-09-26  Nick Clifton  <nickc@redhat.com>
77         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
78         time warning message.
79         (msp430_print_operand_raw): Delete unused letter parameter.
80         (TARGET_PRINT_OPERAND_ADDRESS): Define.
81         (msp430_print_operand_address): New function.
82         (msp430_print_operand): Move address printing code from here to
83         new function.
84         * config/msp430/msp430.md (movsipsi2): Add comment in generated
85         assembler.
86         (zero_extendpsisi2): Likewise.
87         (extendpsisi2): New pattern.
88         (andneghi3): New pattern.
90 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
92         * config/aarch64/aarch64.opt (mlra): New option.
93         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
94         (TARGET_LRA_P): Define.
96 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
98         * expr.c (expand_assignment): Remove obsolete comment.
100 2013-09-25  Jeff Law  <law@redhat.com>
102         * tree-flow.h (thread_through_all_blocks): Prototype moved into
103         tree-ssa-threadupdate.h.
104         (register_jump_thread): Similarly.
105         * tree-ssa-threadupdate.h: New header file.
106         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
107         * tree-vrp.c: Likewise.
108         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
109         (thread_around_empty_blocks): Change type of path vector argument to
110         an edge,type pair from just an edge.  Initialize both elements when
111         appending to a jump threading path.  Tweak references to elements
112         appropriately.
113         (thread_across_edge): Similarly.  Release memory for the elements
114         as needed.
115         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
116         (dump_jump_thread_path): New function broken out from
117         register_jump_thread.
118         (register_jump_thread): Use dump_jump_thread_path.  Change type of
119         path vector entries.  Search the path for NULL edges and dump
120         the path if one is found.  Tweak the conversion of path to 3-edge
121         form to use the block copy type information embedded in the path.
123 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
125         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
127 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
128             Vladimir Makarov  <vmakarov@redhat.com>
130         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
131         from the least significant bit.
132         (strip_address_mutations): Add bitfield operations handling.
133         (must_be_index_p): Add shifting and rotate operations handling.
134         (set_address_base): Use must_be_base_p predicate.
135         (set_address_index): Use must_be_index_p predicate.
137 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
138             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
139             Sergey Lega  <sergey.s.lega@intel.com>
140             Anna Tikhonova  <anna.tikhonova@intel.com>
141             Ilya Tocar  <ilya.tocar@intel.com>
142             Andrey Turetskiy  <andrey.turetskiy@intel.com>
143             Ilya Verbin  <ilya.verbin@intel.com>
144             Kirill Yukhin  <kirill.yukhin@intel.com>
145             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
147         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
148         Use new names.
149         (ix86_expand_vector_move_misalign): Support new unaligned load and
150         stores and use new names.
151         (CODE_FOR_sse2_storedqu): Rename to ...
152         (CODE_FOR_sse2_storedquv16qi): ... this.
153         (CODE_FOR_sse2_loaddqu): Rename to ...
154         (CODE_FOR_sse2_loaddquv16qi): ... this.
155         (CODE_FOR_avx_loaddqu256): Rename to ...
156         (CODE_FOR_avx_loaddquv32qi): ... this.
157         (CODE_FOR_avx_storedqu256): Rename to ...
158         (CODE_FOR_avx_storedquv32qi): ... this.
159         * config/i386/i386.md (fpint_logic): New.
160         * config/i386/sse.md (VMOVE): Extend for AVX512.
161         (VF): Ditto.
162         (VF_128_256): New.
163         (VF_512): Ditto.
164         (VI_UNALIGNED_LOADSTORE): Ditto.
165         (sse2_avx_avx512f): Ditto.
166         (sse2_avx2): Extend for AVX512.
167         (sse4_1_avx2): Ditto.
168         (avx2_avx512f): New.
169         (sse): Extend for AVX512.
170         (sse2): Ditto.
171         (sse4_1): Ditto.
172         (avxsizesuffix): Ditto.
173         (sseintvecmode): Ditto.
174         (ssePSmode): Ditto.
175         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
176         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
177         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
178         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
179         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
180         (<sse2_avx_avx512f>_storedqu<mode): ... this.
181         (<sse>_movnt<mode>): Replace constraint "x" with "v".
182         (STORENT_MODE): Extend for AVX512.
183         (*absneg<mode>2): Replace constraint "x" with "v".
184         (*mul<mode>3): Ditto.
185         (*ieee_smin<mode>3): Ditto.
186         (*ieee_smax<mode>3): Ditto.
187         (avx_cmp<mode>3): Replace VF with VF_128_256.
188         (*<sse>_maskcmp<mode>3_comm): Ditto.
189         (<sse>_maskcmp<mode>3): Ditto.
190         (<sse>_andnot<mode>3): Extend for AVX512.
191         (<code><mode>3, anylogic): Replace VF with VF_128_256.
192         (<code><mode>3, fpint_logic): New.
193         (*<code><mode>3): Extend for AVX512.
194         (avx512flogicsuff): New.
195         (avx512f_<logic><mode>): Ditto.
196         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
197         VF_128_256.
198         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
199         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
200         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
201         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
202         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
203         (xop_vpermil2<mode>3): Ditto.
204         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
205         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
206         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
207         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
209 2013-09-25  Tom Tromey  <tromey@redhat.com>
211         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
212         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
213         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
214         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
215         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
216         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
217         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
218         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
219         (GRAPHITE_HTAB_H): Remove.
221 2013-09-25  Tom Tromey  <tromey@redhat.com>
223         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
225 2013-09-25  Tom Tromey  <tromey@redhat.com>
227         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
229 2013-09-25  Tom Tromey  <tromey@redhat.com>
231         * config/i386/t-i386 (i386.o): Remove.
232         (i386-c.o): Use COMPILE and POSTCOMPILE.
234 2013-09-25  Tom Tromey  <tromey@redhat.com>
236         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
238 2013-09-25  Tom Tromey  <tromey@redhat.com>
240         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
241         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
242         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
243         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
244         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
245         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
246         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
247         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
248         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
249         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
250         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
251         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
252         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
253         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
254         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
255         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
256         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
257         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
258         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
259         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
260         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
261         (tree-ssa-pre.o, tree-ssa-sccvn.o)
262         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
263         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
264         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
265         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
266         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
267         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
268         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
269         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
270         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
271         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
272         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
273         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
274         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
275         (graphite-clast-to-gimple.o, graphite-dependences.o)
276         (graphite-interchange.o, graphite-poly.o)
277         (graphite-scop-detection.o, graphite-sese-to-poly.o)
278         (graphite-optimize-isl.o, tree-vect-loop.o)
279         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
280         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
281         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
282         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
283         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
284         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
285         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
286         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
287         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
288         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
289         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
290         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
291         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
292         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
293         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
294         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
295         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
296         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
297         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
298         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
299         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
300         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
301         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
302         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
303         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
304         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
305         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
306         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
307         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
308         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
309         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
310         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
311         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
312         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
313         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
314         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
315         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
316         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
317         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
318         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
319         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
320         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
321         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
322         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
323         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
324         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
325         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
326         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
327         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
328         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
329         (gcov-dump.o): Remove.
330         (default-c.o): Use COMPILE and POSTCOMPILE.
331         (CFLAGS-gcc.o): New variable.
332         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
334 2013-09-25  Tom Tromey  <tromey@redhat.com>
336         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
337         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
338         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
339         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
340         (c-family/c-pragma.o, c-family/c-pretty-print.o)
341         (c-family/c-semantics.o, c-family/c-ada-spec.o)
342         (c-family/array-notation-common.o, c-family/stub-objc.o)
343         (c-family/c-ubsan.o): Remove.
345 2013-09-25  Tom Tromey  <tromey@redhat.com>
347         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
349 2013-09-25  Tom Tromey  <tromey@redhat.com>
351         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
352         to add -DENABLE_SHARED_LIBGCC.
353         (gcc.o): Don't use subshell.
355 2013-09-25  Tom Tromey  <tromey@redhat.com>
357         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
358         * configure.ac: Don't invoke AM_PROG_CC_C_O.
359         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
360         * configure, config.in: Rebuild.
362 2013-09-25  Tom Tromey  <tromey@redhat.com>
364         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
365         (COMPILE, POSTCOMPILE): New variables.
366         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
367         (DEPFILES): New variable.
368         Include ".Po" files.
369         * configure.ac: Add checks for dependency checking.
370         * configure, aclocal.m4: Regenerate.
372 2013-09-25  Tom Tromey  <tromey@redhat.com>
374         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
375         ($(ALL_HOST_OBJS)): Move order-only dependency to end
376         of file.
378 2013-09-25  Tom Tromey  <tromey@redhat.com>
380         * Makefile.in (generated_files): Add options.h,
381         target-hooks-def.h, insn-opinit.h,
382         common/common-target-hooks-def.h, pass-instances.def,
383         c-family/c-target-hooks-def.h.
385 2013-09-25  Jeff Law  <law@redhat.com>
387         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
388         than foo[foo.length () - 1] to access last member in a vec.
389         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
391 2013-09-25  Richard Biener  <rguenther@suse.de>
393         PR middle-end/58521
394         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
396 2013-09-25  Jan Hubicka  <jh@suse.cz>
398         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
399         test.
401 2013-09-25  Marek Polacek  <polacek@redhat.com>
403         PR sanitizer/58420
404         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
405         when determining the type name.
407 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
409         * config/sh/sh.md: Fix formatting.
411 2013-09-24  Xinliang David Li <davidxl@google.com>
413         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): 
414         Check max peel iterations parameter.
415         * param.def: New parameter.
416         * doc/invoke.texi: Document New parameter.
418 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
420         * gimple-pretty-print.c: Various whitespace tweaks.
421         * tree-core.h: Likewise.
422         * tree-pretty-print.c: Likewise.
423         * tree-ssa-alias.c: Likewise.
424         * tree-ssa-copy.c: Likewise.
425         * tree-ssanames.c: Likewise.
426         * tree-ssanames.h: Likewise.
427         * tree-vrp.c: Likewise.
429 2013-09-24  Alan Modra  <amodra@gmail.com>
431         PR middle-end/57134
432         PR middle-end/57586
433         * stmt.c (expand_asm_operands): Call expand_expr with
434         EXPAND_MEMORY for output operands that disallow regs.  Don't
435         use EXPAND_WRITE on inout operands.
437 2013-09-24  Richard Biener  <rguenther@suse.de>
439         PR middle-end/58513
440         * tree.c (reference_alias_ptr_type): Move ...
441         * alias.c (reference_alias_ptr_type): ... here and implement
442         in terms of the new reference_alias_ptr_type_1.
443         (ref_all_alias_ptr_type_p): New helper.
444         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
445         use ref_all_alias_ptr_type_p.
446         (get_deref_alias_set): Add flag_strict_aliasing check here.
447         (reference_alias_ptr_type_1): New function, split out from ...
448         (get_alias_set): ... here.
449         (alias_ptr_types_compatible_p): New function.
450         * alias.h (reference_alias_ptr_type): Declare.
451         (alias_ptr_types_compatible_p): Likewise.
452         * tree.h (reference_alias_ptr_type): Remove.
453         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
454         to compare MEM_REF alias types.
456 2013-09-24  Richard Biener  <rguenther@suse.de>
458         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
460 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
462         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
463         reload helper function arrays into a single array reg_addr.
464         (reload_fpr_gpr): Likewise.
465         (reload_gpr_vsx): Likewise.
466         (reload_vsx_gpr): Likewise.
467         (struct rs6000_reg_addr): Likewise.
468         (reg_addr): Likewise.
469         (rs6000_debug_reg_global): Change rs6000_vector_reload,
470         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
471         (rs6000_init_hard_regno_mode_ok): Likewise.
472         (rs6000_secondary_reload_direct_move): Likewise.
473         (rs6000_secondary_reload): Likewise.
475         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
476         constraints: wu, ww, and wy.  Repurpose wv constraint added during
477         power8 changes.  Put wg constraint in alphabetical order.
479         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
480         for future work to add ISA 2.07 VSX single precision support.
481         (-mvsx-scalar-double): Change default from -1 to 1, update
482         documentation comment.
483         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
484         (-mupper-regs-df): New debug switch to control whether DF values
485         can go in the traditional Altivec registers.
486         (-mupper-regs-sf): New debug switch to control whether SF values
487         can go in the traditional Altivec registers.
489         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
490         and wy constraints.
491         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
492         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
493         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
494         (rs6000_debug_legitimate_address_p): Print if we are running
495         before, during, or after reload.
496         (rs6000_secondary_reload): Add a comment.
497         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
499         * config/rs6000/constraints.md (wa constraint): Sort w<x>
500         constraints.  Update documentation string.
501         (wd constraint): Likewise.
502         (wf constraint): Likewise.
503         (wg constraint): Likewise.
504         (wn constraint): Likewise.
505         (ws constraint): Likewise.
506         (wt constraint): Likewise.
507         (wx constraint): Likewise.
508         (wz constraint): Likewise.
509         (wu constraint): New constraint for ISA 2.07 SFmode scalar
510         instructions.
511         (ww constraint): Likewise.
512         (wy constraint): Likewise.
513         (wv constraint): Repurpose ISA 2.07 constraint that we not used in
514         the previous submissions.
515         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
517 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
519         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
520         (REG_BR_PROB): Say that the probability is stored in an int_list.
521         * reg-notes.def: Update commentary to mention INT_LIST.
522         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
523         (INT_LIST): New rtx.
524         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
525         * rtlanal.c (int_reg_note_p): New function.
526         (alloc_reg_note): Assert that the note does not have an int argument.
527         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
528         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
529         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
530         rather than an INSN_LIST.  Handle INT_LIST.
531         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
532         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
533         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
534         Manipulate them as ints rather than rtxes.
535         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
536         * regmove.c (copy_src_to_dest): Likewise.
537         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
539         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
540         into the cases that need it.
541         * config/arm/arm.c (arm_unwind_emit): Likewise.
543         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
544         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
545         * loop-doloop.c (add_test, doloop_modify): Likewise.
546         * loop-unswitch.c (compare_and_jump_seq): Likewise.
547         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
548         * predict.c (combine_predictions_for_insn): Likewise.
549         * print-rtl.c (print_rtx): Handle INT_LIST.
550         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
551         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
552         * config/arm/arm.c (emit_unlikely_jump): Likewise.
553         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
554         (ix86_split_fp_branch, predict_jump): Likewise.
555         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
556         * config/sh/sh.c (expand_cbranchsi4): Likewise.
557         * config/spu/spu.c (ea_load_store_inline): Likewise.
559         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
560         value of a REG_BR_PROB note.
561         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
562         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
563         * emit-rtl.c (try_split): Likewise.
564         * predict.c (br_prob_note_reliable_p): Likewise.
565         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
566         * reorg.c (mostly_true_jump): Likewise.
567         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
568         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
569         * config/i386/i386.c (ix86_print_operand): Likewise.
570         * config/ia64/ia64.c (ia64_print_operand): Likewise.
571         * config/mmix/mmix.c (mmix_print_operand): Likewise.
572         * config/rs6000/rs6000.c (output_cbranch): Likewise.
573         * config/s390/s390.c (s390_expand_tbegin): Likewise.
574         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
575         * config/sparc/sparc.c (output_cbranch): Likewise.
576         * config/spu/spu.c (get_branch_target): Likewise.
577         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
578         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
580 2013-09-23  Jan Hubicka  <jh@suse.cz>
582         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
583         for ipa-devirt.
584         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
585         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
586         of external calls
587         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
588         (gimple_fold_call): Dump inconsistent devirtualizations; add
589         sanity check for type based devirtualizations.
590         * ipa-prop.c: Include ipa-utils.h
591         (ipa_intraprocedural_devirtualization): Add sanity check.
592         (try_make_edge_direct_virtual_call): Likewise.
594 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
596         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
597         assignment statements.
599 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
601         * gimple-pretty-print.c (dump_ssaname_info): New function.
602         (dump_gimple_phi): Call it.
603         (pp_gimple_stmt_1): Likewise.
604         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
605         (range_info_def): Declare.
606         * tree-pretty-print.c (pp_double_int): New function.
607         (dump_generic_node): Call it.
608         * tree-pretty-print.h (pp_double_int): Declare.
609         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
610         * tree-ssanames.h (range_info_def): New structure.
611         (value_range_type): Move definition here.
612         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
613         Declare.
614         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
615         initialization.
616         (set_range_info): New function.
617         (get_range_info): Likewise.
618         (duplicate_ssa_name_range_info): Likewise.
619         (duplicate_ssa_name_fn): Check pointer type and call
620         duplicate_ssa_name_range_info.
621         * tree-ssa-copy.c (fini_copy_prop): Likewise.
622         * tree-vrp.c (value_range_type): Remove definition, now in
623         tree-ssanames.h.
624         (vrp_finalize): Call set_range_info to update value range of
625         SSA_NAMEs.
626         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
627         (SSA_NAME_RANGE_INFO): New macro.
629 2013-09-23  Richard Biener  <rguenther@suse.de>
631         PR tree-optimization/58464
632         * tree-ssa-pre.c (phi_trans_lookup): Remove.
633         (phi_trans_add): Change to add conditionally on being not
634         present.
635         (phi_translate_1): Remove recursion detection here.
636         (phi_translate): Pre-seed the cache with NULL to catch
637         recursion here in a more generic way.
638         (bitmap_find_leader): Adjust comment.
639         (get_representative_for): Dump value-numbers.
640         (create_expression_by_pieces): Likewise.
641         (insert_into_preds_of_block): Likewise.
643 2013-09-23  Christian Bruel  <christian.bruel@st.com>
645         PR target/58475
646         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
647         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
649 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
651         Revert r202780:
652         2013-09-20  Renlin Li  <renlin.li@arm.com>
654         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
655         (aarch64_expand_epilogue): Likewise.
656         (aarch64_legitimize_reload_address): Likewise.
658 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
660         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
661         NULL_TREE before pushing them onto the vector.  Likewise for labels.
663 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
665         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
666         comparison operators when -fno-trapping-math is in effect.
667         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
668         comparison operators in TFmode and assert that unsupported operators
669         cannot reach here.
670         (ia64_print_operand): Likewise.
672 2013-09-21  Jan Hubicka  <jh@suse.cz>
674         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
675         (sse_typeless_stores): Enable for core
676         (sse_load0_by_pxor): Likewise.
677         (four_jump_limit): Disable for core.
678         (pad_returns): Likewise.
679         (avoid_vector_decode): Likewise.
680         (fuse_cmp_and_branch): Enable for cores.
681         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
683 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
685         PR middle-end/56791
686         * config/pa/pa.c (pa_option_override): Disable auto increment and
687         decrement instructions until reload is completed.
689         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
690         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
691         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
693 2013-09-20  DJ Delorie  <dj@redhat.com>
694             Nick Clifton  <nickc@redhat.com>
696         * config/rl78/rl78.c: Various whitespace and comment tweaks.
697         (need_to_save): Save bank 0 on interrupts.
698         (characterize_address): Strip far address wrappers.
699         (rl78_as_legitimate_address): Likewise.
700         (transcode_memory_rtx): Likewise.
701         (rl78_peep_movhi_p): Disable this peephole after devirt.
702         (rl78_propogate_register_origins): Forget all origins when a
703         CLOBBER is seen.
704         * config/rl78/rl78-virt.md: Various whitespace tweaks.
705         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
706         peephole2's.
707         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
708         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
709         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
710         relocating.
711         * config/rl78/constraints.md: Various whitespace and paren tweaks.
713 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
715         * config/pa/pa.md: In "scc" insn patterns, change output template to
716         handle const0_rtx in reg_or_0_operand operands.
718 2013-09-20  Martin Husemann  <martin@NetBSD.org>
720         PR target/56875
721         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
722         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
724 2013-09-20  Richard Biener  <rguenther@suse.de>
726         PR middle-end/58484
727         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
728         remembering SSA name version and block index.
729         (new_scev_info_str): Adjust.
730         (hash_scev_info): Likewise.  Also hash the block index.
731         (eq_scev_info): Adjust.
732         (find_var_scev_info): Likewise.
733         (struct instantiate_cache_entry): Remove.
734         (struct instantiate_cache_type): Use a htab to map name, block
735         to chrec.
736         (instantiate_cache_type::~instantiate_cache_type): Adjust.
737         (get_instantiated_value_entry): Likewise.
738         (hash_idx_scev_info, eq_idx_scev_info): New functions.
739         (instantiate_scev_name): Adjust.
741 2013-09-20  Jeff Law  <law@redhat.com>
743         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
745 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
747         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
748         Call aarch64_simd_expand_args to update op[argc].
750 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
752         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside 
753           plugin argument.
755 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
757        * gengtype.c (file_rules): Added rule for *.cc files.
758        (get_output_file_with_visibility): Give fatal message when no
759        rules found.
761 2013-09-20  Renlin Li  <renlin.li@arm.com>
763         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
764         (aarch64_expand_epilogue): Likewise.
765         (aarch64_legitimize_reload_address): Likewise.
767 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
769         PR middle-end/57748
770         * expr.c (expand_assignment): Remove misalignp code path.
772 2013-09-20  Marek Polacek  <polacek@redhat.com>
774         PR sanitizer/58413
775         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
776         TYPE_PRECISION.  Add asserts.
778 2013-09-20  Richard Biener  <rguenther@suse.de>
780         PR tree-optimization/58453
781         * tree-loop-distribution.c (distribute_loop): Apply the cost
782         model for -ftree-loop-distribute-patterns, too.
784 2013-09-20  Richard Biener  <rguenther@suse.de>
786         PR middle-end/58473
787         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
788         make type comparison less strict.
790 2013-09-20  Alan Modra  <amodra@gmail.com>
792         * configure: Regenerate.
793         * aclocal.m4: Regenerate.
795 2013-09-20  Marek Polacek  <polacek@redhat.com>
797         PR other/58467
798         * doc/extend.texi: Document that attribute used is meant to be used
799         on variables with static storage duration.
801 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
803         PR tree-optimization/58472
804         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
805         simd_lane_access set inv_p = false.
806         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
807         the simduid magic VAR_DECL.
809 2013-09-19  Jan Hubicka  <jh@suse.cz>
811         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
813 2013-09-17  Jeff Law  <law@redhat.com>
815         * tree-ssa-dom.c (record_temporary_equivalences): New function
816         split out of dom_opt_dom_walker::after_dom_children.
817         (dom_opt_dom_walker::thread_across_edge): Move common code
818         in here from dom_opt_dom_walker::after_dom_children.
819         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
821 2013-09-19  Jan Hubicka  <jh@suse.cz>
823         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
824         (TARGET_GENERIC): Use PROCESOR_GENERIC
825         (enum processor_type): Unify generic32 and 64.
826         * i386.md (cpu): Likewise.
827         * x86-tune.def (use_leave): Enable for generic32.
828         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
829         * athlon.md: Change generic64 to generic in all occurences.
830         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
831         (ix86_target_macros_internal): Likewise.
832         * driver-i386.c (host_detect_local_cpu): Likewise.
833         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename to ..
834         (generic_memcpy, generic_memset, generic_cost): This one.
835         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
836         (m_GENERIC32, m_GENERIC64): Remove.
837         (m_GENERIC): Turn into one flag.
838         (processor_target): Unify generic tunnings.
839         (ix86_option_override_internal): Replace generic32/64 by generic.
840         (ix86_issue_rate): Likewise.
841         (ix86_adjust_cost): Likewise.
843 2013-09-19  Jan Hubicka  <jh@suse.cz>
845         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
846         of speculative flag.
848 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
850         * omp-low.c (expand_omp_sections): Always pass len - 1 to
851         GOMP_sections_start, even if !exit_reachable.
853 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
855         * lra-constraints.c (need_for_all_save_p): Use macro
856         HARD_REGNO_CALL_PART_CLOBBERED.
857         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
858         set up pseudo conflict hard regs.
860 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
862         PR target/58452
863         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
864         operands.
866 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
868         PR rtl-optimization/58438
869         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
870         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
871         from previous subpasses.
873 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
875         * arm.c (arm_get_frame_offsets): Validate architecture supports
876         LDRD/STRD before accepting the tuning preference.
877         (arm_expand_prologue): Likewise.
878         (arm_expand_epilogue): Likewise.
880 2013-09-18  Richard Biener  <rguenther@suse.de>
882         PR tree-optimization/58417
883         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
884         have chrecs with symbols defined in the loop as operands.
885         (chrec_fold_multiply): Likewise.
886         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
887         parameters before folding binary operations.
888         (struct instantiate_cache_entry_hasher): Remove.
889         (struct instantiate_cache_type): Use a pointer-map.
890         (instantiate_cache_type::instantiate_cache_type): New function.
891         (instantiate_cache_type::get): Likewise.
892         (instantiate_cache_type::set): Likewise.
893         (instantiate_cache_type::~instantiate_cache_type): Adjust.
894         (get_instantiated_value_entry): Likewise.
895         (global_cache): New global.
896         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
897         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
898         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
899         (instantiate_scev_name): Adjust.
900         (instantiate_scev): Construct global instead of local cache.
901         (resolve_mixers): Likewise.
903 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
904             Paolo Carlini  <paolo.carlini@oracle.com>
906         PR c++/58458
907         * doc/implement-cxx.texi: Fix references to the C++ standards.
909 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
911         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
912         * tree-vect-data-refs.c (vect_analyze_data_refs): For
913         simd_lane_access drs, update also DR_ALIGNED_TO.
915 2013-09-18  Marek Polacek  <polacek@redhat.com>
917         PR sanitizer/58411
918         * doc/extend.texi: Document no_sanitize_undefined attribute.
919         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
920         no_sanitize_undefined attribute.
922 2013-09-18  Nick Clifton  <nickc@redhat.com>
924         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
925         (ASM_DECLARE_FUNCTION_NAME): Define.
927 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
929         * compare-elim.c (find_comparison_dom_walker): New class
930         (find_comparisons_in_bb): Rename to
931         find_comparison_dom_walker::before_dom_children
932         (find_comparisons): Adjust
933         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
934         adjust.
935         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
936         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
937         (init_walk_dominator_tree): Remove declaration.
938         (fini_walk_dominator_tree): Remove declaration.
939         * fwprop.c (single_def_use_dom_walker): New class
940         (single_def_use_enter_block): Convert to
941         single_def_use_dom_walker::before_dom_children.
942         (single_def_use_leave_block): Convert to
943         single_def_use_dom_walker::after_dom_children.
944         (build_single_def_use_links): Adjust.
945         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
946         class.
947         (find_candidates_in_block): Convert to
948         find_candidates_dom_walker::before_dom_children.
949         (execute_strength_reduction): Adjust.
950         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
951         (sese_dom_walker): New class.
952         (sese_dom_walker::sese_dom_walker): New constructor.
953         (sese_dom_walker::~sese_dom_walker): New destructor.
954         (build_sese_conditions_before): Convert to
955         sese_dom_walker::before_dom_children.
956         (build_sese_conditions_after): Convert to
957         sese_dom_walker::after_dom_children.
958         (build_poly_scop): Adjust
959         * tree-into-ssa.c (rewrite_dom_walker): New class
960         (rewrite_enter_block): Convert to
961         rewrite_dom_walker::before_dom_children.
962         (rewrite_leave_block): Convert to
963         rewrite_dom_walker::after_dom_children.
964         (rewrite_update_dom_walker): New class.
965         (rewrite_update_enter_block): Convert to
966         rewrite_update_dom_walker::before_dom_children.
967         (rewrite_update_leave_block): Convert to
968         rewrite_update_dom_walker::after_dom_children.
969         (rewrite_blocks, rewrite_into_ssa): Adjust.
970         (mark_def_dom_walker): New class.
971         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
972         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
973         (mark_def_sites_blocks): Convert to
974         mark_def_dom_walker::before_dom_children.
975         (mark_def_site_blocks): Remove.
976         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
977         (tree_ssa_dominator_optimize): Adjust.
978         (dom_thread_across_edge): Convert to method
979         dom_opt_dom_walker::thread_across_edge.
980         (dom_opt_enter_block): Convert to member function
981         dom_opt_dom_walker::before_dom_children.
982         (dom_opt_leave_block): Convert to member function
983         dom_opt_dom_walker::after_dom_children.
984         * tree-ssa-dse.c (dse_dom_walker): New class.
985         (dse_enter_block): Convert to member function
986         dse_dom_walker::before_dom_children.
987         (tree_ssa_dse): Adjust.
988         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
989         (determine_invariantness_stmt): Convert to method
990         invariantness_dom_walker::before_dom_children.
991         (determine_invariantness): Remove
992         (move_computations_dom_walker): New class.
993         (move_computations_stmt): Convert to method
994         move_computations_dom_walker::before_dom_children.
995         (move_computations, tree_ssa_lim): Adjust.
996         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
997         (nt_init_block): Make method
998         notrappping_dom_walker::before_dom_children.
999         (nt_fini_block): Make
1000         method nontrapping_dom_walker::after_dom_children.
1001         (get_non_trapping): Adjust.
1002         * tree-ssa-pre.c (eliminate_dom_walker): New class.
1003         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
1004         (eliminate_leave_block): Make method.
1005         eliminate_dom_walker::after_dom_children.
1006         (eliminate): Adjust
1007         * tree-ssa-strlen.c (strlen_dom_walker): New class.
1008         (strlen_enter_block): Make method
1009         strlen_dom_walker::before_dom_children.
1010         (strlen_leave_block): Make
1011         method strlen_dom_walker::after_dom_children.
1012         (tree_ssa_strlen): Adjust.
1013         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
1014         (tree_ssa_uncprop): Adjust.
1015         (uncprop_leave_block): Make method
1016         uncprop_dom_walker::after_dom_children.
1017         (uncprop_leave_block): Make method
1018         uncprop_dom_walker::before_dom_children.
1020 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
1022         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn
1023         before branch in same basic block.  Check both src and dest of
1024         the move insn.
1026 2013-09-17  Nick Clifton  <nickc@redhat.com>
1028         * config/rl78/rl78-real.md (bf): New pattern.
1029         (bt): New pattern.
1030         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
1031         (rl78_print_operand): Do not put a # before a %B.
1032         * config/rl78/rl78.opt: Tweak doc strings.
1034 2013-09-17  DJ Delorie  <dj@redhat.com>
1036         * config/rl78/constraints.md (Wcv): Allow up to $r31.
1037         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
1038         (rl78_option_override): Likewise, if -mallregs.
1039         (is_virtual_register): Likewise.
1040         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
1041         (REGNO_OK_FOR_BASE_P): Likewise.
1042         * config/rl78/rl78.opt (-mallregs): New.
1044 2013-09-17  Nick Clifton  <nickc@redhat.com>
1046         * config/rl78/rl78.c (need_to_save): Change return type to bool.
1047         For interrupt functions: save all call clobbered registers if the
1048         interrupt handler is not a leaf function.
1049         (rl78_expand_prologue): Always recompute the frame information.
1050         For interrupt functions: only select bank 0 if one of the bank 0
1051         registers is going to be psuhed.
1053 2013-09-17  DJ Delorie  <dj@redhat.com>
1055         * config/rl78/constraints.md: For each W* constraint, rename to C*
1056         and create a W* constraint that checks for an optional ES: prefix
1057         pattern also.
1058         * config/rl78/rl78.md (UNS_ES_ADDR): New.
1059         (es_addr): New.  Used to wrap far addresses.
1060         * config/rl78/rl78-protos.h (rl78_es_addr): New.
1061         (rl78_es_base): New.
1062         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
1063         wrapped far addresses.
1064         (rl78_print_operand_1): Unwrap far addresses before processing.
1065         (rl78_lo16): Wrap far addresses in unspecs.
1066         (rl78_es_addr): New.
1067         (rl78_es_base): New.
1068         (insn_ok_now): Check for not-yet-wrapped far addresses.
1069         (transcode_memory_rtx): Properly re-wrap far addresses.
1071 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
1073         * config/sparc/t-rtems: Add leon3 multilibs.
1075 2013-09-17  Cong Hou  <congh@google.com>
1077         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
1078         when checking the dot production pattern. The type of rhs operand
1079         of multiply is now checked correctly.
1081 2013-09-17  Jeff Law  <law@redhat.com>
1083         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1084         edge implied equivalences into successor phis.
1085         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
1086         here from tree-ssa-threadedge.c.
1087         (mark_threaded_blocks): When threading through a joiner, if both
1088         successors of the joiner's clone reach the same block, verify the
1089         PHI arguments are equal.  If not, cancel the jump threading request.
1090         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
1091         tree-ssa-threadupdate.c
1092         (thread_across_edge): Don't check PHI argument equality when
1093         threading through joiner block here.
1094         
1095 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
1097         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
1098         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
1099         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
1100         make_pass_early_warn_uninitialized): Move to tree-ssa-uninit.c.
1101         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c
1102         (has_undefined_value_p): New.  Pass dependant parts of 
1103         ssa_undefined_value_p.
1104         (uninit_undefined_value_p): Use has_undefined_value_p.
1105         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
1106         make_pass_early_warn_uninitialized): Move from tree-ssa.c
1107         * tree-ssa.h: Adjust prototypes
1109 2013-09-17  Jan Hubicka  <jh@suse.cz>
1111         PR middle-end/58332
1112         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
1113         * ipa-inline.c (can_inline_edge_p): Do not downgrade
1114         FUNCTION_NOT_OPTIMIZED.
1115         * ipa-inline-analysis.c (compute_inline_parameters): Function
1116         not optimized is not inlinable unless it is alwaysinline.
1117         (inline_analyze_function): Force calls in not optimized
1118         function not inlinable.
1120 2013-09-17  Jan Hubicka  <jh@suse.cz>
1122         PR middle-end/58329
1123         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
1124         to return NULL.
1125         * ipa.c (function_and_variable_visibility): Likewise.
1126         * ipa-profile.c (ipa_profile): Likewise.
1128 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1130         PR ipa/58398
1131         * cgraph.c (cgraph_function_body_availability): Check for ifunc
1132         attribute, and don't inline the resolver in this case.
1134 2013-09-17  Teresa Johnson  <tejohnson@google.com>
1136         * coverage.c (get_coverage_counts): Add missing newline.
1138 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1140         PR tree-optimization/58088
1141         * fold-const.c (mask_with_trailing_zeros): New function.
1142         (fold_binary_loc): Make sure we don't recurse infinitely
1143         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
1144         Use mask_with_trailing_zeros where appropriate.
1146 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
1148         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
1149         of 'prev' var to get better distance estimation.
1151 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
1153         * tree-inline.h (struct copy_body_data): Add transform_parameter.
1154         * tree-inline.c (is_parameter_of): New predicate.
1155         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
1156         a parameter has been remapped.
1157         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
1158         (optimize_inline_calls): Initialize transform_parameter.
1159         (copy_gimple_seq_and_replace_locals): Likewise.
1160         (tree_function_versioning): Likewise.
1161         (maybe_inline_call_in_expr): Likewise.
1163 2013-09-17  Nick Clifton  <nickc@redhat.com>
1165         * config/msp430/msp430-protos.h: Add prototypes for new functions.
1166         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
1167         interrupt handlers.
1168         (is_attr_func): New function.
1169         (msp430_is_interrupt_func): New function.
1170         (is_naked_func): New function.
1171         (is_reentrant_func): New function.
1172         (is_critical_func): New function.
1173         (msp430_start_function): Add annotations for function attributes.
1174         (msp430_attr): New function.
1175         (msp430_attribute_table): New.
1176         (msp430_function_section): New function.
1177         (TARGET_ASM_FUNCTION_SECTION): Define.
1178         (msp430_builtin): New enum.
1179         (msp430_init_builtins): New function.
1180         (msp430_builtin_devl): New function.
1181         (msp430_expand_builtin): New function.
1182         (TARGET_INIT_BUILTINS): Define.
1183         (TARGET_EXPAND_BUILTINS): Define.
1184         (TARGET_BUILTIN_DECL): Define.
1185         (msp430_expand_prologue): Add support for naked, interrupt,
1186         critical and reentrant functions.
1187         (msp430_expand_epilogue): Likewise.
1188         (msp430_print_operand): Handle 'O' character.
1189         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
1190         NO_TRAMPOLINES.
1191         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
1192         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
1193         (pushm): Use a 'n' rather than an 'i' constraint.
1194         (msp_return): Add generation of the interrupt return instruction.
1195         (disable_interrupts): New pattern.
1196         (enable_interrupts): New pattern.
1197         (push_intr_state): New pattern.
1198         (pop_intr_state): New pattern.
1199         (bic_SR): New pattern.
1200         (bis_SR): New pattern.
1201         * doc/extend.texi: Document MSP430 function attributes and builtin
1202         functions.
1204 2013-09-17  Richard Biener  <rguenther@suse.de>
1206         PR tree-optimization/58432
1207         * tree-loop-distribution.c (tree_loop_distribution): Also
1208         scan PHIs for outside loop uses and seed a partition from them.
1210 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
1212         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
1213         (restructure_reference): Call backtrace_base_for_ref.
1215 2013-09-17  Alan Modra  <amodra@gmail.com>
1217         PR target/57589
1218         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11 patch.
1220 2013-09-16  DJ Delorie  <dj@redhat.com>
1222         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
1223         vregs location for RL78/G10.
1224         (rl78_expand_prologue): Avoid SEL on G10.
1225         (rl78_expand_epilogue): Likewise.
1226         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
1227         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
1228         __RL78_G10__ when appropriate.
1229         (ASM_SPEC): Pass -mg10 along to the assembler.
1230         * config/rl78/rl78.md (sel_rb): Disable for G10.
1231         * config/rl78/rl78.opt: Add -mg10 option.
1232         * config/rl78/t-rl78: Add -mg10 multilib.
1234 2013-09-16  Xinliang David Li  <davidxl@google.com>
1236         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
1237         * omp-low.c (omp_max_vf): Ditto.
1238         (expand_omp_simd): Ditto.
1239         * tree-vectorizer.c (vectorize_loops): Ditto.
1240         (gate_vect_slp): Ditto.
1241         (gate_increase_alignment): Ditto.
1242         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
1243         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
1244         (gate_tree_vectorize): Name change.
1245         (tree_vectorize): Ditto.
1246         (pass_vectorize::gate): Call new function.
1247         (pass_vectorize::execute): Ditto.
1248         opts.c: O3 default setting change.
1249         (finish_options): Check new flag.
1250         * doc/invoke.texi: Document new flags.
1251         * common.opt: New flags.
1253 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
1255         * doc/tm.texi.in (Cond Exec Macros): Remove node.
1256         (Condition Code): Don't reference it.
1257         * doc/tm.texi: Regenerate.
1259 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
1261         PR middle-end/58418
1262         * lra-constraints.c (undo_optional_reloads): Consider all optional
1263         reload even if it did not get a hard reg.
1265 2013-09-16  Teresa Johnson  <tejohnson@google.com>
1267         * dumpfile.c (dump_loc): Remove newline emission.
1268         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
1269         emission to dump_printf_loc calls where missing.
1270         (vect_mark_for_runtime_alias_test): Ditto.
1271         (vect_analyze_data_ref_dependence): Ditto.
1272         (vect_analyze_data_ref_dependences): Ditto.
1273         (vect_slp_analyze_data_ref_dependence): Ditto.
1274         (vect_slp_analyze_data_ref_dependences): Ditto.
1275         (vect_compute_data_ref_alignment): Ditto.
1276         (vect_update_misalignment_for_peel): Ditto.
1277         (vect_verify_datarefs_alignment): Ditto.
1278         (vector_alignment_reachable_p): Ditto.
1279         (vect_get_data_access_cost): Ditto.
1280         (vect_enhance_data_refs_alignment): Ditto.
1281         (vect_find_same_alignment_drs): Ditto.
1282         (vect_analyze_data_refs_alignment): Ditto.
1283         (vect_analyze_group_access): Ditto.
1284         (vect_analyze_data_ref_access): Ditto.
1285         (vect_analyze_data_ref_accesses): Ditto.
1286         (vect_prune_runtime_alias_test_list): Ditto.
1287         (vect_analyze_data_refs): Ditto.
1288         (vect_create_addr_base_for_vector_ref): Ditto.
1289         (vect_create_data_ref_ptr): Ditto.
1290         (vect_grouped_store_supported): Ditto.
1291         (vect_grouped_load_supported): Ditto.
1292         * value-prof.c (check_counter): Ditto.
1293         (check_ic_target): Ditto.
1294         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
1295         (vect_recog_widen_mult_pattern): Ditto.
1296         (vect_recog_widen_sum_pattern): Ditto.
1297         (vect_recog_over_widening_pattern): Ditto.
1298         (vect_recog_widen_shift_pattern): Ditto.
1299         (vect_recog_rotate_pattern): Ditto.
1300         (vect_recog_vector_vector_shift_pattern): Ditto.
1301         (vect_recog_divmod_pattern): Ditto.
1302         (vect_recog_mixed_size_cond_pattern): Ditto.
1303         (vect_recog_bool_pattern): Ditto.
1304         (vect_pattern_recog_1): Ditto.
1305         (vect_pattern_recog): Ditto.
1306         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
1307         (vect_is_simple_iv_evolution): Ditto.
1308         (vect_analyze_scalar_cycles_1): Ditto.
1309         (vect_get_loop_niters): Ditto.
1310         (vect_analyze_loop_1): Ditto.
1311         (vect_analyze_loop_form): Ditto.
1312         (vect_analyze_loop_operations): Ditto.
1313         (vect_analyze_loop_2): Ditto.
1314         (vect_analyze_loop): Ditto.
1315         (report_vect_op): Ditto.
1316         (vect_is_slp_reduction): Ditto.
1317         (vect_is_simple_reduction_1): Ditto.
1318         (vect_get_known_peeling_cost): Ditto.
1319         (vect_estimate_min_profitable_iters): Ditto.
1320         (vect_model_reduction_cost): Ditto.
1321         (vect_model_induction_cost): Ditto.
1322         (get_initial_def_for_induction): Ditto.
1323         (vect_create_epilog_for_reduction): Ditto.
1324         (vectorizable_reduction): Ditto.
1325         (vectorizable_induction): Ditto.
1326         (vectorizable_live_operation): Ditto.
1327         (vect_loop_kill_debug_uses): Ditto.
1328         (vect_transform_loop): Ditto.
1329         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
1330         (vect_stmt_relevant_p): Ditto.
1331         (process_use): Ditto.
1332         (vect_mark_stmts_to_be_vectorized): Ditto.
1333         (vect_model_simple_cost): Ditto.
1334         (vect_model_promotion_demotion_cost): Ditto.
1335         (vect_model_store_cost): Ditto.
1336         (vect_get_store_cost): Ditto.
1337         (vect_model_load_cost): Ditto.
1338         (vect_get_load_cost): Ditto.
1339         (vect_init_vector_1): Ditto.
1340         (vect_get_vec_def_for_operand): Ditto.
1341         (vect_finish_stmt_generation): Ditto.
1342         (vectorizable_call): Ditto.
1343         (vectorizable_conversion): Ditto.
1344         (vectorizable_assignment): Ditto.
1345         (vectorizable_shift): Ditto.
1346         (vectorizable_operation): Ditto.
1347         (vectorizable_store): Ditto.
1348         (vectorizable_load): Ditto.
1349         (vectorizable_condition): Ditto.
1350         (vect_analyze_stmt): Ditto.
1351         (vect_transform_stmt): Ditto.
1352         (vect_is_simple_use): Ditto.
1353         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
1354         (vect_can_advance_ivs_p): Ditto.
1355         (vect_update_ivs_after_vectorizer): Ditto.
1356         (vect_do_peeling_for_loop_bound): Ditto.
1357         (vect_gen_niters_for_prolog_loop): Ditto.
1358         (vect_update_inits_of_drs): Ditto.
1359         (vect_create_cond_for_alias_checks): Ditto.
1360         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
1361         (vect_build_slp_tree_1): Ditto.
1362         (vect_supported_load_permutation_p): Ditto.
1363         (vect_analyze_slp_instance): Ditto.
1364         (vect_analyze_slp): Ditto.
1365         (vect_make_slp_decision): Ditto.
1366         (vect_detect_hybrid_slp): Ditto.
1367         (vect_bb_vectorization_profitable_p): Ditto.
1368         (vect_slp_analyze_bb_1): Ditto.
1369         (vect_update_slp_costs_according_to_vf): Ditto.
1370         (vect_get_mask_element): Ditto.
1371         (vect_transform_slp_perm_load): Ditto.
1372         (vect_schedule_slp_instance): Ditto.
1373         (vect_schedule_slp): Ditto.
1374         (vect_slp_transform_bb): Ditto.
1375         * profile.c (read_profile_edge_counts): Ditto.
1376         (compute_branch_probabilities): Ditto.
1377         * coverage.c (get_coverage_counts): Ditto.
1379 2013-09-16  Diego Novillo  <dnovillo@google.com>
1381         * tree-core.h: Add missing comment lines from refactoring
1382         of tree.h.
1384 2013-09-16  Jan Hubicka  <jh@suse.cz>
1386         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
1387         abstract functions; for static functions check the presence
1388         of body.
1390 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
1392         * config/aarch64/aarch64-simd-builtins.def (fma): New.
1393         * config/aarch64/aarch64-simd.md
1394         (aarch64_mla_elt<mode>): New.
1395         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1396         (aarch64_mls_elt<mode>): Likewise.
1397         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1398         (aarch64_fma4_elt<mode>): Likewise.
1399         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1400         (aarch64_fma4_elt_to_128v2df): Likewise.
1401         (aarch64_fma4_elt_to_64df): Likewise.
1402         (fnma<mode>4): Likewise.
1403         (aarch64_fnma4_elt<mode>): Likewise.
1404         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1405         (aarch64_fnma4_elt_to_128v2df): Likewise.
1406         (aarch64_fnma4_elt_to_64df): Likewise.
1407         * config/aarch64/iterators.md (VDQSF): New.
1408         * config/aarch64/arm_neon.h
1409         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
1410         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
1412 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
1414         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
1415         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1416         (aarch64_mul3_elt_to_128df): Likewise.
1417         (aarch64_mul3_elt_to_64v2df): Likewise.
1418         * config/aarch64/iterators.md (VEL): Also handle DFmode.
1419         (VMUL): New.
1420         (VMUL_CHANGE_NLANES) Likewise.
1421         (h_con): Likewise.
1422         (f): Likewise.
1423         * config/aarch64/arm_neon.h
1424         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
1426 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
1428         * config/aarch64/arm_neon.h
1429         (vcvtx_high_f32_f64): Fix parameters.
1431 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1432             Uros Bizjak  <ubizjak@gmail.com>
1434         * config/alpha.c: Include tree-ssanames.h.
1436 2013-09-16  Richard Biener  <rguenther@suse.de>
1438         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
1439         (dot_rdg_1): Handle control_dd.
1440         (create_edge_for_control_dependence): New function.
1441         (create_rdg_edges): Add control dependences if asked for.
1442         (build_rdg): Likewise.
1443         (generate_loops_for_partition): If there are not necessary
1444         control stmts remove all their dependencies.
1445         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
1446         (distribute_loop): Pass on control dependences.
1447         (tree_loop_distribution): Compute control dependences and remove
1448         restriction on number of loop nodes.
1450 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
1452         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
1453         for internal calls.
1455 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
1457         * cse.c (try_const_anchors): Punt on CC modes.
1459 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1461         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
1463 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
1465         PR target/58382
1466         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
1467         calls to word_mode.
1469 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
1471         PR target/48094
1472         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
1473         seen.
1474         (darwin_objc1_section): Likewise.
1475         (darwin_file_end): Emit Image Info section when required.
1477 2013-09-14  Jan Hubicka  <jh@suse.cz>
1479         * tree-into-ssa.c (gate_into_ssa): New.
1480         (pass_data_build_ssa): Use it.
1481         * cgraph.h (expand_thunk): Update prototype.
1482         * cgraphunit.c (analyze_function): Expand thunks early.
1483         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
1484         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
1485         set lowered flag; do not add new function.
1486         (assemble_thunks_and_aliases): Update.
1487         * tree-ssa.c (gate_init_datastructures): New gate.
1488         (pass_data_init_datastructures): Use it.
1490 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
1492         PR target/58269
1493         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
1494         xmm register set described in the psABI.
1496 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
1498         * dwarf2out.c (should_emit_struct_debug): Add check
1499         for type_decl variable is not NULL.
1501 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
1503         * config.gcc: Use new winnt-c.c target hooks
1504         * config/t-winnt: New file
1505         * config/winnt-c.c: New file
1506         * doc/tm.texi.in: Document new hook
1507         * doc/tm.texi: Regenerated
1509 2013-09-13  Jan Hubicka  <jh@suse.cz>
1511         PR middle-end/58094
1512         * ipa-inline.c (check_callers): New function.
1513         (check_caller_edge): Remove.
1514         (want_inline_function_to_all_callers_p): Also permit alises that are
1515         called dirrectly.
1516         (inline_to_all_callers): Terminate the walk when devirtualization
1517         introduce new calls.
1519 2013-09-13  Jan Hubicka  <jh@suse.cz>
1521         * ipa-inline-analysis.c (struct growth_data): Add node.
1522         (do_estimate_growth_1): Fix detection of recursion.
1524 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
1526         PR tree-optimization/58392
1527         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
1528         to avoid shadowing of outer loop variable.  If
1529         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
1530         replace_by_duplicate_decl simduid of loops that have it set and
1531         set dest_cfun->has_simduid_loops and/or
1532         dest_cfun->has_force_vect_loops.
1533         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
1534         instead of maybe_lookup_decl.
1535         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
1536         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
1537         call.  Copy over force_vect and copy and remap simduid.  Set
1538         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
1539         (copy_cfg_body): Remove blocks_to_copy argument.  Use
1540         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
1541         caller.  Don't set cfun->has_simduid_loops and/or
1542         cfun->has_force_vect_loops here.
1543         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
1544         caller.
1545         (expand_call_inline, tree_function_versioning): Adjust copy_body
1546         callers.
1548 2013-09-13  Martin Jambor  <mjambor@suse.cz>
1550         PR bootstrap/58388
1551         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
1552         the assert if the edge was a speculative one.
1554 2013-09-13  Richard Biener  <rguenther@suse.de>
1556         * tree-data-ref.h (known_dependences_p): Move here ...
1557         * tree-loop-distribution.c (known_dependences_p): ... from here.
1558         (dump_rdg_component, debug_rdg_component): Remove.
1559         (dump_rdg): Adjust.
1560         (generate_loops_for_partition): Use gimple_uid instead of
1561         relying on matching stmt visit order.
1562         (rdg_build_partitions): Take starting stmt vector.
1563         (ldist_gen): Merge into ...
1564         (distribute_loop): ... this function.  Do not compute starting
1565         vertices vector.
1566         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
1568 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1570         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
1571         Set type attribute correctly. Set predicable_short_it attribute.
1572         (cmpsi_shiftsi): Remove %? from output template.
1574 2013-09-13  Richard Biener  <rguenther@suse.de>
1576         * tree-loop-distribution.c (struct rdg_component,
1577         rdg_defs_used_in_other_loops_p, free_rdg_components,
1578         rdg_build_components): Remove.
1579         (stmts_from_loop): Do not record virtual PHIs.
1580         (generate_loops_for_partition): Skip virtual PHIs.
1581         (build_rdg_partition_for_component): Rename to ...
1582         (build_rdg_partition_for_vertex): ... this and adjust.
1583         (rdg_build_partitions): Take a vector of starting vertices
1584         instead of components.  Remove unnecessary leftover handling.
1585         (ldist_gen): Do not build components or record other stores.
1586         (distribute_loop): Do not distribute loops containing stmts
1587         with side-effects.
1589 2013-09-13  Christian Bruel  <christian.bruel@st.com>
1591         PR target/58314
1592         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
1594 2013-09-13  Kai Tietz  <ktietz@redhat.com>
1596         * config.gcc: Separate cases for mingw and cygwin targets,
1597         and add 64-bit cygwin target case.
1599         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
1600         dll-export inline-functions.
1601         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
1604 2013-09-13  Jeff Law  <law@redhat.com>
1606         PR middle-end/58387
1607         Revert:
1608         2013-09-06  Jeff Law  <law@redhat.com>
1610         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1611         edge implied equivalences into successor phis.
1613 2013-09-12  DJ Delorie  <dj@redhat.com>
1615         * config/rl78/rl78-virt.md: Change from | to \; for asm line
1616         separators.
1618 2013-09-12  Brooks Moses  <bmoses@google.com>
1620         PR driver/42955
1621         * Makefile.in: Do not install driver binaries in $(target)/bin.
1623 2013-09-12  DJ Delorie  <dj@redhat.com>
1625         * config/rl78/rl78.opt (mrelax): New.
1626         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
1627         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
1629         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
1630         between SP and FP.
1631         (rl78_expand_epilogue): Likewise.
1633 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
1635         PR middle-end/58335
1636         * lra-eliminations.c (remove_reg_equal_offset_note): New.
1637         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
1638         pointer elimination with using remove_reg_equal_offset_note.
1640 2013-09-12  DJ Delorie  <dj@redhat.com>
1642         * config/msp430/: New port.
1643         * config.gcc (msp430): Added.
1644         * doc/invoke.texi: Document MSP430 options.
1645         * doc/install.texi: Document msp430-elf
1646         * doc/md.texi: Document msp430-elf
1647         * doc/contrib.texi: Document msp430-elf
1649         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
1650         PSImode.
1652 2013-09-12  Martin Jambor  <mjambor@suse.cz>
1654         PR ipa/58389
1655         * ipa-prop.c (remove_described_reference): Give up if the edge in the
1656         reference descriptor is NULL.
1657         (ipa_edge_removal_hook): If owning a reference descriptor, set its
1658         edge to NULL.
1660 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
1662         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
1663         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
1664         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
1665         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
1666         make_temp_ssa_name): move to tree-ssanames.h
1667         * tree-ssa-alias.h: Move prototype.
1668         * tree-ssa.h: Include tree-ssanames.h.
1669         * tree-ssanames.c (FREE_SSANAMES): Move to here.
1670         * tree-ssanames.h: New.  Move items from tree-flow*.h
1671         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
1673 2013-09-12  Richard Biener  <rguenther@suse.de>
1675         PR tree-optimization/58404
1676         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
1677         propagate non-invariant addresses into dereferences wrapped
1678         in component references.
1680 2013-09-12  Richard Biener  <rguenther@suse.de>
1682         PR tree-optimization/58402
1683         * passes.def: Move pass_late_warn_uninitialized later.
1685 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
1687         * tree-ssa.h: New.  Move content from tree-flow.h and
1688         tree-flow-inline.h.
1689         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
1690         Move prototypes belonging to tree-ssa.c.
1691         * tree-flow-inline.h (redirect_edge_var_map_def,
1692         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
1693         tree-ssa.h.
1694         * gimple.h: Adjust prototypes.
1695         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
1696         to...
1697         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
1698         * tree.h: Move prototype to tree-ssa.h.
1699         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
1700         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
1701         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
1702         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
1703         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
1704         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
1705         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
1706         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
1707         graphite-blocking.c, graphite-clast-to-gimple.c,
1708         graphite-dependences.c, graphite-interchange.c,
1709         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
1710         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
1711         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
1712         ipa-reference.c, ipa-split.c, ipa-utils.c,
1713         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
1714         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
1715         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
1716         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1717         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
1718         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
1719         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
1720         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
1721         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
1722         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
1723         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
1724         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
1725         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
1726         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
1727         value-prof.c, var-tracking.c,
1728         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
1730 2013-09-12  Richard Biener  <rguenther@suse.de>
1732         PR tree-optimization/58396
1733         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
1734         (build_rdg): Take a loop-nest parameter, fix memleaks.
1735         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
1737 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1739         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
1740         for SLM.
1742 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
1744         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
1745         The FMA instruction names should have a 'v' prefix.
1747 2013-09-12  Richard Biener  <rguenther@suse.de>
1749         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
1750         (dot_rdg): Use popen instead of system in optional code.
1751         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
1752         (already_processed_vertex_p): Adjust.
1753         (has_anti_or_output_dependence, predecessor_has_mem_write,
1754         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
1755         rdg_flag_uses): Remove.
1756         (rdg_flag_vertex): Simplify.
1757         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
1758         remove recursion.
1759         (build_rdg_partition_for_component): Process the first vertex
1760         of a component only.
1761         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
1763 2013-09-12  Alan Modra  <amodra@gmail.com>
1765         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
1767 2013-09-11  DJ Delorie  <dj@redhat.com>
1768             Nick Clifton  <nickc@redhat.com>
1770         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
1771         (rl78_stack_based_mem): New.
1772         * config/rl78/constraints.md (Iv08): New.
1773         (Iv16): New.
1774         (Iv24): New.
1775         (Is09): New.
1776         (Is17): New.
1777         (Is25): New.
1778         (ISsi): New.
1779         (IShi): New.
1780         (ISqi): New.
1781         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
1782         (movhi): Likewise.
1783         (movsi): Change from expand to insn-and-split.
1784         (ashrsi3): Clobber AX.
1785         (lshrsi3): New.
1786         (ashlsi3): New.
1787         (cbranchsi4): New.
1788         * config/rl78/rl78.md (CC_REG): Fix.
1789         (addsi3): Allow memory and immediate operands.
1790         (addsi3_internal): Split into...
1791         (addsi3_internal_virt): ...new, and ...
1792         (addsi3_internal_real): ...new.
1793         (subsi): New.
1794         (subsi3_internal_virt): New.
1795         (subsi3_internal_real): New.
1796         (mulsi3): Add memory operand.
1797         (mulsi3_rl78): Likewise.
1798         (mulsi3_g13): Likewise.
1799         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
1800         (cbranchqi4_real): Add more constraint options.
1801         (cbranchhi4_real): Expand pattern.
1802         (cbranchhi4_real_signed): New.
1803         (cbranchhi4_real_inverted): New.
1804         (cbranchsi4_real_lt): New.
1805         (cbranchsi4_real_ge): New.
1806         (cbranchsi4_real_signed): New.
1807         (cbranchsi4_real): New.
1808         (peephole2): New.
1809         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
1810         constant shifts.
1811         (lshrsi3_virt): Likewise.
1812         (ashlsi3_virt): Likewise.
1813         (cbranchqi4_virt_signed): New.
1814         (cbranchhi4_virt_signed): New.
1815         (cbranchsi4_virt): New.
1816         * config/rl78/rl78.c: Whitespace fixes throughout.
1817         (move_elim_pass): New.
1818         (pass_data_rl78_move_elim): New.
1819         (pass_rl78_move_elim): New.
1820         (make_pass_rl78_move_elim): New.
1821         (rl78_devirt_info): Run devirt earlier.
1822         (rl78_move_elim_info): New.
1823         (rl78_asm_file_start): Register it.
1824         (rl78_split_movsi): New.
1825         (rl78_as_legitimate_address): Allow virtual base registers when
1826         appropriate.
1827         (rl78_addr_space_convert): Remove spurious debug stuff.
1828         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
1829         (rl78_print_operand): More cases for not printing '#'.
1830         (rl78_expand_compare): Remove most of the logic.
1831         (content_memory): New.
1832         (clear_content_memory): New.
1833         (get_content_index): New.
1834         (get_content_name): New.
1835         (display_content_memory): New.
1836         (update_content): New.
1837         (record_content): New.
1838         (already_contains): New.
1839         (insn_ok_now): Re-recog insns with virtual registers.
1840         (add_postponed_content_update): New.
1841         (process_postponed_content_update): New.
1842         (gen_and_emit_move): New.
1843         (transcode_memory_rtx): Record new location content.
1844         Use gen_and_emit_move.
1845         (force_into_acc): New.
1846         (move_to_acc): Use gen_and_emit_move.
1847         (move_from_acc): Likewise.
1848         (move_acc_to_reg): Likewise.
1849         (move_to_x): Likewise.
1850         (move_to_hl): Likewise.
1851         (move_to_de): Likewise.
1852         (rl78_alloc_physical_registers_op1): Record location content.
1853         (has_constraint): New.
1854         (rl78_alloc_physical_registers_op2): Record location content.
1855         Optimize use of HL.
1856         (rl78_alloc_physical_registers_ro1): Likewise.
1857         (rl78_alloc_physical_registers_cmp): Likewise.
1858         (rl78_alloc_physical_registers_umul): Likewise.
1859         (rl78_alloc_address_registers_macax): New.
1860         (rl78_alloc_physical_registers): Initialize and set location
1861         content memory as needed.
1862         (rl78_reorg): Make sure split2 is called.
1863         (rl78_rtx_costs): New.
1865 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
1867         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
1868         for (not (neg ...)) and (neg (not ...)) cases.
1870 2013-09-11  Richard Biener  <rguenther@suse.de>
1872         PR middle-end/58377
1873         * passes.def: Split critical edges before late uninit warning passes.
1874         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
1876 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
1878         PR tree-optimization/58385
1879         * fold-const.c (build_range_check): If both low and high are NULL,
1880         use omit_one_operand_loc to preserve exp side-effects.
1882 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1884         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
1886 2013-09-11  Richard Biener  <rguenther@suse.de>
1888         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
1889         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
1890         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
1891         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
1892         stmts_from_loop, known_dependences_p, build_empty_rdg,
1893         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
1894         * tree-loop-distribution.c: ... here.
1895         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
1896         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
1897         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
1898         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
1899         * tree-loop-distribution.c: ... here.
1900         * tree-loop-distribution.c: Include gimple-pretty-print.h.
1901         (struct partition_s): Add loops member.
1902         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
1903         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
1904         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
1906 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1907             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1908             Sergey Lega  <sergey.s.lega@intel.com>
1909             Anna Tikhonova  <anna.tikhonova@intel.com>
1910             Ilya Tocar  <ilya.tocar@intel.com>
1911             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1912             Ilya Verbin  <ilya.verbin@intel.com>
1913             Kirill Yukhin  <kirill.yukhin@intel.com>
1914             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1916         * config/i386/constraints.md (k): New.
1917         (Yk): Ditto.
1918         * config/i386/i386.c (const regclass_map): Add new mask registers.
1919         (dbx_register_map): Ditto.
1920         (dbx64_register_map): Ditto.
1921         (svr4_dbx_register_map): Ditto.
1922         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
1923         disabled.
1924         (ix86_preferred_reload_class): Disable constants for mask registers.
1925         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
1926         (ix86_hard_regno_mode_ok): Support new mask registers.
1927         (x86_order_regs_for_local_alloc): Ditto.
1928         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
1929         (FIXED_REGISTERS): Add new mask registers.
1930         (CALL_USED_REGISTERS): Ditto.
1931         (REG_ALLOC_ORDER): Ditto.
1932         (VALID_MASK_REG_MODE): New.
1933         (FIRST_MASK_REG): Ditto.
1934         (LAST_MASK_REG): Ditto.
1935         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
1936         (MAYBE_MASK_CLASS_P): New.
1937         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
1938         (REG_CLASS_CONTENTS): Ditto.
1939         (MASK_REGNO_P): New.
1940         (ANY_MASK_REG_P): Ditto.
1941         (HI_REGISTER_NAMES): Add new mask registers.
1942         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
1943         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
1944         mask registers.
1945         (attribute "type"): Add mskmov, msklog.
1946         (attribute "length_immediate"): Support them.
1947         (attribute "memory"): Ditto.
1948         (attribute "prefix_0f"): Ditto.
1949         (*movhi_internal): Support new mask registers.
1950         (*movqi_internal): Ditto.
1951         (define_split): Split out clobber pattern is a logic
1952         insn on mask registers.
1953         (*k<logic><mode>): New.
1954         (*andhi_1): Extend to support mask regs.
1955         (*andqi_1): Extend to support mask regs.
1956         (kandn<mode>): New.
1957         (define_split): Split and-not to and and not if operands
1958         are not mask regs.
1959         (*<code><mode>_1): Separate HI mode to new pattern...
1960         (*<code>hi_1): This.
1961         (*<code>qi_1): Extend to support mask regs.
1962         (kxnor<mode>): New.
1963         (kortestzhi): Ditto.
1964         (kortestchi): Ditto.
1965         (kunpckhi): Ditto.
1966         (*one_cmpl<mode>2_1): Remove HImode and handle it...
1967         (*one_cmplhi2_1): ...Here, now with mask registers support.
1968         (*one_cmplqi2_1): Support new mask registers.
1969         (HI/QImode arithmetics splitter): Don't split if mask registers
1970         are used.
1971         (HI/QImode not splitter): Ditto.
1972         * config/i386/predicated.md (mask_reg_operand): New.
1973         (general_reg_operand): Ditto.
1975 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1977         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
1978         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
1980 2013-09-10  Jeff Law  <law@redhat.com>
1982         PR tree-optimization/58380
1983         * tree-ssa-threadupdate.c (thread_block): Recognize another case
1984         of threading through a buried loop header.
1986         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
1987         return value for single successor case.
1989 2013-09-10  Jan Hubicka  <jh@suse.cz>
1991         * ipa-devirt.c (ipa_devirt): Enable with LTO.
1993 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
1995         PR target/58361
1996         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
1997         support conditional execution.
1998         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
2000 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
2002         * lra.c (lra): Clear lra_optional_reload_pseudos before every
2003         constraint pass.
2004         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
2005         Check destination too to check move insn.
2006         (undo_optional_reloads): Add check that the original peudo did not
2007         changed its allocation and the optional reload was inherited on last
2008         inheritance pass.  Break loop after deciding to keep optional reload.
2009         (lra_undo_inheritance): Add check that inherited pseudo still in
2010         memory.
2012 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
2014         * config/aarch64/aarch64.md (generic_sched): New.
2015         * config/aarch64/aarch64-generic.md (load): Make conditional
2016         on generic_sched attribute.
2017         (nonload): Likewise.
2019 2013-09-10  Jan Hubicka  <jh@suse.cz>
2021         * lto-cgraph.c: Include ipa-utils.h.
2022         (compute_ltrans_boundary): Also add possible targets into the boundary.
2024 2013-09-10  Jan Hubicka  <jh@suse.cz>
2026         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
2027         VAR_DECL of vtable rather than full expression.
2029 2013-09-10  Jan Hubicka  <jh@suse.cz>
2030             Paolo Carlini  <paolo.carlini@oracle.com>
2032         * cgraphunit.c (analyze_functions): Save input_location, set it
2033         to UNKNOWN_LOCATION and restore it at the end.
2035 2013-09-10  Martin Jambor  <mjambor@suse.cz>
2037         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
2038         represented by a thunk.
2040 2013-09-10  Jeff Law  <law@redhat.com>
2042         PR tree-optimization/58343
2043         * tree-ssa-threadupdate.c (thread_block): Identify and disable
2044         jump threading requests through loop headers buried in the middle
2045         of a jump threading path.
2047         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
2048         in return value/type.
2050 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
2052         PR rtl-optimization/58365
2053         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
2054         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
2055         it differs.
2057 2013-09-10  Richard Biener  <rguenther@suse.de>
2059         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
2060         * tree-data-ref.c (create_rdg_vertices): Collect all data
2061         references, signal failure to the caller, use data-ref API.
2062         (build_rdg): Compute data references only once.  Maintain lifetime
2063         of data references and data dependences from within RDG.
2064         (free_rdg): Free dependence relations.
2065         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
2066         inventing extra dependences.
2067         (distribute_loop): Update for RDG API changes.
2069 2013-09-10  Kai Tietz  <ktietz@redhat.com>
2071         * doc/invoke.texi (fms-extensions): Document changed
2072         behavior for ms-abi targets.
2073         * config/i386/i386.c (ix86_option_override_internal):
2074         Set default value of option -fms-extension for ms-abi targets.
2076 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
2078         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
2080 2013-09-10  Alan Modra  <amodra@gmail.com>
2082         PR target/58330
2083         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
2085 2013-09-10  Alan Modra  <amodra@gmail.com>
2087         * config/rs6000/predicates.md (add_cint_operand): New.
2088         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
2089         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
2090         using add_cint_operand.
2091         (largetoc_high_plus_aix): Likewise.
2093 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
2095         PR tree-optimization/58364
2096         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
2097         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
2098         the current range can't be an unconditional true or false.
2100 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
2102         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
2104 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
2106         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
2108 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
2110         PR c++/43452
2111         * doc/invoke.texi (-Wdelete-incomplete): Document it.
2113 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
2115         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
2116         NO_REGS for immediate that can't be moved directly into FP_REGS.
2118 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2120         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
2121         comparison with negated operand.
2122         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
2123         RTL form.
2125 2013-09-09  Richard Biener  <rguenther@suse.de>
2127         PR middle-end/58326
2128         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
2129         of a subloop record all its block as affecting loop-closed SSA form.
2131 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2133         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
2134         of an rtx/bitpos pair.
2135         (store_fixed_bit_field): Update accordingly.
2137 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2139         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
2140         GEN_INT.
2141         * builtins.c (expand_errno_check): Likewise.
2142         * dwarf2cfi.c (init_return_column_size): Likewise.
2143         * except.c (sjlj_mark_call_sites): Likewise.
2144         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
2145         * lra-constraints.c (emit_inc): Likewise.
2146         * ree.c (combine_set_extension): Likewise.
2147         * regmove.c (fixup_match_2): Likewise.
2148         * reload1.c (inc_for_reload): Likewise.
2150 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2152         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
2153         (force_to_mode, simplify_shift_const_1, simplify_comparison):
2154         Use gen_int_mode with the mode of the associated simplify_* call.
2155         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
2156         * expmed.c (expand_shift_1): Likewise.
2157         * function.c (instantiate_virtual_regs_in_insn): Likewise.
2158         * loop-iv.c (iv_number_of_iterations): Likewise.
2159         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
2160         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
2162 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2164         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
2165         of the associated expand_* call.
2166         (asan_emit_stack_protection): Likewise.
2167         * builtins.c (round_trampoline_addr): Likewise.
2168         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
2169         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
2170         (emit_store_flag): Likewise.
2171         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
2172         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
2173         Likewise.
2174         * function.c (instantiate_virtual_regs_in_insn): Likewise.
2175         * ifcvt.c (noce_try_store_flag_constants): Likewise.
2176         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
2177         * modulo-sched.c (generate_prolog_epilog): Likewise.
2178         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
2179         (expand_ctz, expand_ffs, expand_unop): Likewise.
2181 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2183         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
2184         of the associated gen_rtx_* call.
2185         * caller-save.c (init_caller_save): Likewise.
2186         * combine.c (find_split_point, make_extraction): Likewise.
2187         (make_compound_operation): Likewise.
2188         * dwarf2out.c (mem_loc_descriptor): Likewise.
2189         * explow.c (plus_constant, probe_stack_range): Likewise.
2190         * expmed.c (expand_mult_const): Likewise.
2191         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
2192         * reload1.c (init_reload): Likewise.
2193         * valtrack.c (cleanup_auto_inc_dec): Likewise.
2194         * var-tracking.c (adjust_mems): Likewise.
2195         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
2196         rather than gen_rtx_fmt_ee.
2198 2013-09-09  Jan Hubicka  <jh@suse.cz>
2200         PR middle-end/58294
2201         * value-prof.c (gimple_ic): Copy also abnormal edges.
2203 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2205         * asan.c (asan_shadow_cst): Use gen_int_mode.
2207 2013-09-08  Jan Hubicka  <jh@suse.cz>
2209         * ipa-profile.c: Add toplevel comment.
2210         (ipa_propagate_frequency_1): Be more conservative when profile is read.
2211         (contains_hot_call_p): New function.
2212         (ipa_propagate_frequency): Set frequencies based on counts when
2213         profile is read.
2214         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
2215         profile; do not tamper with profile after inlining if it is read.
2217 2013-09-08  Jan Hubicka  <jh@suse.cz>
2219         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
2220         speculative edges.
2222 2013-09-08  Jan Hubicka  <jh@suse.cz>
2224         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
2225         summary generation.
2227 2013-09-08  Jeff Law  <law@redhat.com>
2229         PR bootstrap/58340
2230         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
2231         of 'found'.
2233 2013-09-08  Andi Kleen  <ak@linux.intel.com>
2235         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
2237 2013-09-08  Jan Hubicka  <jh@suse.cz>
2239         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
2241 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
2243         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
2244         for non-debug insns.
2245         * lra.c (new_insn_reg): Take the containing insn as a parameter.
2246         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
2247         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
2248         accordingly.
2250 2013-09-08  Jan Hubicka  <jh@suse.cz>
2252         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
2253         targets and devirtualize to BUILT_IN_UNREACHABLE.
2254         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
2255         * ipa.c (walk_polymorphic_call_targets): New function.
2256         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
2257         functions; use the new timevar.
2258         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
2259         was removed from the program.
2260         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
2261         not consider it in the walk when its vtable is dead.
2262         (possible_polymorphic_call_targets_1): Pass anonymous flag to
2263         record_binfo.
2264         (devirt_variable_node_removal_hook): New function.
2265         (possible_polymorphic_call_targets): Also register
2266         devirt_variable_node_removal_hook.
2267         (ipa_devirt): Do not do non-speculative devirtualization.
2268         (gate_ipa_devirt): One execute if devirtualizing speculatively.
2270 2013-09-08  Jan Hubicka  <jh@suse.cz>
2272         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
2273         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
2274         varpool_remove_variable_insertion_hook): Declare.
2275         * varpool.c (varpool_node_hook_list): New structure.
2276         (first_varpool_node_removal_hook,
2277         first_varpool_variable_insertion_hook): New variables.
2278         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
2279         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
2280         varpool_remove_variable_insertion_hook,
2281         varpool_call_variable_insertion_hooks): New functions.
2282         (varpool_remove_node): Use it.
2284 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
2286         PR c++/54941
2287         * diagnostic.c (diagnostic_build_prefix): When s.file is
2288         "<built-in>" don't output line and column numbers.
2290 2013-09-06  Jan Hubicka  <jh@suse.cz>
2292         * cgraphunit.c (expand_thunk): Get body before touching arguments.
2293         * lto-streamer-out.c: Stream thunks, too.
2294         * lto-streamer-in.c (input_function): Pop cfun here
2295         (lto_read_body): Instead of here.
2297 2013-09-06  Caroline Tice  <cmtice@google.com>
2299         * doc/install.texi: Add documentation for the --enable-vtable-verify
2300         and the --disable-libvtv configure options.
2302 2013-09-06  Jeff Law  <law@redhat.com>
2304         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
2305         edge implied equivalences into successor phis.
2307 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
2309         * resource.c (mark_referenced_resources): Handle COND_EXEC.
2311 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
2313         * resource.c (mark_target_live_regs): Compute resources taking
2314         into account if a call is predicated or not.
2316 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
2318         * toplev.c (output_stack_usage): Be prepared for suffixes created by
2319         the compiler in the function names.
2321 2013-09-06  Jan Hubicka  <jh@suse.cz>
2323         PR middle-end/58094
2324         * ipa-inline.c (has_caller_p): New function.
2325         (want_inline_function_to_all_callers_p): Use it.
2326         (sum_callers, inline_to_all_callers): Break out from ...
2327         (ipa_inline): ... here.
2329 2013-09-06  Jan Hubicka  <jh@suse.cz>
2331         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
2332         only when AVX is enabled.
2334 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2336         * config/aarch64/aarch64.md
2337         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
2338         is fpsimd_<load/store>2.
2339         (load_pair<mode>): Likewise.
2340         (store_pair<mode>): Likewise.
2342 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2344         * config/arm/types.md (type): Add "mrs" type.
2345         * config/aarch64/aarch64.md
2346         (aarch64_load_tp_hard): Make type "mrs".
2347         * config/arm/arm.md
2348         (load_tp_hard): Make type "mrs".
2349         * config/arm/cortex-a15.md: Update with new attributes.
2350         * config/arm/cortex-a5.md: Update with new attributes.
2351         * config/arm/cortex-a53.md: Update with new attributes.
2352         * config/arm/cortex-a7.md: Update with new attributes.
2353         * config/arm/cortex-a8.md: Update with new attributes.
2354         * config/arm/cortex-a9.md: Update with new attributes.
2355         * config/arm/cortex-m4.md: Update with new attributes.
2356         * config/arm/cortex-r4.md: Update with new attributes.
2357         * config/arm/fa526.md: Update with new attributes.
2358         * config/arm/fa606te.md: Update with new attributes.
2359         * config/arm/fa626te.md: Update with new attributes.
2360         * config/arm/fa726te.md: Update with new attributes.
2362 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2364         * config/aarch64/aarch64.md
2365         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
2366         (*movtf_aarch64): Likewise.
2367         * config/arm/arm.md
2368         (thumb1_movdi_insn): Use "multiple" for type where more than one
2369         instruction is used for a move.
2370         (*arm32_movhf): Likewise.
2371         (*thumb_movdf_insn): Likewise.
2373 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2375         * config/arm/types.md (type): Rename fcpys to fmov.
2376         * config/arm/vfp.md
2377         (*arm_movsi_vfp): Rename type fcpys as fmov.
2378         (*thumb2_movsi_vfp): Likewise
2379         (*movhf_vfp_neon): Likewise
2380         (*movhf_vfp): Likewise
2381         (*movsf_vfp): Likewise
2382         (*thumb2_movsf_vfp): Likewise
2383         (*movsfcc_vfp): Likewise
2384         (*thumb2_movsfcc_vfp): Likewise
2385         * config/aarch64/aarch64-simd.md
2386         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
2387         * config/aarch64/aarch64.md
2388         (*movsi_aarch64): Replace type mov_reg with fmovs.
2389         (*movdi_aarch64): Likewise
2390         (*movsf_aarch64): Likewise
2391         (*movdf_aarch64): Likewise
2392         * config/arm/arm.c
2393         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
2394         * config/arm/iwmmxt.md
2395         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
2396         * config/arm/arm1020e.md: Update with new attributes.
2397         * config/arm/cortex-a15-neon.md: Update with new attributes.
2398         * config/arm/cortex-a5.md: Update with new attributes.
2399         * config/arm/cortex-a53.md: Update with new attributes.
2400         * config/arm/cortex-a7.md: Update with new attributes.
2401         * config/arm/cortex-a8-neon.md: Update with new attributes.
2402         * config/arm/cortex-a9.md: Update with new attributes.
2403         * config/arm/cortex-m4-fpu.md: Update with new attributes.
2404         * config/arm/cortex-r4f.md: Update with new attributes.
2405         * config/arm/marvell-pj4.md: Update with new attributes.
2406         * config/arm/vfp11.md: Update with new attributes.
2408 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2410         * config/aarch64/aarch64.md
2411         (*madd<mode>): Fix type attribute.
2412         (*maddsi_uxtw): Likewise.
2413         (*msub<mode>): Likewise.
2414         (*msubsi_uxtw): Likewise.
2415         (<su_optab>maddsidi4): Likewise.
2416         (<su_optab>msubsidi4): Likewise.
2418 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2420         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
2421         * config/arm/arm.md (core_cycles): Remove fdiv.
2422         * config/arm/vfp.md:
2423         (*sqrtsf2_vfp): Update for attribute changes.
2424         (*sqrtdf2_vfp): Likewise.
2425         * config/aarch64/aarch64.md:
2426         (sqrt<mode>2): Update for attribute changes.
2427         * config/arm/arm1020e.md: Update with new attributes.
2428         * config/arm/cortex-a15-neon.md: Update with new attributes.
2429         * config/arm/cortex-a5.md: Update with new attributes.
2430         * config/arm/cortex-a53.md: Update with new attributes.
2431         * config/arm/cortex-a7.md: Update with new attributes.
2432         * config/arm/cortex-a8-neon.md: Update with new attributes.
2433         * config/arm/cortex-a9.md: Update with new attributes.
2434         * config/arm/cortex-m4-fpu.md: Update with new attributes.
2435         * config/arm/cortex-r4f.md: Update with new attributes.
2436         * config/arm/marvell-pj4.md: Update with new attributes.
2437         * config/arm/vfp11.md: Update with new attributes.
2439 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2441         * config/arm/types.md
2442         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
2443         * config/aarch64/aarch64.md
2444         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
2445         new attributes.
2446         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
2447         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
2448         (float<GPI:mode><GPF:mode>2): Likewise.
2449         * config/arm/vfp.md
2450         (*truncsisf2_vfp): Update with new attributes.
2451         (*truncsidf2_vfp): Likewise.
2452         (fixuns_truncsfsi2): Likewise.
2453         (fixuns_truncdfsi2): Likewise.
2454         (*floatsisf2_vfp): Likewise.
2455         (*floatsidf2_vfp): Likewise.
2456         (floatunssisf2): Likewise.
2457         (floatunssidf2): Likewise.
2458         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
2459         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
2460         * config/arm/arm1020e.md: Update with new attributes.
2461         * config/arm/cortex-a15-neon.md: Update with new attributes.
2462         * config/arm/cortex-a5.md: Update with new attributes.
2463         * config/arm/cortex-a53.md: Update with new attributes.
2464         * config/arm/cortex-a7.md: Update with new attributes.
2465         * config/arm/cortex-a8-neon.md: Update with new attributes.
2466         * config/arm/cortex-a9.md: Update with new attributes.
2467         * config/arm/cortex-m4-fpu.md: Update with new attributes.
2468         * config/arm/cortex-r4f.md: Update with new attributes.
2469         * config/arm/marvell-pj4.md: Update with new attributes.
2470         * config/arm/vfp11.md: Update with new attributes.
2472 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2474         * config/aarch64/arm_neon.h
2475         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
2476         (vqtbx<1,2,3,4><q>_s8): Likewise.
2478 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2480         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
2481         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
2482         (add<mode>3): Add type attribute.
2483         (add<mode>3): Likewise.
2484         (usadd<mode>3): Likewise.
2485         (ssadd<mode>3): Likewise.
2486         (sub<mode>3): Likewise.
2487         (sub<mode>3): Likewise.
2488         (ussub<mode>3): Likewise.
2489         (sssub<mode>3): Likewise.
2490         (ssmulsa3): Likewise.
2491         (usmulusa3): Likewise.
2492         (arm_usatsihi): Likewise.
2493         * config/arm/vfp.md
2494         (*movdi_vfp): Add types for all instructions.
2495         (*movdi_vfp_cortexa8): Likewise.
2496         (*movhf_vfp_neon): Likewise.
2497         (*movhf_vfp): Likewise.
2498         (*movdf_vfp): Likewise.
2499         (*thumb2_movdf_vfp): Likewise.
2500         (*thumb2_movdfcc_vfp): Likewise.
2501         * config/arm/arm.md: Add type attribute to all insn patterns.
2502         (*thumb1_adddi3): Add type attribute.
2503         (*arm_adddi3): Likewise.
2504         (*adddi_sesidi_di): Likewise.
2505         (*adddi_zesidi_di): Likewise.
2506         (*thumb1_addsi3): Likewise.
2507         (addsi3_compare0): Likewise.
2508         (*addsi3_compare0_scratch): Likewise.
2509         (*compare_negsi_si): Likewise.
2510         (cmpsi2_addneg): Likewise.
2511         (*addsi3_carryin_<optab>): Likewise.
2512         (*addsi3_carryin_alt2_<optab>): Likewise.
2513         (*addsi3_carryin_clobercc_<optab>): Likewise.
2514         (*subsi3_carryin): Likewise.
2515         (*subsi3_carryin_const): Likewise.
2516         (*subsi3_carryin_compare): Likewise.
2517         (*subsi3_carryin_compare_const): Likewise.
2518         (*arm_subdi3): Likewise.
2519         (*thumb_subdi3): Likewise.
2520         (*subdi_di_zesidi): Likewise.
2521         (*subdi_di_sesidi): Likewise.
2522         (*subdi_zesidi_di): Likewise.
2523         (*subdi_sesidi_di): Likewise.
2524         (*subdi_zesidi_ze): Likewise.
2525         (thumb1_subsi3_insn): Likewise.
2526         (*arm_subsi3_insn): Likewise.
2527         (*anddi3_insn): Likewise.
2528         (*anddi_zesidi_di): Likewise.
2529         (*anddi_sesdi_di): Likewise.
2530         (*ne_zeroextracts): Likewise.
2531         (*ne_zeroextracts): Likewise.
2532         (*ite_ne_zeroextr): Likewise.
2533         (*ite_ne_zeroextr): Likewise.
2534         (*anddi_notdi_di): Likewise.
2535         (*anddi_notzesidi): Likewise.
2536         (*anddi_notsesidi): Likewise.
2537         (andsi_notsi_si): Likewise.
2538         (thumb1_bicsi3): Likewise.
2539         (*iordi3_insn): Likewise.
2540         (*iordi_zesidi_di): Likewise.
2541         (*iordi_sesidi_di): Likewise.
2542         (*thumb1_iorsi3_insn): Likewise.
2543         (*xordi3_insn): Likewise.
2544         (*xordi_zesidi_di): Likewise.
2545         (*xordi_sesidi_di): Likewise.
2546         (*arm_xorsi3): Likewise.
2547         (*andsi_iorsi3_no): Likewise.
2548         (*smax_0): Likewise.
2549         (*smax_m1): Likewise.
2550         (*arm_smax_insn): Likewise.
2551         (*smin_0): Likewise.
2552         (*arm_smin_insn): Likewise.
2553         (*arm_umaxsi3): Likewise.
2554         (*arm_uminsi3): Likewise.
2555         (*minmax_arithsi): Likewise.
2556         (*minmax_arithsi_): Likewise.
2557         (*satsi_<SAT:code>): Likewise.
2558         (arm_ashldi3_1bit): Likewise.
2559         (arm_ashrdi3_1bit): Likewise.
2560         (arm_lshrdi3_1bit): Likewise.
2561         (*arm_negdi2): Likewise.
2562         (*thumb1_negdi2): Likewise.
2563         (*arm_negsi2): Likewise.
2564         (*thumb1_negsi2): Likewise.
2565         (*negdi_extendsid): Likewise.
2566         (*negdi_zero_extend): Likewise.
2567         (*arm_abssi2): Likewise.
2568         (*thumb1_abssi2): Likewise.
2569         (*arm_neg_abssi2): Likewise.
2570         (*thumb1_neg_abss): Likewise.
2571         (one_cmpldi2): Likewise.
2572         (extend<mode>di2): Likewise.
2573         (*compareqi_eq0): Likewise.
2574         (*arm_extendhisi2addsi): Likewise.
2575         (*arm_movdi): Likewise.
2576         (*thumb1_movdi_insn): Likewise.
2577         (*arm_movt): Likewise.
2578         (*thumb1_movsi_insn): Likewise.
2579         (pic_add_dot_plus_four): Likewise.
2580         (pic_add_dot_plus_eight): Likewise.
2581         (tls_load_dot_plus_eight): Likewise.
2582         (*thumb1_movhi_insn): Likewise.
2583         (*thumb1_movsf_insn): Likewise.
2584         (*movdf_soft_insn): Likewise.
2585         (*thumb_movdf_insn): Likewise.
2586         (cbranchsi4_insn): Likewise.
2587         (cbranchsi4_scratch): Likewise.
2588         (*negated_cbranchsi4): Likewise.
2589         (*tbit_cbranch): Likewise.
2590         (*tlobits_cbranch): Likewise.
2591         (*tstsi3_cbranch): Likewise.
2592         (*cbranchne_decr1): Likewise.
2593         (*addsi3_cbranch): Likewise.
2594         (*addsi3_cbranch_scratch): Likewise.
2595         (*arm_cmpdi_insn): Likewise.
2596         (*arm_cmpdi_unsig): Likewise.
2597         (*arm_cmpdi_zero): Likewise.
2598         (*thumb_cmpdi_zero): Likewise.
2599         (*deleted_compare): Likewise.
2600         (*mov_scc): Likewise.
2601         (*mov_negscc): Likewise.
2602         (*mov_notscc): Likewise.
2603         (*cstoresi_eq0_thumb1_insn): Likewise.
2604         (cstoresi_nltu_thumb1): Likewise.
2605         (cstoresi_ltu_thu): Likewise.
2606         (thumb1_addsi3_addgeu): Likewise.
2607         (*arm_jump): Likewise.
2608         (*thumb_jump): Likewise.
2609         (*check_arch2): Likewise.
2610         (arm_casesi_internal): Likewise.
2611         (thumb1_casesi_dispatch): Likewise.
2612         (*arm_indirect_jump): Likewise.
2613         (*thumb1_indirect_jump): Likewise.
2614         (nop): Likewise.
2615         (*and_scc): Likewise.
2616         (*ior_scc): Likewise.
2617         (*compare_scc): Likewise.
2618         (*cond_move): Likewise.
2619         (*cond_arith): Likewise.
2620         (*cond_sub): Likewise.
2621         (*cmp_ite0): Likewise.
2622         (*cmp_ite1): Likewise.
2623         (*cmp_and): Likewise.
2624         (*cmp_ior): Likewise.
2625         (*ior_scc_scc): Likewise.
2626         (*ior_scc_scc_cmp): Likewise.
2627         (*and_scc_scc): Likewise.
2628         (*and_scc_scc_cmp): Likewise.
2629         (*and_scc_scc_nod): Likewise.
2630         (*negscc): Likewise.
2631         (movcond_addsi): Likewise.
2632         (movcond): Likewise.
2633         (*ifcompare_plus_move): Likewise.
2634         (*if_plus_move): Likewise.
2635         (*ifcompare_move_plus): Likewise.
2636         (*if_move_plus): Likewise.
2637         (*ifcompare_arith_arith): Likewise.
2638         (*if_arith_arith): Likewise.
2639         (*ifcompare_arith_move): Likewise.
2640         (*if_arith_move): Likewise.
2641         (*ifcompare_move_arith): Likewise.
2642         (*if_move_arith): Likewise.
2643         (*ifcompare_move_not): Likewise.
2644         (*if_move_not): Likewise.
2645         (*ifcompare_not_move): Likewise.
2646         (*if_not_move): Likewise.
2647         (*ifcompare_shift_move): Likewise.
2648         (*if_shift_move): Likewise.
2649         (*ifcompare_move_shift): Likewise.
2650         (*if_move_shift): Likewise.
2651         (*ifcompare_shift_shift): Likewise.
2652         (*ifcompare_not_arith): Likewise.
2653         (*ifcompare_arith_not): Likewise.
2654         (*if_arith_not): Likewise.
2655         (*ifcompare_neg_move): Likewise.
2656         (*if_neg_move): Likewise.
2657         (*ifcompare_move_neg): Likewise.
2658         (*if_move_neg): Likewise.
2659         (prologue_thumb1_interwork): Likewise.
2660         (*cond_move_not): Likewise.
2661         (*sign_extract_onebit): Likewise.
2662         (*not_signextract_onebit): Likewise.
2663         (stack_tie): Likewise.
2664         (align_4): Likewise.
2665         (align_8): Likewise.
2666         (consttable_end): Likewise.
2667         (consttable_1): Likewise.
2668         (consttable_2): Likewise.
2669         (consttable_4): Likewise.
2670         (consttable_8): Likewise.
2671         (consttable_16): Likewise.
2672         (*thumb1_tablejump): Likewise.
2673         (prefetch): Likewise.
2674         (force_register_use): Likewise.
2675         (thumb_eh_return): Likewise.
2676         (load_tp_hard): Likewise.
2677         (load_tp_soft): Likewise.
2678         (tlscall): Likewise.
2679         (*arm_movtas_ze): Likewise.
2680         (*arm_rev): Likewise.
2681         (*arm_revsh): Likewise.
2682         (*arm_rev16): Likewise.
2683         * config/arm/thumb2.md
2684         (*thumb2_smaxsi3): Likewise.
2685         (*thumb2_sminsi3): Likewise.
2686         (*thumb32_umaxsi3): Likewise.
2687         (*thumb2_uminsi3): Likewise.
2688         (*thumb2_negdi2): Likewise.
2689         (*thumb2_abssi2): Likewise.
2690         (*thumb2_neg_abss): Likewise.
2691         (*thumb2_movsi_insn): Likewise.
2692         (tls_load_dot_plus_four): Likewise.
2693         (*thumb2_movhi_insn): Likewise.
2694         (*thumb2_mov_scc): Likewise.
2695         (*thumb2_mov_negs): Likewise.
2696         (*thumb2_mov_negs): Likewise.
2697         (*thumb2_mov_nots): Likewise.
2698         (*thumb2_mov_nots): Likewise.
2699         (*thumb2_movsicc_): Likewise.
2700         (*thumb2_movsfcc_soft_insn): Likewise.
2701         (*thumb2_indirect_jump): Likewise.
2702         (*thumb2_and_scc): Likewise.
2703         (*thumb2_ior_scc): Likewise.
2704         (*thumb2_ior_scc_strict_it): Likewise.
2705         (*thumb2_cond_move): Likewise.
2706         (*thumb2_cond_arith): Likewise.
2707         (*thumb2_cond_ari): Likewise.
2708         (*thumb2_cond_sub): Likewise.
2709         (*thumb2_negscc): Likewise.
2710         (*thumb2_movcond): Likewise.
2711         (thumb2_casesi_internal): Likewise.
2712         (thumb2_casesi_internal_pic): Likewise.
2713         (*thumb2_alusi3_short): Likewise.
2714         (*thumb2_mov<mode>_shortim): Likewise.
2715         (*thumb2_addsi_short): Likewise.
2716         (*thumb2_subsi_short): Likewise.
2717         (thumb2_addsi3_compare0): Likewise.
2718         (*thumb2_cbz): Likewise.
2719         (*thumb2_cbnz): Likewise.
2720         (*thumb2_one_cmplsi2_short): Likewise.
2721         (*thumb2_negsi2_short): Likewise.
2722         (*orsi_notsi_si): Likewise.
2723         * config/arm/arm1020e.md: Update with new attributes.
2724         * config/arm/arm1026ejs.md: Update with new attributes.
2725         * config/arm/arm1136jfs.md: Update with new attributes.
2726         * config/arm/arm926ejs.md: Update with new attributes.
2727         * config/arm/cortex-a15.md: Update with new attributes.
2728         * config/arm/cortex-a5.md: Update with new attributes.
2729         * config/arm/cortex-a53.md: Update with new attributes.
2730         * config/arm/cortex-a7.md: Update with new attributes.
2731         * config/arm/cortex-a8.md: Update with new attributes.
2732         * config/arm/cortex-a9.md: Update with new attributes.
2733         * config/arm/cortex-m4.md: Update with new attributes.
2734         * config/arm/cortex-r4.md: Update with new attributes.
2735         * config/arm/fa526.md: Update with new attributes.
2736         * config/arm/fa606te.md: Update with new attributes.
2737         * config/arm/fa626te.md: Update with new attributes.
2738         * config/arm/fa726te.md: Update with new attributes.
2740 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2742         * config/aarch64/aarch64-simd.md
2743         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
2744         <vwx> iterator to ensure correct register choice.
2745         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
2746         (aarch64_sqdmull_n<mode>): Likewise.
2747         (aarch64_sqdmull2_n<mode>_internal): Likewise.
2748         * config/aarch64/arm_neon.h
2749         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
2750         (vml<as><q>_n_<su>16): Likewise.
2751         (vml<as>l_high_lane<q>_<su>16): Likewise.
2752         (vml<as>l_high_n_<su>16): Likewise.
2753         (vml<as>l_lane<q>_<su>16): Likewise.
2754         (vml<as>l_n_<su>16): Likewise.
2755         (vmul<q>_lane<q>_<su>16): Likewise.
2756         (vmul<q>_n_<su>16): Likewise.
2757         (vmull_lane<q>_<su>16): Likewise.
2758         (vmull_n_<su>16): Likewise.
2759         (vmull_high_lane<q>_<su>16): Likewise.
2760         (vmull_high_n_<su>16): Likewise.
2761         (vqrdmulh<q>_n_s16): Likewise.
2763 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
2765         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
2766         have the correct lane parameter.
2768 2013-09-06  Richard Biener <rguenther@suse.de>
2770         * cfganal.c (control_dependences::~control_dependences):
2771         Properly free all of the vector.
2773 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
2775         PR target/58269
2776         * config/i386/i386.c (ix86_conditional_register_usage):
2777         Proper initialize extended SSE registers.
2779 2013-09-06  Jan Hubicka  <jh@suse.cz>
2781         PR tree-optimization/58311
2782         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
2784 2013-09-06  Jan Hubicka  <jh@suse.cz>
2786         * Makefile.in (tree-sra.o): Update dependencies.
2787         * tree-sra.c: Include ipa-utils.h
2788         (scan_function): Use recursive_call_p.
2789         (has_caller_p): New function.
2790         (cgraph_for_node_and_aliases): Count also callers of aliases.
2792 2013-09-06  Jan Hubicka  <jh@suse.cz>
2794         PR middle-end/58094
2795         * cgraph.h (symtab_semantically_equivalent_p): Declare.
2796         * tree-tailcall.c: Include ipa-utils.h.
2797         (find_tail_calls): Use it.
2798         * ipa-pure-const.c (check_call): Likewise.
2799         * ipa-utils.c (recursive_call_p): New function.
2800         * ipa-utils.h (recursive_call_p): Dclare.
2801         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
2802         (symtab_semantically_equivalent_p): New function.
2803         * Makefile.in (tree-tailcall.o): Update dependencies.
2805 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
2807         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
2808         non-inlinable part.
2810 2013-09-06  Richard Biener  <rguenther@suse.de>
2812         * lto-streamer.h (lto_global_var_decls): Remove.
2813         * Makefile.in (OBJS): Remove lto-symtab.o.
2814         (lto-symtab.o): Remove.
2815         (GTFILES): Remove lto-symtab.c
2816         * lto-symtab.c: Move to lto/
2818 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2820         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
2821         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
2822         (UNSPEC_FPINT_RINT): New constant definitions.
2823         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
2824         definition with 2 attributes.
2825         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
2826         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
2827         definitions.
2829 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2831         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
2832         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
2833         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
2834         attribute to "z196_alone".
2835         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
2836         "zEC12_simple".
2838 2013-09-06  Richard Biener  <rguenther@suse.de>
2840         * basic-block.h (class control_dependences): New.
2841         * tree-ssa-dce.c (control_dependence_map): Remove.
2842         (cd): New global.
2843         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
2844         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
2845         find_pdom, find_control_dependence, find_all_control_dependences):
2846         Move to cfganal.c.
2847         (mark_control_dependent_edges_necessary,
2848         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
2849         tree_dce_done, perform_tree_ssa_dce): Adjust.
2850         * cfganal.c (set_control_dependence_map_bit,
2851         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
2852         find_all_control_dependences): Move from tree-ssa-dce.c and
2853         implement as methods of control_dependences class.
2854         (control_dependences::control_dependences): New.
2855         (control_dependences::~control_dependences): Likewise.
2856         (control_dependences::get_edges_dependent_on): Likewise.
2857         (control_dependences::get_edge): Likewise.
2859 2013-09-04  Jan Hubicka  <jh@suse.cz>
2861         * tree.c (types_same_for_odr): Drop overactive check.
2862         * ipa-devirt.c (hash_type_name): Likewise.
2864 2013-09-04  Jan Hubicka  <jh@suse.cz>
2866         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
2867         (analyze_functions): ... here.
2869 2013-09-04  Jan Hubicka  <jh@suse.cz>
2871         PR middle-end/58201
2872         * cgraphunit.c (analyze_functions): Clear AUX fields
2873         after processing; initialize assembler name has.
2875 2013-09-05  Jeff Law  <law@redhat.com>
2877         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
2878         from thread_around_empty_block.  Record threading path into PATH.
2879         Recurse if threading through the initial block is successful.
2880         (thread_across_edge): Corresponding changes to slightly simplify.
2882 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2884         * config/aarch64/aarch64.md
2885         (type): Remove frecpe, frecps, frecpx.
2886         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
2887         fix to be a TARGET_SIMD instruction.
2888         (aarch64_frecps): Remove.
2889         * config/aarch64/aarch64-simd.md
2890         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
2891         (aarch64_frecps<mode>): Handle all float/vector of float modes.
2893 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2894             Sofiane Naci  <sofiane.naci@arm.com>
2896         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
2897         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
2898         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
2899         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
2900         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
2901         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
2902         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
2903         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
2904         into "clz, "rbit".  Rename "shift" to "shift_imm".
2905         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
2906         changes.  Update for attribute changes all occurrences of arlo_* and
2907         shift* types.
2908         * config/arm/arm-fixed.md: Update for attribute changes
2909         all occurrences of arlo_* types.
2910         * config/arm/thumb2.md: Update for attribute changes all occurrences
2911         of arlo_* types.
2912         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
2913         (cortexa7_older_only): Likewise.
2914         (cortexa7_younger):  Likewise.
2915         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
2916         (1020alu_shift_op): Likewise.
2917         (1020alu_shift_reg_op): Likewise.
2918         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
2919         (alu_shift_op): Likewise.
2920         (alu_shift_reg_op): Likewise.
2921         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
2922         (11_alu_shift_op): Likewise.
2923         (11_alu_shift_reg_op): Likewise.
2924         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
2925         (9_alu_shift_reg_op): Likewise.
2926         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
2927         attribute changes.
2928         (cortex_a15_alu_shift): Likewise.
2929         (cortex_a15_alu_shift_reg): Likewise.
2930         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
2931         attribute changes.
2932         (cortex_a5_alu_shift): Likewise.
2933         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
2934         attribute changes.
2935         (cortex_a53_alu_shift): Likewise.
2936         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
2937         attribute changes.
2938         (cortex_a7_alu_reg): Likewise.
2939         (cortex_a7_alu_shift): Likewise.
2940         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
2941         attribute changes.
2942         (cortex_a8_alu_shift): Likewise.
2943         (cortex_a8_alu_shift_reg): Likewise.
2944         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
2945         (cortex_a9_dp_shift): Likewise.
2946         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
2947         attribute changes.
2948         * config/arm/cortex-r4.md
2949         (cortex_r4_alu): Update for attribute changes.
2950         (cortex_r4_mov): Likewise.
2951         (cortex_r4_alu_shift_reg): Likewise.
2952         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
2953         (526_alu_shift_op): Likewise.
2954         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
2955         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
2956         (626te_alu_shift_op): Likewise.
2957         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
2958         (726te_alu_shift_op): Likewise.
2959         (726te_alu_shift_reg_op): Likewise.
2960         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
2961         (mp626_alu_shift_op): Likewise.
2962         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
2963         (pj4_alu_conds): Likewise.
2964         (pj4_shift): Likewise.
2965         (pj4_shift_conds): Likewise.
2966         (pj4_alu_shift): Likewise.
2967         (pj4_alu_shift_conds): Likewise.
2968         * config/aarch64/aarch64.md: Update for attribute change
2969         all occurrences of arlo_* and shift* types.
2971 2013-09-05  Mike Stump  <mikestump@comcast.net>
2973         * tree.h: Move documentation for tree_function_decl to tree-core.h
2974         with the declaration.
2976 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
2978         PR target/58139
2979         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
2980         looking for widest mode.
2982 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
2984         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
2986 2013-09-05  Richard Biener  <rguenther@suse.de>
2988         PR tree-optimization/58137
2989         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
2990         Do not create vectors of pointers.
2991         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
2992         types for the components of the vector initializer.
2993         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
2994         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
2996 2013-09-05  Martin Jambor  <mjambor@suse.cz>
2998         * ipa-prop.c (remove_described_reference): Accept missing references,
2999         return false if that hppens, otherwise return true.
3000         (cgraph_node_for_jfunc): New function.
3001         (try_decrement_rdesc_refcount): Likewise.
3002         (try_make_edge_direct_simple_call): Use them.
3003         (ipa_edge_removal_hook): Remove references from rdescs.
3004         (ipa_edge_duplication_hook): Clone rdescs and their references
3005         when the new edge has the same caller as the old one.
3006         * cgraph.c (cgraph_resolve_speculation): Remove speculative
3007         reference before removing any edges.
3009 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
3011         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
3012         initial store.
3013         * thumb2.md (thumb2_storewb_parisi): New pattern.
3015 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
3017         * config/aarch64/aarch64-option-extensions.def: Add
3018         AARCH64_OPT_EXTENSION of 'crc'.
3019         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
3020         (AARCH64_ISA_CRC): Ditto.
3021         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
3022         description of the CRC extension.
3024 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3026         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
3027         * config/rs6000/linux.h: Ditto.
3028         * alpha/linux.h: Ditto.
3029         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
3030         no_c99_libc_has_function.
3031         * config/c6x/uclinux-elf.h: Ditto.
3032         * config/lm32/uclinux-elf.h: Ditto.
3033         * config/m68k/uclinux.h: Ditto.
3034         * config/moxie/uclinux.h: Ditto.
3035         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
3036         (crisv32-*-linux*, cris-*-linux*): Ditto.
3037         * config/bfin/bfin.c: Include "tm_p.h".
3039 2013-09-05  Richard Biener  <rguenther@suse.de>
3041         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
3042         check for a definition without a basic-block.
3044 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
3045             Sofiane Naci <sofiane.naci@arm.com>
3047         * config/aarch64/aarch64.md
3048         (*movti_aarch64): Rename r_2_f and f_2_r.
3049         (*movsf_aarch64): Likewise.
3050         (*movdf_aarch64): Likewise.
3051         (*movtf_aarch64): Likewise.
3052         (aarch64_movdi_<mode>low): Likewise.
3053         (aarch64_movdi_<mode>high): Likewise.
3054         (aarch64_mov<mode>high_di): Likewise.
3055         (aarch64_mov<mode>low_di): Likewise.
3056         (aarch64_movtilow_tilow): Likewise.
3057         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
3058         values to config/arm/types.md
3059         (attribute "conds"): Update for attribute change.
3060         (anddi3_insn): Likewise.
3061         (iordi3_insn): Likewise.
3062         (xordi3_insn): Likewise.
3063         (one_cmpldi2): Likewise.
3064         * config/arm/types.md (type): Add Neon types.
3065         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
3066         use "type" attribute.
3067         (movmisalign<mode>_neon_store): Likewise.
3068         (movmisalign<mode>_neon_load): Likewise.
3069         (vec_set<mode>_internal): Likewise.
3070         (vec_setv2di_internal): Likewise.
3071         (vec_extract<mode>): Likewise.
3072         (vec_extractv2di): Likewise.
3073         (add<mode>3_neon): Likewise.
3074         (adddi3_neon): Likewise.
3075         (sub<mode>3_neon): Likewise.
3076         (subdi3_neon): Likewise.
3077         (mul<mode>3_neon): Likewise.
3078         (mul<mode>3add<mode>_neon): Likewise.
3079         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
3080         (fma<VCVTF:mode>4)): Likewise.
3081         (fma<VCVTF:mode>4_intrinsic): Likewise.
3082         (fmsub<VCVTF:mode>4)): Likewise.
3083         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
3084         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
3085         (ior<mode>3): Likewise.
3086         (and<mode>3): Likewise.
3087         (anddi3_neon): Likewise.
3088         (orn<mode>3_neon): Likewise.
3089         (orndi3_neon): Likewise.
3090         (bic<mode>3_neon): Likewise.
3091         (bicdi3_neon): Likewise.
3092         (xor<mode>3): Likewise.
3093         (one_cmpl<mode>2): Likewise.
3094         (abs<mode>2): Likewise.
3095         (neg<mode>2): Likewise.
3096         (umin<mode>3_neon): Likewise.
3097         (umax<mode>3_neon): Likewise.
3098         (smin<mode>3_neon): Likewise.
3099         (smax<mode>3_neon): Likewise.
3100         (vashl<mode>3): Likewise.
3101         (vashr<mode>3_imm): Likewise.
3102         (vlshr<mode>3_imm): Likewise.
3103         (ashl<mode>3_signed): Likewise.
3104         (ashl<mode>3_unsigned): Likewise.
3105         (neon_load_count): Likewise.
3106         (ashldi3_neon_noclobber): Likewise.
3107         (signed_shift_di3_neon): Likewise.
3108         (unsigned_shift_di3_neon): Likewise.
3109         (ashrdi3_neon_imm_noclobber): Likewise.
3110         (lshrdi3_neon_imm_noclobber): Likewise.
3111         (widen_ssum<mode>3): Likewise.
3112         (widen_usum<mode>3): Likewise.
3113         (quad_halves_<code>v4si): Likewise.
3114         (quad_halves_<code>v4sf): Likewise.
3115         (quad_halves_<code>v8hi): Likewise.
3116         (quad_halves_<code>v16qi): Likewise.
3117         (reduc_splus_v2di): Likewise.
3118         (neon_vpadd_internal<mode>): Likewise.
3119         (neon_vpsmin<mode>): Likewise.
3120         (neon_vpsmax<mode>): Likewise.
3121         (neon_vpumin<mode>): Likewise.
3122         (neon_vpumax<mode>): Likewise.
3123         (ss_add<mode>_neon): Likewise.
3124         (us_add<mode>_neon): Likewise.
3125         (ss_sub<mode>_neon): Likewise.
3126         (us_sub<mode>_neon): Likewise.
3127         (neon_vadd<mode>_unspec): Likewise.
3128         (neon_vaddl<mode>): Likewise.
3129         (neon_vaddw<mode>): Likewise.
3130         (neon_vhadd<mode>): Likewise.
3131         (neon_vqadd<mode>): Likewise.
3132         (neon_vaddhn<mode>): Likewise.
3133         (neon_vmul<mode>): Likewise.
3134         (neon_vmla<mode>): Likewise.
3135         (neon_vmlal<mode>): Likewise.
3136         (neon_vmls<mode>): Likewise.
3137         (neon_vmlsl<mode>): Likewise.
3138         (neon_vqdmulh<mode>): Likewise.
3139         (neon_vqdmlal<mode>): Likewise.
3140         (neon_vqdmlsl<mode>): Likewise.
3141         (neon_vmull<mode>): Likewise.
3142         (neon_vqdmull<mode>): Likewise.
3143         (neon_vsub<mode>_unspec): Likewise.
3144         (neon_vsubl<mode>): Likewise.
3145         (neon_vsubw<mode>): Likewise.
3146         (neon_vqsub<mode>): Likewise.
3147         (neon_vhsub<mode>): Likewise.
3148         (neon_vsubhn<mode>): Likewise.
3149         (neon_vceq<mode>): Likewise.
3150         (neon_vcge<mode>): Likewise.
3151         (neon_vcgeu<mode>): Likewise.
3152         (neon_vcgt<mode>): Likewise.
3153         (neon_vcgtu<mode>): Likewise.
3154         (neon_vcle<mode>): Likewise.
3155         (neon_vclt<mode>): Likewise.
3156         (neon_vcage<mode>): Likewise.
3157         (neon_vcagt<mode>): Likewise.
3158         (neon_vtst<mode>): Likewise.
3159         (neon_vabd<mode>): Likewise.
3160         (neon_vabdl<mode>): Likewise.
3161         (neon_vaba<mode>): Likewise.
3162         (neon_vabal<mode>): Likewise.
3163         (neon_vmax<mode>): Likewise.
3164         (neon_vmin<mode>): Likewise.
3165         (neon_vpaddl<mode>): Likewise.
3166         (neon_vpadal<mode>): Likewise.
3167         (neon_vpmax<mode>): Likewise.
3168         (neon_vpmin<mode>): Likewise.
3169         (neon_vrecps<mode>): Likewise.
3170         (neon_vrsqrts<mode>): Likewise.
3171         (neon_vqabs<mode>): Likewise.
3172         (neon_vqneg<mode>): Likewise.
3173         (neon_vcls<mode>): Likewise.
3174         (clz<mode>2): Likewise.
3175         (popcount<mode>2): Likewise.
3176         (neon_vrecpe): Likewise.
3177         (neon_vrsqrte): Likewise.
3178         (neon_vget_lane<mode>_sext_internal): Likewise.
3179         (neon_vget_lane<mode>_zext_internal): Likewise.
3180         (neon_vdup_n<mode>): Likewise.
3181         (neon_vdup_nv2di): Likewise.
3182         (neon_vdpu_lane<mode>_internal): Likewise.
3183         (neon_vswp<mode>): Likewise.
3184         (float<mode><V_cvtto>2): Likewise.
3185         (floatuns<mode><V_cvtto>2): Likewise.
3186         (fix_trunc<mode><V_cvtto>)2): Likewise
3187         (fixuns_trunc<mode><V_cvtto)2): Likewise.
3188         (neon_vcvt<mode>): Likewise.
3189         (neon_vcvtv4sfv4hf): Likewise.
3190         (neon_vcvtv4hfv4sf): Likewise.
3191         (neon_vcvt_n<mode>): Likewise.
3192         (neon_vmovn<mode>): Likewise.
3193         (neon_vqmovn<mode>): Likewise.
3194         (neon_vqmovun<mode>): Likewise.
3195         (neon_vmovl<mode>): Likewise.
3196         (neon_vmul_lane<mode>): Likewise.
3197         (neon_vmull_lane<mode>): Likewise.
3198         (neon_vqdmull_lane<mode>): Likewise.
3199         (neon_vqdmulh_lane<mode>): Likewise.
3200         (neon_vmla_lane<mode>): Likewise.
3201         (neon_vmlal_lane<mode>): Likewise.
3202         (neon_vqdmlal_lane<mode>): Likewise.
3203         (neon_vmls_lane<mode>): Likewise.
3204         (neon_vmlsl_lane<mode>): Likewise.
3205         (neon_vqdmlsl_lane<mode>): Likewise.
3206         (neon_vext<mode>): Likewise.
3207         (neon_vrev64<mode>): Likewise.
3208         (neon_vrev32<mode>): Likewise.
3209         (neon_vrev16<mode>): Likewise.
3210         (neon_vbsl<mode>_internal): Likewise.
3211         (neon_vshl<mode>): Likewise.
3212         (neon_vqshl<mode>): Likewise.
3213         (neon_vshr_n<mode>): Likewise.
3214         (neon_vshrn_n<mode>): Likewise.
3215         (neon_vqshrn_n<mode>): Likewise.
3216         (neon_vqshrun_n<mode>): Likewise.
3217         (neon_vshl_n<mode>): Likewise.
3218         (neon_vqshl_n<mode>): Likewise.
3219         (neon_vqshlu_n<mode>): Likewise.
3220         (neon_vshll_n<mode>): Likewise.
3221         (neon_vsra_n<mode>): Likewise.
3222         (neon_vsri_n<mode>): Likewise.
3223         (neon_vsli_n<mode>): Likewise.
3224         (neon_vtbl1v8qi): Likewise.
3225         (neon_vtbl2v8qi): Likewise.
3226         (neon_vtbl3v8qi): Likewise.
3227         (neon_vtbl4v8qi): Likewise.
3228         (neon_vtbx1v8qi): Likewise.
3229         (neon_vtbx2v8qi): Likewise.
3230         (neon_vtbx3v8qi): Likewise.
3231         (neon_vtbx4v8qi): Likewise.
3232         (neon_vtrn<mode>_internal): Likewise.
3233         (neon_vzip<mode>_internal): Likewise.
3234         (neon_vuzp<mode>_internal): Likewise.
3235         (neon_vld1<mode>): Likewise.
3236         (neon_vld1_lane<mode>): Likewise.
3237         (neon_vld1_dup<mode>): Likewise.
3238         (neon_vld1_dupv2di): Likewise.
3239         (neon_vst1<mode>): Likewise.
3240         (neon_vst1_lane<mode>): Likewise.
3241         (neon_vld2<mode>): Likewise.
3242         (neon_vld2_lane<mode>): Likewise.
3243         (neon_vld2_dup<mode>): Likewise.
3244         (neon_vst2<mode>): Likewise.
3245         (neon_vst2_lane<mode>): Likewise.
3246         (neon_vld3<mode>): Likewise.
3247         (neon_vld3qa<mode>): Likewise.
3248         (neon_vld3qb<mode>): Likewise.
3249         (neon_vld3_lane<mode>): Likewise.
3250         (neon_vld3_dup<mode>): Likewise.
3251         (neon_vst3<mode>): Likewise.
3252         (neon_vst3qa<mode>): Likewise.
3253         (neon_vst3qb<mode>): Likewise.
3254         (neon_vst3_lane<mode>): Likewise.
3255         (neon_vld4<mode>): Likewise.
3256         (neon_vld4qa<mode>): Likewise.
3257         (neon_vld4qb<mode>): Likewise.
3258         (neon_vld4_lane<mode>): Likewise.
3259         (neon_vld4_dup<mode>): Likewise.
3260         (neon_vst4<mode>): Likewise.
3261         (neon_vst4qa<mode>): Likewise.
3262         (neon_vst4qb<mode>): Likewise.
3263         (neon_vst4_lane<mode>): Likewise.
3264         (neon_vec_unpack<US>_lo_<mode>): Likewise.
3265         (neon_vec_unpack<US>_hi_<mode>): Likewise.
3266         (neon_vec_<US>mult_lo_<mode>): Likewise.
3267         (neon_vec_<US>mult_hi_<mode>): Likewise.
3268         (neon_vec_<US>shiftl_<mode>): Likewise.
3269         (neon_unpack<US>_<mode>): Likewise.
3270         (neon_vec_<US>mult_<mode>): Likewise.
3271         (vec_pack_trunc_<mode>): Likewise.
3272         (neon_vec_pack_trunk_<mode>): Likewise.
3273         (neon_vabd<mode>_2): Likewise.
3274         (neon_vabd<mode>_3): Likewise.
3275         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3276         (thumb2_movsi_vfp): Likewise.
3277         (movdi_vfp): Likewise.
3278         (movdi_vfp_cortexa8): Likewise.
3279         (movhf_vfp_neon): Likewise.
3280         (movhf_vfp): Likewiwse.
3281         (movsf_vfp): Likewiwse.
3282         (thumb2_movsf_vfp): Likewiwse.
3283         (movdf_vfp): Likewise.
3284         (thumb2_movdf_vfp): Likewise.
3285         (movsfcc_vfp): Likewise.
3286         (thumb2_movsfcc_vfp): Likewise.
3287         (movdfcc_vfp): Likewise.
3288         (thumb2_movdfcc_vfp): Likewise.
3289         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
3290         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
3291         (v10_v2c): Likewise.
3292         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
3293         attribute change.
3294         (cortex_a15_neon_int_2): Likewise.
3295         (cortex_a15_neon_int_3): Likewise.
3296         (cortex_a15_neon_int_4): Likewise.
3297         (cortex_a15_neon_int_5): Likewise.
3298         (cortex_a15_neon_vqneg_vqabs): Likewise.
3299         (cortex_a15_neon_vmov): Likewise.
3300         (cortex_a15_neon_vaba): Likewise.
3301         (cortex_a15_neon_vaba_qqq): Likewise.
3302         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3303         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3304         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
3305         scalar_64_32_long_scalar): Likewise.
3306         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3307         (cortex_a15_neon_mla_qqq_8_16): Likewise.
3308         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3309         lotype_qdd_64_32_long): Likewise.
3310         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3311         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3312         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
3313         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3314         (cortex_a15_neon_shift_1): Likewise.
3315         (cortex_a15_neon_shift_2): Likewise.
3316         (cortex_a15_neon_shift_3): Likewise.
3317         (cortex_a15_neon_vshl_ddd): Likewise.
3318         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3319         (cortex_a15_neon_vsra_vrsra): Likewise.
3320         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
3321         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
3322         (cortex_a15_neon_fp_vmul_ddd): Likewise.
3323         (cortex_a15_neon_fp_vmul_qqd): Likewise.
3324         (cortex_a15_neon_fp_vmla_ddd): Likewise.
3325         (cortex_a15_neon_fp_vmla_qqq): Likewise.
3326         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3327         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3328         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3329         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3330         (cortex_a15_neon_bp_simple): Likewise.
3331         (cortex_a15_neon_bp_2cycle): Likewise.
3332         (cortex_a15_neon_bp_3cycle): Likewise.
3333         (cortex_a15_neon_vld1_1_2_regs): Likewise.
3334         (cortex_a15_neon_vld1_3_4_regs): Likewise.
3335         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3336         (cortex_a15_neon_vld2_4_regs): Likewise.
3337         (cortex_a15_neon_vld3_vld4): Likewise.
3338         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3339         (cortex_a15_neon_vst1_3_4_regs): Likewise.
3340         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
3341         (cortex_a15_neon_vst3_vst4): Likewise.
3342         (cortex_a15_neon_vld1_vld2_lane): Likewise.
3343         (cortex_a15_neon_vld3_vld4_lane" 10
3344         (cortex_a15_neon_vst1_vst2_lane): Likewise.
3345         (cortex_a15_neon_vst3_vst4_lane): Likewise.
3346         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3347         (cortex_a15_neon_ldm_2): Likewise.0
3348         (cortex_a15_neon_stm_2): Likewise.
3349         (cortex_a15_neon_mcr): Likewise.
3350         (cortex_a15_neon_mcr_2_mcrr): Likewise.
3351         (cortex_a15_neon_mrc): Likewise.
3352         (cortex_a15_neon_mrrc): Likewise.
3353         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3354         change.
3355         (cortex_a15_alu_shift): Likewise.
3356         (cortex_a15_alu_shift_reg): Likewise.
3357         (cortex_a15_mult32): Likewise.
3358         (cortex_a15_mult64): Likewise.
3359         (cortex_a15_block): Likewise.
3360         (cortex_a15_branch): Likewise.
3361         (cortex_a15_load1): Likewise.
3362         (cortex_a15_load3): Likewise.
3363         (cortex_a15_store1): Likewise.
3364         (cortex_a15_store3): Likewise.
3365         (cortex_a15_call): Likewise.
3366         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
3367         (cortex_a5_f2r): Likewise.
3368         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
3369         change.
3370         (cortex_a53_f2r): Likewise.
3371         * config/arm/cortex-a7.md
3372         (cortex_a7_branch): Update for attribute change.
3373         (cortex_a7_call): Likewise.
3374         (cortex_a7_alu_imm): Likewise.
3375         (cortex_a7_alu_reg): Likewise.
3376         (cortex_a7_alu_shift): Likewise.
3377         (cortex_a7_mul): Likewise.
3378         (cortex_a7_load1): Likewise.
3379         (cortex_a7_store1): Likewise.
3380         (cortex_a7_load2): Likewise.
3381         (cortex_a7_store2): Likewise.
3382         (cortex_a7_load3): Likewise.
3383         (cortex_a7_store3): Likewise.
3384         (cortex_a7_load4): Likewise.
3385         (cortex_a7_store4): Likewise.
3386         (cortex_a7_fpalu): Likewise.
3387         (cortex_a7_fconst): Likewise.
3388         (cortex_a7_fpmuls): Likewise.
3389         (cortex_a7_neon_mul): Likewise.
3390         (cortex_a7_fpmacs): Likewise.
3391         (cortex_a7_neon_mla: Likewise.
3392         (cortex_a7_fpmuld: Likewise.
3393         (cortex_a7_fpmacd: Likewise.
3394         (cortex_a7_fpfmad: Likewise.
3395         (cortex_a7_fdivs: Likewise.
3396         (cortex_a7_fdivd: Likewise.
3397         (cortex_a7_r2f: Likewise.
3398         (cortex_a7_f2r: Likewise.
3399         (cortex_a7_f_flags: Likewise.
3400         (cortex_a7_f_loads: Likewise.
3401         (cortex_a7_f_loadd: Likewise.
3402         (cortex_a7_f_stores: Likewise.
3403         (cortex_a7_f_stored: Likewise.
3404         (cortex_a7_neon): Likewise.
3405         * config/arm/cortex-a8-neon.md
3406         (cortex_a8_neon_mrc): Update for attribute change.
3407         (cortex_a8_neon_mrrc): Likewise.
3408         (cortex_a8_neon_int_1): Likewise.
3409         (cortex_a8_neon_int_2): Likewise.
3410         (cortex_a8_neon_int_3): Likewise.
3411         (cortex_a8_neon_int_4): Likewise.
3412         (cortex_a8_neon_int_5): Likewise.
3413         (cortex_a8_neon_vqneg_vqabs): Likewise.
3414         (cortex_a8_neon_vmov): Likewise.
3415         (cortex_a8_neon_vaba): Likewise.
3416         (cortex_a8_neon_vaba_qqq): Likewise.
3417         (cortex_a8_neon_vsma): Likewise.
3418         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3419         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3420         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
3421         long_scalar): Likewise.
3422         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3423         (cortex_a8_neon_mla_qqq_8_16): Likewise.
3424         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3425         long_scalar_qdd_64_32_long): Likewise.
3426         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3427         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3428         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
3429         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3430         (cortex_a8_neon_shift_1): Likewise.
3431         (cortex_a8_neon_shift_2): Likewise.
3432         (cortex_a8_neon_shift_3): Likewise.
3433         (cortex_a8_neon_vshl_ddd): Likewise.
3434         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3435         (cortex_a8_neon_vsra_vrsra): Likewise.
3436         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
3437         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
3438         (cortex_a8_neon_fp_vsum): Likewise.
3439         (cortex_a8_neon_fp_vmul_ddd): Likewise.
3440         (cortex_a8_neon_fp_vmul_qqd): Likewise.
3441         (cortex_a8_neon_fp_vmla_ddd): Likewise.
3442         (cortex_a8_neon_fp_vmla_qqq): Likewise.
3443         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
3444         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
3445         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3446         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3447         (cortex_a8_neon_bp_simple): Likewise.
3448         (cortex_a8_neon_bp_2cycle): Likewise.
3449         (cortex_a8_neon_bp_3cycle): Likewise.
3450         (cortex_a8_neon_ldr): Likewise.
3451         (cortex_a8_neon_str): Likewise.
3452         (cortex_a8_neon_vld1_1_2_regs): Likewise.
3453         (cortex_a8_neon_vld1_3_4_regs): Likewise.
3454         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3455         (cortex_a8_neon_vld2_4_regs): Likewise.
3456         (cortex_a8_neon_vld3_vld4): Likewise.
3457         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3458         (cortex_a8_neon_vst1_3_4_regs): Likewise.
3459         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
3460         (cortex_a8_neon_vst3_vst4): Likewise.
3461         (cortex_a8_neon_vld1_vld2_lane): Likewise.
3462         (cortex_a8_neon_vld3_vld4_lane): Likewise.
3463         (cortex_a8_neon_vst1_vst2_lane): Likewise.
3464         (cortex_a8_neon_vst3_vst4_lane): Likewise.
3465         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
3466         (cortex_a8_neon_mcr): Likewise.
3467         (cortex_a8_neon_mcr_2_mcrr): Likewise.
3468         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3469         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
3470         change.
3471         (ca9_neon_mrrc): Likewise.
3472         (cortex_a9_neon_int_1): Likewise.
3473         (cortex_a9_neon_int_2): Likewise.
3474         (cortex_a9_neon_int_3): Likewise.
3475         (cortex_a9_neon_int_4): Likewise.
3476         (cortex_a9_neon_int_5): Likewise.
3477         (cortex_a9_neon_vqneg_vqabs): Likewise.
3478         (cortex_a9_neon_vmov): Likewise.
3479         (cortex_a9_neon_vaba): Likewise.
3480         (cortex_a9_neon_vaba_qqq): Likewise.
3481         (cortex_a9_neon_vsma): Likewise.
3482         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3483         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3484         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
3485         long_scalar): Likewise.
3486         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3487         (cortex_a9_neon_mla_qqq_8_16): Likewise.
3488         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3489         long_scalar_qdd_64_32_long): Likewise.
3490         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3491         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3492         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3493         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3494         (cortex_a9_neon_shift_1): Likewise.
3495         (cortex_a9_neon_shift_2): Likewise.
3496         (cortex_a9_neon_shift_3): Likewise.
3497         (cortex_a9_neon_vshl_ddd): Likewise.
3498         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3499         (cortex_a9_neon_vsra_vrsra): Likewise.
3500         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3501         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3502         (cortex_a9_neon_fp_vsum): Likewise.
3503         (cortex_a9_neon_fp_vmul_ddd): Likewise.
3504         (cortex_a9_neon_fp_vmul_qqd): Likewise.
3505         (cortex_a9_neon_fp_vmla_ddd): Likewise.
3506         (cortex_a9_neon_fp_vmla_qqq): Likewise.
3507         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3508         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3509         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3510         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3511         (cortex_a9_neon_bp_simple): Likewise.
3512         (cortex_a9_neon_bp_2cycle): Likewise.
3513         (cortex_a9_neon_bp_3cycle): Likewise.
3514         (cortex_a9_neon_ldr): Likewise.
3515         (cortex_a9_neon_str): Likewise.
3516         (cortex_a9_neon_vld1_1_2_regs): Likewise.
3517         (cortex_a9_neon_vld1_3_4_regs): Likewise.
3518         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3519         (cortex_a9_neon_vld2_4_regs): Likewise.
3520         (cortex_a9_neon_vld3_vld4): Likewise.
3521         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3522         (cortex_a9_neon_vst1_3_4_regs): Likewise.
3523         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3524         (cortex_a9_neon_vst3_vst4): Likewise.
3525         (cortex_a9_neon_vld1_vld2_lane): Likewise.
3526         (cortex_a9_neon_vld3_vld4_lane): Likewise.
3527         (cortex_a9_neon_vst1_vst2_lane): Likewise.
3528         (cortex_a9_neon_vst3_vst4_lane): Likewise.
3529         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3530         (cortex_a9_neon_mcr): Likewise.
3531         (cortex_a9_neon_mcr_2_mcrr): Likewise.
3532         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3533         (cortex_a9_fps): Likewise.
3534         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
3535         change.
3536         (cortex_m4_fmuls): Likewise.
3537         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
3538         change.
3539         (cortex_r4_mrc): Likewise.
3540         * config/arm/iterators.md: Update comment referring to neon_type.
3541         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
3542         (iwmmxt_movsi_insn): Likewise.
3543         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
3544         attribute change.
3545         (pj4_core_to_vfp): Likewise.
3546         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
3547         attribute change.
3548         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
3549         (vfp_fstore): Likewise.
3550         * doc/md.texi: Change references to neon_type to refer to type.
3552 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
3554         * tree.h (DECL_BUILT_IN): Fix typo in comment.
3556 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
3558         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
3559         lglobl if not weak.
3561 2013-09-04  Easwaran Raman  <eraman@google.com>
3563         PR middle-end/57370
3564         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
3565         (build_and_add_sum): Use it.
3566         (appears_later_in_bb): Simplify code.
3568 2013-09-04  Teresa Johnson  <tejohnson@google.com>
3570         * dumpfile.c (dump_finish): Don't close stderr/stdout.
3572 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
3574         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
3576 2013-09-04  Jan Hubicka  <jh@suse.cz>
3578         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
3579         * ipa-devirt.c: Include diganostic.h
3580         (odr_type_d): Add types and types_set.
3581         (hash_type_name): Work for types with vtables during LTO.
3582         (odr_hasher::remove): Fix comment; destroy types_set.
3583         (add_type_duplicate): New function,
3584         (get_odr_type): Use it.
3585         (dump_type_inheritance_graph): Dump type duplicates.
3586         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
3587         graph.
3588         * tree.c (types_same_for_odr): Give exact answers on types with
3589         virtual tables.
3591 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
3593         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
3594         explaining their differences.
3596 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
3598         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
3600 2013-09-03  Jeff Law  <law@redhat.com>
3602         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
3603         when not threading through a joiner block.  Pass joiner/no joiner
3604         state to register_jump_thread.
3605         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
3606         state from argument rather than implying on path length.
3607         Dump the entire jump thread path into debugging dump.
3608         * tree-flow.h (register_jump_thread): Update prototype.
3610 2013-08-29  Xinliang David Li  <davidxl@google.com>
3612         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
3613         Remove a trivial gcc_assert.
3615 2013-08-29  Xinliang David Li  <davidxl@google.com>
3617         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
3618         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
3619         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
3620         Delay base decl alignment adjustment.
3621         * tree-vectorizer.c (vect_destroy_datarefs): New function.
3622         * tree-vectorizer.h: New data structure.
3623         (set_dr_misalignment): New function.
3624         (dr_misalignment): Ditto.
3625         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
3626         (vectorizable_load): Ditto.
3627         (ensure_base_align): New function.
3628         (vectorize_loops): Add dbg_cnt support.
3629         (execute_vect_slp): Ditto.
3630         * dbgcnt.def: New debug counter.
3631         * Makefile: New dependency.
3633 2013-09-03  Meador Inge  <meadori@codesourcery.com>
3635         Revert:
3637         2013-08-30  Meador Inge  <meadori@codesourcery.com>
3639         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3641 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
3643         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
3644         function descriptor.
3646 2013-09-03  Richard Biener  <rguenther@suse.de>
3648         * tree-affine.c (add_elt_to_tree): Fix association issue,
3649         avoid useless converts and make sure to always return a
3650         properly typed result.
3652 2013-09-03  Richard Biener  <rguenther@suse.de>
3654         PR middle-end/57656
3655         * fold-const.c (negate_expr_p): Fix division case.
3656         (negate_expr): Likewise.
3658 2013-09-03  Richard Biener  <rguenther@suse.de>
3660         PR lto/58285
3661         * tree-streamer-out.c: Include tm.h.
3662         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
3664 2013-09-03  Jan Hubicka  <jh@suse.cz>
3666         * tree-profile.c (tree_profiling): Cleanup CFG when done.
3668 2013-09-03  Alan Modra  <amodra@gmail.com>
3670         * config.gcc (powerpc*-*-linux*): Add support for little-endian
3671         multilibs to big-endian target and vice versa.
3672         * config/rs6000/t-linux64: Use := assignment on all vars.
3673         (MULTILIB_EXTRA_OPTS): Remove fPIC.
3674         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
3675         * config/rs6000/t-linux64le: New file.
3676         * config/rs6000/t-linux64bele: New file.
3677         * config/rs6000/t-linux64lebe: New file.
3679 2013-09-02  Jan Hubicka  <jh@suse.cz>
3681         * ipa-inline-transform.c (inline_transform): Do not
3682         optimize_inline_calls when not optimizing.
3684 2013-09-02  Jan Hubicka  <jh@suse.cz>
3686         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
3687         duplicated nodes for assembler names.
3688         * symtab.c (symtab_unregister_node): Do not attempt to unlink
3689         hard registers from assembler name hash.
3691 2013-09-02  Jan Hubicka  <jh@suse.cz>
3693         * ipa-split.c (execute_split_functions): Split externally visible
3694         functions called once.
3696 2013-09-02  Martin Jambor  <mjambor@suse.cz>
3698         PR ipa/58106
3699         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
3700         linked list.  When finding the correct duplicate, also consider also
3701         the caller in additon to its inlined_to node.
3703 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
3705         * config/aarch64/aarch64-simd-builtins.def
3706         (dup_lane_scalar): Remove.
3707         * config/aarch64/aarch64-simd.md
3708         (aarch64_simd_dup): Add 'w->w' alternative.
3709         (aarch64_dup_lane<mode>): Allow for VALL.
3710         (aarch64_dup_lane_scalar<mode>): Remove.
3711         (aarch64_dup_lane_<vswap_width_name><mode>): New.
3712         (aarch64_get_lane_signed<mode>): Add w->w altenative.
3713         (aarch64_get_lane_unsigned<mode>): Likewise.
3714         (aarch64_get_lane<mode>): Likewise.
3715         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
3716         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
3717         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
3718         (VCON): Change container of V2SF.
3719         (vswap_width_name): Likewise.
3720         * config/aarch64/arm_neon.h
3721         (__aarch64_vdup_lane_any): New.
3722         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
3723         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
3724         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
3726 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
3728         PR middle-end/56382
3729         * expr.c (emit_move_complex): Do not move complex FP values as parts if
3730         the source or the destination is a single hard register.
3732 2013-09-02  Richard Biener  <rguenther@suse.de>
3734         PR middle-end/57511
3735         * tree-scalar-evolution.c (instantiate_scev_name): Allow
3736         non-linear SCEVs.
3738 2013-09-02  Richard Biener  <rguenther@suse.de>
3740         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
3741         arithmetic to sizetype.
3743 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
3745         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
3746         Find auto-increment use both before and after candidate.
3748 2013-09-02  Marek Polacek  <polacek@redhat.com>
3750         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
3752 2013-09-01  Jan Hubicka  <jh@suse.cz>
3754         * Makefile.in: Add ipa-profile.o
3755         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
3756         * cgraph.c (struct cgraph_propagate_frequency_data,
3757         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
3758         ipa-profile.c; replace cgraph_ by ipa_ prefix.
3759         * cgraph.h (cgraph_propagate_frequency): Remove.
3760         * ipa-inline-analysis.c: Include ipa-utils.h;
3761         drop duplicated cfgloop.h.
3762         (inline_update_callee_summaries): Update.
3763         * ipa-profile.c: New file.
3764         * ipa-utils.h (ipa_propagate_frequency): Declare.
3765         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
3766         data-streamer.h, value-prof.h.
3767         (symtab_remove_unreachable_nodes): Update profile.
3768         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
3769         account_time_size, cmp_counts, dump_histogram,
3770         ipa_profile_generate_summary, ipa_profile_write_summary,
3771         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
3772         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
3773         Move to ipa-profile.c.
3775 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
3777         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
3779 2013-09-01  Jan Hubicka  <jh@suse.cz>
3781         * common.opt (fdevirtualize-speculatively): New function.
3782         * invoke.texi (fdevirtualize-speculatively): Document.
3783         * ipa-devirt.c: Include ipa-inline.h
3784         (likely_target_p): New function.
3785         (ipa_devirt): New function.
3786         (gate_ipa_devirt): New function.
3787         (pass_data_ipa_devirt): New static var.
3788         (pass_ipa_devirt): Likewise.
3789         (make_pass_ipa_devirt): New function.
3790         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
3791         (common_handle_option): Disable devirtualization when
3792         value range profiling is available.
3793         * passes.def (pass_ipa_devirt): Add.
3794         * timever.def (TV_IPA_DEVIRT): New timevar.
3795         * tree-pass.h (make_pass_ipa_devirt):
3797 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
3799         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
3800         include sanitize(undefined).
3802 2013-08-31  Diego Novillo  <dnovillo@google.com>
3804         * Makefile.in (TREE_CORE_H): Define.
3805         (TREE_H): Use.
3806         (GTFILES): Add tree-core.h.
3807         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
3808         size the array.
3809         * tree-core.h: New file.
3810         Move all data structures, enum, typedefs, global
3811         declarations and constants from ...
3812         * tree.h: ... here.
3814 2013-08-31  Jan Hubicka  <jh@suse.cz>
3816         * bulitins.c (expand_builtin): Do not early exit for gcov
3817         instrumented functions.
3819 2013-08-31  Marek Polacek  <polacek@redhat.com>
3821         * ubsan.c: Include tm_p.h.
3823 2013-08-31  Jan Hubicka  <jh@suse.cz>
3825         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
3826         warning.
3828         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
3829         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
3830         * tree-cfg.c (verify_gimple_label): ... here.
3831         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
3832         (ipa_merge_profiles): New function.
3833         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
3834         (lto_input_function_body): Likewise.
3835         * ipa-utils.h (ipa_merge_profiles): Declare.
3836         * lto-streamer.h (lto_input_function_body): Update prototype.
3837         (emit_label_in_global_context_p): Remove.
3838         * lto-symtab.c: Include ipa-utils.h
3839         (lto_cgraph_replace_node): Use ipa_merge_profiles.
3841 2013-08-31  Jan Hubicka  <jh@suse.cz>
3843         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
3845 2013-08-31  Jan Hubicka  <jh@suse.cz>
3847         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
3849 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
3851         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
3852         "cmp" RTX before signed_comparison_operator check to account
3853         for "code" changes.
3855 2013-08-30  Jan Hubicka  <jh@suse.cz>
3857         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
3858         (detect_type_change_1): Rename to ...
3859         (detect_type_change): ... this one; early return on non-polymorphic
3860         types.
3861         (detect_type_change_ssa): Add comp_type parameter; update
3862         use of detect_type_change.
3863         (compute_complex_assign_jump_func): Add param_type parameter;
3864         update use of detect_type_change_ssa.
3865         (compute_complex_ancestor_jump_func): Likewise.
3866         (ipa_get_callee_param_type): New function.
3867         (ipa_compute_jump_functions_for_edge): Compute parameter type;
3868         update calls to the jump function computation functions.
3870 2013-08-30  Teresa Johnson  <tejohnson@google.com>
3871             Steven Bosscher  <steven@gcc.gnu.org>
3873         * cfgrtl.c (fixup_new_cold_bb): New routine.
3874         (commit_edge_insertions): Invoke fixup_partitions.
3875         (find_partition_fixes): New routine.
3876         (fixup_partitions): Ditto.
3877         (verify_hot_cold_block_grouping): Update comments.
3878         (rtl_verify_edges): Invoke find_partition_fixes.
3879         (rtl_verify_bb_pointers): Update comments.
3880         (rtl_verify_bb_layout): Ditto.
3881         * basic-block.h (probably_never_executed_edge_p): Declare.
3882         (fixup_partitions): Ditto.
3883         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
3884         * bb-reorder.c (sanitize_hot_paths): New function.
3885         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
3886         sanitize_hot_paths.
3887         * predict.c (probably_never_executed_edge_p): New routine.
3888         * cfg.c (check_bb_profile): Add partition insanity warnings.
3890 2013-08-30  Meador Inge  <meadori@codesourcery.com>
3892         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3894 2013-08-30  Marek Polacek  <polacek@redhat.com>
3896         * Makefile.in (ubsan.o): Add.
3897         (c-family/c-ubsan.o): Add.
3898         (builtins.o): Add ubsan.h dependency.
3899         * ubsan.h: New file.
3900         * ubsan.c: New file.
3901         * common.opt: Add -fsanitize=undefined option.
3902         (flag_sanitize): Add variable.
3903         (fsanitize=): Add option.  Add Driver.
3904         (fsanitize=thread): Remove option.
3905         (fsanitize=address): Likewise.
3906         (static-libubsan): New option.
3907         * doc/invoke.texi: Document the new flag and -static-libubsan.
3908         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
3909         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
3910         * builtin-attrs.def (ATTR_COLD): Define.
3911         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3912         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
3913         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
3914         * flag-types.h (sanitize_code): New enum.
3915         * opts.c (common_handle_option): Parse command line arguments
3916         of -fsanitize=.  Add -fsanitize=unreachable option.
3917         * varasm.c (get_variable_section): Adjust.
3918         (assemble_noswitch_variable): Likewise.
3919         (assemble_variable): Likewise.
3920         (output_constant_def_contents): Likewise.
3921         (categorize_decl_for_section): Likewise.
3922         (place_block_symbol): Likewise.
3923         (output_object_block): Likewise.
3924         * builtins.def: Likewise.
3925         * toplev.c (compile_file): Likewise.
3926         (process_options): Likewise.
3927         * cppbuiltin.c: Likewise.
3928         * tsan.c (tsan_pass): Likewise.
3929         (tsan_gate): Likewise.
3930         (tsan_gate_O0): Likewise.
3931         * cfgexpand.c (partition_stack_vars): Likewise.
3932         (expand_stack_vars): Likewise.
3933         (defer_stack_allocation): Likewise.
3934         (expand_used_vars): Likewise.
3935         * cfgcleanup.c (old_insns_match_p): Likewise.
3936         * asan.c (asan_finish_file): Likewise.
3937         (asan_instrument): Likewise.
3938         (gate_asan): Likewise.
3939         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
3940         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3941         (asan_global_struct): Use pointer_sized_int_node instead
3942         calling build_nonstandard_integer_type.
3943         (initialize_sanitizer_builtins): Likewise.
3944         (asan_finish_file): Likewise.
3945         * gcc.c: Document %{%:function(args):X}.
3946         (static_spec_functions): Add sanitize.
3947         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
3948         store funcval != NULL there.
3949         (do_spec_1): Adjust handle_spec_function caller.
3950         (handle_braces): Allow %:function(args) as condition.
3951         (sanitize_spec_function): New function.
3952         (ADD_STATIC_LIBUBSAN_LIBS): Define.
3953         (LIBUBSAN_SPEC): Likewise.
3954         (LIBUBSAN_EARLY_SPEC): Likewise.
3955         (SANITIZER_SPEC): Handle libubsan.
3956         (SANITIZER_EARLY_SPEC): Likewise.
3957         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
3958         instead of fsanitize=address.
3959         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
3960         instead of fsanitize=address*.
3961         * builtins.c: Include ubsan.h.
3962         (fold_builtin_0): Instrument __builtin_unreachable.
3963         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
3964         instead of flag_asan.
3965         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
3966         (pointer_sized_int_node): Define.
3967         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
3969 2013-08-30  Mike Stump  <mikestump@comcast.net>
3971         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
3972         with RE patterns.
3974 2013-08-29  Jan Hubicka  <jh@suse.cz>
3976         * cgraph.c (cgraph_function_body_availability): Handle weakref
3977         correctly.
3978         * passes.def: Remove pass_fixup_cfg.
3979         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
3980         track when we need to remove functions.
3981         (gate_ipa_inline): Execute inlining always; add comment why.
3982         (pass_data_ipa_inline): Remove TODO_remove_functions.
3983         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
3984         do not produce summaries.
3985         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
3986         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
3987         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
3988         constant pool and vtable.
3990 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
3992         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
3993         New arm_neon.h's internal macros to specify 64-bit constants.
3994         Avoid using stdint.h's macros.
3996 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
3998         * recog.c (verify_changes): Verify that changes[i].old is non-zero
3999         before applying REG_P.
4001 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
4003         PR tree-optimization/58277
4004         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
4005         after seeing too many stmts with vdef in between dombb and current
4006         bb, invalidate everything.
4008 2013-08-30  Richard Biener  <rguenther@suse.de>
4010         * fold-const.c (fold_single_bit_test): Fix overflow test.
4012 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
4014         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
4015         and which can live in a register, always retrieve the value on entry.
4016         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
4017         passed by invisible reference specially.
4018         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
4019         (vt_add_function_parameter): Correctly deal with a parameter passed by
4020         invisible reference.
4022 2013-08-30  Jan Hubicka  <jh@suse.cz>
4024         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
4026 2013-08-30  Richard Biener  <rguenther@suse.de>
4028         PR tree-optimization/58228
4029         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
4030         allow invariant loads in nested loop vectorization.
4032 2013-08-30  Richard Biener  <rguenther@suse.de>
4034         PR tree-optimization/58223
4035         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
4036         (has_anti_or_output_dependence): ... this and adjust to also
4037         look for output dependences.
4038         (mark_nodes_having_upstream_mem_writes): Adjust.
4039         (rdg_flag_uses): Likewise.
4041 2013-08-30  Richard Biener  <rguenther@suse.de>
4043         PR tree-optimization/58010
4044         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
4045         assert that we have a loop-closed PHI.
4047 2013-08-29  Jan Hubicka  <jh@suse.cz>
4049         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
4050         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
4051         * lto-section-in.c (lto_free_function_in_decl_state): New function.
4052         (lto_free_function_in_decl_state_for_node): New function.
4054 2013-08-29  Xinliang David Li  <davidxl@google.com>
4056         * loop-unroll.c (report_unroll_peel): Minor message change.
4057         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
4058         Emit alignment peeling message with default -fopt-info.
4059         (vect_loop_versioning): Emit loop version info message.
4060         * tree-vectorizer.c (vectorize_loops): Minor message change.
4061         (execute_vect_slp): Ditto.
4063 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
4065         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
4066         of the clone from the DECL_NAME of the original function.
4068 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
4070         * passes.c (register_pass): Add overload.
4071         * tree-pass.h (register_pass): Forward declare it.  Add comment.
4073 2013-08-29  Jan Hubicka  <jh@suse.cz>
4075         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
4076         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
4077         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
4078         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
4079         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
4080         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
4081         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
4082         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
4084 2013-08-29  Teresa Johnson  <tejohnson@google.com>
4086         * dumpfile.c (dump_loc): Output column number.
4087         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
4088         * doc/invoke.texi: Document optall -fopt-info flag.
4089         * profile.c (read_profile_edge_counts): Use new dump framework.
4090         (compute_branch_probabilities): Ditto.
4091         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
4092         when pass not in any opt group.
4093         * pass_manager.h (pass_manager::get_pass_profile): New method.
4094         * value-prof.c (check_counter): Use new dump framework.
4095         (check_ic_target): Ditto.
4096         * coverage.c (get_coverage_counts): Ditto.
4097         (coverage_init): Setup new dump framework.
4099 2013-08-29  Richard Biener  <rguenther@suse.de>
4101         PR tree-optimization/58246
4102         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
4103         handle the dominance check inside a basic-block.
4105 2013-08-29  Richard Biener  <rguenther@suse.de>
4107         PR middle-end/57287
4108         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
4109         of default defs that appear in abnormal PHI nodes.
4111 2013-08-29  Richard Biener  <rguenther@suse.de>
4113         PR tree-optimization/57685
4114         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
4115         single-use operands to avoid exponential complexity.
4117 2013-08-28  Dehao Chen  <dehao@google.com>
4119         * ipa-inline.c (edge_badness): Fix integer underflow.
4121 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
4123         * gtm-builtins.def (_ITM_free): Declare leaf.
4125 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
4127         PR target/58067
4128         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
4129         (*tls_local_dynamic_base_64_largepic): Likewise.
4130         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
4131         Remove predicate from call operand.
4132         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
4133         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
4135 2013-08-28  Jeff Law  <law@redhat.com>
4137         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
4138         checks for the number of predecessors and successors allowed.
4139         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
4140         which require copying a joiner block if there is a request which
4141         is a subpath that requires no joiner block copying.
4143 2013-08-28  Jan Hubicka  <jh@suse.cz>
4145         * lto-streamer-out.c (DFS_write_tree_body): Drop
4146         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
4147         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
4148         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
4149         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
4150         Do not read DECL_ERROR_ISSUED.
4151         (unpack_ts_decl_with_vis_value_fields): Do not read
4152         DECL_DEFER_OUTPUT.
4153         (lto_input_ts_binfo_tree_pointers): Do not read
4154         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
4155         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
4156         write DECL_ERROR_ISSUED..
4157         (pack_ts_decl_with_vis_value_fields): Do not write
4158         DECL_DEFER_OUTPUT.
4159         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
4160         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
4161         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
4162         * tree.h (tree_decl_common): Update comment.
4163         (DECL_ERROR_ISSUED): Remove.
4165 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
4167         PR middle-end/58257
4168         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
4170 2013-08-28  Jan Hubicka  <jh@suse.cz>
4172         * builtins.def (free): Declare leaf.
4174 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
4176         * gdbhooks.py: New.
4177         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
4178         * configure: Regenerate.
4180 2013-08-27  Martin Jambor  <mjambor@suse.cz>
4182         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
4183         (ipa_ancestor_jf_data): Likewise.
4184         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
4185         (ipa_get_jf_pass_through_type_preserved): New function.
4186         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
4187         (ipa_get_jf_ancestor_type_preserved): New function.
4188         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
4189         (ipa_get_jf_ancestor_result): Likewise.
4190         (propagate_vals_accross_pass_through): Use
4191         ipa_get_jf_pass_through_result to do all the value mappings.
4192         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
4193         type_preserved flag.
4194         (ipa_set_jf_cst_copy): New function.
4195         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
4196         (ipa_set_jf_arith_pass_through): Likewise.
4197         (ipa_set_ancestor_jf): Likewise.
4198         (compute_complex_assign_jump_func): Set type_preserved instead of
4199         punting.
4200         (ipa_compute_jump_functions_for_edge): Likewise.
4201         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
4202         (update_jump_functions_after_inlining): Update type_preserved.
4203         Explicitely create jump functions when combining one with pass_through.
4204         (ipa_write_jump_function): Stream the type_preserved flags.
4205         (ipa_read_jump_function): Likewise.
4207 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
4208             Aldy Hernandez  <aldyh@redhat.com>
4210         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
4211         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
4212         * function.h (struct function): Add has_force_vect_loops and
4213         has_simduid_loops.
4214         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
4215         * gimple.c (gimple_build_omp_critical): Add KIND argument and
4216         handle it.
4217         * gimple.def: Update CLAUSES comments.
4218         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
4219         (gimple_build_omp_for): Add argument to prototype.
4220         (gimple_omp_for_kind): New.
4221         (gimple_omp_for_set_kind): New.
4222         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
4223         GOVD_DATA_SHARE_CLASS.
4224         (enum omp_region_type): Add ORT_SIMD.
4225         (gimple_add_tmp_var): Handle ORT_SIMD.
4226         (gimplify_var_or_parm_decl): Same.
4227         (is_gimple_stmt): Same.
4228         (omp_firstprivatize_variable): Same.
4229         (omp_add_variable): Only use splay_tree_insert if lookup failed.
4230         (omp_notice_variable): Handle ORT_SIMD.
4231         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
4232         (omp_check_private): Handle ORT_SIMD.
4233         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
4234         OMP_CLAUSE_SAFELEN.
4235         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
4236         Handle OMP_CLAUSE_LASTPRIVATE.
4237         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
4238         OMP_CLAUSE_SAFELEN.
4239         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
4240         (gimplify_expr): Handle OMP_SIMD.
4241         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
4242         (expand_GOMP_SIMD_VF): New.
4243         (expand_GOMP_SIMD_LAST_LANE): New.
4244         * internal-fn.def (GOMP_SIMD_LANE): New.
4245         (GOMP_SIMD_VF): New.
4246         (GOMP_SIMD_LAST_LANE): New.
4247         * omp-low.c: Include target.h.
4248         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
4249         OMP_CLAUSE_SAFELEN.
4250         (check_omp_nesting_restrictions): Same.
4251         (omp_max_vf): New.
4252         (lower_rec_simd_input_clauses): New.
4253         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
4254         OMP_CLAUSE_LINEAR.
4255         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
4256         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
4257         (expand_omp_build_assign): New.
4258         (expand_omp_for_init_counts): New.
4259         (expand_omp_for_init_vars): New.
4260         (extract_omp_for_update_vars): New.
4261         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
4262         and rewrite accordingly.
4263         (expand_omp_simd): New.
4264         (expand_omp_for): Use expand_omp_simd.
4265         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
4266         (lower_omp_for): Do not lower the body.
4267         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
4268         in their own loops.
4269         * tree-flow.h (find_omp_clause): Remove prototype.
4270         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
4271         forcing vectorization of the loop, or if flag_tree_vectorize.
4272         (gate_tree_if_conversion): Similarly.
4273         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
4274         gimple_build_omp_for.
4275         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
4276         * tree-parloops (create_parallel_loop): Pass kind argument to
4277         gimple_build_omp_for.
4278         * tree-pretty-print.c (dump_omp_clause): Add cases for
4279         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
4280         OMP_CLAUSE__SIMDUID_.
4281         (dump_generic_node): Handle OMP_SIMD.
4282         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
4283         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
4284         unroll OMP_SIMD loops here.
4285         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
4286         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
4287         loop->safelen.
4288         (vect_analyze_data_refs): Handle simd loops.
4289         * tree-vect-loop.c (vectorizable_live_operation): Handle
4290         IFN_GOMP_SIMD*.
4291         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
4292         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
4293         (vectorizable_load): Same.
4294         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
4295         (struct simduid_to_vf): New.
4296         (simduid_to_vf::hash): New.
4297         (simduid_to-vf::equal): New.
4298         (struct simd_array_to_simduid): New.
4299         (simd_array_to_simduid::hash): New.
4300         (simd_array_to_simduid::equal): New.
4301         (adjust_simduid_builtins): New.
4302         (struct note_simd_array_uses_struct): New.
4303         (note_simd_array_uses_cb): New.
4304         (note_simd_array_uses): New.
4305         (vectorize_loops): Handle simd hints and adjust simd builtins
4306         accordingly.
4307         * tree-vectorizer.h (struct _stmt_vec_info): Add
4308         simd_lane_access_p field.
4309         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
4310         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
4311         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
4312         (omp_clause_code_name): Same.
4313         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
4314         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
4315         * tree.def (OMP_SIMD): New entry.
4316         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
4317         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
4318         (OMP_CLAUSE_DECL): Adjust range for new clauses.
4319         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
4320         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
4321         (OMP_CLAUSE_LINEAR_STEP): New.
4322         (OMP_CLAUSE_SAFELEN_EXPR): New.
4323         (OMP_CLAUSE__SIMDUID__DECL): New.
4324         (find_omp_clause): New prototype.
4326 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
4328         * config/i386/driver-i386.c (host_detect_local_cpu): Update
4329         Haswell processor detection.
4331 2013-08-27  Christian Widmer  <shadow@umbrox.de>
4333         PR target/57927
4334         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
4335         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
4336         AVX2 capable processors.
4338 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
4340         * config/aarch64/arm_neon.h: Replace all inline asm implementations
4341         of vget_low_* with implementations in terms of other intrinsics.
4343 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
4345         PR middle-end/57219
4346         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
4347         values to -1, 0 and 1.
4349 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
4351         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
4352         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
4353         (<optab><mode>3_insn): Remove.
4354         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
4355         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
4356         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
4357         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
4358         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
4359         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
4360         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
4361         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
4362         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
4363         (ror<mode>3_insn): Likewise.
4364         * config/aarch64/predicates.md (aarch64_simd_register): New.
4366 2013-08-27  Richard Biener  <rguenther@suse.de>
4368         PR tree-optimization/57521
4369         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
4370         one edge is non-critical.
4371         (find_phi_replacement_condition): Make sure to use a non-critical
4372         edge.  Cleanup and remove old bug workarounds.
4373         (bb_postdominates_preds): Remove.
4374         (if_convertible_loop_p_1): Do not compute post-dominators.
4375         (combine_blocks): Do not free post-dominators.
4376         (main_tree_if_conversion): Likewise.
4377         (pass_data_if_conversion): Add TODO_verify_ssa.
4379 2013-08-27  DJ Delorie  <dj@redhat.com>
4381         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
4383 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4385         * function.c (assign_parm_find_data_types): Set passed_mode and
4386         nominal_mode to the TYPE_MODE of nominal_type for the built
4387         pointer type in case of the struct-pass-by-reference.
4389 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
4391         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
4392         (UINT16_TYPE): Change default to "unsigned int".
4394         * config/avr/avr.opt (mfract-convert-truncate): New option.
4395         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
4396         is set, round negative fractional integers according to n1169
4397         when converting to integer types.
4399 2013-08-26  Jan Hubicka  <jh@suse.cz>
4401         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
4402         methods can not be called indirectly when their address is not taken.
4404 2013-08-26  Jan Hubicka  <jh@suse.cz>
4406         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
4407         ctor_for_folding.
4409 2013-08-26  Jan Hubicka  <jh@suse.cz>
4411         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
4412         can be unshared.
4414 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
4416         * reload.c (find_valid_class): Allow classes that do not include
4417         FIRST_PSEUDO_REGISTER - 1.
4419 2013-08-26  Jan Hubicka  <jh@suse.cz>
4421         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
4422         fix edge count/frequency when speculation failed; fix type check
4423         for the direct call.
4425 2013-08-26  Jan Hubicka  <jh@suse.cz>
4427         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
4429 2013-08-26  Jan Hubicka  <jh@suse.cz>
4431         * ipa-inline-transform.c (inline_transform): Be ready for basic block
4432         to be changed by edge redirection.
4434 2013-08-26  Jan Hubicka  <jh@suse.cz>
4436         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
4437         formating; add sanity check.
4438         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
4439         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
4440         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
4441         (resolve_noninline_speculation): Update callee keys, too.
4443 2013-08-26  Jan Hubicka  <jh@suse.cz>
4445         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
4446         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
4448 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
4450         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
4451         into proper place.
4453 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
4455         * config/i386/i386.c (ix86_debug_options): Remove prototype.
4456         (x86_64_elf_select_section): Ditto.
4457         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
4458         arguments.
4459         (ix86_pass_by_reference): Ditto.
4460         (output_set_got): Ditto.
4461         (ix86_unary_operator_ok): Ditto.
4462         (ix86_expand_builtin): Ditto.
4464 2013-08-23  Jan Hubicka  <jh@suse.cz>
4466         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
4468 2013-08-23  Jan Hubicka  <jh@suse.cz>
4470         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
4471         (tree_decl_with_vis): Add FINAL field.
4473 2013-08-23  Jeff Law  <law@redhat.com>
4475         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
4476         the debugging dump when the expression is fully redundant.
4478 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4480         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
4481         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
4482         * pretty-print.c (pp_formatted_text_data): Likewise.
4483         (pp_write_text_to_stream): Likewise.
4484         (pp_write_text_as_dot_label_to_stream): Likewise.
4485         (pp_append_r): Likewise.
4486         (pp_format): Likewise.
4487         (pp_flush): Likewise.
4488         (pp_clear_output_area): Likewise.
4489         (pp_append_text): Likewise.
4490         (pp_formatted_text): Likewise.
4491         (pp_remaining_character_count_for_line): Likewise.
4492         (pp_newline): Likewise.
4493         (pp_character): Likewise.
4494         (output_buffer::~output_buffer): Define.
4495         (pretty_printer::~pretty_printer): Destruct output buffer.
4496         * pretty-print.h (output_buffer::~output_buffer): Declare.
4497         (pretty_printer::~pretty_printer): Declare virtual.
4499 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
4501         PR other/57324
4502         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
4503         HOST_WIDE_INT_M1U): New macros.
4504         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
4505         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
4506         unsigned -1 for lshift.
4507         * cse.c (cse_insn): Likewise.
4508         * double-int.c (rshift_double, lshift_double): Likewise.
4509         * builtins.c (fold_builtin_bitop): Likewise.
4510         * combine.c (force_to_mode): Likewise.
4511         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
4512         * simplify-rtx.c (simplify_const_unary_operation,
4513         simplify_const_binary_operation): Likewise.
4514         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
4515         check_va_list_escapes): Likewise.
4516         * rtlanal.c (nonzero_bits1): Likewise.
4517         * expmed.c (expand_smod_pow2): Likewise.
4518         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
4520 2013-08-23  Jan Hubicka  <jh@suse.cz>
4522         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
4523         as having address taken.
4525 2013-08-23  Jan Hubicka  <jh@suse.cz>
4527         * ipa-utils.h (method_class_type): Declare.
4528         * ipa-devirt.c (method_class_type): Export.
4530         * cgraphunit.c (analyze_functions): Do basic devirtualization;
4531         do not walk base classes of anonymous types.
4533 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
4535         PR rtl-optimization/58220
4536         PR regression/58221
4537         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
4538         handle SEQUENCE insns properly.
4540 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4542         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
4543         definition.
4544         (pp_newline_and_indent): Likewise.
4545         (pp_separate_with): Likewise.
4546         * pretty-print.c (pp_newline_and_flush): Define.
4547         (pp_newline_and_indent): Likewise.
4548         (pp_separate_with): Likewise.
4550 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
4552         PR target/58218
4553         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
4554         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
4556 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
4558         * config/i386/predicates.md (ext_sse_reg_operand): New.
4559         * config/i386/i386.md (*movti_internal): Use
4560         predicate to determine if EVEX is needed.
4561         (*movsi_internal): Ditto.
4562         (*movdf_internal): Ditto.
4563         (*movsf_internal): Ditto.
4564         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
4566 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
4568         PR tree-optimization/58209
4569         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
4570         (find_tail_calls): Give up for pointer result types if m is non-NULL.
4571         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
4572         emit POINTER_PLUS_EXPR.
4573         (create_tailcall_accumulator): For pointer result type accumulate in
4574         sizetype type.
4576 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
4578         * configure.ac: Add backslashes missing from the last change.
4579         * configure: Regenerate.
4581 2013-08-22  Jan Hubicka  <jh@susue.cz>
4583         * ipa.c (function_and_variable_visibility): First remember function
4584         was global and then make it local.
4586 2013-08-22  Julian Brown  <julian@codesourcery.com>
4588         * configure.ac: Add aarch64 to list of arches which use "nop" in
4589         debug_line test.
4590         * configure: Regenerate.
4592 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4594         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
4595         gnu_libc_has_function.
4596         * config/s390/tpf.h: Likewise.
4598 2013-08-22  Jan Hubicka  <jh@susue.cz>
4600         * timevar.c (validate_phases): Add cast.
4602 2013-08-22  Jan Hubicka  <jh@susue.cz>
4604         * timevar.c (validate_phases): Use size_t for memory.
4605         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
4607 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4609         * pretty-print.h (output_buffer::output_buffer): Declare.
4610         (pretty_printer::pretty_printer): Likewise.
4611         (pp_construct): Remove.
4612         * pretty-print.c (output_buffer::output_buffer): Define.
4613         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
4614         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
4615         (print_gimple_expr): Likewise.
4616         (print_gimple_seq): Likewise.
4617         (gimple_dump_bb): Likewise.
4618         * sched-vis.c (dump_value_slim): Likewise.
4619         (dump_insn_slim): Likewise.
4620         (dump_rtl_slim): Likewise.
4621         (str_pattern_slim): Likewise.
4622         * tree-mudflap.c (mf_varname_tree): Likewise.
4623         * graph.c (print_graph_cfg): Likewise.
4624         (start_graph_dump): Likewise.
4625         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
4626         placement-new.
4627         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
4628         pretty printer initialization.
4629         * coretypes.h (diagnostic_context): Remove superflous type alias
4630         declaration.
4631         (pretty_printer): Likewise.  Declare directly as a class.
4632         (pretty_print_info): Remove declaration as class.
4633         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
4634         and pp_clear_output_area.
4635         (asan_add_global): Likewise.
4637 2013-08-22  Jan Hubicka  <jh@suse.cz>
4639         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
4640         * ipa-utils.h (update_type_inheritance_graph): Declare.
4641         (possible_polymorphic_call_target_p): Declare.
4642         (possible_polymorphic_call_target_p): New.
4643         * ipa-devirt.c: Update toplevel comments.
4644         (cached_polymorphic_call_targets): Move up.
4645         (odr_type_d): Move ID down.
4646         (polymorphic_type_binfo_p): Update comment.
4647         (odr_hasher::remove): Likewise;
4648         (get_odr_type): Set anonymous_namespace.
4649         (dump_odr_type): Dump it.
4650         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
4651         (maybe_record_node): Record node in cached_polymorphic_call_targets.
4652         (record_binfo): Add comment.
4653         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
4654         built.
4655         (devirt_node_removal_hook): Do not iCE when cache is freed.
4656         (possible_polymorphic_call_target_p): New predicate.
4657         (update_type_inheritance_graph): New function.
4659 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4660             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4661             Sergey Lega  <sergey.s.lega@intel.com>
4662             Anna Tikhonova  <anna.tikhonova@intel.com>
4663             Ilya Tocar  <ilya.tocar@intel.com>
4664             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4665             Ilya Verbin  <ilya.verbin@intel.com>
4666             Kirill Yukhin  <kirill.yukhin@intel.com>
4667             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4669         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
4670         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
4671         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
4672         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
4673         (OPTION_MASK_ISA_AVX2_UNSET): Update.
4674         (OPTION_MASK_ISA_AVX512F_UNSET): New.
4675         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
4676         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
4677         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
4678         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
4679         OPT_mavx512pf, OPT_mavx512er cases.
4680         * config/i386/constraints.md (v): New constraint.
4681         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
4682         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
4683         (bit_AVX512CD): New.
4684         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4685         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
4686         * config/i386/i386-c.c (ix86_target_macros_internal):
4687         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
4688         __AVX512PF__.
4689         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
4690         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
4691         * config/i386/i386.c (regclass_map, dbx_register_map)
4692         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
4693         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
4694         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
4695         -mavx512pf options.
4696         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
4697         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
4698         -mavx512pf options.  Fix formatting.
4699         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
4700         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
4701         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
4702         -mavx512cd, -mavx512pf options.
4703         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
4704         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
4705         (ix86_preferred_output_reload_class): Replace SSE_REGS with
4706         ALL_SSE_REGS.
4707         (ix86_hard_regno_mode_ok): Support 512-bit registers.
4708         (ix86_set_reg_reg_cost): Ditto.
4709         (x86_order_regs_for_local_alloc): Ditto.
4710         (MAX_VECT_LEN): Extend to 64-byte.
4711         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
4712         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
4713         (TARGET_AVX512ER, TARGET_AVX512CD): New.
4714         (BIGGEST_ALIGNMENT): Extend to 512-bits.
4715         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
4716         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
4717         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
4718         (SSE_REG_MODE_P): Support new modes.
4719         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
4720         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
4721         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
4722         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
4723         (REG_CLASS_CONTENTS): Add new registers.
4724         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
4725         (EXT_REX_SSE_REGNO_P): New.
4726         (HI_REGISTER_NAMES): Add new registers.
4727         * config/i386/i386.md: Define constants for new registers.
4728         (mode): Add new 512-bit modes.
4729         (prefix): Support evex prefix.
4730         (isa): Support avx512f, noavx512f, fma_avx512f.
4731         (ssemodesuffix): Add new 512-bit modes.
4732         (movxi): New.
4733         (*movxi_internal_avx512f): Ditto.
4734         (*movdi_internal): Replace constraint "x" with the new constraint "v".
4735         Support MODE_XI.
4736         (*movsi_internal): Likewise.
4737         (*movdf_internal): Likewise.
4738         (*movsf_internal): Likewise.
4739         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
4740         (<code><mode>3): Likewise.
4741         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
4742         New.
4743         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
4744         with the new constraint "v".
4745         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
4746         modes.
4747         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
4748         with the new constraint "v".
4749         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
4750         (<sse2>_storedqu<avxsizesuffix>): Likewise.
4751         (*<plusminus_insn><mode>3): Likewise.
4752         (<sse>_vm<plusminus_insn><mode>3): Likewise.
4753         (*mul<mode>3): Likewise.
4754         (<sse>_vmmul<mode>3): Likewise.
4755         (<sse>_div<mode>3): Likewise.
4756         (<sse>_vmdiv<mode>3): Likewise.
4757         (<sse>_sqrt<mode>2): Likewise.
4758         (<sse>_vmsqrt<mode>2): Likewise.
4759         (*<code><mode>3_finite): Likewise.
4760         (*<code><mode>3) <smaxmin>: Likewise.
4761         (<sse>_vm<code><mode>3): Likewise.
4762         (*<code><mode>3) <any_logic>: Likewise.
4763         (*fma_fmadd_<mode>): Likewise.
4764         (*fma_fmsub_<mode>): Likewise.
4765         (*fma_fnmadd_<mode>): Likewise.
4766         (*fma_fnmsub_<mode>): Likewise.
4767         (*fma_fmaddsub_<mode>): Likewise.
4768         (*fma_fmsubadd_<mode>): Likewise.
4769         (*fmai_fmadd_<mode>): Likewise.
4770         (*fmai_fmsub_<mode>): Likewise.
4771         (*fmai_fnmadd_<mode>): Likewise.
4772         (*fmai_fnmsub_<mode>): Likewise.
4773         (sse_cvtsi2ss): Likewise.
4774         (sse_cvtsi2ssq): Likewise.
4775         (sse_cvtss2si): Likewise.
4776         (sse_cvtss2si_2): Likewise.
4777         (sse_cvtss2siq): Likewise.
4778         (sse_cvtss2siq_2): Likewise.
4779         (sse_cvttss2si): Likewise.
4780         (sse_cvtss2siq_2): Likewise.
4781         (float<sseintvecmodelower><mode>2): Likewise.
4782         (sse2_cvtsd2si_2): Likewise.
4783         (sse2_cvtsd2siq_2): Likewise.
4784         (*<plusminus_insn><mode>3): Likewise.
4785         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
4786         (*<sse4_1_avx2>_mul<mode>3): Likewise.
4787         (ashr<mode>3): Likewise.
4788         (<shift_insn><mode>3): Likewise.
4789         (avx2_<code><mode>3): Likewise.
4790         (*avx2_<code><mode>3): Likewise.
4791         (*andnot<mode>3): Likewise.
4792         (*<code><mode>3) <any_logic>: Likewise.
4793         (abs<mode>2): Likewise.
4794         (avx2_permvar<mode>): Likewise.
4795         (avx2_perm<mode>_1): Likewise.
4796         (*avx_vpermilp<mode>): Likewise.
4797         (avx_vpermilvar<mode>3): Likewise.
4798         (avx2_ashrv<mode>): Likewise.
4799         (avx2_<shift_insn>v<mode>): Likewise.
4800         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
4801         -mavx512cd.
4802         * doc/rtl.texi: Document XImode.
4804 2013-08-21  Jeff Law  <law@redhat.com>
4806         * tree-flow.h (register_jump_thread): Pass vector of edges
4807         instead of each important edge.
4808         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
4809         thread path into a vector and pass that to register_jump_thread.
4810         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
4811         passed in edge vector to the current 3-edge form.
4813         Revert:
4814         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4816         * dce.c (fini_dce): Call df_analyze again just in case
4817         delete_unmarked_insns removed anything.
4819 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
4821         * reload.h (struct reg_equivs): Rename to ..
4822         (struct reg_equivs_s): .. this.
4824 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
4826         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
4828 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4830         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
4832 2013-08-21  Jeff Law  <law@redhat.com>
4834         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
4835         simplify assignments too.  If the RHS collapses to a singleton
4836         range, then return the value for the range.
4838 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
4840         * config/i386/sse.md (V16): Rename to...
4841         (VMOVE): this.
4842         (mov<mode>): Update iterator name.
4843         (*mov<mode>_internal): Ditto.
4844         (push<mode>1): Ditto.
4845         (movmisalign<mode>): Ditto.
4847 2013-08-20  Jan Hubicka  <jh@suse.cz>
4849         PR bootstrap/58186
4850         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
4851         entry for direct edges.
4852         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
4854 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4856         Revert my last two changes, r201865 and r201864:
4858         Revert r201865:
4859         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4861         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4862         instances can own GC refs.
4864         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4865         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4866         (gcc::context::gt_pch_nx): Likewise.
4867         (gcc::context::gt_pch_nx):  Likewise.
4868         * ggc.h (gt_ggc_mx <T>): New.
4869         (gt_pch_nx_with_op <T>): New.
4870         (gt_pch_nx <T>): New.
4871         * passes.c (opt_pass::gt_ggc_mx): New.
4872         (opt_pass::gt_pch_nx): New.
4873         (opt_pass::gt_pch_nx_with_op): New.
4874         (pass_manager::gt_ggc_mx): New.
4875         (pass_manager::gt_pch_nx): New.
4876         (pass_manager::gt_pch_nx_with_op): New.
4877         (pass_manager::operator new): Use
4878         ggc_internal_cleared_alloc_stat rather than xcalloc.
4879         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4880         (pass_manager::gt_ggc_mx): New.
4881         (pass_manager::gt_pch_nx): New.
4882         (pass_manager::gt_pch_nx_with_op): New.
4883         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4884         (opt_pass::operator new): New.
4885         (opt_pass::gt_ggc_mx): New.
4886         (opt_pass::gt_pch_nx): New.
4887         (opt_pass::gt_pch_nx_with_op): New.
4889         Revert r201864:
4890         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4892         * Makefile.in (GTFILES): Add context.h.
4893         * context.c (gcc::context::operator new): New.
4894         (gcc::context::gt_ggc_mx): New.
4895         (gcc::context::gt_pch_nx): New.
4896         (gcc::context::gt_pch_nx): New.
4897         * context.h (gcc::context): Add GTY((user)) marking.
4898         (gcc::context::operator new): New.
4899         (gcc::context::gt_ggc_mx): New.
4900         (gcc::context::gt_pch_nx): New.
4901         (gcc::context::gt_pch_nx): New.
4902         (g): Add GTY marking.
4903         (gt_ggc_mx (gcc::context *)): New.
4904         (gt_pch_nx (gcc::context *)): New.
4905         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4906         void *cookie)): New.
4907         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4909 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4911         * dce.c (fini_dce): Call df_analyze again just in case
4912         delete_unmarked_insns removed anything.
4914 2013-08-20  Teresa Johnson  <tejohnson@google.com>
4916         PR rtl-optimizations/57451
4917         * final.c (reemit_insn_block_notes): Prevent lexical blocks
4918         from crossing split section boundaries.
4920 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
4922         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
4923         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
4924         with MULTLIB_DEFAULTS.
4926 2013-08-20  Nick Clifton  <nickc@redhat.com>
4928         * target.def (narrow_volatile_bitfield): Note that the default
4929         value is false, not !TARGET_STRICT_ALIGN.
4930         * doc/tm.texi: Regenerate.
4932 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
4934         Fix LIB_SPEC for systems without libpthread.
4936         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
4937         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
4938         for Android.
4939         * config/i386/linux-common.h: Likewise.
4940         * config/mips/linux-common.h: Likewise.
4942 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
4944         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
4945         checks.
4947 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4949         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4950         instances can own GC refs.
4952         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4953         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4954         (gcc::context::gt_pch_nx): Likewise.
4955         (gcc::context::gt_pch_nx):  Likewise.
4956         * ggc.h (gt_ggc_mx <T>): New.
4957         (gt_pch_nx_with_op <T>): New.
4958         (gt_pch_nx <T>): New.
4959         * passes.c (opt_pass::gt_ggc_mx): New.
4960         (opt_pass::gt_pch_nx): New.
4961         (opt_pass::gt_pch_nx_with_op): New.
4962         (pass_manager::gt_ggc_mx): New.
4963         (pass_manager::gt_pch_nx): New.
4964         (pass_manager::gt_pch_nx_with_op): New.
4965         (pass_manager::operator new): Use
4966         ggc_internal_cleared_alloc_stat rather than xcalloc.
4967         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4968         (pass_manager::gt_ggc_mx): New.
4969         (pass_manager::gt_pch_nx): New.
4970         (pass_manager::gt_pch_nx_with_op): New.
4971         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4972         (opt_pass::operator new): New.
4973         (opt_pass::gt_ggc_mx): New.
4974         (opt_pass::gt_pch_nx): New.
4975         (opt_pass::gt_pch_nx_with_op): New.
4977 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4979         * Makefile.in (GTFILES): Add context.h.
4980         * context.c (gcc::context::operator new): New.
4981         (gcc::context::gt_ggc_mx): New.
4982         (gcc::context::gt_pch_nx): New.
4983         (gcc::context::gt_pch_nx): New.
4984         * context.h (gcc::context): Add GTY((user)) marking.
4985         (gcc::context::operator new): New.
4986         (gcc::context::gt_ggc_mx): New.
4987         (gcc::context::gt_pch_nx): New.
4988         (gcc::context::gt_pch_nx): New.
4989         (g): Add GTY marking.
4990         (gt_ggc_mx (gcc::context *)): New.
4991         (gt_pch_nx (gcc::context *)): New.
4992         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4993         void *cookie)): New.
4994         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4996 2013-08-20  Alan Modra  <amodra@gmail.com>
4998         PR target/57865
4999         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
5000         (rs6000_emit_epilogue): Likewise.
5002 2013-08-19  Dehao Chen  <dehao@google.com>
5004         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
5006 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
5007             Jakub Jelinek  <jakub@redhat.com>
5009         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
5010         (BUILT_IN_FABSD64): Likewise.
5011         (BUILT_IN_FABSD128): Likewise.
5012         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
5013         (fold_builtin_1): Likewise.
5014         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
5015         destination and source operands.
5016         (*abstd2_fpr): Likewise.
5017         (*nabstd2_fpr): Likewise.
5019 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
5021         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
5022         JUMP_P and INSN_P.
5024 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
5026         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
5027         incomplete.
5029 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5031         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
5032         * builtins.c (default_libc_has_function): New.
5033         (gnu_libc_has_function): Ditto.
5034         (no_c99_libc_has_function): Ditto.
5035         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
5036         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
5037         (fold_builtin_sincos): Likewise.
5038         (fold_builtin_cexp): Likewise.
5039         * builtins.def (DEF_C94_BUILTIN): Likewise.
5040         (DEF_C99_BUILTIN): Likewise.
5041         (DEF_C99_C90RES_BUILTIN): Likewise.
5042         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
5043         definitions to using this define.
5044         * config/darwin-protos.h (darwin_libc_has_function): New.
5045         * config/darwin.c (darwin_libc_has_function): Ditto.
5046         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
5047         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
5048         * config/darwin.h: Ditto.
5049         * config/elfos.h: Ditto.
5050         * config/freebsd.h: Ditto.
5051         * config/i386/cygming.h: Ditto.
5052         * config/i386/djgpp.h: Ditto.
5053         * config/i386/i386-interix.h: Ditto.
5054         * config/microblaze/microblaze.h: Ditto.
5055         * config/mmix/mmix.h: Ditto.
5056         * config/gnu-user.h: Ditto.
5057         * config/ia64/hpux.h: Ditto.
5058         * config/pa/pa-hpux.h: Ditto.
5059         * config/pdp11/pdp11.h: Ditto.
5060         * config/picochip/picochip.h: Ditto.
5061         * config/linux.h: Ditto.
5062         * config/netbsd.h: Ditto.
5063         * config/openbsd.h: Ditto.
5064         * config/rs6000/aix43.h: Ditto.
5065         * config/rs6000/aix51.h: Ditto.
5066         * config/rs6000/aix52.h: Ditto.
5067         * config/rs6000/aix53.h: Ditto.
5068         * config/rs6000/aix61.h: Ditto.
5069         * config/rs6000/darwin.h: Ditto.
5070         * config/rs6000/linux.h: Ditto.
5071         * config/rs6000/linux64.h: Ditto.
5072         * config/s390/tpf.h: Ditto.
5073         * config/sol2-10.h: Ditto.
5074         * config/sol2.h: Ditto.
5075         * config/vms/vms.h: Ditto.
5076         * config/vxworks.h: Ditto.
5077         * config/linux-android.c (linux_android_libc_has_function):
5078         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
5079         * config/linux-protos.h (linux_android_libc_has_function):
5080         New declaration.
5081         * config/i386/i386.c (ix86_libc_has_function): New.
5082         * config/i386/i386-protos.h
5083         (ix86_libc_has_function): New declaration.
5084         * config/i386/i386.md
5085         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
5086         ("isinf<mode>2): Likewise.
5087         * convert.c (convert_to_integer): Using new target hook
5088         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
5089         TARGET_C99_FUNCTIONS.
5090         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
5091         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
5092         * coretypes.h (function_class): New enum for different
5093         classes of functions.
5094         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
5095         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
5096         (TARGET_HAS_SINCOS): Likewise.
5097         (TARGET_LIBC_HAS_FUNCTION): New.
5098         * doc/tm.texi: Regenerated.
5099         * targhooks.h (default_libc_has_function): New declaration.
5100         (no_c99_libc_has_function): Ditto.
5101         (gnu_libc_has_function): Ditto.
5102         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
5103         and TARGET_HAS_SINCOS.
5105 2013-08-18  Jan Hubicka  <jh@suse.cz>
5107         * Makeifle-in (ipa-devirt.o): New.
5108         (GTFILES): Add ipa-utils.h and ipa-devirt.c
5109         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
5110         (analyze_functions): Look into possible targets of polymorphic call.
5111         * dumpfile.c (dump_files): Add type-inheritance dump.
5112         * dumpfile.h (TDI_inheritance): New.
5113         * ipa-devirt.c: New file.
5114         * ipa-utils.h (odr_type_d): Forward declare.
5115         (odr_type): New type.
5116         (build_type_inheritance_graph): Declare.
5117         (possible_polymorphic_call_targets): Declare and introduce inline
5118         variant when only edge is pased.
5119         (dump_possible_polymorphic_call_targets): Likewise.
5120         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
5121         * tree.c (type_in_anonymous_namespace_p): Break out from ...
5122         (types_same_for_odr): ... here.
5123         * tree.h (type_in_anonymous_namespace_p): Declare.
5125 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
5127         PR tree-optimization/58006
5128         * tree-parloops.c (take_address_of): Don't ICE if get_name
5129         returns NULL.
5130         (eliminate_local_variables_stmt): Remove clobber stmts.
5132 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
5134         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
5135         error message is issued for an alias to undefined symbol.
5137 2013-08-18  Jan Hubicka  <jh@suse.cz>
5139         * cgraph.c (cgraph_create_indirect_edge): Discover
5140         polymorphic calls and record basic info into indirect_info.
5141         * gimple-fold.c (gimple_fold_call): When doing BINFO based
5142         devirtualization, ignore objc function calls.
5143         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
5144         call with no parm index info.
5145         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
5146         * tree.c (virtual_method_call_p): New function.
5147         * tree.h (virtual_method_call_p): Declare.
5149 2013-08-16  Jan Hubicka  <jh@suse.cz>
5151         PR middle-end/58179
5152         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
5154 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
5156         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
5157         attribute.
5159 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
5161         * gengtype.c (type_for_name): Add special-case support for
5162         locating types within the "gcc::" namespace.
5163         (open_base_files): Emit a "using namespace gcc" directive.
5165 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
5167         PR target/58160
5168         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
5169         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
5171         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
5172         array instead of each individual operand as a separate argument.
5173         (emit_fusion_gpr_load): Likewise.
5174         (expand_fusion_gpr_load): Add new function declaration.
5176         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
5177         signature to have the operands passed as an array, instead of as
5178         separate arguments.  Allow ZERO_EXTEND to be in the memory
5179         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
5180         depend on the register live/dead flags when peepholes are run.
5181         (expand_fusion_gpr_load): New function to be called from the
5182         peephole2 pass, to change the register that addis sets to be the
5183         target register.
5184         (emit_fusion_gpr_load): Change the calling signature to have the
5185         operands passed as an array, instead of as separate arguments.
5186         Allow ZERO_EXTEND to be in the memory address, and also
5187         SIGN_EXTEND if -mpower8-fusion-sign.
5189         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
5190         unspec enumeration.
5191         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
5192         adjust the register addis loads up in the peephole2 pass.  Do not
5193         depend on the register live/dead state when the peephole pass is done.
5195 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
5197         * gengtype.c (create_user_defined_type): Ensure that the kind
5198         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
5199         declaration is seen before the GTY((user)) marking.
5201 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5203         PR target/58105
5204         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
5206 2013-08-16  Jan Hubicka  <jh@suse.cz>
5208         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
5209         arugment expected_type.
5210         (gimple_fold_call): Use it.
5211         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
5212         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
5213         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
5214         (try_make_edge_direct_virtual_call): Likewise.
5215         * tree.c (obj_type_ref_class): New.
5216         * tree.h (obj_type_ref_class): Use it.
5218 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5220         * sched-vis.c (rtl_slim_pp_initialized): Remove.
5221         (rtl_slim_pp): Likewise.
5222         (init_rtl_slim_pretty_print): Likewise.
5223         (dump_value_slim):  Don't call it.  Use local pretty printer.
5224         (dump_insn_slim): Likewise.
5225         (dump_rtl_slim): Likewise.
5226         (str_pattern_slim): Likewise.
5227         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
5228         Simplify.
5230 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
5232         PR tree-optimization/58164
5233         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
5234         walk gimple_goto_dest of GIMPLE_GOTO.
5236         PR tree-optimization/58165
5237         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
5238         bi_call must be the last stmt in a bb, don't split_block, instead
5239         use fallthru edge from it and give up if there is none.
5240         Release conds vector when returning early.
5242 2013-08-14  Xinliang David Li  <davidxl@google.com>
5244         * config/i386/i386.c (ix86_option_override_internal):
5245         Remove unused variable and field.
5247 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5249         PR target/57949
5250         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
5251         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
5252         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
5253         and Linux, correct BLKmode alignment when 128-bit alignment is
5254         required and compatibility flag is not set.
5255         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
5256         for zero-size arguments when compatibility flag is not set.
5258 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
5260         PR tree-optimization/58145
5261         * tree-sra.c (build_ref_for_offset): If prev_base has
5262         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
5264 2013-08-14  Xinliang David Li  <davidxl@google.com>
5266         * config/i386/i386.c (ix86_option_override_internal):
5267         Fix uninitialized variable error.
5269 2013-08-14  Xinliang David Li  <davidxl@google.com>
5271         * config/i386/i386.opt: Define two new options.
5272         * config/i386/x86-tune.def: Add arch selector field in macros.
5273         * config/i386/i386.h: Adjust macro definition.
5274         * config/i386/i386.c (ix86_option_override_internal):
5275         Refactor the code.
5276         (parse_mtune_ctrl_str): New function.
5277         (set_ix86_tune_features): New function.
5278         (ix86_function_specific_restore): Call the new helper function.
5280 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
5282         PR rtl-optimization/57662
5283         * sel-sched.c (code_motion_process_successors): When the current insn
5284         is removed after the recursive traversal, break from the loop.
5285         Add comments and debug printouts.
5287 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
5288             Alexandre Oliva  <aoliva@redhat.com>
5290         PR target/58067
5291         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
5292         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
5293         there also UNSPEC_PLTOFF.
5295 2013-08-14  Marek Polacek  <polacek@redhat.com>
5297         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
5298         NUM_CONDITIONS bit positions.
5300 2013-08-13  Cary Coutant  <ccoutant@google.com>
5302         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
5303         (attr_checksum): Hash vector contents instead of pointer.
5304         (attr_checksum_ordered): Likewise.
5306 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
5308         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
5309         when Pmode != word_mode.  Add length_address attribute.
5310         (sse3_monitor_<mode>): Merge from sse3_monitor and
5311         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
5312         Pmode != word_mode.  Update insn length attribute.
5313         * config/i386/i386.c (ix86_option_override_internal): Update
5314         ix86_gen_monitor selection for merged sse3_monitor insn.
5316 2013-08-13  Julian Brown  <julian@codesourcery.com>
5318         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
5319         perform invalid legitimization on greater-than-word-size modes for
5320         TARGET_E500_DOUBLE.
5322 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
5324         * ira.c (setup_class_translate_array): Use aclass instead of cl
5325         for classes not fully covered by allocno classes.
5327 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
5329         PR tree-optimization/57661
5330         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
5331         * tree-inline.c (tree_function_versioning): Initialize it.
5332         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
5333         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
5334         that is not being copied.
5336         PR sanitizer/56417
5337         * asan.c (instrument_strlen_call): Fix typo in comment.
5338         Use char * type even for the lhs of POINTER_PLUS_EXPR.
5340 2013-08-13  Steve Ellcey  <sellcey@mips.com>
5342         * config/mips/mips.md (prefetch): Use lw instead of ld on
5343         loongson in 32bit mode.
5345 2013-08-13  Nick Clifton  <nickc@redhat.com>
5347         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
5349 2013-08-13  Jan Hubicka  <jh@suse.cz>
5351         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
5352         introduced edge; fix typo in sanity check.
5353         (cgraph_resolve_speculation): Export; improve diagnostic.
5354         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
5355         speculation at type mismatch.
5356         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
5357         (cgraph_resolve_speculation): Declare.
5358         (symtab_can_be_discarded): New function.
5359         * value-prof.c (gimple_ic_transform): Remove actual transform code.
5360         * ipa-inline-transform.c (speculation_removed): New global var.
5361         (clone_inlined_nodes): See if speculation can be removed.
5362         (inline_call): If speculations was removed, we growths may not match.
5363         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
5364         (speculation_useful_p): New function.
5365         (resolve_noninline_speculation): New function.
5366         (inline_small_functions): Resolve useless speculations.
5367         * ipa-inline.h (speculation_useful_p): Declare
5368         * ipa.c (can_replace_by_local_alias): Simplify.
5369         (ipa_profile): Produce speculative calls in non-lto, too;
5370         add simple cost model; produce local aliases.
5372 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
5374         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
5375         PASS_MANAGER_H.
5377 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
5379         * config/i386/i386.c (ix86_function_versions): Use error + inform.
5381 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
5383         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
5384         iterator instead of X87MODEF.
5386 2013-08-12  Perez Read  <netfirewall@gmail.com>
5388         PR target/58132
5389         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
5390         operand 0 for intel asm alternative.
5391         (*movabs<mode>_2): Ditto for operand 1.
5393 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
5395         * config/aarch64/arm_none.h
5396         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
5398 2013-08-12  Nick Clifton  <nickc@redhat.com>
5400         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
5402 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5404         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
5405         expand for QI/HImode operand to produce more effictive code for
5406         unsigned char(short) --> float(double) conversion.
5408 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
5410         * doc/invoke.texi: Mention that -ftls-model does not force the final
5411         model.
5413 2013-08-12  Marek Polacek  <polacek@redhat.com>
5414             Marc Glisse  <marc.glisse@inria.fr>
5416         PR tree-optimization/57980
5417         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
5418         when creating -1 constant.
5420 2013-08-10  Jan Hubicka  <jh@suse.cz>
5422         Workaround binutils PR14342.
5423         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
5424         (gimple_init_edge_profiler): Likewise.
5425         (gimple_gen_ic_func_profiler): Likewise.
5427 2013-08-09  Jan Hubicka  <jh@suse.cz>
5429         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
5431 2013-08-09  Xinliang David Li  <davidxl@google.com>
5433         * config/i386/stringop.def: New file.
5434         * config/i386/stringop.opt: New file.
5435         * config/i386/i386-opts.h: Include stringopt.def.
5436         * config/i386/i386.opt: Include stringopt.opt.
5437         * config/i386/i386.c (ix86_option_override_internal):
5438         Override default size based stringop inline strategies with options.
5439         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
5440         New function.
5442 2013-08-09  Jan Hubicka  <jh@suse.cz>
5444         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
5446 2013-08-09  Jan Hubicka  <jh@suse.cz>
5448         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
5449         CGRAPH_FREQ_MAX.
5450         (dump_cgraph_node): Dump profile-id.
5451         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
5452         and common_target_probability.
5453         * lto-cgraph.c (lto_output_edge): Stream common targets.
5454         (lto_output_node): Stream profile ids.
5455         (input_node): Stream profile ids.
5456         (input_edge): Stream common targets.
5457         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
5458         * ipa.c: Include value-prof.h
5459         (ipa_profile_generate_summary): Turn indirect call statement histograms
5460         into common targets.
5461         (ipa_profile): Turn common targets into speculative edges.
5463 2013-08-09  Jan Hubicka  <jh@suse.cz>
5465         * cgraph.h (cgraph_node): Add profile_id.
5466         * value-prof.c (cgraph_node_map): Turn into pointer_map.
5467         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
5468         (del_node_map): Update.
5469         (find_func_by_funcdef_no): Replace by ...
5470         (find_func_by_profile_id): ... this one.
5471         (gimple_ic_transform): Do not remove useful histograms when
5472         speculation is not done; dump info when indirect call removal
5473         can happen at LTO.
5474         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
5475         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
5476         (__gcov_indirect_call_profiler_v2): .. this one.
5477         * profile.h (init_node_map): Update.
5478         * coverage.c (coverage_compute_profile_id): New function.
5479         * coverage.h (coverage_compute_profile_id): Declare.
5480         * tree-profile.c (init_ic_make_global_vars): Make
5481         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
5482         (gimple_init_edge_profiler): Update prototype of
5483         __gcov_indirect_call_profiler.
5484         (gimple_gen_ic_func_profiler): Simplify.
5485         (tree_profiling): Use init_node_map
5487 2013-08-09  Jan Hubicka  <jh@suse.cz>
5489         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
5490         non-speculative refs.
5491         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
5492         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
5493         (cgraph_set_call_stmt): Likewise.
5494         (cgraph_create_edge_1): Fix release checking compilatoin;
5495         clear lto_stmt_uid.
5496         (cgraph_free_edge): Free indirect info.
5497         (cgraph_turn_edge_to_speculative): New function.
5498         (cgraph_speculative_call_info): New function.
5499         (cgraph_make_edge_direct): Return direct edge; handle speculation.
5500         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
5501         (dump_cgraph_node): Dump speculation.
5502         (verify_edge_count_and_frequency): Accept speculative edges.
5503         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
5504         (verify_cgraph_node): Handle speculation.
5505         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
5506         (cgraph_set_call_stmt): Update prototype.
5507         (cgraph_make_edge_direct): Update prototype.
5508         (cgraph_speculative_call_info): Declare.
5509         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
5510         to change; update call of ipa_find_references.
5511         * ipa-ref.c (ipa_record_reference): Fix return value; clear
5512         lto_stmt_uid and speculative flags.
5513         (ipa_dump_references): Dump speculation.
5514         (ipa_clone_references): Clone speculative flag.
5515         (ipa_clone_referring): Likewise.
5516         (ipa_clone_ref): New function.
5517         (ipa_find_reference): Look into lto_stmt_uids
5518         (ipa_clear_stmts_in_references): Do not clear speculative calls.
5519         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
5520         (ipa_find_reference): Update declaration.
5521         (ipa_clone_ref): Declare.
5522         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
5523         stream speculative flag.
5524         (lto_output_ref): Stream statements uids and speculation.
5525         (input_ref): Likewise.
5526         (input_edge): Stream speuclation.
5527         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
5528         (cgraph_set_call_stmt_including_clones): Handle speculation.
5529         * ipa-inline.c (heap_edge_removal_hook): New function.
5530         (inline_small_functions): Register it.
5531         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
5532         also initialize refs.
5533         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
5534         edge to change.
5535         (try_make_edge_direct_simple_call): Likewise.
5536         (try_make_edge_direct_simple_call): Likewise.
5537         (update_indirect_edges_after_inlining): Likewise.
5538         (remove_described_reference): Look proper lto_stmt_uid.
5539         (propagate_controlled_uses): Likewise.
5540         (propagate_controlled_uses): Liekwise.
5541         * tree-inline.c (copy_bb): Copy speculative edges.
5542         (redirect_all_calls): New function.
5543         (copy_cfg_body): Do redirection after loop info is updated.
5544         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
5546 2013-08-09  Jan Hubicka  <jh@suse.cz>
5548         * lto-streamer-out.c (output_function): Renumber PHIs.
5549         * lto-streamer-in.c (input_function): Likewise.
5551 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
5553         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
5554         (get_lane_unsigned): Likewise.
5555         (dup_lane_scalar): Likewise.
5556         (get_lane): enable for VALL.
5557         * config/aarch64/aarch64-simd.md
5558         (aarch64_dup_lane_scalar<mode>): Remove.
5559         (aarch64_get_lane_signed<mode>): Likewise.
5560         (aarch64_get_lane_unsigned<mode>): Likewise.
5561         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
5562         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
5563         (aarch64_get_lane<mode>): Enable for all vector modes.
5564         (aarch64_get_lanedi): Remove misleading constraints.
5565         * config/aarch64/arm_neon.h
5566         (__aarch64_vget_lane_any): Define.
5567         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
5568         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
5569         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
5570         * config/aarch64/iterators.md (VDQQH): New.
5571         (VDQQHS): Likewise.
5572         (vwcore): Likewise.
5574 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
5576         * configure.ac: Add GAS check for LEON instructions on SPARC.
5577         * configure: Regenerate.
5578         * config.in: Likewise.
5579         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
5580         sparc*-*-* block.
5581         * config/sparc/sparc.opt (LEON, LEON3): New masks.
5582         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
5583         for LEON or LEON3.
5584         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
5585         (AS_LEON_FLAG): New macro.
5586         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
5587         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
5588         Deal with LEON and LEON3 for the memory model.
5589         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
5590         (atomic_compare_and_swap<mode>_1): Likewise.
5591         (*atomic_compare_and_swap<mode>_1): Likewise.
5593 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5595         * config/arm/neon.md (vcond): Fix floating-point vector
5596         comparisons against 0.
5598 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
5600         * lra-constraints.c (emit_spill_move): Remove assert.
5601         (process_alt_operands): Add more debugging
5602         output.  Increase reject for spilling into memory.  Decrease
5603         reject for reloading scratch.
5604         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
5606 2013-08-08  Steve Ellcey  <sellcey@mips.com>
5608         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
5609         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
5610         micromips incompatible.  Add nan2008.
5611         (MULTILIB_DIRNAMES): Add nan2008.
5612         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
5613         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
5614         and micromips incompatible.  Add nan2008.
5615         (MULTILIB_DIRNAMES): Add nan2008.
5616         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
5618 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
5620         PR rtl-optimization/58079
5621         * combine.c (combine_simplify_rtx): Avoid using SUBST if
5622         simplify_comparison has widened a comparison with an integer.
5624 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5626         * config/arm/neon.md (movmisalign<mode>): Disable when we
5627         don't allow unaligned accesses.
5628         (*movmisalign<mode>_neon_store): Likewise.
5629         (*movmisalign<mode>_neon_load): Likewise.
5630         (*movmisalign<mode>_neon_store): Likewise.
5631         (*movmisalign<mode>_neon_load): Likewise.
5633 2013-08-08  Jan Hubicka  <jh@suse.cz>
5635         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
5636         (make_pass_rebuild_cgraph_edges): Also clear references.
5637         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
5638         * ipa-inline-transform.c (inline_transform): Remove all references
5639         after inlining.
5640         * cgraphunit.c (expand_function): Remove all references after
5641         expansion.
5642         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
5643         (ipa_find_reference): Rewrite to iterator.
5644         (remove_stmt_references): Likewise.
5645         (ipa_clear_stmts_in_references): New function.
5646         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
5647         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
5648         clear references.
5649         * ipa-split.c (split_function): Remove references in split function.
5651 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
5653         PR target/57431
5654         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
5655         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
5657 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
5659         PR target/56979
5660         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
5661         suggested mode for the assignment isn't compatible with the
5662         registers required.
5664 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5666         PR target/58065
5667         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
5669 2013-08-07  Xinliang David Li  <davidxl@google.com>
5671         * config/i386/i386.opt: New option -mtune-ctrl=.
5672         * config/i386/x86-tune.def: New file.
5673         * config/i386/i386.h: include x86-tune.def.
5674         * config/i386/i386.c (ix86_option_override_internal):
5675         Parsing -mtune-ctrl= option and set tune features.
5677 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
5679         PR other/12081
5680         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
5681         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
5682         to gen_2arg_fn_t.
5684 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
5686         * rtl.h (update_alignments): Declare.
5687         * final.c (grow_label_align): New function extracted from...
5688         (shorten_branches): ...here.  Call it.
5689         (update_alignments): New function.
5690         * reorg.c (sibling_labels): New variable.
5691         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
5692         the new label along with it onto the sibling_labels vector.
5693         (fill_simple_delay_slots): Adjust call to get_label_before.
5694         (fill_slots_from_thread): Likewise.
5695         (relax_delay_slots): Likewise.
5696         (make_return_insns): Likewise.
5697         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
5699 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
5701         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
5702         document its semantics.
5703         (diagnostic_report_diagnostic): Adjust accordingly.
5705 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
5707         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
5708         (sparc_option_override): ...and port to new C++ pass API.
5709         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
5711 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
5713         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
5715 2013-08-06  Caroline Tice  <cmtice@google.com>
5717         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
5718         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
5719         * tree-pass.h: Add pass_vtable_verify.
5720         * varasm.c (assemble_variable): Add code to properly set the comdat
5721         section and name for the .vtable_map_vars section.
5722         (assemble_vtyv_preinit_initializer): New function.
5723         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
5724         LINK_ONCE flag.
5725         * output.h: Add function decl for assemble_vtv_preinit_initializer.
5726         * vtable-verify.c: New file.
5727         * vtable-verify.h: New file.
5728         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
5729         initialiation levels.
5730         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
5731         * passes.def: Insert pass_vtable_verify.
5732         * aclocal.m4: Reorder includes.
5733         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
5734         -fvtv-counts options.
5735         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
5736         as appropriate, if -fvtable-verify=... is used.
5737         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
5738         -fvtable-verify=... is used.
5739         * Makefile.in (OBJS):  Add vtable-verify.o to list.
5740         (vtable-verify.o): Add new build rule.
5741         (GTFILES): Add vtable-verify.c to list.
5742         * common.opt (fvtable-verify=): New flag.
5743         (vtv_priority): Values for fvtable-verify= flag.
5744         (fvtv-counts): New flag.
5745         (fvtv-debug): New flag.
5746         * tree.h (save_vtable_map_decl): New extern function decl.
5748 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
5750         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
5751         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
5752         (pass_data_rl78_devirt): ...new pass_data instance and...
5753         (make_pass_rl78_devirt): ...new function.
5754         (rl78_asm_file_start): Port pass registration to new C++ API.
5756 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
5758         * coretypes.h (rtl_opt_pass): Add.
5759         (gcc::context): Add.
5760         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
5761         (epiphany_init): Port to new C++ pass API.
5762         (epiphany_optimize_mode_switching): Likewise.
5763         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
5764         (pass_manager::get_pass_mode_switching): New.
5765         (pass_manager::get_pass_peephole2): New.
5766         * mode-switching.c (pass_mode_switching): Add clone method.
5767         * recog.c (pass_peephole2): Add clone method.
5768         (pass_split_all_insns): Add clone method.
5770 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
5772         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
5773         (mips_option_override): ...here, porting to new C++ API for passes.
5775 2013-08-06  Jan Hubicka  <jh@suse.cz>
5777         * cgraph.c (cgraph_get_body): New function based on lto.c
5778         implementation.
5779         * cgraph.h (cgraph_get_body): Declare.
5780         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
5781         LTO paths.
5782         * cgraphunit.c (expand_function): Get body prior expanding.
5783         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
5784         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
5785         really need.
5786         * passes.c (do_per_function_toporder): Get body.
5787         * tree-inline.c (expand_call_inline): Get body prior inlining it.
5788         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
5790 2013-08-06  Martin Jambor  <mjambor@suse.cz>
5792         PR fortran/57987
5793         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
5794         is not re-finalized.  Rename second parameter to no_collect.
5796 2013-08-06  Martin Jambor  <mjambor@suse.cz>
5798         PR middle-end/58041
5799         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
5800         MEM_REF has proper alignment information.
5802 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
5804         PR other/12081
5805         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
5806         class insn_gen_fn.
5807         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
5808         rtx (*) (rtx, ...) with insn_gen_fn.
5809         * genoutput.c (output_insn_data): Cast gen_? function pointers to
5810         insn_gen_fn::stored_funcptr.  Add initializer braces.
5812 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5814         Rewrite how instances of passes are cloned to remove assumptions
5815         about their sizes (thus allowing pass subclasses to have
5816         additional data fields, albeit non-GC-managed ones at this point).
5818         * passes.c (make_pass_instance): Now that passes have clone
5819         methods, rewrite this function to eliminate XNEW and memcpy
5820         calls that used hardcoded sizes.  Since this function no longer
5821         creates pass instances, rename it to...
5822         (add_pass_instance): ...this.  Document the old way that passes were
5823         numbered and flagged, and rework this function to continue using it.
5824         (next_pass_1): Add an initial_pass argument for use by
5825         add_pass_instance.
5826         (position_pass): When adding multiple instances of a pass, use
5827         the pass's clone method, rather than relying on the XNEW/memcpy
5828         within the former make_pass_instance (now add_pass_instance).
5829         (pass_manager::pass_manager): When invoking next_pass_1, also supply
5830         the initial instance of the current pass within the pass manager.
5832 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5834         This is the automated part of the conversion of passes from C
5835         structs to C++ classes.
5837         Patch autogenerated by refactor_passes.py from
5838         https://github.com/davidmalcolm/gcc-refactoring-scripts
5839         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
5841         * asan.c (pass_asan): Convert from a global struct to a subclass of
5842         gimple_opt_pass along with...
5843         (pass_data_asan): ...new pass_data instance and...
5844         (make_pass_asan): ...new function.
5845         (pass_asan_O0): Convert from a global struct to a subclass of
5846         gimple_opt_pass along with...
5847         (pass_data_asan_O0): ...new pass_data instance and...
5848         (make_pass_asan_O0): ...new function.
5849         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
5850         subclass of rtl_opt_pass along with...
5851         (pass_data_inc_dec): ...new pass_data instance and...
5852         (make_pass_inc_dec): ...new function.
5853         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
5854         a subclass of rtl_opt_pass along with...
5855         (pass_data_reorder_blocks): ...new pass_data instance and...
5856         (make_pass_reorder_blocks): ...new function.
5857         (pass_duplicate_computed_gotos): Convert from a global struct to a
5858         subclass of rtl_opt_pass along with...
5859         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
5860         (make_pass_duplicate_computed_gotos): ...new function.
5861         (pass_partition_blocks): Convert from a global struct to a subclass of
5862         rtl_opt_pass along with...
5863         (pass_data_partition_blocks): ...new pass_data instance and...
5864         (make_pass_partition_blocks): ...new function.
5865         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
5866         struct to a subclass of rtl_opt_pass along with...
5867         (pass_data_branch_target_load_optimize1): ...new pass_data instance
5868         and...
5869         (make_pass_branch_target_load_optimize1): ...new function.
5870         (pass_branch_target_load_optimize2): Convert from a global struct to a
5871         subclass of rtl_opt_pass along with...
5872         (pass_data_branch_target_load_optimize2): ...new pass_data instance
5873         and...
5874         (make_pass_branch_target_load_optimize2): ...new function.
5875         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
5876         of rtl_opt_pass along with...
5877         (pass_data_jump): ...new pass_data instance and...
5878         (make_pass_jump): ...new function.
5879         (pass_jump2): Convert from a global struct to a subclass of
5880         rtl_opt_pass along with...
5881         (pass_data_jump2): ...new pass_data instance and...
5882         (make_pass_jump2): ...new function.
5883         * cfgexpand.c (pass_expand): Convert from a global struct to a
5884         subclass of rtl_opt_pass along with...
5885         (pass_data_expand): ...new pass_data instance and...
5886         (make_pass_expand): ...new function.
5887         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
5888         of rtl_opt_pass along with...
5889         (pass_data_free_cfg): ...new pass_data instance and...
5890         (make_pass_free_cfg): ...new function.
5891         (pass_into_cfg_layout_mode): Convert from a global struct to a
5892         subclass of rtl_opt_pass along with...
5893         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
5894         (make_pass_into_cfg_layout_mode): ...new function.
5895         (pass_outof_cfg_layout_mode): Convert from a global struct to a
5896         subclass of rtl_opt_pass along with...
5897         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
5898         (make_pass_outof_cfg_layout_mode): ...new function.
5899         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
5900         struct to a subclass of gimple_opt_pass along with...
5901         (pass_data_build_cgraph_edges): ...new pass_data instance and...
5902         (make_pass_build_cgraph_edges): ...new function.
5903         (pass_rebuild_cgraph_edges): Convert from a global struct to a
5904         subclass of gimple_opt_pass along with...
5905         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
5906         (make_pass_rebuild_cgraph_edges): ...new function.
5907         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
5908         subclass of gimple_opt_pass along with...
5909         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
5910         and...
5911         (make_pass_remove_cgraph_callee_edges): ...new function.
5912         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
5913         struct to a subclass of rtl_opt_pass along with...
5914         (pass_data_stack_adjustments): ...new pass_data instance and...
5915         (make_pass_stack_adjustments): ...new function.
5916         * combine.c (pass_combine): Convert from a global struct to a subclass
5917         of rtl_opt_pass along with...
5918         (pass_data_combine): ...new pass_data instance and...
5919         (make_pass_combine): ...new function.
5920         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
5921         global struct to a subclass of rtl_opt_pass along with...
5922         (pass_data_compare_elim_after_reload): ...new pass_data instance
5923         and...
5924         (make_pass_compare_elim_after_reload): ...new function.
5925         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
5926         of rtl_opt_pass along with...
5927         (pass_data_rtl_cprop): ...new pass_data instance and...
5928         (make_pass_rtl_cprop): ...new function.
5929         * cse.c (pass_cse): Convert from a global struct to a subclass of
5930         rtl_opt_pass along with...
5931         (pass_data_cse): ...new pass_data instance and...
5932         (make_pass_cse): ...new function.
5933         (pass_cse2): Convert from a global struct to a subclass of
5934         rtl_opt_pass along with...
5935         (pass_data_cse2): ...new pass_data instance and...
5936         (make_pass_cse2): ...new function.
5937         (pass_cse_after_global_opts): Convert from a global struct to a
5938         subclass of rtl_opt_pass along with...
5939         (pass_data_cse_after_global_opts): ...new pass_data instance and...
5940         (make_pass_cse_after_global_opts): ...new function.
5941         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
5942         of rtl_opt_pass along with...
5943         (pass_data_ud_rtl_dce): ...new pass_data instance and...
5944         (make_pass_ud_rtl_dce): ...new function.
5945         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
5946         rtl_opt_pass along with...
5947         (pass_data_fast_rtl_dce): ...new pass_data instance and...
5948         (make_pass_fast_rtl_dce): ...new function.
5949         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
5950         a subclass of rtl_opt_pass along with...
5951         (pass_data_df_initialize_opt): ...new pass_data instance and...
5952         (make_pass_df_initialize_opt): ...new function.
5953         (pass_df_initialize_no_opt): Convert from a global struct to a
5954         subclass of rtl_opt_pass along with...
5955         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
5956         (make_pass_df_initialize_no_opt): ...new function.
5957         (pass_df_finish): Convert from a global struct to a subclass of
5958         rtl_opt_pass along with...
5959         (pass_data_df_finish): ...new pass_data instance and...
5960         (make_pass_df_finish): ...new function.
5961         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
5962         rtl_opt_pass along with...
5963         (pass_data_rtl_dse1): ...new pass_data instance and...
5964         (make_pass_rtl_dse1): ...new function.
5965         (pass_rtl_dse2): Convert from a global struct to a subclass of
5966         rtl_opt_pass along with...
5967         (pass_data_rtl_dse2): ...new pass_data instance and...
5968         (make_pass_rtl_dse2): ...new function.
5969         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
5970         subclass of rtl_opt_pass along with...
5971         (pass_data_dwarf2_frame): ...new pass_data instance and...
5972         (make_pass_dwarf2_frame): ...new function.
5973         * except.c (pass_set_nothrow_function_flags): Convert from a global
5974         struct to a subclass of rtl_opt_pass along with...
5975         (pass_data_set_nothrow_function_flags): ...new pass_data instance
5976         and...
5977         (make_pass_set_nothrow_function_flags): ...new function.
5978         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
5979         subclass of rtl_opt_pass along with...
5980         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
5981         and...
5982         (make_pass_convert_to_eh_region_ranges): ...new function.
5983         * final.c (pass_compute_alignments): Convert from a global struct to a
5984         subclass of rtl_opt_pass along with...
5985         (pass_data_compute_alignments): ...new pass_data instance and...
5986         (make_pass_compute_alignments): ...new function.
5987         (pass_final): Convert from a global struct to a subclass of
5988         rtl_opt_pass along with...
5989         (pass_data_final): ...new pass_data instance and...
5990         (make_pass_final): ...new function.
5991         (pass_shorten_branches): Convert from a global struct to a subclass of
5992         rtl_opt_pass along with...
5993         (pass_data_shorten_branches): ...new pass_data instance and...
5994         (make_pass_shorten_branches): ...new function.
5995         (pass_clean_state): Convert from a global struct to a subclass of
5996         rtl_opt_pass along with...
5997         (pass_data_clean_state): ...new pass_data instance and...
5998         (make_pass_clean_state): ...new function.
5999         * function.c (pass_instantiate_virtual_regs): Convert from a global
6000         struct to a subclass of rtl_opt_pass along with...
6001         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
6002         (make_pass_instantiate_virtual_regs): ...new function.
6003         (pass_leaf_regs): Convert from a global struct to a subclass of
6004         rtl_opt_pass along with...
6005         (pass_data_leaf_regs): ...new pass_data instance and...
6006         (make_pass_leaf_regs): ...new function.
6007         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
6008         subclass of rtl_opt_pass along with...
6009         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
6010         and...
6011         (make_pass_thread_prologue_and_epilogue): ...new function.
6012         (pass_match_asm_constraints): Convert from a global struct to a
6013         subclass of rtl_opt_pass along with...
6014         (pass_data_match_asm_constraints): ...new pass_data instance and...
6015         (make_pass_match_asm_constraints): ...new function.
6016         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
6017         subclass of rtl_opt_pass along with...
6018         (pass_data_rtl_fwprop): ...new pass_data instance and...
6019         (make_pass_rtl_fwprop): ...new function.
6020         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
6021         rtl_opt_pass along with...
6022         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
6023         (make_pass_rtl_fwprop_addr): ...new function.
6024         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
6025         rtl_opt_pass along with...
6026         (pass_data_rtl_pre): ...new pass_data instance and...
6027         (make_pass_rtl_pre): ...new function.
6028         (pass_rtl_hoist): Convert from a global struct to a subclass of
6029         rtl_opt_pass along with...
6030         (pass_data_rtl_hoist): ...new pass_data instance and...
6031         (make_pass_rtl_hoist): ...new function.
6032         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
6033         subclass of gimple_opt_pass along with...
6034         (pass_data_lower_cf): ...new pass_data instance and...
6035         (make_pass_lower_cf): ...new function.
6036         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
6037         from a global struct to a subclass of gimple_opt_pass along with...
6038         (pass_data_strength_reduction): ...new pass_data instance and...
6039         (make_pass_strength_reduction): ...new function.
6040         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
6041         of rtl_opt_pass along with...
6042         (pass_data_rtl_ifcvt): ...new pass_data instance and...
6043         (make_pass_rtl_ifcvt): ...new function.
6044         (pass_if_after_combine): Convert from a global struct to a subclass of
6045         rtl_opt_pass along with...
6046         (pass_data_if_after_combine): ...new pass_data instance and...
6047         (make_pass_if_after_combine): ...new function.
6048         (pass_if_after_reload): Convert from a global struct to a subclass of
6049         rtl_opt_pass along with...
6050         (pass_data_if_after_reload): ...new pass_data instance and...
6051         (make_pass_if_after_reload): ...new function.
6052         * init-regs.c (pass_initialize_regs): Convert from a global struct to
6053         a subclass of rtl_opt_pass along with...
6054         (pass_data_initialize_regs): ...new pass_data instance and...
6055         (make_pass_initialize_regs): ...new function.
6056         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
6057         of ipa_opt_pass_d along with...
6058         (pass_data_ipa_cp): ...new pass_data instance and...
6059         (make_pass_ipa_cp): ...new function.
6060         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
6061         global struct to a subclass of gimple_opt_pass along with...
6062         (pass_data_inline_parameters): ...new pass_data instance and...
6063         (make_pass_inline_parameters): ...new function.
6064         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
6065         subclass of gimple_opt_pass along with...
6066         (pass_data_early_inline): ...new pass_data instance and...
6067         (make_pass_early_inline): ...new function.
6068         (pass_ipa_inline): Convert from a global struct to a subclass of
6069         ipa_opt_pass_d along with...
6070         (pass_data_ipa_inline): ...new pass_data instance and...
6071         (make_pass_ipa_inline): ...new function.
6072         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
6073         struct to a subclass of gimple_opt_pass along with...
6074         (pass_data_local_pure_const): ...new pass_data instance and...
6075         (make_pass_local_pure_const): ...new function.
6076         (pass_ipa_pure_const): Convert from a global struct to a subclass of
6077         ipa_opt_pass_d along with...
6078         (pass_data_ipa_pure_const): ...new pass_data instance and...
6079         (make_pass_ipa_pure_const): ...new function.
6080         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
6081         to a subclass of ipa_opt_pass_d along with...
6082         (pass_data_ipa_reference): ...new pass_data instance and...
6083         (make_pass_ipa_reference): ...new function.
6084         * ipa-split.c (pass_split_functions): Convert from a global struct to
6085         a subclass of gimple_opt_pass along with...
6086         (pass_data_split_functions): ...new pass_data instance and...
6087         (make_pass_split_functions): ...new function.
6088         (pass_feedback_split_functions): Convert from a global struct to a
6089         subclass of gimple_opt_pass along with...
6090         (pass_data_feedback_split_functions): ...new pass_data instance and...
6091         (make_pass_feedback_split_functions): ...new function.
6092         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
6093         global struct to a subclass of simple_ipa_opt_pass along with...
6094         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
6095         instance and...
6096         (make_pass_ipa_function_and_variable_visibility): ...new function.
6097         (pass_ipa_free_inline_summary): Convert from a global struct to a
6098         subclass of simple_ipa_opt_pass along with...
6099         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
6100         (make_pass_ipa_free_inline_summary): ...new function.
6101         (pass_ipa_whole_program_visibility): Convert from a global struct to a
6102         subclass of ipa_opt_pass_d along with...
6103         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
6104         and...
6105         (make_pass_ipa_whole_program_visibility): ...new function.
6106         (pass_ipa_profile): Convert from a global struct to a subclass of
6107         ipa_opt_pass_d along with...
6108         (pass_data_ipa_profile): ...new pass_data instance and...
6109         (make_pass_ipa_profile): ...new function.
6110         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
6111         ipa_opt_pass_d along with...
6112         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
6113         (make_pass_ipa_cdtor_merge): ...new function.
6114         * ira.c (pass_ira): Convert from a global struct to a subclass of
6115         rtl_opt_pass along with...
6116         (pass_data_ira): ...new pass_data instance and...
6117         (make_pass_ira): ...new function.
6118         (pass_reload): Convert from a global struct to a subclass of
6119         rtl_opt_pass along with...
6120         (pass_data_reload): ...new pass_data instance and...
6121         (make_pass_reload): ...new function.
6122         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
6123         subclass of rtl_opt_pass along with...
6124         (pass_data_cleanup_barriers): ...new pass_data instance and...
6125         (make_pass_cleanup_barriers): ...new function.
6126         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
6127         of rtl_opt_pass along with...
6128         (pass_data_loop2): ...new pass_data instance and...
6129         (make_pass_loop2): ...new function.
6130         (pass_rtl_loop_init): Convert from a global struct to a subclass of
6131         rtl_opt_pass along with...
6132         (pass_data_rtl_loop_init): ...new pass_data instance and...
6133         (make_pass_rtl_loop_init): ...new function.
6134         (pass_rtl_loop_done): Convert from a global struct to a subclass of
6135         rtl_opt_pass along with...
6136         (pass_data_rtl_loop_done): ...new pass_data instance and...
6137         (make_pass_rtl_loop_done): ...new function.
6138         (pass_rtl_move_loop_invariants): Convert from a global struct to a
6139         subclass of rtl_opt_pass along with...
6140         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
6141         (make_pass_rtl_move_loop_invariants): ...new function.
6142         (pass_rtl_unswitch): Convert from a global struct to a subclass of
6143         rtl_opt_pass along with...
6144         (pass_data_rtl_unswitch): ...new pass_data instance and...
6145         (make_pass_rtl_unswitch): ...new function.
6146         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
6147         subclass of rtl_opt_pass along with...
6148         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
6149         and...
6150         (make_pass_rtl_unroll_and_peel_loops): ...new function.
6151         (pass_rtl_doloop): Convert from a global struct to a subclass of
6152         rtl_opt_pass along with...
6153         (pass_data_rtl_doloop): ...new pass_data instance and...
6154         (make_pass_rtl_doloop): ...new function.
6155         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
6156         a subclass of rtl_opt_pass along with...
6157         (pass_data_lower_subreg): ...new pass_data instance and...
6158         (make_pass_lower_subreg): ...new function.
6159         (pass_lower_subreg2): Convert from a global struct to a subclass of
6160         rtl_opt_pass along with...
6161         (pass_data_lower_subreg2): ...new pass_data instance and...
6162         (make_pass_lower_subreg2): ...new function.
6163         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
6164         struct to a subclass of ipa_opt_pass_d along with...
6165         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
6166         (make_pass_ipa_lto_gimple_out): ...new function.
6167         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
6168         of ipa_opt_pass_d along with...
6169         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
6170         (make_pass_ipa_lto_finish_out): ...new function.
6171         * mode-switching.c (pass_mode_switching): Convert from a global struct
6172         to a subclass of rtl_opt_pass along with...
6173         (pass_data_mode_switching): ...new pass_data instance and...
6174         (make_pass_mode_switching): ...new function.
6175         * modulo-sched.c (pass_sms): Convert from a global struct to a
6176         subclass of rtl_opt_pass along with...
6177         (pass_data_sms): ...new pass_data instance and...
6178         (make_pass_sms): ...new function.
6179         * omp-low.c (pass_expand_omp): Convert from a global struct to a
6180         subclass of gimple_opt_pass along with...
6181         (pass_data_expand_omp): ...new pass_data instance and...
6182         (make_pass_expand_omp): ...new function.
6183         (pass_lower_omp): Convert from a global struct to a subclass of
6184         gimple_opt_pass along with...
6185         (pass_data_lower_omp): ...new pass_data instance and...
6186         (make_pass_lower_omp): ...new function.
6187         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
6188         of gimple_opt_pass along with...
6189         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
6190         (make_pass_diagnose_omp_blocks): ...new function.
6191         * passes.c (pass_early_local_passes): Convert from a global struct to
6192         a subclass of simple_ipa_opt_pass along with...
6193         (pass_data_early_local_passes): ...new pass_data instance and...
6194         (make_pass_early_local_passes): ...new function.
6195         (pass_all_early_optimizations): Convert from a global struct to a
6196         subclass of gimple_opt_pass along with...
6197         (pass_data_all_early_optimizations): ...new pass_data instance and...
6198         (make_pass_all_early_optimizations): ...new function.
6199         (pass_all_optimizations): Convert from a global struct to a subclass
6200         of gimple_opt_pass along with...
6201         (pass_data_all_optimizations): ...new pass_data instance and...
6202         (make_pass_all_optimizations): ...new function.
6203         (pass_all_optimizations_g): Convert from a global struct to a subclass
6204         of gimple_opt_pass along with...
6205         (pass_data_all_optimizations_g): ...new pass_data instance and...
6206         (make_pass_all_optimizations_g): ...new function.
6207         (pass_rest_of_compilation): Convert from a global struct to a subclass
6208         of rtl_opt_pass along with...
6209         (pass_data_rest_of_compilation): ...new pass_data instance and...
6210         (make_pass_rest_of_compilation): ...new function.
6211         (pass_postreload): Convert from a global struct to a subclass of
6212         rtl_opt_pass along with...
6213         (pass_data_postreload): ...new pass_data instance and...
6214         (make_pass_postreload): ...new function.
6215         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
6216         subclass of rtl_opt_pass along with...
6217         (pass_data_gcse2): ...new pass_data instance and...
6218         (make_pass_gcse2): ...new function.
6219         * postreload.c (pass_postreload_cse): Convert from a global struct to
6220         a subclass of rtl_opt_pass along with...
6221         (pass_data_postreload_cse): ...new pass_data instance and...
6222         (make_pass_postreload_cse): ...new function.
6223         * predict.c (pass_profile): Convert from a global struct to a subclass
6224         of gimple_opt_pass along with...
6225         (pass_data_profile): ...new pass_data instance and...
6226         (make_pass_profile): ...new function.
6227         (pass_strip_predict_hints): Convert from a global struct to a subclass
6228         of gimple_opt_pass along with...
6229         (pass_data_strip_predict_hints): ...new pass_data instance and...
6230         (make_pass_strip_predict_hints): ...new function.
6231         * recog.c (pass_peephole2): Convert from a global struct to a subclass
6232         of rtl_opt_pass along with...
6233         (pass_data_peephole2): ...new pass_data instance and...
6234         (make_pass_peephole2): ...new function.
6235         (pass_split_all_insns): Convert from a global struct to a subclass of
6236         rtl_opt_pass along with...
6237         (pass_data_split_all_insns): ...new pass_data instance and...
6238         (make_pass_split_all_insns): ...new function.
6239         (pass_split_after_reload): Convert from a global struct to a subclass
6240         of rtl_opt_pass along with...
6241         (pass_data_split_after_reload): ...new pass_data instance and...
6242         (make_pass_split_after_reload): ...new function.
6243         (pass_split_before_regstack): Convert from a global struct to a
6244         subclass of rtl_opt_pass along with...
6245         (pass_data_split_before_regstack): ...new pass_data instance and...
6246         (make_pass_split_before_regstack): ...new function.
6247         (pass_split_before_sched2): Convert from a global struct to a subclass
6248         of rtl_opt_pass along with...
6249         (pass_data_split_before_sched2): ...new pass_data instance and...
6250         (make_pass_split_before_sched2): ...new function.
6251         (pass_split_for_shorten_branches): Convert from a global struct to a
6252         subclass of rtl_opt_pass along with...
6253         (pass_data_split_for_shorten_branches): ...new pass_data instance
6254         and...
6255         (make_pass_split_for_shorten_branches): ...new function.
6256         * ree.c (pass_ree): Convert from a global struct to a subclass of
6257         rtl_opt_pass along with...
6258         (pass_data_ree): ...new pass_data instance and...
6259         (make_pass_ree): ...new function.
6260         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
6261         subclass of rtl_opt_pass along with...
6262         (pass_data_stack_regs): ...new pass_data instance and...
6263         (make_pass_stack_regs): ...new function.
6264         (pass_stack_regs_run): Convert from a global struct to a subclass of
6265         rtl_opt_pass along with...
6266         (pass_data_stack_regs_run): ...new pass_data instance and...
6267         (make_pass_stack_regs_run): ...new function.
6268         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
6269         subclass of rtl_opt_pass along with...
6270         (pass_data_cprop_hardreg): ...new pass_data instance and...
6271         (make_pass_cprop_hardreg): ...new function.
6272         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
6273         subclass of rtl_opt_pass along with...
6274         (pass_data_reginfo_init): ...new pass_data instance and...
6275         (make_pass_reginfo_init): ...new function.
6276         * regmove.c (pass_regmove): Convert from a global struct to a subclass
6277         of rtl_opt_pass along with...
6278         (pass_data_regmove): ...new pass_data instance and...
6279         (make_pass_regmove): ...new function.
6280         * regrename.c (pass_regrename): Convert from a global struct to a
6281         subclass of rtl_opt_pass along with...
6282         (pass_data_regrename): ...new pass_data instance and...
6283         (make_pass_regrename): ...new function.
6284         * reorg.c (pass_delay_slots): Convert from a global struct to a
6285         subclass of rtl_opt_pass along with...
6286         (pass_data_delay_slots): ...new pass_data instance and...
6287         (make_pass_delay_slots): ...new function.
6288         (pass_machine_reorg): Convert from a global struct to a subclass of
6289         rtl_opt_pass along with...
6290         (pass_data_machine_reorg): ...new pass_data instance and...
6291         (make_pass_machine_reorg): ...new function.
6292         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
6293         of rtl_opt_pass along with...
6294         (pass_data_sched): ...new pass_data instance and...
6295         (make_pass_sched): ...new function.
6296         (pass_sched2): Convert from a global struct to a subclass of
6297         rtl_opt_pass along with...
6298         (pass_data_sched2): ...new pass_data instance and...
6299         (make_pass_sched2): ...new function.
6300         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
6301         to a subclass of rtl_opt_pass along with...
6302         (pass_data_stack_ptr_mod): ...new pass_data instance and...
6303         (make_pass_stack_ptr_mod): ...new function.
6304         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
6305         to a subclass of rtl_opt_pass along with...
6306         (pass_data_rtl_store_motion): ...new pass_data instance and...
6307         (make_pass_rtl_store_motion): ...new function.
6308         * tracer.c (pass_tracer): Convert from a global struct to a subclass
6309         of gimple_opt_pass along with...
6310         (pass_data_tracer): ...new pass_data instance and...
6311         (make_pass_tracer): ...new function.
6312         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
6313         to a subclass of gimple_opt_pass along with...
6314         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
6315         (make_pass_diagnose_tm_blocks): ...new function.
6316         (pass_lower_tm): Convert from a global struct to a subclass of
6317         gimple_opt_pass along with...
6318         (pass_data_lower_tm): ...new pass_data instance and...
6319         (make_pass_lower_tm): ...new function.
6320         (pass_tm_init): Convert from a global struct to a subclass of
6321         gimple_opt_pass along with...
6322         (pass_data_tm_init): ...new pass_data instance and...
6323         (make_pass_tm_init): ...new function.
6324         (pass_tm_mark): Convert from a global struct to a subclass of
6325         gimple_opt_pass along with...
6326         (pass_data_tm_mark): ...new pass_data instance and...
6327         (make_pass_tm_mark): ...new function.
6328         (pass_tm_edges): Convert from a global struct to a subclass of
6329         gimple_opt_pass along with...
6330         (pass_data_tm_edges): ...new pass_data instance and...
6331         (make_pass_tm_edges): ...new function.
6332         (pass_tm_memopt): Convert from a global struct to a subclass of
6333         gimple_opt_pass along with...
6334         (pass_data_tm_memopt): ...new pass_data instance and...
6335         (make_pass_tm_memopt): ...new function.
6336         (pass_ipa_tm): Convert from a global struct to a subclass of
6337         simple_ipa_opt_pass along with...
6338         (pass_data_ipa_tm): ...new pass_data instance and...
6339         (make_pass_ipa_tm): ...new function.
6340         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
6341         subclass of gimple_opt_pass along with...
6342         (pass_data_call_cdce): ...new pass_data instance and...
6343         (make_pass_call_cdce): ...new function.
6344         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
6345         subclass of gimple_opt_pass along with...
6346         (pass_data_build_cfg): ...new pass_data instance and...
6347         (make_pass_build_cfg): ...new function.
6348         (pass_split_crit_edges): Convert from a global struct to a subclass of
6349         gimple_opt_pass along with...
6350         (pass_data_split_crit_edges): ...new pass_data instance and...
6351         (make_pass_split_crit_edges): ...new function.
6352         (pass_warn_function_return): Convert from a global struct to a
6353         subclass of gimple_opt_pass along with...
6354         (pass_data_warn_function_return): ...new pass_data instance and...
6355         (make_pass_warn_function_return): ...new function.
6356         (pass_warn_function_noreturn): Convert from a global struct to a
6357         subclass of gimple_opt_pass along with...
6358         (pass_data_warn_function_noreturn): ...new pass_data instance and...
6359         (make_pass_warn_function_noreturn): ...new function.
6360         (pass_warn_unused_result): Convert from a global struct to a subclass
6361         of gimple_opt_pass along with...
6362         (pass_data_warn_unused_result): ...new pass_data instance and...
6363         (make_pass_warn_unused_result): ...new function.
6364         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
6365         a subclass of gimple_opt_pass along with...
6366         (pass_data_merge_phi): ...new pass_data instance and...
6367         (make_pass_merge_phi): ...new function.
6368         * tree-complex.c (pass_lower_complex): Convert from a global struct to
6369         a subclass of gimple_opt_pass along with...
6370         (pass_data_lower_complex): ...new pass_data instance and...
6371         (make_pass_lower_complex): ...new function.
6372         (pass_lower_complex_O0): Convert from a global struct to a subclass of
6373         gimple_opt_pass along with...
6374         (pass_data_lower_complex_O0): ...new pass_data instance and...
6375         (make_pass_lower_complex_O0): ...new function.
6376         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
6377         subclass of gimple_opt_pass along with...
6378         (pass_data_lower_eh): ...new pass_data instance and...
6379         (make_pass_lower_eh): ...new function.
6380         (pass_refactor_eh): Convert from a global struct to a subclass of
6381         gimple_opt_pass along with...
6382         (pass_data_refactor_eh): ...new pass_data instance and...
6383         (make_pass_refactor_eh): ...new function.
6384         (pass_lower_resx): Convert from a global struct to a subclass of
6385         gimple_opt_pass along with...
6386         (pass_data_lower_resx): ...new pass_data instance and...
6387         (make_pass_lower_resx): ...new function.
6388         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
6389         of gimple_opt_pass along with...
6390         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
6391         (make_pass_lower_eh_dispatch): ...new function.
6392         (pass_cleanup_eh): Convert from a global struct to a subclass of
6393         gimple_opt_pass along with...
6394         (pass_data_cleanup_eh): ...new pass_data instance and...
6395         (make_pass_cleanup_eh): ...new function.
6396         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
6397         to a subclass of simple_ipa_opt_pass along with...
6398         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
6399         (make_pass_ipa_lower_emutls): ...new function.
6400         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
6401         a subclass of gimple_opt_pass along with...
6402         (pass_data_if_conversion): ...new pass_data instance and...
6403         (make_pass_if_conversion): ...new function.
6404         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
6405         subclass of gimple_opt_pass along with...
6406         (pass_data_build_ssa): ...new pass_data instance and...
6407         (make_pass_build_ssa): ...new function.
6408         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
6409         global struct to a subclass of gimple_opt_pass along with...
6410         (pass_data_loop_distribution): ...new pass_data instance and...
6411         (make_pass_loop_distribution): ...new function.
6412         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
6413         subclass of gimple_opt_pass along with...
6414         (pass_data_mudflap_1): ...new pass_data instance and...
6415         (make_pass_mudflap_1): ...new function.
6416         (pass_mudflap_2): Convert from a global struct to a subclass of
6417         gimple_opt_pass along with...
6418         (pass_data_mudflap_2): ...new pass_data instance and...
6419         (make_pass_mudflap_2): ...new function.
6420         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
6421         subclass of gimple_opt_pass along with...
6422         (pass_data_mudflap_1): ...new pass_data instance and...
6423         (make_pass_mudflap_1): ...new function.
6424         (pass_mudflap_2): Convert from a global struct to a subclass of
6425         gimple_opt_pass along with...
6426         (pass_data_mudflap_2): ...new pass_data instance and...
6427         (make_pass_mudflap_2): ...new function.
6428         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
6429         gimple_opt_pass along with...
6430         (pass_data_nrv): ...new pass_data instance and...
6431         (make_pass_nrv): ...new function.
6432         (pass_return_slot): Convert from a global struct to a subclass of
6433         gimple_opt_pass along with...
6434         (pass_data_return_slot): ...new pass_data instance and...
6435         (make_pass_return_slot): ...new function.
6436         * tree-object-size.c (pass_object_sizes): Convert from a global struct
6437         to a subclass of gimple_opt_pass along with...
6438         (pass_data_object_sizes): ...new pass_data instance and...
6439         (make_pass_object_sizes): ...new function.
6440         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
6441         global struct to a subclass of gimple_opt_pass along with...
6442         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
6443         and...
6444         (make_pass_cleanup_cfg_post_optimizing): ...new function.
6445         (pass_fixup_cfg): Convert from a global struct to a subclass of
6446         gimple_opt_pass along with...
6447         (pass_data_fixup_cfg): ...new pass_data instance and...
6448         (make_pass_fixup_cfg): ...new function.
6449         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
6450         (make_pass_mudflap_1): ...new function.
6451         (pass_mudflap_2): Replace declaration with that of...
6452         (make_pass_mudflap_2): ...new function.
6453         (pass_asan): Replace declaration with that of...
6454         (make_pass_asan): ...new function.
6455         (pass_asan_O0): Replace declaration with that of...
6456         (make_pass_asan_O0): ...new function.
6457         (pass_tsan): Replace declaration with that of...
6458         (make_pass_tsan): ...new function.
6459         (pass_tsan_O0): Replace declaration with that of...
6460         (make_pass_tsan_O0): ...new function.
6461         (pass_lower_cf): Replace declaration with that of...
6462         (make_pass_lower_cf): ...new function.
6463         (pass_refactor_eh): Replace declaration with that of...
6464         (make_pass_refactor_eh): ...new function.
6465         (pass_lower_eh): Replace declaration with that of...
6466         (make_pass_lower_eh): ...new function.
6467         (pass_lower_eh_dispatch): Replace declaration with that of...
6468         (make_pass_lower_eh_dispatch): ...new function.
6469         (pass_lower_resx): Replace declaration with that of...
6470         (make_pass_lower_resx): ...new function.
6471         (pass_build_cfg): Replace declaration with that of...
6472         (make_pass_build_cfg): ...new function.
6473         (pass_early_tree_profile): Replace declaration with that of...
6474         (make_pass_early_tree_profile): ...new function.
6475         (pass_cleanup_eh): Replace declaration with that of...
6476         (make_pass_cleanup_eh): ...new function.
6477         (pass_sra): Replace declaration with that of...
6478         (make_pass_sra): ...new function.
6479         (pass_sra_early): Replace declaration with that of...
6480         (make_pass_sra_early): ...new function.
6481         (pass_early_ipa_sra): Replace declaration with that of...
6482         (make_pass_early_ipa_sra): ...new function.
6483         (pass_tail_recursion): Replace declaration with that of...
6484         (make_pass_tail_recursion): ...new function.
6485         (pass_tail_calls): Replace declaration with that of...
6486         (make_pass_tail_calls): ...new function.
6487         (pass_tree_loop): Replace declaration with that of...
6488         (make_pass_tree_loop): ...new function.
6489         (pass_tree_loop_init): Replace declaration with that of...
6490         (make_pass_tree_loop_init): ...new function.
6491         (pass_lim): Replace declaration with that of...
6492         (make_pass_lim): ...new function.
6493         (pass_tree_unswitch): Replace declaration with that of...
6494         (make_pass_tree_unswitch): ...new function.
6495         (pass_predcom): Replace declaration with that of...
6496         (make_pass_predcom): ...new function.
6497         (pass_iv_canon): Replace declaration with that of...
6498         (make_pass_iv_canon): ...new function.
6499         (pass_scev_cprop): Replace declaration with that of...
6500         (make_pass_scev_cprop): ...new function.
6501         (pass_empty_loop): Replace declaration with that of...
6502         (make_pass_empty_loop): ...new function.
6503         (pass_record_bounds): Replace declaration with that of...
6504         (make_pass_record_bounds): ...new function.
6505         (pass_graphite): Replace declaration with that of...
6506         (make_pass_graphite): ...new function.
6507         (pass_graphite_transforms): Replace declaration with that of...
6508         (make_pass_graphite_transforms): ...new function.
6509         (pass_if_conversion): Replace declaration with that of...
6510         (make_pass_if_conversion): ...new function.
6511         (pass_loop_distribution): Replace declaration with that of...
6512         (make_pass_loop_distribution): ...new function.
6513         (pass_vectorize): Replace declaration with that of...
6514         (make_pass_vectorize): ...new function.
6515         (pass_slp_vectorize): Replace declaration with that of...
6516         (make_pass_slp_vectorize): ...new function.
6517         (pass_complete_unroll): Replace declaration with that of...
6518         (make_pass_complete_unroll): ...new function.
6519         (pass_complete_unrolli): Replace declaration with that of...
6520         (make_pass_complete_unrolli): ...new function.
6521         (pass_parallelize_loops): Replace declaration with that of...
6522         (make_pass_parallelize_loops): ...new function.
6523         (pass_loop_prefetch): Replace declaration with that of...
6524         (make_pass_loop_prefetch): ...new function.
6525         (pass_iv_optimize): Replace declaration with that of...
6526         (make_pass_iv_optimize): ...new function.
6527         (pass_tree_loop_done): Replace declaration with that of...
6528         (make_pass_tree_loop_done): ...new function.
6529         (pass_ch): Replace declaration with that of...
6530         (make_pass_ch): ...new function.
6531         (pass_ccp): Replace declaration with that of...
6532         (make_pass_ccp): ...new function.
6533         (pass_phi_only_cprop): Replace declaration with that of...
6534         (make_pass_phi_only_cprop): ...new function.
6535         (pass_build_ssa): Replace declaration with that of...
6536         (make_pass_build_ssa): ...new function.
6537         (pass_build_alias): Replace declaration with that of...
6538         (make_pass_build_alias): ...new function.
6539         (pass_build_ealias): Replace declaration with that of...
6540         (make_pass_build_ealias): ...new function.
6541         (pass_dominator): Replace declaration with that of...
6542         (make_pass_dominator): ...new function.
6543         (pass_dce): Replace declaration with that of...
6544         (make_pass_dce): ...new function.
6545         (pass_dce_loop): Replace declaration with that of...
6546         (make_pass_dce_loop): ...new function.
6547         (pass_cd_dce): Replace declaration with that of...
6548         (make_pass_cd_dce): ...new function.
6549         (pass_call_cdce): Replace declaration with that of...
6550         (make_pass_call_cdce): ...new function.
6551         (pass_merge_phi): Replace declaration with that of...
6552         (make_pass_merge_phi): ...new function.
6553         (pass_split_crit_edges): Replace declaration with that of...
6554         (make_pass_split_crit_edges): ...new function.
6555         (pass_pre): Replace declaration with that of...
6556         (make_pass_pre): ...new function.
6557         (pass_profile): Replace declaration with that of...
6558         (make_pass_profile): ...new function.
6559         (pass_strip_predict_hints): Replace declaration with that of...
6560         (make_pass_strip_predict_hints): ...new function.
6561         (pass_lower_complex_O0): Replace declaration with that of...
6562         (make_pass_lower_complex_O0): ...new function.
6563         (pass_lower_complex): Replace declaration with that of...
6564         (make_pass_lower_complex): ...new function.
6565         (pass_lower_vector): Replace declaration with that of...
6566         (make_pass_lower_vector): ...new function.
6567         (pass_lower_vector_ssa): Replace declaration with that of...
6568         (make_pass_lower_vector_ssa): ...new function.
6569         (pass_lower_omp): Replace declaration with that of...
6570         (make_pass_lower_omp): ...new function.
6571         (pass_diagnose_omp_blocks): Replace declaration with that of...
6572         (make_pass_diagnose_omp_blocks): ...new function.
6573         (pass_expand_omp): Replace declaration with that of...
6574         (make_pass_expand_omp): ...new function.
6575         (pass_expand_omp_ssa): Replace declaration with that of...
6576         (make_pass_expand_omp_ssa): ...new function.
6577         (pass_object_sizes): Replace declaration with that of...
6578         (make_pass_object_sizes): ...new function.
6579         (pass_strlen): Replace declaration with that of...
6580         (make_pass_strlen): ...new function.
6581         (pass_fold_builtins): Replace declaration with that of...
6582         (make_pass_fold_builtins): ...new function.
6583         (pass_stdarg): Replace declaration with that of...
6584         (make_pass_stdarg): ...new function.
6585         (pass_early_warn_uninitialized): Replace declaration with that of...
6586         (make_pass_early_warn_uninitialized): ...new function.
6587         (pass_late_warn_uninitialized): Replace declaration with that of...
6588         (make_pass_late_warn_uninitialized): ...new function.
6589         (pass_cse_reciprocals): Replace declaration with that of...
6590         (make_pass_cse_reciprocals): ...new function.
6591         (pass_cse_sincos): Replace declaration with that of...
6592         (make_pass_cse_sincos): ...new function.
6593         (pass_optimize_bswap): Replace declaration with that of...
6594         (make_pass_optimize_bswap): ...new function.
6595         (pass_optimize_widening_mul): Replace declaration with that of...
6596         (make_pass_optimize_widening_mul): ...new function.
6597         (pass_warn_function_return): Replace declaration with that of...
6598         (make_pass_warn_function_return): ...new function.
6599         (pass_warn_function_noreturn): Replace declaration with that of...
6600         (make_pass_warn_function_noreturn): ...new function.
6601         (pass_cselim): Replace declaration with that of...
6602         (make_pass_cselim): ...new function.
6603         (pass_phiopt): Replace declaration with that of...
6604         (make_pass_phiopt): ...new function.
6605         (pass_forwprop): Replace declaration with that of...
6606         (make_pass_forwprop): ...new function.
6607         (pass_phiprop): Replace declaration with that of...
6608         (make_pass_phiprop): ...new function.
6609         (pass_tree_ifcombine): Replace declaration with that of...
6610         (make_pass_tree_ifcombine): ...new function.
6611         (pass_dse): Replace declaration with that of...
6612         (make_pass_dse): ...new function.
6613         (pass_nrv): Replace declaration with that of...
6614         (make_pass_nrv): ...new function.
6615         (pass_rename_ssa_copies): Replace declaration with that of...
6616         (make_pass_rename_ssa_copies): ...new function.
6617         (pass_sink_code): Replace declaration with that of...
6618         (make_pass_sink_code): ...new function.
6619         (pass_fre): Replace declaration with that of...
6620         (make_pass_fre): ...new function.
6621         (pass_check_data_deps): Replace declaration with that of...
6622         (make_pass_check_data_deps): ...new function.
6623         (pass_copy_prop): Replace declaration with that of...
6624         (make_pass_copy_prop): ...new function.
6625         (pass_vrp): Replace declaration with that of...
6626         (make_pass_vrp): ...new function.
6627         (pass_uncprop): Replace declaration with that of...
6628         (make_pass_uncprop): ...new function.
6629         (pass_return_slot): Replace declaration with that of...
6630         (make_pass_return_slot): ...new function.
6631         (pass_reassoc): Replace declaration with that of...
6632         (make_pass_reassoc): ...new function.
6633         (pass_rebuild_cgraph_edges): Replace declaration with that of...
6634         (make_pass_rebuild_cgraph_edges): ...new function.
6635         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
6636         (make_pass_remove_cgraph_callee_edges): ...new function.
6637         (pass_build_cgraph_edges): Replace declaration with that of...
6638         (make_pass_build_cgraph_edges): ...new function.
6639         (pass_local_pure_const): Replace declaration with that of...
6640         (make_pass_local_pure_const): ...new function.
6641         (pass_tracer): Replace declaration with that of...
6642         (make_pass_tracer): ...new function.
6643         (pass_warn_unused_result): Replace declaration with that of...
6644         (make_pass_warn_unused_result): ...new function.
6645         (pass_diagnose_tm_blocks): Replace declaration with that of...
6646         (make_pass_diagnose_tm_blocks): ...new function.
6647         (pass_lower_tm): Replace declaration with that of...
6648         (make_pass_lower_tm): ...new function.
6649         (pass_tm_init): Replace declaration with that of...
6650         (make_pass_tm_init): ...new function.
6651         (pass_tm_mark): Replace declaration with that of...
6652         (make_pass_tm_mark): ...new function.
6653         (pass_tm_memopt): Replace declaration with that of...
6654         (make_pass_tm_memopt): ...new function.
6655         (pass_tm_edges): Replace declaration with that of...
6656         (make_pass_tm_edges): ...new function.
6657         (pass_split_functions): Replace declaration with that of...
6658         (make_pass_split_functions): ...new function.
6659         (pass_feedback_split_functions): Replace declaration with that of...
6660         (make_pass_feedback_split_functions): ...new function.
6661         (pass_strength_reduction): Replace declaration with that of...
6662         (make_pass_strength_reduction): ...new function.
6663         (pass_ipa_lower_emutls): Replace declaration with that of...
6664         (make_pass_ipa_lower_emutls): ...new function.
6665         (pass_ipa_function_and_variable_visibility): Replace declaration with
6666         that of...
6667         (make_pass_ipa_function_and_variable_visibility): ...new function.
6668         (pass_ipa_tree_profile): Replace declaration with that of...
6669         (make_pass_ipa_tree_profile): ...new function.
6670         (pass_early_local_passes): Replace declaration with that of...
6671         (make_pass_early_local_passes): ...new function.
6672         (pass_ipa_whole_program_visibility): Replace declaration with that
6673         of...
6674         (make_pass_ipa_whole_program_visibility): ...new function.
6675         (pass_ipa_lto_gimple_out): Replace declaration with that of...
6676         (make_pass_ipa_lto_gimple_out): ...new function.
6677         (pass_ipa_increase_alignment): Replace declaration with that of...
6678         (make_pass_ipa_increase_alignment): ...new function.
6679         (pass_ipa_inline): Replace declaration with that of...
6680         (make_pass_ipa_inline): ...new function.
6681         (pass_ipa_free_lang_data): Replace declaration with that of...
6682         (make_pass_ipa_free_lang_data): ...new function.
6683         (pass_ipa_free_inline_summary): Replace declaration with that of...
6684         (make_pass_ipa_free_inline_summary): ...new function.
6685         (pass_ipa_cp): Replace declaration with that of...
6686         (make_pass_ipa_cp): ...new function.
6687         (pass_ipa_reference): Replace declaration with that of...
6688         (make_pass_ipa_reference): ...new function.
6689         (pass_ipa_pure_const): Replace declaration with that of...
6690         (make_pass_ipa_pure_const): ...new function.
6691         (pass_ipa_pta): Replace declaration with that of...
6692         (make_pass_ipa_pta): ...new function.
6693         (pass_ipa_lto_finish_out): Replace declaration with that of...
6694         (make_pass_ipa_lto_finish_out): ...new function.
6695         (pass_ipa_tm): Replace declaration with that of...
6696         (make_pass_ipa_tm): ...new function.
6697         (pass_ipa_profile): Replace declaration with that of...
6698         (make_pass_ipa_profile): ...new function.
6699         (pass_ipa_cdtor_merge): Replace declaration with that of...
6700         (make_pass_ipa_cdtor_merge): ...new function.
6701         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
6702         of...
6703         (make_pass_cleanup_cfg_post_optimizing): ...new function.
6704         (pass_init_datastructures): Replace declaration with that of...
6705         (make_pass_init_datastructures): ...new function.
6706         (pass_fixup_cfg): Replace declaration with that of...
6707         (make_pass_fixup_cfg): ...new function.
6708         (pass_expand): Replace declaration with that of...
6709         (make_pass_expand): ...new function.
6710         (pass_instantiate_virtual_regs): Replace declaration with that of...
6711         (make_pass_instantiate_virtual_regs): ...new function.
6712         (pass_rtl_fwprop): Replace declaration with that of...
6713         (make_pass_rtl_fwprop): ...new function.
6714         (pass_rtl_fwprop_addr): Replace declaration with that of...
6715         (make_pass_rtl_fwprop_addr): ...new function.
6716         (pass_jump): Replace declaration with that of...
6717         (make_pass_jump): ...new function.
6718         (pass_jump2): Replace declaration with that of...
6719         (make_pass_jump2): ...new function.
6720         (pass_lower_subreg): Replace declaration with that of...
6721         (make_pass_lower_subreg): ...new function.
6722         (pass_cse): Replace declaration with that of...
6723         (make_pass_cse): ...new function.
6724         (pass_fast_rtl_dce): Replace declaration with that of...
6725         (make_pass_fast_rtl_dce): ...new function.
6726         (pass_ud_rtl_dce): Replace declaration with that of...
6727         (make_pass_ud_rtl_dce): ...new function.
6728         (pass_rtl_dce): Replace declaration with that of...
6729         (make_pass_rtl_dce): ...new function.
6730         (pass_rtl_dse1): Replace declaration with that of...
6731         (make_pass_rtl_dse1): ...new function.
6732         (pass_rtl_dse2): Replace declaration with that of...
6733         (make_pass_rtl_dse2): ...new function.
6734         (pass_rtl_dse3): Replace declaration with that of...
6735         (make_pass_rtl_dse3): ...new function.
6736         (pass_rtl_cprop): Replace declaration with that of...
6737         (make_pass_rtl_cprop): ...new function.
6738         (pass_rtl_pre): Replace declaration with that of...
6739         (make_pass_rtl_pre): ...new function.
6740         (pass_rtl_hoist): Replace declaration with that of...
6741         (make_pass_rtl_hoist): ...new function.
6742         (pass_rtl_store_motion): Replace declaration with that of...
6743         (make_pass_rtl_store_motion): ...new function.
6744         (pass_cse_after_global_opts): Replace declaration with that of...
6745         (make_pass_cse_after_global_opts): ...new function.
6746         (pass_rtl_ifcvt): Replace declaration with that of...
6747         (make_pass_rtl_ifcvt): ...new function.
6748         (pass_into_cfg_layout_mode): Replace declaration with that of...
6749         (make_pass_into_cfg_layout_mode): ...new function.
6750         (pass_outof_cfg_layout_mode): Replace declaration with that of...
6751         (make_pass_outof_cfg_layout_mode): ...new function.
6752         (pass_loop2): Replace declaration with that of...
6753         (make_pass_loop2): ...new function.
6754         (pass_rtl_loop_init): Replace declaration with that of...
6755         (make_pass_rtl_loop_init): ...new function.
6756         (pass_rtl_move_loop_invariants): Replace declaration with that of...
6757         (make_pass_rtl_move_loop_invariants): ...new function.
6758         (pass_rtl_unswitch): Replace declaration with that of...
6759         (make_pass_rtl_unswitch): ...new function.
6760         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
6761         (make_pass_rtl_unroll_and_peel_loops): ...new function.
6762         (pass_rtl_doloop): Replace declaration with that of...
6763         (make_pass_rtl_doloop): ...new function.
6764         (pass_rtl_loop_done): Replace declaration with that of...
6765         (make_pass_rtl_loop_done): ...new function.
6766         (pass_web): Replace declaration with that of...
6767         (make_pass_web): ...new function.
6768         (pass_cse2): Replace declaration with that of...
6769         (make_pass_cse2): ...new function.
6770         (pass_df_initialize_opt): Replace declaration with that of...
6771         (make_pass_df_initialize_opt): ...new function.
6772         (pass_df_initialize_no_opt): Replace declaration with that of...
6773         (make_pass_df_initialize_no_opt): ...new function.
6774         (pass_reginfo_init): Replace declaration with that of...
6775         (make_pass_reginfo_init): ...new function.
6776         (pass_inc_dec): Replace declaration with that of...
6777         (make_pass_inc_dec): ...new function.
6778         (pass_stack_ptr_mod): Replace declaration with that of...
6779         (make_pass_stack_ptr_mod): ...new function.
6780         (pass_initialize_regs): Replace declaration with that of...
6781         (make_pass_initialize_regs): ...new function.
6782         (pass_combine): Replace declaration with that of...
6783         (make_pass_combine): ...new function.
6784         (pass_if_after_combine): Replace declaration with that of...
6785         (make_pass_if_after_combine): ...new function.
6786         (pass_ree): Replace declaration with that of...
6787         (make_pass_ree): ...new function.
6788         (pass_partition_blocks): Replace declaration with that of...
6789         (make_pass_partition_blocks): ...new function.
6790         (pass_match_asm_constraints): Replace declaration with that of...
6791         (make_pass_match_asm_constraints): ...new function.
6792         (pass_regmove): Replace declaration with that of...
6793         (make_pass_regmove): ...new function.
6794         (pass_split_all_insns): Replace declaration with that of...
6795         (make_pass_split_all_insns): ...new function.
6796         (pass_fast_rtl_byte_dce): Replace declaration with that of...
6797         (make_pass_fast_rtl_byte_dce): ...new function.
6798         (pass_lower_subreg2): Replace declaration with that of...
6799         (make_pass_lower_subreg2): ...new function.
6800         (pass_mode_switching): Replace declaration with that of...
6801         (make_pass_mode_switching): ...new function.
6802         (pass_sms): Replace declaration with that of...
6803         (make_pass_sms): ...new function.
6804         (pass_sched): Replace declaration with that of...
6805         (make_pass_sched): ...new function.
6806         (pass_ira): Replace declaration with that of...
6807         (make_pass_ira): ...new function.
6808         (pass_reload): Replace declaration with that of...
6809         (make_pass_reload): ...new function.
6810         (pass_clean_state): Replace declaration with that of...
6811         (make_pass_clean_state): ...new function.
6812         (pass_branch_prob): Replace declaration with that of...
6813         (make_pass_branch_prob): ...new function.
6814         (pass_value_profile_transformations): Replace declaration with that
6815         of...
6816         (make_pass_value_profile_transformations): ...new function.
6817         (pass_postreload_cse): Replace declaration with that of...
6818         (make_pass_postreload_cse): ...new function.
6819         (pass_gcse2): Replace declaration with that of...
6820         (make_pass_gcse2): ...new function.
6821         (pass_split_after_reload): Replace declaration with that of...
6822         (make_pass_split_after_reload): ...new function.
6823         (pass_branch_target_load_optimize1): Replace declaration with that
6824         of...
6825         (make_pass_branch_target_load_optimize1): ...new function.
6826         (pass_thread_prologue_and_epilogue): Replace declaration with that
6827         of...
6828         (make_pass_thread_prologue_and_epilogue): ...new function.
6829         (pass_stack_adjustments): Replace declaration with that of...
6830         (make_pass_stack_adjustments): ...new function.
6831         (pass_peephole2): Replace declaration with that of...
6832         (make_pass_peephole2): ...new function.
6833         (pass_if_after_reload): Replace declaration with that of...
6834         (make_pass_if_after_reload): ...new function.
6835         (pass_regrename): Replace declaration with that of...
6836         (make_pass_regrename): ...new function.
6837         (pass_cprop_hardreg): Replace declaration with that of...
6838         (make_pass_cprop_hardreg): ...new function.
6839         (pass_reorder_blocks): Replace declaration with that of...
6840         (make_pass_reorder_blocks): ...new function.
6841         (pass_branch_target_load_optimize2): Replace declaration with that
6842         of...
6843         (make_pass_branch_target_load_optimize2): ...new function.
6844         (pass_leaf_regs): Replace declaration with that of...
6845         (make_pass_leaf_regs): ...new function.
6846         (pass_split_before_sched2): Replace declaration with that of...
6847         (make_pass_split_before_sched2): ...new function.
6848         (pass_compare_elim_after_reload): Replace declaration with that of...
6849         (make_pass_compare_elim_after_reload): ...new function.
6850         (pass_sched2): Replace declaration with that of...
6851         (make_pass_sched2): ...new function.
6852         (pass_stack_regs): Replace declaration with that of...
6853         (make_pass_stack_regs): ...new function.
6854         (pass_stack_regs_run): Replace declaration with that of...
6855         (make_pass_stack_regs_run): ...new function.
6856         (pass_df_finish): Replace declaration with that of...
6857         (make_pass_df_finish): ...new function.
6858         (pass_compute_alignments): Replace declaration with that of...
6859         (make_pass_compute_alignments): ...new function.
6860         (pass_duplicate_computed_gotos): Replace declaration with that of...
6861         (make_pass_duplicate_computed_gotos): ...new function.
6862         (pass_variable_tracking): Replace declaration with that of...
6863         (make_pass_variable_tracking): ...new function.
6864         (pass_free_cfg): Replace declaration with that of...
6865         (make_pass_free_cfg): ...new function.
6866         (pass_machine_reorg): Replace declaration with that of...
6867         (make_pass_machine_reorg): ...new function.
6868         (pass_cleanup_barriers): Replace declaration with that of...
6869         (make_pass_cleanup_barriers): ...new function.
6870         (pass_delay_slots): Replace declaration with that of...
6871         (make_pass_delay_slots): ...new function.
6872         (pass_split_for_shorten_branches): Replace declaration with that of...
6873         (make_pass_split_for_shorten_branches): ...new function.
6874         (pass_split_before_regstack): Replace declaration with that of...
6875         (make_pass_split_before_regstack): ...new function.
6876         (pass_convert_to_eh_region_ranges): Replace declaration with that
6877         of...
6878         (make_pass_convert_to_eh_region_ranges): ...new function.
6879         (pass_shorten_branches): Replace declaration with that of...
6880         (make_pass_shorten_branches): ...new function.
6881         (pass_set_nothrow_function_flags): Replace declaration with that of...
6882         (make_pass_set_nothrow_function_flags): ...new function.
6883         (pass_dwarf2_frame): Replace declaration with that of...
6884         (make_pass_dwarf2_frame): ...new function.
6885         (pass_final): Replace declaration with that of...
6886         (make_pass_final): ...new function.
6887         (pass_rtl_seqabstr): Replace declaration with that of...
6888         (make_pass_rtl_seqabstr): ...new function.
6889         (pass_release_ssa_names): Replace declaration with that of...
6890         (make_pass_release_ssa_names): ...new function.
6891         (pass_early_inline): Replace declaration with that of...
6892         (make_pass_early_inline): ...new function.
6893         (pass_inline_parameters): Replace declaration with that of...
6894         (make_pass_inline_parameters): ...new function.
6895         (pass_update_address_taken): Replace declaration with that of...
6896         (make_pass_update_address_taken): ...new function.
6897         (pass_convert_switch): Replace declaration with that of...
6898         (make_pass_convert_switch): ...new function.
6899         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
6900         to a subclass of simple_ipa_opt_pass along with...
6901         (pass_data_ipa_tree_profile): ...new pass_data instance and...
6902         (make_pass_ipa_tree_profile): ...new function.
6903         * tree-sra.c (pass_sra_early): Convert from a global struct to a
6904         subclass of gimple_opt_pass along with...
6905         (pass_data_sra_early): ...new pass_data instance and...
6906         (make_pass_sra_early): ...new function.
6907         (pass_sra): Convert from a global struct to a subclass of
6908         gimple_opt_pass along with...
6909         (pass_data_sra): ...new pass_data instance and...
6910         (make_pass_sra): ...new function.
6911         (pass_early_ipa_sra): Convert from a global struct to a subclass of
6912         gimple_opt_pass along with...
6913         (pass_data_early_ipa_sra): ...new pass_data instance and...
6914         (make_pass_early_ipa_sra): ...new function.
6915         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
6916         subclass of gimple_opt_pass along with...
6917         (pass_data_ccp): ...new pass_data instance and...
6918         (make_pass_ccp): ...new function.
6919         (pass_fold_builtins): Convert from a global struct to a subclass of
6920         gimple_opt_pass along with...
6921         (pass_data_fold_builtins): ...new pass_data instance and...
6922         (make_pass_fold_builtins): ...new function.
6923         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
6924         subclass of gimple_opt_pass along with...
6925         (pass_data_copy_prop): ...new pass_data instance and...
6926         (make_pass_copy_prop): ...new function.
6927         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
6928         global struct to a subclass of gimple_opt_pass along with...
6929         (pass_data_rename_ssa_copies): ...new pass_data instance and...
6930         (make_pass_rename_ssa_copies): ...new function.
6931         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
6932         subclass of gimple_opt_pass along with...
6933         (pass_data_dce): ...new pass_data instance and...
6934         (make_pass_dce): ...new function.
6935         (pass_dce_loop): Convert from a global struct to a subclass of
6936         gimple_opt_pass along with...
6937         (pass_data_dce_loop): ...new pass_data instance and...
6938         (make_pass_dce_loop): ...new function.
6939         (pass_cd_dce): Convert from a global struct to a subclass of
6940         gimple_opt_pass along with...
6941         (pass_data_cd_dce): ...new pass_data instance and...
6942         (make_pass_cd_dce): ...new function.
6943         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
6944         subclass of gimple_opt_pass along with...
6945         (pass_data_dominator): ...new pass_data instance and...
6946         (make_pass_dominator): ...new function.
6947         (pass_phi_only_cprop): Convert from a global struct to a subclass of
6948         gimple_opt_pass along with...
6949         (pass_data_phi_only_cprop): ...new pass_data instance and...
6950         (make_pass_phi_only_cprop): ...new function.
6951         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
6952         subclass of gimple_opt_pass along with...
6953         (pass_data_dse): ...new pass_data instance and...
6954         (make_pass_dse): ...new function.
6955         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
6956         a subclass of gimple_opt_pass along with...
6957         (pass_data_forwprop): ...new pass_data instance and...
6958         (make_pass_forwprop): ...new function.
6959         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
6960         struct to a subclass of gimple_opt_pass along with...
6961         (pass_data_tree_ifcombine): ...new pass_data instance and...
6962         (make_pass_tree_ifcombine): ...new function.
6963         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
6964         subclass of gimple_opt_pass along with...
6965         (pass_data_ch): ...new pass_data instance and...
6966         (make_pass_ch): ...new function.
6967         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
6968         subclass of gimple_opt_pass along with...
6969         (pass_data_tree_loop): ...new pass_data instance and...
6970         (make_pass_tree_loop): ...new function.
6971         (pass_tree_loop_init): Convert from a global struct to a subclass of
6972         gimple_opt_pass along with...
6973         (pass_data_tree_loop_init): ...new pass_data instance and...
6974         (make_pass_tree_loop_init): ...new function.
6975         (pass_lim): Convert from a global struct to a subclass of
6976         gimple_opt_pass along with...
6977         (pass_data_lim): ...new pass_data instance and...
6978         (make_pass_lim): ...new function.
6979         (pass_tree_unswitch): Convert from a global struct to a subclass of
6980         gimple_opt_pass along with...
6981         (pass_data_tree_unswitch): ...new pass_data instance and...
6982         (make_pass_tree_unswitch): ...new function.
6983         (pass_predcom): Convert from a global struct to a subclass of
6984         gimple_opt_pass along with...
6985         (pass_data_predcom): ...new pass_data instance and...
6986         (make_pass_predcom): ...new function.
6987         (pass_vectorize): Convert from a global struct to a subclass of
6988         gimple_opt_pass along with...
6989         (pass_data_vectorize): ...new pass_data instance and...
6990         (make_pass_vectorize): ...new function.
6991         (pass_graphite): Convert from a global struct to a subclass of
6992         gimple_opt_pass along with...
6993         (pass_data_graphite): ...new pass_data instance and...
6994         (make_pass_graphite): ...new function.
6995         (pass_graphite_transforms): Convert from a global struct to a subclass
6996         of gimple_opt_pass along with...
6997         (pass_data_graphite_transforms): ...new pass_data instance and...
6998         (make_pass_graphite_transforms): ...new function.
6999         (pass_check_data_deps): Convert from a global struct to a subclass of
7000         gimple_opt_pass along with...
7001         (pass_data_check_data_deps): ...new pass_data instance and...
7002         (make_pass_check_data_deps): ...new function.
7003         (pass_iv_canon): Convert from a global struct to a subclass of
7004         gimple_opt_pass along with...
7005         (pass_data_iv_canon): ...new pass_data instance and...
7006         (make_pass_iv_canon): ...new function.
7007         (pass_scev_cprop): Convert from a global struct to a subclass of
7008         gimple_opt_pass along with...
7009         (pass_data_scev_cprop): ...new pass_data instance and...
7010         (make_pass_scev_cprop): ...new function.
7011         (pass_record_bounds): Convert from a global struct to a subclass of
7012         gimple_opt_pass along with...
7013         (pass_data_record_bounds): ...new pass_data instance and...
7014         (make_pass_record_bounds): ...new function.
7015         (pass_complete_unroll): Convert from a global struct to a subclass of
7016         gimple_opt_pass along with...
7017         (pass_data_complete_unroll): ...new pass_data instance and...
7018         (make_pass_complete_unroll): ...new function.
7019         (pass_complete_unrolli): Convert from a global struct to a subclass of
7020         gimple_opt_pass along with...
7021         (pass_data_complete_unrolli): ...new pass_data instance and...
7022         (make_pass_complete_unrolli): ...new function.
7023         (pass_parallelize_loops): Convert from a global struct to a subclass
7024         of gimple_opt_pass along with...
7025         (pass_data_parallelize_loops): ...new pass_data instance and...
7026         (make_pass_parallelize_loops): ...new function.
7027         (pass_loop_prefetch): Convert from a global struct to a subclass of
7028         gimple_opt_pass along with...
7029         (pass_data_loop_prefetch): ...new pass_data instance and...
7030         (make_pass_loop_prefetch): ...new function.
7031         (pass_iv_optimize): Convert from a global struct to a subclass of
7032         gimple_opt_pass along with...
7033         (pass_data_iv_optimize): ...new pass_data instance and...
7034         (make_pass_iv_optimize): ...new function.
7035         (pass_tree_loop_done): Convert from a global struct to a subclass of
7036         gimple_opt_pass along with...
7037         (pass_data_tree_loop_done): ...new pass_data instance and...
7038         (make_pass_tree_loop_done): ...new function.
7039         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
7040         struct to a subclass of gimple_opt_pass along with...
7041         (pass_data_cse_reciprocals): ...new pass_data instance and...
7042         (make_pass_cse_reciprocals): ...new function.
7043         (pass_cse_sincos): Convert from a global struct to a subclass of
7044         gimple_opt_pass along with...
7045         (pass_data_cse_sincos): ...new pass_data instance and...
7046         (make_pass_cse_sincos): ...new function.
7047         (pass_optimize_bswap): Convert from a global struct to a subclass of
7048         gimple_opt_pass along with...
7049         (pass_data_optimize_bswap): ...new pass_data instance and...
7050         (make_pass_optimize_bswap): ...new function.
7051         (pass_optimize_widening_mul): Convert from a global struct to a
7052         subclass of gimple_opt_pass along with...
7053         (pass_data_optimize_widening_mul): ...new pass_data instance and...
7054         (make_pass_optimize_widening_mul): ...new function.
7055         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
7056         subclass of gimple_opt_pass along with...
7057         (pass_data_phiopt): ...new pass_data instance and...
7058         (make_pass_phiopt): ...new function.
7059         (pass_cselim): Convert from a global struct to a subclass of
7060         gimple_opt_pass along with...
7061         (pass_data_cselim): ...new pass_data instance and...
7062         (make_pass_cselim): ...new function.
7063         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
7064         subclass of gimple_opt_pass along with...
7065         (pass_data_phiprop): ...new pass_data instance and...
7066         (make_pass_phiprop): ...new function.
7067         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
7068         subclass of gimple_opt_pass along with...
7069         (pass_data_pre): ...new pass_data instance and...
7070         (make_pass_pre): ...new function.
7071         (pass_fre): Convert from a global struct to a subclass of
7072         gimple_opt_pass along with...
7073         (pass_data_fre): ...new pass_data instance and...
7074         (make_pass_fre): ...new function.
7075         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
7076         subclass of gimple_opt_pass along with...
7077         (pass_data_reassoc): ...new pass_data instance and...
7078         (make_pass_reassoc): ...new function.
7079         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
7080         subclass of gimple_opt_pass along with...
7081         (pass_data_sink_code): ...new pass_data instance and...
7082         (make_pass_sink_code): ...new function.
7083         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
7084         subclass of gimple_opt_pass along with...
7085         (pass_data_strlen): ...new pass_data instance and...
7086         (make_pass_strlen): ...new function.
7087         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
7088         struct to a subclass of gimple_opt_pass along with...
7089         (pass_data_build_alias): ...new pass_data instance and...
7090         (make_pass_build_alias): ...new function.
7091         (pass_build_ealias): Convert from a global struct to a subclass of
7092         gimple_opt_pass along with...
7093         (pass_data_build_ealias): ...new pass_data instance and...
7094         (make_pass_build_ealias): ...new function.
7095         (pass_ipa_pta): Convert from a global struct to a subclass of
7096         simple_ipa_opt_pass along with...
7097         (pass_data_ipa_pta): ...new pass_data instance and...
7098         (make_pass_ipa_pta): ...new function.
7099         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
7100         subclass of gimple_opt_pass along with...
7101         (pass_data_uncprop): ...new pass_data instance and...
7102         (make_pass_uncprop): ...new function.
7103         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
7104         global struct to a subclass of gimple_opt_pass along with...
7105         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
7106         (make_pass_late_warn_uninitialized): ...new function.
7107         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
7108         to a subclass of gimple_opt_pass along with...
7109         (pass_data_init_datastructures): ...new pass_data instance and...
7110         (make_pass_init_datastructures): ...new function.
7111         (pass_early_warn_uninitialized): Convert from a global struct to a
7112         subclass of gimple_opt_pass along with...
7113         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
7114         (make_pass_early_warn_uninitialized): ...new function.
7115         (pass_update_address_taken): Convert from a global struct to a
7116         subclass of gimple_opt_pass along with...
7117         (pass_data_update_address_taken): ...new pass_data instance and...
7118         (make_pass_update_address_taken): ...new function.
7119         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
7120         struct to a subclass of gimple_opt_pass along with...
7121         (pass_data_release_ssa_names): ...new pass_data instance and...
7122         (make_pass_release_ssa_names): ...new function.
7123         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
7124         subclass of gimple_opt_pass along with...
7125         (pass_data_stdarg): ...new pass_data instance and...
7126         (make_pass_stdarg): ...new function.
7127         * tree-switch-conversion.c (pass_convert_switch): Convert from a
7128         global struct to a subclass of gimple_opt_pass along with...
7129         (pass_data_convert_switch): ...new pass_data instance and...
7130         (make_pass_convert_switch): ...new function.
7131         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
7132         to a subclass of gimple_opt_pass along with...
7133         (pass_data_tail_recursion): ...new pass_data instance and...
7134         (make_pass_tail_recursion): ...new function.
7135         (pass_tail_calls): Convert from a global struct to a subclass of
7136         gimple_opt_pass along with...
7137         (pass_data_tail_calls): ...new pass_data instance and...
7138         (make_pass_tail_calls): ...new function.
7139         * tree-vect-generic.c (pass_lower_vector): Convert from a global
7140         struct to a subclass of gimple_opt_pass along with...
7141         (pass_data_lower_vector): ...new pass_data instance and...
7142         (make_pass_lower_vector): ...new function.
7143         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
7144         gimple_opt_pass along with...
7145         (pass_data_lower_vector_ssa): ...new pass_data instance and...
7146         (make_pass_lower_vector_ssa): ...new function.
7147         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
7148         to a subclass of gimple_opt_pass along with...
7149         (pass_data_slp_vectorize): ...new pass_data instance and...
7150         (make_pass_slp_vectorize): ...new function.
7151         (pass_ipa_increase_alignment): Convert from a global struct to a
7152         subclass of simple_ipa_opt_pass along with...
7153         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
7154         (make_pass_ipa_increase_alignment): ...new function.
7155         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
7156         gimple_opt_pass along with...
7157         (pass_data_vrp): ...new pass_data instance and...
7158         (make_pass_vrp): ...new function.
7159         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
7160         subclass of simple_ipa_opt_pass along with...
7161         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
7162         (make_pass_ipa_free_lang_data): ...new function.
7163         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
7164         gimple_opt_pass along with...
7165         (pass_data_tsan): ...new pass_data instance and...
7166         (make_pass_tsan): ...new function.
7167         (pass_tsan_O0): Convert from a global struct to a subclass of
7168         gimple_opt_pass along with...
7169         (pass_data_tsan_O0): ...new pass_data instance and...
7170         (make_pass_tsan_O0): ...new function.
7171         * var-tracking.c (pass_variable_tracking): Convert from a global
7172         struct to a subclass of rtl_opt_pass along with...
7173         (pass_data_variable_tracking): ...new pass_data instance and...
7174         (make_pass_variable_tracking): ...new function.
7175         * web.c (pass_web): Convert from a global struct to a subclass of
7176         rtl_opt_pass along with...
7177         (pass_data_web): ...new pass_data instance and...
7178         (make_pass_web): ...new function.
7179         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
7180         declaration with that of...
7181         (make_pass_mode_switch_use): ...new function.
7182         (pass_resolve_sw_modes): Replace declaration with that of...
7183         (make_pass_resolve_sw_modes): ...new function.
7184         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
7185         from a global struct to a subclass of rtl_opt_pass along with...
7186         (pass_data_mode_switch_use): ...new pass_data instance and...
7187         (make_pass_mode_switch_use): ...new function.
7188         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
7189         from a global struct to a subclass of rtl_opt_pass along with...
7190         (pass_data_resolve_sw_modes): ...new pass_data instance and...
7191         (make_pass_resolve_sw_modes): ...new function.
7192         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
7193         struct to a subclass of rtl_opt_pass along with...
7194         (pass_data_insert_vzeroupper): ...new pass_data instance and...
7195         (make_pass_insert_vzeroupper): ...new function.
7196         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
7197         global struct to a subclass of rtl_opt_pass along with...
7198         (pass_data_work_around_errata): ...new pass_data instance and...
7199         (make_pass_work_around_errata): ...new function.
7200         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
7201         struct to a subclass of rtl_opt_pass along with...
7202         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
7203         (make_pass_mips_machine_reorg2): ...new function.
7205 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
7207         * passes.c (pass_manager::operator new): New.
7209 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
7211         Handwritten part of conversion of passes to C++ classes.
7213         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
7214         (toplev.o): Add dep on PASS_MANAGER_H.
7215         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
7216         of early local pases to reflect this moving from a global to a
7217         member of gcc::pass_manager.
7218         (cgraph_add_new_function): Likewise.
7219         * lto-cgraph.c (lto_output_node): Update for conversion of
7220         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
7221         * passes.c (opt_pass::clone): New.
7222         (opt_pass::gate): New.
7223         (opt_pass::execute): New.
7224         (opt_pass::opt_pass): New.
7225         (pass_manager::execute_early_local_passes): New.
7226         (pass_manager::execute_pass_mode_switching): new.
7227         (finish_optimization_passes): Convert to...
7228         (pass_manager::finish_optimization_passes): ...this.
7229         (finish_optimization_passes): Update for conversion of passes to
7230         C++ classes.
7231         (register_dump_files_1): Use has_gate since we cannot portably
7232         check a vtable entry against NULL.
7233         (dump_one_pass): Likewise.
7234         (ipa_write_summaries_2): Likewise.
7235         (ipa_write_optimization_summaries_1): Likewise.
7236         (ipa_read_summaries_1): Likewise.
7237         (ipa_read_optimization_summaries_1): Likewise.
7238         (execute_ipa_stmt_fixups): Likewise.
7239         (pass_manager::pass_manager): Rewrite pass-creation, invoking
7240         pass-creation functions rather than wiring up globals, and
7241         storing the results in fields of pass_manager generated using
7242         pass-instances.def.
7243         (pass_manager::dump_profile_report): Update for conversion of
7244         passes to C++ classes.
7245         (pass_manager::execute_ipa_summary_passes): Likewise.
7246         (execute_one_ipa_transform_pass): Likewise.
7247         (execute_one_pass): Use has_gate and has_execute since we cannot
7248         portably check a vtable entry against NULL.
7249         * pass_manager.h (pass_manager::finish_optimization_passes): New.
7250         (pass_manager): Use pass-instances.def to add fields for the
7251         various pass instances.
7252         * toplev.c (finalize): Update for move of
7253         finish_optimization_passes to a method of gcc::pass_manager.
7254         * toplev.h (finish_optimization_passes): Move to method of class
7255         pass_manager.
7256         * tree-pass.h (struct pass_data): New.
7257         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
7258         (opt_pass::gate): Convert to virtual function.
7259         (opt_pass::~opt_pass): New.
7260         (opt_pass::clone): New.
7261         (opt_pass::execute): Convert to virtual function.
7262         (opt_pass::opt_pass): New.
7263         (opt_pass::ctxt_): new.
7264         (gimple_opt_pass): Convert to subclass of opt_pass.
7265         (gimple_opt_pass::gimple_opt_pass): New.
7266         (rtl_opt_pass): Convert to subclass of opt_pass.
7267         (rtl_opt_pass::rtl_opt_pass): New.
7268         (ipa_opt_pass_d): Convert to subclass of opt_pass.
7269         (ipa_opt_pass_d::ipa_opt_pass_d): New.
7270         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
7271         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
7272         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
7273         invocation of pass_mode_switching to reflect this moving from a
7274         global to a member of gcc::pass_manager.
7275         (ix86_option_override): Rework how pass_insert_vzeroupper is
7276         added to the pass_manager to reflect autogenerated changes.
7277         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
7279 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
7281         PR rtl-optimization/57708
7282         * recog.c (peep2_find_free_register): Validate all regs in a
7283         multi-reg mode.
7285 2013-08-05  Jan Hubicka  <jh@suse.cz>
7287         PR lto/57602
7288         * cgraph.c (verify_cgraph_node): Accept local flags from other
7289         partitions.
7290         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
7291         (function_and_variable_visibility): Likewise.
7292         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
7294 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7296         * graph.c (init_graph_slim_pretty_print): Remove.
7297         (print_graph_cfg): Do not call it.  Use local pretty printer.
7298         (start_graph_dump): Likewise.
7300 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7302         * gimple-pretty-print.c (buffer): Remove.
7303         (initialized): Likewise.
7304         (maybe_init_pretty_print): Likewise.
7305         (print_gimple_stmt): Do not call it.  Use non-static local
7306         pretty_printer variable.
7307         (print_gimple_expr): Likewise.
7308         (print_gimple_seq): Likewise.
7309         (gimple_dump_bb): Likewise.
7311 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7313         * asan.c (asan_pp): Remove.
7314         (asan_pp_initialized): Likewise.
7315         (asan_pp_initialize): Likewise.
7316         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
7317         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
7318         (asan_add_global): Likewise.
7320 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7322         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
7323         * diagnostic.h (diagnostic_flush_buffer): Adjust.
7324         * pretty-print.c (pp_formatted_text_data): Likewise.
7325         (pp_indent): Rename from pp_base_indent.
7326         (pp_format): Rename from pp_base_format.
7327         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
7328         (pp_format_verbatim): Rename from pp_base_format_verbatim.
7329         (pp_flush): Rename from pp_base_flush.
7330         (pp_set_line_maximum_length): Rename from
7331         pp_base_set_line_maximum_length.
7332         (pp_clear_output_area): Rename from pp_base_clear_output_area.
7333         (pp_set_prefix): Rename from pp_base_set_prefix.
7334         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
7335         (pp_emit_prefix): Rename from pp_base_emit_prefix.
7336         (pp_append_text): Rename from pp_base_append_text.
7337         (pp_formatted_text): Rename from pp_base_formatted_text.
7338         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
7339         (pp_remaining_character_count_for_line): Rename from
7340         pp_base_remaining_character_count_for_line.
7341         (pp_newline): Rename from pp_base_newline.
7342         (pp_character): Rename from pp_base_character.
7343         (pp_string): Rename from pp_base_string.
7344         (pp_maybe_space): Rename from pp_base_maybe_space.
7345         * asan.c (asan_pp_string): Adjust.
7346         (asan_emit_stack_protection): Likewise.
7347         (asan_add_global): Likewise.
7348         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
7349         * tree-mudflap.c (mf_varname_tree): Likewise.
7350         * tree-pretty-print.c (pp_tree_identifier): Rename from
7351         pp_base_tree_identifier.
7352         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
7353         Declare as function.
7355 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7357         * pretty-print.h (pp_bar_bar): New.
7358         (pp_ampersand_ampersand): Likewise.
7359         (pp_less_equal): Likewise.
7360         (pp_greater_equal): Likewise.
7361         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
7362         printer functions instead of pp_string or operators and punctuators.
7363         (dump_gimple_call): Likewise.
7364         (dump_gimple_omp_for): Likewise.
7365         (dump_gimple_transaction): Likewise.
7366         (dump_gimple_phi): Likewise.
7367         (pp_gimple_stmt_1): Likewise.
7368         * sched-vis.c (print_insn): Likewise.
7369         * tree-mudflap.c (mf_varname_tree): Likewise.
7370         * tree-pretty-print.c (dump_block_node): Likewise.
7371         (dump_generic_node): Likewise.
7373 2013-08-02  Jan Hubicka  <jh@suse.cz>
7375         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
7376         boundaries.
7377         * lto-streamer-out.c (tree_is_indexable): Results decls and
7378         parm decls are not indexable.
7379         (DFS_write_tree_body): Do not follow args and results.
7380         (hash_tree): Likewise.
7381         (output_functions): Rearrange so struct function is needed
7382         only when real body is output; be able to also ouptut abstract
7383         functions; output DECL_ARGUMENTS and DECL_RESULT.
7384         (lto_output): When not in WPA, ale store abstract functions.
7385         (write_symbol): Do not care about RESULT_DECL.
7386         (output_symbol_p): Handle correctly sbtract decls.
7387         * lto-streamer-in.c (input_function): Rearrange so struct
7388         function can be NULL at entry; allow streaming of
7389         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
7390         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
7391         sanity check during LTO.
7392         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
7393         RESULT_DECl and DECL_ARGUMENTS.
7394         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
7395         Likewise.
7397 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7399         * pretty-print.h (pp_underscore): New.
7400         (pp_comma): Tidy.
7401         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
7402         printer functions instead of pp_character.
7403         (dump_binary_rhs): Likewise.
7404         (dump_ternary_rhs): Likewise.
7405         (dump_gimple_call_args): Likewise.
7406         (pp_points_to_solution): Likewise.
7407         (dump_gimple_call): Likewise.
7408         (dump_gimple_switch): Likewise.
7409         (dump_gimple_cond): Likewise.
7410         (dump_gimple_bind): Likewise.
7411         (dump_gimple_try): Likewise.
7412         (dump_gimple_omp_for): Likewise.
7413         (dump_gimple_omp_continue): Likewise.
7414         (dump_gimple_omp_single): Likewise.
7415         (dump_gimple_omp_sections): Likewise.
7416         (dump_gimple_omp_block): Likewise.
7417         (dump_gimple_omp_critical): Likewise.
7418         (dump_gimple_transaction): Likewise.
7419         (dump_gimple_asm): Likewise.
7420         (dump_gimple_phi): Likewise.
7421         (dump_gimple_omp_parallel): Likewise.
7422         (dump_gimple_omp_task): Likewise.
7423         (dump_gimple_omp_atomic_load): Likewise.
7424         (dump_gimple_omp_atomic_store): Likewise.
7425         (dump_gimple_mem_ops): Likewise.
7426         (pp_gimple_stmt_1): Likewise.
7427         (pp_cfg_jump): Likewise.
7428         (dump_implicit_edges): Likewise.
7429         (gimple_dump_bb_for_graph): Likewise.
7430         * graph.c (draw_cfg_node): Likewise.
7431         * langhooks.c (lhd_print_error_function): Likewise.
7432         * sched-vis.c (print_exp): Likewise.
7433         (print_value): Likewise.
7434         (print_pattern): Likewise.
7435         (print_insn): Likewise.
7436         (rtl_dump_bb_for_graph): Likewise.
7437         * tree-pretty-print.c (dump_function_declaration): Likewise.
7438         (dump_array_domain): Likewise.
7439         (dump_omp_clause): Likewise.
7440         (dump_location): Likewise.
7441         (dump_generic_node): Likewise.
7442         (print_struct_decl): Likewise.
7443         * diagnostic.c (diagnostic_show_locus): Use pp_space.
7445 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7447         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
7448         candidate table when replacing a candidate statement.
7449         (replace_rhs_if_not_dup): Likewise.
7450         (replace_one_candidate): Likewise.
7452 2013-08-02  Jan Hubicka  <jh@suse.cz>
7453             Martin Liska  <marxin.liska@gmail.com>
7455         * cgraphunit.c (add_new_function): Fix logic when adding from
7456         late IPA pass.
7457         (assemble_thunk): Rename to ...
7458         (expand_thunk); .. this one; export; get it working with
7459         general functions; make produced gimple valid.
7460         * cgraph.h (expand_thunk): Declare.
7462 2013-08-02  Jan Hubicka  <jh@suse.cz>
7464         * ipa-cp.c (gather_context_independent_values): Use
7465         ipa_get_param_move_cost.
7466         (get_replacement_map): Remove PARAM; move parameter folding
7467         into tree-inline.c
7468         (create_specialized_node): Update.
7469         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
7470         assert that we have gimple body; update move_cost.
7471         (count_formal_params): Assert that we have gimple body.
7472         (ipa_dump_param): New function.
7473         (ipa_alloc_node_params): Break out from ...
7474         (ipa_initialize_node_params): ... here.
7475         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
7476         (ipa_write_node_info): Stream move costs.
7477         (ipa_read_node_info): Read move costs.
7478         (ipa_update_after_lto_read): Do not recompute node params.
7479         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
7480         (ipa_get_param): Check we are not in WPA.
7481         (ipa_get_param_move_cost): New.
7482         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
7483         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
7484         parm numbers to be present.
7486 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
7488         PR rtl-optimization/58048
7489         * lra-constraints.c (process_alt_operands): Don't check asm
7490         operand on register.
7492 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
7494         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
7495         the implied StoreLoad barrier for atomic operations if before.
7497 2013-08-02  Jan Hubicka  <jh@suse.cz>
7498             Martin Liska  <marxin.liska@gmail.com>
7500         * cgraph.c (cgraph_function_body_availability): Do not check
7501         cgraph flags.
7502         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
7503         symtab_node_availability): Declare.
7504         * ipa.c (can_replace_by_local_alias): New.
7505         (function_and_variable_visibility): Use it.
7506         * symtab.c (symtab_for_node_and_aliases,
7507         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
7509 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
7511         PR rtl-optimization/57963
7512         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
7513         (lra_constraints): Use them.
7515 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
7517         * config/arm/types.md (define_attr "type"): Add "load_acq"
7518         and "store_rel".
7519         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
7520         changes.
7521         (cortex_a53_store1): Likewise.
7523 2013-08-01  Jan Hubicka  <jh@suse.cz>
7525         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
7526         partitions are not needed.
7528 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
7530         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
7531         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
7532         MAYBE_NON_Q_CLASS_P where appropriate.
7534 2013-08-01  Jan Hubicka  <jh@suse.cz>
7536         * cgraph.h (release_function_body): Declare.
7537         * tree.c (free_lang_data_in_decl): Free, parameters and return values
7538         of unused delcarations.
7540 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7542         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
7543         RTL form when subtracting a constant.
7545 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7547         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
7548         Generate canonical plus rtx with negated immediate instead of minus
7549         where appropriate.
7550         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
7552 2013-08-01  Jan Hubicka  <jh@suse.cz>
7554         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
7555         (cgraph_release_function_body): Likewise.
7556         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
7557         * cgraph.h (cgrpah_node): Rename abstract_and_needed
7558         to used_as_abstract_origin.
7559         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
7560         symbols used as abstract origins.
7561         * cgraphunit.c (analyze_functions): Update.
7562         * ipa.c (symtab_remove_unreachable_nodes): Recompute
7563         used_as_abstract_origin.
7564         * tree-inline.c (tree_function_versioning): Update
7565         used_as_abstract_origin; be ready for DECL_RESULT and
7566         DECL_ARGUMENTS to be NULL.
7568         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
7569         for abstract functions.
7570         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
7571         real symbols.
7573 2013-08-01  Jan Hubicka  <jh@suse.cz>
7575         * profile.c (compute_value_histograms): Fix thinko.
7577 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
7579         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
7580         aarch-common-protos.h to extra_headers.
7581         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
7582         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
7583         * config/aarch64/t-aarch64 (aarch-common.o): Define.
7585 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
7587         * config/aarch64/aarch64.md (define_attr "type"): Delete.
7588         Include "../arm/types.md".  Define "type" attribute for all patterns.
7589         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
7590         attribute changes.
7592 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
7594         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
7595         to support power8 load fusion.
7596         (fusion_gpr_mem_load): Likewise.
7598         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
7600         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
7601         declarations for power8 load fusion.
7602         (emit_fusion_gpr_load): Likewise.
7604         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
7605         tuning for power8, turn on fusion mode by default.  Turn on sign
7606         extending fusion mode if normal fusion mode is on, and we are at
7607         -O2 or -O3.
7608         (fusion_gpr_load_p): New function, return true if we can fuse an
7609         addis instruction with a dependent load to a GPR.
7610         (emit_fusion_gpr_load): Emit the instructions for power8 load
7611         fusion to GPRs.
7613         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
7614         (VSX load fusion peepholes): New peepholes to fuse together an
7615         addi instruction with a VSX load instruction.
7617         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
7618         peepholes to fuse an addis instruction with a load to a GPR base
7619         register.  If we are supporting sign extending fusions, convert
7620         sign extending loads to zero extending loads and add an explicit
7621         sign extension.
7623 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
7625         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
7626         aarch-common-protos.h to extra_headers.
7627         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
7628         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
7629         (arm_early_store_addr_dep): Likewise.
7630         (arm_no_early_alu_shift_dep): Likewise.
7631         (arm_no_early_alu_shift_value_dep): Likewise.
7632         (arm_no_early_mul_dep): Likewise.
7633         (arm_no_early_store_addr_dep): Likewise.
7634         (arm_mac_accumulator_is_mul_result): Likewise.
7635         (arm_mac_accumulator_is_result): Likewise.
7636         * config/arm/aarch-common.c: ... here.  New file.
7637         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
7638         here to ...
7639         (arm_early_store_addr_dep): Likewise.
7640         (arm_no_early_alu_shift_dep): Likewise.
7641         (arm_no_early_alu_shift_value_dep): Likewise.
7642         (arm_no_early_mul_dep): Likewise.
7643         (arm_no_early_store_addr_dep): Likewise.
7644         (arm_mac_accumulator_is_mul_result): Likewise.
7645         (arm_mac_accumulator_is_result): Likewise.
7646         * config/arm/aarch-common-protos.h: ... here.  New file.
7647         * config/arm/t-arm (aarch-common.o): Define.
7649 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
7651         * config/arm/arm.md: Include new file "types.md".
7652         (define_attr "type"): Move from here to ...
7653         (define_attr "mul32"): Likewise.
7654         (define_attr "mul64"): Likewise.
7655         * config/arm/types.md: ... here.  New file.
7657 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7659         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
7660         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
7662 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7664         * gen-pass-instances.awk: Fix offset of substr().
7666 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
7668         * Makefile.in (pass-instances.def): New.
7669         (passes.o): Replace dependency on passes.def with one on
7670         pass-instances.def
7672         * gen-pass-instances.awk: New.
7674         * passes.c (pass_manager::pass_manager): Use pass-instances.def
7675         rather than passes.def, updating local definition of NEXT_PASS
7676         macro to add an extra NUM parameter (currently unused).
7678 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
7680         * Makefile.in (PASS_MANAGER_H): New.
7681         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
7682         (passes.o): Likewise.
7683         (statistics.o): Likewise.
7684         (cgraphunit.o): Likewise.
7685         (context.o): Depend on PASS_MANAGER_H.
7687         * pass_manager.h: New.
7689         * cgraphunit.c (cgraph_add_new_function): Update for moves
7690         of globals to fields of pass_manager.
7691         (analyze_function): Likewise.
7692         (expand_function): Likewise.
7693         (ipa_passes): Likewise.
7694         (compile): Likewise.
7696         * context.c (context::context): New.
7697         * context.h  (context::context): New.
7698         (context::get_passes): New.
7699         (context::passes_): New.
7701         * lto-cgraph.c (input_node): Update for moves of globals to
7702         fields of pass_manager.
7704         * passes.c (all_passes): Remove, in favor of a field of the
7705         same name within the new class pass_manager.
7706         (all_small_ipa_passes): Likewise.
7707         (all_lowering_passes): Likewise.
7708         (all_regular_ipa_passes): Likewise.
7709         (all_late_ipa_passes): Likewise.
7710         (all_lto_gen_passes): Likewise.
7711         (passes_by_id): Likewise.
7712         (passes_by_id_size): Likewise.
7713         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7714         the new class pass_manager.
7715         (set_pass_for_id): Convert to...
7716         (pass_manager::set_pass_for_id): ...method.
7717         (get_pass_for_id): Convert to...
7718         (pass_manager::get_pass_for_id): ...method.
7719         (register_one_dump_file): Move body of implementation into...
7720         (pass_manager::register_one_dump_file): ...here.
7721         (register_dump_files_1): Convert to...
7722         (pass_manager::register_dump_files_1): ...method.
7723         (register_dump_files): Convert to...
7724         (pass_manager::register_dump_files): ...method.
7725         (create_pass_tab): Update for moves of globals to fields of
7726         pass_manager.
7727         (dump_passes): Move body of implementation into...
7728         (pass_manager::dump_passes): ...here.
7729         (register_pass): Move body of implementation into...
7730         (pass_manager::register_pass): ...here.
7731         (init_optimization_passes): Convert into...
7732         (pass_manager::pass_manager): ...constructor for new
7733         pass_manager class, and initialize the pass_lists array.
7734         (check_profile_consistency): Update for moves of globals to
7735         fields of pass_manager.
7736         (dump_profile_report): Move body of implementation into...
7737         (pass_manager::dump_profile_report): ...here.
7738         (ipa_write_summaries_1): Update for moves of pass lists from
7739         being globals to fields of pass_manager.
7740         (ipa_write_optimization_summaries): Likewise.
7741         (ipa_read_summaries):  Likewise.
7742         (ipa_read_optimization_summaries): Likewise.
7743         (execute_all_ipa_stmt_fixups): Likewise.
7745         * statistics.c (statistics_fini): Update for moves of globals to
7746         fields of pass_manager.
7748         * toplev.c (general_init): Replace call to
7749         init_optimization_passes with construction of the pass_manager
7750         instance.
7752         * tree-pass.h (all_passes): Remove, in favor of a field of the
7753         same name within the new class pass_manager.
7754         (all_small_ipa_passes): Likewise.
7755         (all_lowering_passes): Likewise.
7756         (all_regular_ipa_passes): Likewise.
7757         (all_lto_gen_passes): Likewise.
7758         (all_late_ipa_passes): Likewise.
7759         (passes_by_id): Likewise.
7760         (passes_by_id_size): Likewise.
7761         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7762         the new class pass_manager.
7763         (get_pass_for_id): Remove.
7765 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
7767         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
7768         configs.
7770 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
7772         * arm.md (mulhi3): New expand pattern.
7774 2013-07-30  Jan Hubicka  <jh@suse.cz>
7775             Martin Liska  <marxin.liska@gmail.com>
7777         * profile.c (compute_value_histograms): Do not ICE when
7778         there is mismatch only on some counters.
7780 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7782         PR rtl-optimization/57637
7783         * function.c (move_insn_for_shrink_wrap): Also check the
7784         GEN set of the LIVE problem for the liveness analysis
7785         if it exists, otherwise give up.
7787 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7789         PR tree-optimization/57993
7790         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
7791         replaced statement in the candidate table.
7792         (phi_add_costs): Return infinite cost when the hidden basis does
7793         not dominate all phis on which the candidate is dependent.
7794         (replace_one_candidate): Record replaced statement in the
7795         candidate table.
7797 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
7799         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
7800         (ashlv2si3): New expander.
7801         (*ashlv2si3_i): New define_insn_and_split.
7802         * predicates.md (float_operation): Allow patterns with three
7803         basic sub-patterns.
7805         PR rtl-optimization/58021
7806         * mode-switching.c (create_pre_exit): Always split off preceding
7807         insns if we are not at the basic block head.
7809 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
7811         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
7812         (UCLIBC_DYNAMIC_LINKER): New macro.
7813         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
7814         `-mnan=2008'.
7815         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
7816         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
7817         `-mnan=2008'.
7818         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
7819         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
7820         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
7821         for SF and DF modes.  Use ieee_quad_format for TF mode.
7822         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
7823         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
7824         (mips_option_override): Handle `-mnan=legacy'.
7825         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
7826         `-mabs=2008' and `-mnan=2008'.
7827         (OPTION_DEFAULT_SPECS): Add "nan" default.
7828         (ASM_SPEC): Handle `-mnan='.
7829         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
7830         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
7831         comment accordingly.
7832         (neg<mode>2): Likewise.
7833         * config/mips/mips.opt (mabs, mnan): New options.
7834         * doc/install.texi (Configuration): Document `--with-nan=' option.
7835         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
7836         `-mnan=' options.
7837         (MIPS Options): Document them.
7838         * config.gcc <mips*-*-*>: Handle `--with-nan='.
7839         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
7840         * configure: Regenerate.
7841         * config.in: Regenerate.
7843 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
7845         * config/i386/i386.md (float post-reload splitters): Do not check
7846         for subregs of SSE registers.
7848 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
7849             H.J. Lu  <hongjiu.lu@intel.com>
7851         PR target/57954
7852         PR target/57988
7853         * config/i386/i386.md (post-reload splitter
7854         to avoid partial SSE reg dependency stalls): New pattern.
7856 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
7858         * config/s390/s390.md ("movcc"): Swap load and store instructions.
7860 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
7862         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
7863         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
7865 2013-07-26  Cary Coutant  <ccoutant@google.com>
7867         * dwarf2out.c (die_checksum_ordered): Don't include template
7868         instantiations in signature.
7869         (is_template_parameter): New function.
7870         (is_template_instantiation): New function.
7871         (generate_skeleton_bottom_up): Don't include template instantiations
7872         in type unit DIE.
7873         (generate_skeleton): Likewise.
7874         (break_out_comdat_types): Move recursive call to break out nested
7875         types earlier.
7876         (prune_unused_types_mark_generic_parms_dies): Call
7877         is_template_parameter.
7879 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
7881         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
7882         uses vector registers.
7883         * config/aarch64/iterators.md: Add attributes rtn and vas.
7885 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7886             Richard Earnshaw  <richard.earnshaw@arm.com>
7888         * combine.c (simplify_comparison): Re-canonicalize operands
7889         where appropriate.
7890         * config/arm/arm.md (movcond_addsi): New splitter.
7892 2013-07-25  Sterling Augustine  <saugustine@google.com>
7894         * dwarf2out.c (size_of_pubnames): Move code to...
7895         (include_pubname_in_output): ...here.  New.
7896         (want_pubnames): Rearrange.
7897         (output_pubnames): Call include_pubname_in_output.  Move assertion.
7899 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7901         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
7903 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7905         PR target/38836
7906         * doc/extend.texi: Remove obsolete builtins. Fix
7907         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
7909 2013-07-25  Jan Hubicka  <jh@suse.cz>
7911         * cgraph.c (release_function_body): Break out from ...
7912         (cgraph_release_function_body): ... this one; also release DECL_RESULT
7913         and DECL_ARGUMENTS.
7914         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
7915         old_tree in the map.
7916         (create_specialized_node): Update.
7917         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
7918         into index.
7919         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
7920         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
7921         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
7922         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
7923         DECL_RESULT.
7925 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7927         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
7928         addsi3_carryin_alt2_<optab>): Correct output template.
7930 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7932         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
7933         Adjust for arm_restrict_it.
7934         Remove trailing whitespace.
7936 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
7938         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
7939         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
7941         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
7943 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
7945         PR rtl-optimization/57960
7946         * lra-constraints.c (process_alt_operands): Use the right mode
7947         when checking strict_low.
7949 2013-07-25  Jan Hubicka  <jh@suse.cz>
7951         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
7952         * cgraph.c (cgraph_remove_node): Do not release function body
7953         when in cgraph streaming.
7954         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
7955         in other partitions are not considered reachable; fix handling of
7956         clones.
7958 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7960         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
7962 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7964         PR target/19599
7965         PR target/57731
7966         PR target/57837
7967         * config/arm/arm.md ("*sibcall_insn): Replace use of
7968         Ss with US. Adjust output for v5 and v4t.
7969         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
7971         * config/arm/constraints.md ("Ss"): Rename to US.
7973 2013-07-25  Terry Guo  <terry.guo@arm.com>
7975         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
7976         shift_add/shift_sub0/shift_sub1 RTXs.
7978 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
7979             Anton Blanchard  <anton@au1.ibm.com>
7981         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
7982         (altivec_vpks<VI_char>ss): Likewise.
7983         (altivec_vpks<VI_char>us): Likewise.
7984         (altivec_vpku<VI_char>us): Likewise.
7985         (altivec_vpku<VI_char>um): Likewise.
7987 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
7989         Introduce context class.
7991         * Makefile.in (CONTEXT_H): New.
7992         (OBJS): Add context.o.
7993         (toplev.o): Add CONTEXT_H to dependencies.
7994         (context.o): New.
7996         * toplev.c (general_init): Create the singleton gcc::context instance.
7998         * context.c: New.
8000         * context.h: New.
8002 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
8004         PR rtl-optimization/57968
8005         * mode-switching.c (create_pre_exit): Allow instructions that
8006         don't set a return register to need a non-exit mode.
8008 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
8009             Anton Blanchard  <anton@au1.ibm.com>
8011         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
8012         operands to vperm for little endian.
8013         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
8014         of lvsl to create the control mask for a vperm for little endian.
8016 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8017             Anton Blanchard  <anton@au1.ibm.com>
8019         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
8020         two operands for little-endian.
8022 2013-07-23  Steve Ellcey  <sellcey@mips.com>
8024         * config/mips/mips.c (mips_case_values_threshold): New.
8025         (TARGET_CASE_VALUES_THRESHOLD): Define.
8027 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8028             Anton Blanchard  <anton@au1.ibm.com>
8030         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
8031         selection of field for vector splat in little endian mode.
8033 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
8035         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
8036         expanders to rs6000.md.
8037         (ior<mode>3): Likewise.
8038         (and<mode>3): Likewise.
8039         (one_cmpl<mode>2): Likewise.
8040         (nor<mode>3): Likewise.
8041         (andc<mode>3): Likewise.
8042         (eqv<mode>3): Likewise.
8043         (nand<mode>3): Likewise.
8044         (orc<mode>3): Likewise.
8046         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
8047         declaration.
8049         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
8050         to split multi-word logical operations.
8051         (rs6000_split_logical_di): Likewise.
8052         (rs6000_split_logical): Likewise.
8054         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
8055         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
8056         and allow TImode operations in 32-bit.
8057         (vsx_and<mode>3_64bit): Likewise.
8058         (vsx_ior<mode>3_32bit): Likewise.
8059         (vsx_ior<mode>3_64bit): Likewise.
8060         (vsx_xor<mode>3_32bit): Likewise.
8061         (vsx_xor<mode>3_64bit): Likewise.
8062         (vsx_one_cmpl<mode>2_32bit): Likewise.
8063         (vsx_one_cmpl<mode>2_64bit): Likewise.
8064         (vsx_nor<mode>3_32bit): Likewise.
8065         (vsx_nor<mode>3_64bit): Likewise.
8066         (vsx_andc<mode>3_32bit): Likewise.
8067         (vsx_andc<mode>3_64bit): Likewise.
8068         (vsx_eqv<mode>3_32bit): Likewise.
8069         (vsx_eqv<mode>3_64bit): Likewise.
8070         (vsx_nand<mode>3_32bit): Likewise.
8071         (vsx_nand<mode>3_64bit): Likewise.
8072         (vsx_orc<mode>3_32bit): Likewise.
8073         (vsx_orc<mode>3_64bit): Likewise.
8075         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
8076         logical types in GPRs.
8078         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
8079         logical insns to rs6000.md, and allow TImode operations in
8080         32-bit.
8081         (altivec_ior<mode>3): Likewise.
8082         (altivec_xor<mode>3): Likewise.
8083         (altivec_one_cmpl<mode>2): Likewise.
8084         (altivec_nor<mode>3): Likewise.
8085         (altivec_andc<mode>3): Likewise.
8087         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
8088         attributes for moving the 128-bit logical operations into
8089         rs6000.md.
8090         (BOOL_REGS_OUTPUT): Likewise.
8091         (BOOL_REGS_OP1): Likewise.
8092         (BOOL_REGS_OP2): Likewise.
8093         (BOOL_REGS_UNARY): Likewise.
8094         (BOOL_REGS_AND_CR0): Likewise.
8095         (one_cmpl<mode>2): Add support for DI logical operations on
8096         32-bit, splitting the operations to 32-bit.
8097         (anddi3): Likewise.
8098         (iordi3): Likewise.
8099         (xordi3): Likewise.
8100         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
8101         changes to combine the 32/64-bit code, allow logical operations on
8102         TI mode in 32-bit, and to use similar match_operator patterns like
8103         scalar mode uses.  Combine the Altivec and VSX code for logical
8104         operations, and move it here.
8105         (ior<mode>3, 128-bit types): Likewise.
8106         (xor<mode>3, 128-bit types): Likewise.
8107         (one_cmpl<mode>3, 128-bit types): Likewise.
8108         (nor<mode>3, 128-bit types): Likewise.
8109         (andc<mode>3, 128-bit types): Likewise.
8110         (eqv<mode>3, 128-bit types): Likewise.
8111         (nand<mode>3, 128-bit types): Likewise.
8112         (orc<mode>3, 128-bit types): Likewise.
8113         (and<mode>3_internal): Likewise.
8114         (bool<mode>3_internal): Likewise.
8115         (boolc<mode>3_internal1): Likewise.
8116         (boolc<mode>3_internal2): Likewise.
8117         (boolcc<mode>3_internal1): Likewise.
8118         (boolcc<mode>3_internal2): Likewise.
8119         (eqv<mode>3_internal1): Likewise.
8120         (eqv<mode>3_internal2): Likewise.
8121         (one_cmpl1<mode>3_internal): Likewise.
8123 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
8125         * config/microblaze/microblaze.c (microblaze_expand_prologue):
8126         Rename flag_stack_usage to flag_stack_usage_info.
8128 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
8130         * config/microblaze/sync.md: New file.
8131         * config/microblaze/microblaze.md: Include sync.md
8132         * config/microblaze/microblaze.c: Add print_operand 'y'.
8133         * config/microblaze/constraints.md: Add memory_contraint
8134         'Q' which is a single register.
8136 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
8138         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
8140 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
8142         * reload.c (find_reloads): Exit loop once we find this operand
8143         cannot be reloaded somehow for this alternative.
8145         * reload.c (find_reloads): Exit loop once we find a hard register.
8147         * rtlanal.c (computed_jump_p): Exit loop once we find label
8148         reference is used.
8150         * i386.c (ix86_pad_returns): Exit loop after setting replace.
8152         * cfgloopmanip.c (remove_path): Exit loop after setting
8153         irred_invalidated.
8155         * gensupport.c (subst_dup): Avoid loop if code is not
8156         MATCH_DUP nor MATCH_OP_DUP.
8158 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
8160         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
8162 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8164         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
8165         true for SP_REGNUM if mode == ptr_mode.
8166         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
8167         with value R0_REGNUM + 31.
8169 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8171         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
8172         pad pointer-typed argument downward.
8174 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8176         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
8177         and __ILP32__ when the ILP32 model is in use.
8179 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8181         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
8182         (aarch64_load_symref_appropriately): In the case of
8183         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
8184         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
8185         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
8186         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
8187         if 'mode' doesn't equal to 'ptr_mode'.
8188         (aarch64_output_mi_thunk): Add an assertion on the alignment of
8189         'vcall_offset'; change to call aarch64_emit_move differently depending
8190         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
8191         to calculate the upper bound of 'vcall_offset'.
8192         (aarch64_cannot_force_const_mem): Change to also return true if
8193         mode != ptr_mode.
8194         (aarch64_legitimize_reload_address): In the case of large
8195         displacements, add new local variable 'xmode' and an assertion
8196         based on it; change to use 'xmode' to generate the new rtx and
8197         reload.
8198         (aarch64_asm_trampoline_template): Change to generate the template
8199         differently depending on TARGET_ILP32 or not; change to use
8200         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
8201         (aarch64_trampoline_size): Removed.
8202         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
8203         and replace immediate literals with it.  Change to use 'ptr_mode'
8204         instead of 'DImode' and call convert_memory_address if the mode
8205         of 'fnaddr' doesn't equal to 'ptr_mode'.
8206         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
8207         to output symbol.
8208         (aarch64_elf_asm_destructor): Likewise.
8209         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
8210         on TARGET_ILP32 instead of aarch64_trampoline_size.
8211         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
8212         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
8213         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
8214         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
8215         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
8216         (storewb_pair<GPI:mode>_<P:mode>): ... this.
8217         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
8218         depending on the value of 'mode'.
8219         (add_losym_<mode>): New.
8220         (ldr_got_small_<mode>): New, based on ldr_got_small.
8221         (ldr_got_small): Remove.
8222         (ldr_got_small_sidi): New.
8223         * config/aarch64/iterators.md (P): New.
8224         (PTR): Change to 'ptr_mode' in the condition.
8226 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8228         * config.gcc (aarch64*-*-*): Support --with-abi.
8229         (aarch64*-*-elf): Support --with-multilib-list.
8230         (aarch64*-*-linux*): Likewise.
8231         (supported_defaults): Add abi to aarch64*-*-*.
8232         * configure.ac: Mention AArch64 for --with-multilib-list.
8233         * configure: Re-generated.
8234         * config/aarch64/biarchilp32.h: New file.
8235         * config/aarch64/biarchlp64.h: New file.
8236         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
8237         (ABI_SPEC): Ditto.
8238         (MULTILIB_DEFAULTS): Ditto.
8239         (DRIVER_SELF_SPECS): Ditto.
8240         (ASM_SPEC): Update to also substitute -mabi.
8241         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
8242         file whose name depends on -mabi= and -mbig-endian.
8243         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
8244         TARGET_ILP32.
8245         (POINTER_SIZE): New define.
8246         (POINTERS_EXTEND_UNSIGNED): Ditto.
8247         (enum aarch64_abi_type): New enumeration tag.
8248         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
8249         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
8250         (TARGET_ILP32): New define.
8251         * config/aarch64/aarch64.opt (mabi): New.
8252         (aarch64_abi): New.
8253         (ilp32, lp64): New values for -mabi.
8254         * config/aarch64/t-aarch64 (comma): New define.
8255         (MULTILIB_OPTIONS): Ditto.
8256         (MULTILIB_DIRNAMES): Ditto.
8257         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
8258         * doc/invoke.texi: Document -mabi for AArch64.
8260 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
8262         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
8264 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
8265             Anton Blanchard  <anton@au1.ibm.com>
8267         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
8268         endianness when selecting field to splat.
8270 2013-07-22  Eric Christopher  <echristo@gmail.com>
8272         * dwarf2out.c (die_odr_checksum): New function to use
8273         CHECKSUM_ macros and ULEB128 for DIE tag.
8274         (generate_type_signature): Use.
8276 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
8278         * config.gcc (sparc*-*-*): Accept leon3 processor.
8279         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
8280         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
8281         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
8282         * config/sparc/sparc.opt (enum processor_type): Add leon3.
8283         (mfix-ut699): Adjust comment.
8284         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
8285         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
8286         (CPP_CPU_SPEC): Likewise.
8287         (ASM_CPU_SPEC): Likewise.
8288         * config/sparc/sparc.c (leon3_cost): New constant.
8289         (sparc_option_override): Add leon3 support.
8290         (mem_ref): New function.
8291         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
8292         (sparc_do_work_around_errata): Look into the instruction in the delay
8293         slot and adjust accordingly.  Add fix for the data cache nullify issues
8294         of the UT699.  Change insertion position for the NOP.
8295         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
8296         (leon3_load): New reservation.
8297         (leon_store): Bump latency to 2.
8298         (grfpu): New automaton.
8299         (grfpu_alu): New unit.
8300         (grfpu_ds): Likewise.
8301         (leon_fp_alu): Adjust.
8302         (leon_fp_mult): Delete.
8303         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
8304         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
8305         * config/sparc/sparc.md (cpu): Add leon3.
8306         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
8307         (swapsi): Likewise.
8308         (atomic_test_and_set): Likewise.
8309         (ldstub): Likewise.
8311 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
8313         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
8314         default for R5900 targets.
8315         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
8316         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
8317         * config/mips/mips.c (mips_option_override): Report an error for
8318         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
8319         for -march=r5900 -mhard-float.
8321 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
8323         * df-problems.c (can_move_insns_across): Exit loop once we
8324         find a non-fixed, non-global register.
8326         * ipa-pure-const.c (propagate_nothrow): Exit loop after
8327         setting can_throw.
8329         * omega.c (omega_eliminate_red): Break after setting red_found.
8330         (omega_problem_has_red_equations): Similarly after setting found.
8331         (omega_query_variable): Similarly after setting coupled.
8333 2013-07-22  Marek Polacek  <polacek@redhat.com>
8335         * gimplify.c: Don't include gimple.h twice.
8337 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8339         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
8340         instead of TARGET_THUMB1.
8341         (Pz): New constraint.
8342         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
8343         encodings.
8344         (compare_negsi_si): Likewise.
8345         (compare_addsi2_op0): Likewise.
8346         (compare_addsi2_op1): Likewise.
8347         (addsi3_carryin_<optab>): Likewise.
8348         (addsi3_carryin_alt2_<optab>): Likewise.
8349         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
8350         for arm_restrict_it.
8351         (subsi3_carryin): Likewise.
8352         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
8353         (minmax_arithsi): Disable for arm_restrict_it.
8354         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
8355         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
8356         (satsi_<SAT:code>_shift): Likewise.
8357         (arm_shiftsi3): Add alternative for 16-bit encoding.
8358         (arm32_movhf): Disable for arm_restrict_it.
8359         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
8360         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
8362 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
8364         * config/arm/arm.md (attribute "insn"): Delete.
8365         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
8366         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
8367         (not_shiftsi): Update for attribute change.
8368         (not_shiftsi_compare0): Likewise.
8369         (not_shiftsi_compare0_scratch): Likewise.
8370         (arm_one_cmplsi2): Likewise.
8371         (thumb1_one_cmplsi2): Likewise.
8372         (notsi_compare0): Likewise.
8373         (notsi_compare0_scratch): Likewise.
8374         (thumb1_movdi_insn): Likewise.
8375         (arm_movsi_insn): Likewise.
8376         (movhi_insn_arch4): Likewise.
8377         (movhi_bytes): Likewise.
8378         (arm_movqi_insn): Likewise.
8379         (thumb1_movqi_insn): Likewise.
8380         (arm32_movhf): Likewise.
8381         (thumb1_movhf): Likewise.
8382         (arm_movsf_soft_insn): Likewise.
8383         (thumb1_movsf_insn): Likewise.
8384         (thumb_movdf_insn): Likewise.
8385         (movsicc_insn): Likewise.
8386         (movsfcc_soft_insn): Likewise.
8387         (and_scc): Likewise.
8388         (cond_move): Likewise.
8389         (if_move_not): Likewise.
8390         (if_not_move): Likewise.
8391         (if_shift_move): Likewise.
8392         (if_move_shift): Likewise.
8393         (if_shift_shift): Likewise.
8394         (if_not_arith): Likewise.
8395         (if_arith_not): Likewise.
8396         (cond_move_not): Likewise.
8397         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
8398         (neon_mov<mode>): Likewise.
8399         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
8400         (thumb2_movsi_vfp): Likewise.
8401         (movsf_vfp): Likewise.
8402         (thumb2_movsf_vfp): Likewise.
8403         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8404         change.
8405         (cortexa7_older_only): Likewise.
8406         (cortexa7_younger): Likewise.
8407         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
8408         (1020alu_shift_op): Likewise.
8409         (1020alu_shift_reg_op): Likewise.
8410         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
8411         (alu_shift_op): Likewise.
8412         (alu_shift_reg_op): Likewise.
8413         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
8414         (11_alu_shift_op): Likewise.
8415         (11_alu_shift_reg_op): Likewise.
8416         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
8417         (9_alu_shift_reg_op): Likewise.
8418         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
8419         change.
8420         (cortex_a15_alu_shift): Likewise.
8421         (cortex_a15_alu_shift_reg): Likewise.
8422         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
8423         (cortex_a5_alu_shift): Likewise.
8424         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
8425         change.
8426         (cortex_a53_alu_shift): Likewise.
8427         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
8428         change.
8429         (cortex_a7_alu_reg): Likewise.
8430         (cortex_a7_alu_shift): Likewise.
8431         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
8432         (cortex_a8_alu_shift): Likewise.
8433         (cortex_a8_alu_shift_reg): Likewise.
8434         (cortex_a8_mov): Likewise.
8435         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
8436         (cortex_a9_dp_shift): Likewise.
8437         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
8438         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
8439         (cortex_r4_mov): Likewise.
8440         (cortex_r4_alu_shift): Likewise.
8441         (cortex_r4_alu_shift_reg): Likewise.
8442         * config/arm/fa526.md (526_alu_op): Update for attribute change.
8443         (526_alu_shift_op): Likewise.
8444         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
8445         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
8446         (626te_alu_shift_op): Likewise.
8447         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
8448         (726te_alu_op): Likewise.
8449         (726te_alu_shift_op): Likewise.
8450         (726te_alu_shift_reg_op): Likewise.
8451         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
8452         (mp626_alu_shift_op): Likewise.
8453         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
8454         (pj4_alu_e1_conds): Likewise.
8455         (pj4_alu): Likewise.
8456         (pj4_alu_conds): Likewise.
8457         (pj4_shift): Likewise.
8458         (pj4_shift_conds): Likewise.
8459         (pj4_alu_shift): Likewise.
8460         (pj4_alu_shift_conds): Likewise.
8462 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8464         * config/arm/predicates.md (shiftable_operator_strict_it):
8465         New predicate.
8466         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
8467         Disable cond_exec version for arm_restrict_it.
8468         (thumb2_smaxsi3): Convert to generate cond_exec.
8469         (thumb2_sminsi3): Likewise.
8470         (thumb32_umaxsi3): Likewise.
8471         (thumb2_uminsi3): Likewise.
8472         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
8473         (thumb2_neg_abssi2): Likewise.
8474         (thumb2_mov_scc): Add alternative for 16-bit encoding.
8475         (thumb2_movsicc_insn): Adjust alternatives.
8476         (thumb2_mov_negscc): Disable for arm_restrict_it.
8477         (thumb2_mov_negscc_strict_it): New pattern.
8478         (thumb2_mov_notscc_strict_it): New pattern.
8479         (thumb2_mov_notscc): Disable for arm_restrict_it.
8480         (thumb2_ior_scc): Likewise.
8481         (thumb2_ior_scc_strict_it): New pattern.
8482         (thumb2_cond_move): Adjust for arm_restrict_it.
8483         (thumb2_cond_arith): Disable for arm_restrict_it.
8484         (thumb2_cond_arith_strict_it): New pattern.
8485         (thumb2_cond_sub): Adjust for arm_restrict_it.
8486         (thumb2_movcond): Likewise.
8487         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
8488         (thumb2_zero_extendhisi2_v6): Likewise.
8489         (thumb2_zero_extendqisi2_v6): Likewise.
8490         (orsi_notsi_si): Likewise.
8491         (orsi_not_shiftsi_si): Likewise.
8493 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
8495         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
8496         instruction sequence is 1 byte shorter.
8498 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
8500         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
8501         it is not needed after split.
8503 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
8505         PR target/51784
8506         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
8507         second label for nonlocal goto receivers. Don't output pic base labels
8508         unless we're producing PIC; mark that action unreachable().
8509         (ix86_save_reg): If the function contains a nonlocal label, save the
8510         PIC base reg.
8511         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
8512         * config/darwin.c (emitted_pic_label_num): New GTY.
8513         (update_pic_label_number_if_needed): New.
8514         (machopic_output_function_base_name): Adjust for nonlocal receiver
8515         case.
8516         (machopic_should_output_picbase_label): New.
8517         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
8518         (nonlocal_goto_receiver): New insn and split.
8520 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
8522         * config/aarch64/aarch64-builtins.c
8523         (aarch64_fold_builtin): Fold abs in all modes.
8524         * config/aarch64/aarch64-simd-builtins.def
8525         (abs): Enable for all modes.
8526         * config/aarch64/arm_neon.h
8527         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
8528         (vabs_f64): Add missing intrinsic.
8530 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
8532         * config/aarch64/arm_neon.h (vabs_s64): New function
8534 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
8536         PR target/57516
8537         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
8538         * config/avr/avr.md (adjust_len): Add `round'.
8539         * config/avr/avr-protos.h (avr_out_round): New prototype.
8540         (avr_out_plus): Add `out_label' argument.
8541         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
8542         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
8543         Handle the case where `insn' is just a pattern.
8544         (avr_out_bitop): Handle the case where `insn' is just a pattern.
8545         (avr_out_round): New function.
8546         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
8548 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
8550         * config/microblaze/microblaze.c (microblaze_expand_prologue):
8551         Add check for flag_stack_usage to handle -fstack-usage support
8553 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
8555         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
8556         interaction for new Power8 flags and VSX.
8558 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
8560         PR middle-end/57698
8561         * tree-inline.c (expand_call_inline): Emit errors during
8562         early_inlining only if optimization is not turned on.
8564 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8566         * passes.def: New.
8568         * passes.c (init_optimization_passes): Move the construction of
8569         the pass hierarchy into a new passes.def file.
8571         * Makefile.in (passes.o): Add dependency on passes.def.
8573 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8575         * passes.c (init_optimization_passes): Introduce macros for
8576         constructing the tree of passes (INSERT_PASSES_AFTER,
8577         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
8578         TERMINATE_PASS_LIST).
8580 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
8581             Wei Mi  <wmi@google.com>
8583         PR rtl-optimization/57878
8584         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
8585         top.
8586         (reload_pseudo_compare_func): Check nregs first for reload
8587         pseudos.
8589 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8591         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
8593 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
8595         * read-rtl.c (validate_const_int): Once an invalid character is
8596         seen, quit the loop.
8598         * gengtype.c (write_roots): Similarly once we find the "deletable"
8599         or "if_marked" option.
8601 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
8603         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
8604         "xtab" and "sat".  Move value "clz" from here to ...
8605         (attriubte "type"): ... here.
8606         (satsi_<SAT:code>): Delete "insn" attribute.
8607         (satsi_<SAT:code>_shift): Likewise.
8608         (arm_zero_extendqisi2addsi): Likewise.
8609         (arm_extendqisi2addsi): Likewise.
8610         (clzsi2): Update for attribute changes.
8611         (rbitsi2): Likewise.
8612         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
8613         attribute.
8614         (arm_usatsihi): Likewise.
8615         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
8617 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
8619         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
8620         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
8621         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
8622         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
8623         in alphabetical order.
8624         (attribute "core_cycles"): Update for attribute changes.
8625         (arm_addsi3): Likewise.
8626         (addsi3_compare0): Likewise.
8627         (addsi3_compare0_scratch): Likewise.
8628         (addsi3_compare_op1): Likewise.
8629         (addsi3_compare_op2): Likewise.
8630         (compare_addsi2_op0): Likewise.
8631         (compare_addsi2_op1): Likewise.
8632         (addsi3_carryin_shift_<optab>): Likewise.
8633         (subsi3_carryin_shift): Likewise.
8634         (rsbsi3_carryin_shift): Likewise.
8635         (arm_subsi3_insn): Likewise.
8636         (subsi3_compare0): Likewise.
8637         (subsi3_compare): Likewise.
8638         (arm_andsi3_insn): Likewise.
8639         (thumb1_andsi3_insn): Likewise.
8640         (andsi3_compare0): Likewise.
8641         (andsi3_compare0_scratch): Likewise.
8642         (zeroextractsi_compare0_scratch
8643         (andsi_not_shiftsi_si): Likewise.
8644         (iorsi3_insn): Likewise.
8645         (iorsi3_compare0): Likewise.
8646         (iorsi3_compare0_scratch): Likewise.
8647         (arm_xorsi3): Likewise.
8648         (thumb1_xorsi3_insn): Likewise.
8649         (xorsi3_compare0): Likewise.
8650         (xorsi3_compare0_scratch): Likewise.
8651         (satsi_<SAT:code>_shift): Likewise.
8652         (rrx): Likewise.
8653         (arm_shiftsi3): Likewise.
8654         (shiftsi3_compare0): Likewise.
8655         (not_shiftsi): Likewise.
8656         (not_shiftsi_compare0): Likewise.
8657         (not_shiftsi_compare0_scratch): Likewise.
8658         (arm_one_cmplsi2): Likewise.
8659         (thumb_one_complsi2): Likewise.
8660         (notsi_compare0): Likewise.
8661         (notsi_compare0_scratch): Likewise.
8662         (thumb1_zero_extendhisi2): Likewise.
8663         (arm_zero_extendhisi2): Likewise.
8664         (arm_zero_extendhisi2_v6): Likewise.
8665         (arm_zero_extendhisi2addsi): Likewise.
8666         (thumb1_zero_extendqisi2): Likewise.
8667         (thumb1_zero_extendqisi2_v6): Likewise.
8668         (arm_zero_extendqisi2): Likewise.
8669         (arm_zero_extendqisi2_v6): Likewise.
8670         (arm_zero_extendqisi2addsi): Likewise.
8671         (thumb1_extendhisi2): Likewise.
8672         (arm_extendhisi2): Likewise.
8673         (arm_extendhisi2_v6): Likewise.
8674         (arm_extendqisi): Likewise.
8675         (arm_extendqisi_v6): Likewise.
8676         (arm_extendqisi2addsi): Likewise.
8677         (thumb1_extendqisi2): Likewise.
8678         (thumb1_movdi_insn): Likewise.
8679         (arm_movsi_insn): Likewise.
8680         (movsi_compare0): Likewise.
8681         (movhi_insn_arch4): Likewise.
8682         (movhi_bytes): Likewise.
8683         (arm_movqi_insn): Likewise.
8684         (thumb1_movqi_insn): Likewise.
8685         (arm32_movhf): Likewise.
8686         (thumb1_movhf): Likewise.
8687         (arm_movsf_soft_insn): Likewise.
8688         (thumb1_movsf_insn): Likewise.
8689         (movdf_soft_insn): Likewise.
8690         (thumb_movdf_insn): Likewise.
8691         (arm_cmpsi_insn): Likewise.
8692         (cmpsi_shiftsi): Likewise.
8693         (cmpsi_shiftsi_swp): Likewise.
8694         (arm_cmpsi_negshiftsi_si): Likewise.
8695         (movsicc_insn): Likewise.
8696         (movsfcc_soft_insn): Likewise.
8697         (arith_shiftsi): Likewise.
8698         (arith_shiftsi_compare0
8699         (arith_shiftsi_compare0_scratch
8700         (sub_shiftsi): Likewise.
8701         (sub_shiftsi_compare0
8702         (sub_shiftsi_compare0_scratch
8703         (and_scc): Likewise.
8704         (cond_move): Likewise.
8705         (if_plus_move): Likewise.
8706         (if_move_plus): Likewise.
8707         (if_move_not): Likewise.
8708         (if_not_move): Likewise.
8709         (if_shift_move): Likewise.
8710         (if_move_shift): Likewise.
8711         (if_shift_shift): Likewise.
8712         (if_not_arith): Likewise.
8713         (if_arith_not): Likewise.
8714         (cond_move_not): Likewise.
8715         (thumb1_ashlsi3): Set type attribute.
8716         (thumb1_ashrsi3): Likewise.
8717         (thumb1_lshrsi3): Likewise.
8718         (thumb1_rotrsi3): Likewise.
8719         (shiftsi3_compare0_scratch): Likewise.
8720         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
8721         (neon_mov<mode>): Likewise.
8722         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
8723         attribute changes.
8724         (thumb2_movsi_insn): Likewise.
8725         (thumb2_cmpsi_neg_shiftsi): Likewise.
8726         (thumb2_extendqisi_v6): Likewise.
8727         (thumb2_zero_extendhisi2_v6): Likewise.
8728         (thumb2_zero_extendqisi2_v6): Likewise.
8729         (thumb2_shiftsi3_short): Likewise.
8730         (thumb2_addsi3_compare0_scratch): Likewise.
8731         (orsi_not_shiftsi_si): Likewise.
8732         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
8733         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
8734         changes.
8735         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
8736         (1020alu_shift_op): Likewise.
8737         (1020alu_shift_reg_op): Likewise.
8738         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
8739         (alu_shift_op): Likewise.
8740         (alu_shift_reg_op): Likewise.
8741         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
8742         (11_alu_shift_op): Likewise.
8743         (11_alu_shift_reg_op): Likewise.
8744         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
8745         (9_alu_shift_reg_op): Likewise.
8746         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
8747         changes.
8748         (cortex_a15_alu_shift): Likewise.
8749         (cortex_a15_alu_shift_reg): Likewise.
8750         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
8751         changes.
8752         (cortex_a5_alu_shift): Likewise.
8753         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
8754         changes.
8755         (cortex_a53_alu_shift): Likewise.
8756         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
8757         changes.
8758         (cortex_a7_alu_reg): Likewise.
8759         (cortex_a7_alu_shift): Likewise.
8760         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
8761         changes.
8762         (cortex_a8_alu_shift): Likewise.
8763         (cortex_a8_alu_shift_reg): Likewise.
8764         (cortex_a8_mov): Likewise.
8765         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
8766         (cortex_a9_dp_shift): Likewise.
8767         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
8768         changes.
8769         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
8770         changes.
8771         (cortex_r4_mov): Likewise.
8772         (cortex_r4_alu_shift): Likewise.
8773         (cortex_r4_alu_shift_reg): Likewise.
8774         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
8775         (526_alu_shift_op): Likewise.
8776         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
8777         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
8778         (626te_alu_shift_op): Likewise.
8779         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
8780         (726te_alu_op): Likewise.
8781         (726te_alu_shift_op): Likewise.
8782         (726te_alu_shift_reg_op): Likewise.
8783         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
8784         (mp626_alu_shift_op): Likewise.
8785         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
8786         (pj4_alu_e1_conds): Likewise.
8787         (pj4_alu): Likewise.
8788         (pj4_alu_conds): Likewise.
8789         (pj4_shift): Likewise.
8790         (pj4_shift_conds): Likewise.
8791         (pj4_alu_shift): Likewise.
8792         (pj4_alu_shift_conds): Likewise.
8793         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8794         changes.
8795         (cortexa7_older_only): Likewise.
8796         (cortexa7_younger): Likewise.
8798 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8800         * ipa-pure-const.c (generate_summary): Rename to...
8801         (pure_const_generate_summary): ... this.
8803 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
8805         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
8807 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
8809         PR target/57909
8810         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
8811         usage in HI mode.
8813 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8815         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
8816         enabled without -march=zEC12.
8817         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
8818         flags to be set.
8820 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8822         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
8823         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
8824         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
8825         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
8826         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
8827         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
8828         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
8829         ISA_HAS_FP4.
8830         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
8831         and ISA_HAS_NMADD3_NMSUB3.
8832         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
8833         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
8834         (nmsub4<mode>, nmsub3<mode>): Likewise.
8835         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
8837 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8839         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
8840         TARGET_MIPS5400 checking.
8842 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
8843             Peter Bergner  <bergner@vnet.ibm.com>
8845         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
8846         registers in the comment.
8847         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
8848         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
8849         rather than FIRST_PSEUDO_REGISTERS.
8851 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
8853         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
8854         enable extra ISA flags with TARGET_HTM.
8856 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8858         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
8859         Fix comment typos.
8861 2013-07-15  Cong Hou  <congh@google.com>
8863         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
8864         in compare function for sorting.
8866 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
8868         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
8869         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
8870         * config/rs6000/rs6000.opt: Add -mhtm option.
8871         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
8872         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
8873         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8874         __HTM__ if the HTM instructions are available.
8875         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
8876         htm_spr_reg_operand): New define_predicates.
8877         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
8878         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
8879         Include htm.md.
8880         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
8881         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
8882         HTM builtin functions.
8883         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
8884         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
8885         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
8886         (rs6000_builtin_mask_calculate): Likewise.
8887         (rs6000_option_override_internal): Likewise.
8888         (bdesc_htm): Add new HTM builtin support.
8889         (htm_spr_num): New function.
8890         (htm_spr_regno): Likewise.
8891         (rs6000_htm_spr_icode): Likewise.
8892         (htm_expand_builtin): Likewise.
8893         (htm_init_builtins): Likewise.
8894         (rs6000_expand_builtin): Add support for HTM builtin functions.
8895         (rs6000_init_builtins): Likewise.
8896         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
8897         option.
8898         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
8899         (TARGET_HTM, MASK_HTM): Define macros.
8900         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
8901         (FIXED_REGISTERS): Likewise.
8902         (CALL_USED_REGISTERS): Likewise.
8903         (CALL_REALLY_USED_REGISTERS): Likewise.
8904         (REG_ALLOC_ORDER): Likewise.
8905         (enum reg_class): Likewise.
8906         (REG_CLASS_NAMES): Likewise.
8907         (REG_CLASS_CONTENTS): Likewise.
8908         (REGISTER_NAMES): Likewise.
8909         (ADDITIONAL_REGISTER_NAMES): Likewise.
8910         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
8911         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
8912         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
8913         * config/rs6000/htm.md: New file.
8914         * config/rs6000/htmintrin.h: New file.
8915         * config/rs6000/htmxlintrin.h: New file.
8917 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8919         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8920         Define SYMBOL_TINY_GOT, update comment.
8921         * config/aarch64/aarch64.c
8922         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
8923         (aarch64_expand_mov_immediate): Likewise.
8924         (aarch64_print_operand): Likewise.
8925         (aarch64_classify_symbol): Likewise.
8926         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
8927         (ldr_got_tiny): Define.
8929 2013-07-13  Tobias Grosser  <tobias@grosser.es>
8931         PR tree-optimization/54094
8932         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
8933           scheduling dimension for the parallelism check from the polyhedral
8934           information in the AST.
8935         * graphite-dependences.c (carries_deps): Do not assume the schedule is
8936           in 2D + 1 form.
8938 2013-07-13  Jason Merrill  <jason@redhat.com>
8940         * print-tree.c (debug_vec_tree): Use debug_raw.
8941         (debug_raw (vec<tree, va_gc> &)): New.
8942         (debug_raw (vec<tree, va_gc> *)): New.
8943         * tree.h: Declare them.
8945 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
8947         * ifcvt.c (ifcvt_after_combine): New static variable.
8948         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
8949         for size.
8950         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
8951         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
8952         rest_of_handle_if_after_reload): Pass new argument for if_convert.
8954 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
8956         * config/mips/mips.c (mips_expand_call): Remove empty statement.
8958 2013-07-12  Michael Matz  <matz@suse.de>
8960         PR middle-end/55771
8961         * convert.c (convert_to_real): Reject non-float inner types.
8963 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
8965         * config/aarch64/aarch64-protos.h
8966         (aarch64_simd_immediate_valid_for_move): Remove.
8967         * config/aarch64/aarch64.c (simd_immediate_info): New member.
8968         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
8969         cases.
8970         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
8972 2013-07-11  Steve Ellcey  <sellcey@mips.com>
8974         * config/mips/mips.c (mips_conditional_register_usage): Do not
8975         use t[0-7] registers in MIPS16 mode when optimizing for size.
8977 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
8979         * config/i386/i386.c (dispatch_function_versions): Fix array
8980         indexing of function_version_info to match actual_versions.
8982 2013-07-11  Teresa Johnson  <tejohnson@google.com>
8984         * vec.h (struct va_gc): Move release out-of-line.
8985         (va_gc::release): Call ggc_free on released vec.
8987 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
8989         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
8990         Require GOT register as additional operand in UNSPEC.
8991         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
8992         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8993         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8994         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
8995         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
8996         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
8997         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
8999 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
9001         PR target/57631
9002         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
9003         name seen by assembler/linker rather if available.
9005 2013-07-11  Andreas Schwab  <schwab@suse.de>
9007         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
9009 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
9011         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
9013 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
9015         * doc/tm.texi.in: Move hook documentation to ....
9016         * target.def: ... here.
9018         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
9019         text on @hook line.
9020         * doc/tm.texi: Regenerate.
9022 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
9024         PR c++/57869
9025         * doc/invoke.texi: Document -Wconditionally-supported.
9027 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
9029         PR target/57844
9030         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
9031         of my_fp.
9033 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
9035         PR target/57506
9036         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
9037         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
9038         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
9039         Remove duplicate devices.
9040         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
9041         * config/avr/t-multilib: Regenerate.
9042         * config/avr/avr-tables.opt: Regenerate.
9043         * doc/avr-mmcu.texi: Regenerate.
9045 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
9047         PR target/56987
9048         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
9050 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
9052         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
9053         the cost of MULT when optimizing for size.
9055 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9057         * config/cr16/cr16-protos.h: Don't include target.h.
9059 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
9061         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
9062         adjust register size for TDmode and TFmode for VSX registers.
9064 2013-07-08  Kai Tietz  <ktietz@redhat.com>
9066         PR target/56892
9067         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
9068         hook_bool_const_tree_true.
9070 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9072         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
9073         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
9074         * config/s390/s390.md: Define FPR*_REGNUM constants.
9075         Fix FPR2_REGNUM constant (18 -> 17).
9076         ("*trunc<BFP:mode><DFP_ALL:mode>2")
9077         ("*trunc<DFP_ALL:mode><BFP:mode>2")
9078         ("trunc<BFP:mode><DFP_ALL:mode>2")
9079         ("trunc<DFP_ALL:mode><BFP:mode>2")
9080         ("*extend<BFP:mode><DFP_ALL:mode>2")
9081         ("*extend<DFP_ALL:mode><BFP:mode>2")
9082         ("extend<BFP:mode><DFP_ALL:mode>2")
9083         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
9084         FPR4_REGNUM.
9086 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
9088         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
9090 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9092         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
9093         and cfun_fpr_bit_p to cfun_fpr_save_p.
9094         (s390_frame_area, s390_register_info, s390_frame_info)
9095         (s390_emit_prologue, s390_emit_epilogue)
9096         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
9097         register numbers.
9098         * config/s390/s390.h: Define *_REGNUM macros for floating point
9099         register numbers.
9101 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
9103         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
9105 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
9107         PR rtl-optimization/57786
9108         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
9109         and break out of the loop when it is set to false.
9111 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
9113         PR target/57819
9114         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
9115         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
9116         (const_int 63)) 0)).
9117         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
9118         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
9119         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
9121         PR rtl-optimization/57829
9122         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
9123         mask bits outside of mode are just sign-extension from mode to HWI.
9125 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
9127         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
9128         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
9129         adjust_address instead of change_address to keep info about alignment.
9130         (emit_strmov): Remove.
9131         (emit_memmov): New function.
9132         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
9133         (expand_movmem_epilogue): Likewise and return updated rtx for
9134         destination.
9135         (expand_constant_movmem_prologue): Likewise and return updated rtx for
9136         destination and source.
9137         (decide_alignment): Refactor, handle vector_loop.
9138         (ix86_expand_movmem): Likewise.
9139         (ix86_expand_setmem): Likewise.
9140         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
9142 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
9144         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
9145         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
9147 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
9149         * config/i386/sse.md (sse_movlhps): Change alternative 3
9150         of operand 2 to "m".
9152 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
9154         PR target/57807
9155         * config/i386/sse.md (iptr): New mode attribute.
9156         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
9157         (<sse>_vm<plusminus_insn><mode>3): Ditto.
9158         (<sse>_vmmul<mode>3): Ditto.
9159         (<sse>_vmdiv<mode>3): Ditto.
9160         (sse_vmrcpv4sf2): Ditto.
9161         (<sse>_vmsqrt<mode>2): Ditto.
9162         (sse_vmrsqrtv4sf2): Ditto.
9163         (<sse>_vm<code><mode>3): Ditto.
9164         (avx_vmcmp<mode>3): Ditto.
9165         (<sse>_vmmaskcmp<mode>3): Ditto.
9166         (<sse>_comi): Ditto.
9167         (<sse>_ucomi): Ditto.
9168         (*xop_vmfrcz_<mode>): Ditto.
9169         (*fmai_fmadd_<mode>): Ditto.
9170         (*fmai_fmsub_<mode>): Ditto.
9171         (*fmai_fnmadd_<mode>): Ditto.
9172         (*fmai_fnmsub_<mode>): Ditto.
9173         (*fma4i_vmfmadd_<mode>): Ditto.
9174         (*fma4i_vmfmsub_<mode>): Ditto.
9175         (*fma4i_vmfnmadd_<mode>): Ditto.
9176         (*fma4i_vmfnmsub_<mode>): Ditto.
9177         (*xop_vmfrcz_<mode>): Ditto.
9178         (sse_cvtps2pi): Ditto.
9179         (sse_cvttps2pi): Ditto.
9180         (sse_cvtss2si): Ditto.
9181         (sse_cvtss2si_2): Ditto.
9182         (sse_cvtss2siq_2): Ditto.
9183         (sse_cvttss2si): Ditto.
9184         (sse_cvttss2siq): Ditto.
9185         (sse_cvtsd2si): Ditto.
9186         (sse_cvtsd2si_2): Ditto.
9187         (sse_cvtsd2siq_2): Ditto.
9188         (sse_cvttsd2si): Ditto.
9189         (sse_cvttsd2siq): Ditto.
9190         (sse_cvtsd2ss): Ditto.
9191         (sse_cvtss2sd): Ditto.
9192         (avx2_pbroadcast<mode>): Ditto.
9193         (avx2_pbroadcast<mode>_1): Ditto.
9194         (*avx_vperm_broadcast_v4sf): Ditto.
9196         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
9197         (sse_movlhps): Ditto.
9198         (sse_storehps): Ditto.
9199         (sse_loadhps): Ditto.
9200         (sse_storelps): Ditto.
9201         (sse_loadlps): Ditto.
9202         (*vec_concatv4sf): Ditto.
9203         (*vec_interleave_highv2df): Ditto.
9204         (*vec_interleave_lowv2df): Ditto.
9205         (*vec_extractv2df_1_sse): Ditto.
9206         (*vec_extractv2df_0_sse): Ditto.
9207         (sse2_storelpd): Ditto.
9208         (sse2_loadlpd): Ditto.
9209         (sse2_movsd): Ditto.
9210         (*vec_concatv4si): Ditto.
9211         (vec_concatv2di): Ditto.
9213         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
9214         for Intel asm dialect.
9215         (mmx_punpcklwd): Ditto.
9216         (mmx_punpckldq): Ditto.
9218         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
9219         for intel assembler dialect.
9221 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
9223         PR target/29776
9224         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
9225         for BUILT_IN_C{LZ,LRSB}*.
9226         * tree.h (CASE_INT_FN): Add FN##IMAX case.
9227         * tree-vrp.c (extract_range_basic): Handle
9228         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
9229         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
9230         fall thru to code calling set_value*.
9231         * builtins.c (expand_builtin): Remove *IMAX cases.
9232         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
9233         if width is bigger than 2*HWI.
9235 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
9237         PR rtl-optimization/55342
9238         * lra-int.h (lra_subreg_reload_pseudos): New.
9239         * lra.c: Add undoing optional reloads to the block diagram.
9240         (lra_subreg_reload_pseudos): New.
9241         (lra_optional_reload_pseudos): Change comments.
9242         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
9243         lra_optional_reload_pseudos after undo transformations.
9244         * lra-assigns.c (pseudo_prefix_title): New.
9245         (lra_setup_reg_renumber): Use it.
9246         (spill_for): Ditto.  Check subreg reload pseudos too.
9247         (assign_by_spills): Consider subreg reload pseudos too.
9248         * lra-constraints.c (simplify_operand_subreg): Use
9249         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
9250         (curr_insn_transform): Recognize and do optional reloads.
9251         (undo_optional_reloads): New.
9252         (lra_undo_inheritance): Call undo_optional_reloads.
9254 2013-07-05  Thomas Quinot  <quinot@adacore.com>
9256         * tree-complex.c (expand_complex_operations_1): Fix typo.
9258 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
9260         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
9261         (tune_params): New member 'const vec_costs'.
9262         * config/aarch64/aarch64.c (generic_vector_cost): New.
9263         (generic_tunings): New member 'generic_vector_cost'.
9264         (aarch64_builtin_vectorization_cost): New.
9265         (aarch64_add_stmt_cost): New.
9266         (TARGET_VECTORIZE_ADD_STMT_COST): New.
9267         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
9269 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
9271         PR target/57777
9272         * config/i386/predicates.md (vsib_address_operand): Disallow
9273         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
9275 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
9277         PR middle-end/55030
9278         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
9279         expand_builtin_setjmp_receiver.
9280         (expand_label): Adjust, call expand_builtin_setjmp_receiver
9281         with NULL for the label parameter.
9282         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
9283         the frame-pointer.  Adjust comments.
9284         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
9285         only if LABEL is non-NULL.
9287 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
9289         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
9290         (ARM_ABI_AAPCS64): Ditto.
9291         (arm_abi): Ditto.
9292         (ARM_DEFAULT_ABI): Ditto.
9294 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
9296         * config/aarch64/aarch64-builtins.c
9297         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
9298         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
9299         (st1): Likewise.
9300         * config/aarch64/aarch64-simd.md
9301         (aarch64_ld1<VALL:mode>): New.
9302         (aarch64_st1<VALL:mode>): Likewise.
9303         * config/aarch64/arm_neon.h
9304         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
9306 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
9308         * config/i386/i386.c (gate_insert_vzeroupper): Check if
9309         target ISA is AVX.
9310         (ix86_option_override_internal):Turn on all -mavx target flags by
9311         default as they are dependent on AVX anyway.
9313 2013-07-02  Cary Coutant  <ccoutant@google.com>
9315         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
9316         deterministic hash.
9317         (loc_checksum_ordered): Likewise.
9318         (hash_loc_operands): Remove inline keyword.
9320 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
9322         PR tree-optimization/57741
9323         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
9324         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
9325         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
9326         Allow REAL_CST step_exprs if flag_associative_math.
9327         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
9329 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
9331         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
9333 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
9335         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
9337 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9339         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
9340         encoding.
9341         (iorsi3_insn): Likewise.
9342         (arm_xorsi3): Likewise.
9344 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
9346         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
9347         to ...
9348         (attribute "type"): ... here, and prefix with "wmmx_".
9349         (attribute "core_cycles"): Update for attribute changes.
9350         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
9351         (tbcstv4hi): Likewise.
9352         (tbcstv2si): Likewise.
9353         (iwmmxt_iordi3): Likewise.
9354         (iwmmxt_xordi3): Likewise.
9355         (iwmmxt_anddi3): Likewise.
9356         (iwmmxt_nanddi3): Likewise.
9357         (iwmmxt_arm_movdi): Likewise.
9358         (iwmmxt_movsi_insn): Likewise.
9359         (mov<mode>_internal): Likewise.
9360         (and<mode>3_iwmmxt): Likewise.
9361         (ior<mode>3_iwmmxt): Likewise.
9362         (xor<mode>3_iwmmxt): Likewise.
9363         (add<mode>3_iwmmxt): Likewise.
9364         (ssaddv8qi3): Likewise.
9365         (ssaddv4hi3): Likewise.
9366         (ssaddv2si3): Likewise.
9367         (usaddv8qi3): Likewise.
9368         (usaddv4hi3): Likewise.
9369         (usaddv2si3): Likewise.
9370         (sub<mode>3_iwmmxt): Likewise.
9371         (sssubv8qi3): Likewise.
9372         (sssubv4hi3): Likewise.
9373         (sssubv2si3): Likewise.
9374         (ussubv8qi3): Likewise.
9375         (ussubv4hi3): Likewise.
9376         (ussubv2si3): Likewise.
9377         (mulv4hi3_iwmmxt): Likewise.
9378         (smulv4hi3_highpart): Likewise.
9379         (umulv4hi3_highpart): Likewise.
9380         (iwmmxt_wmacs): Likewise.
9381         (iwmmxt_wmacsz): Likewise.
9382         (iwmmxt_wmacu): Likewise.
9383         (iwmmxt_wmacuz): Likewise.
9384         (iwmmxt_clrdi): Likewise.
9385         (iwmmxt_clrv8qi): Likewise.
9386         (iwmmxt_clr4hi): Likewise.
9387         (iwmmxt_clr2si): Likewise.
9388         (iwmmxt_uavgrndv8qi3): Likewise.
9389         (iwmmxt_uavgrndv4hi3): Likewise.
9390         (iwmmxt_uavgv8qi3): Likewise.
9391         (iwmmxt_uavgv4hi3): Likewise.
9392         (iwmmxt_tinsrb): Likewise.
9393         (iwmmxt_tinsrh): Likewise.
9394         (iwmmxt_tinsrw): Likewise.
9395         (iwmmxt_textrmub): Likewise.
9396         (iwmmxt_textrmsb): Likewise.
9397         (iwmmxt_textrmuh): Likewise.
9398         (iwmmxt_textrmsh): Likewise.
9399         (iwmmxt_textrmw): Likewise.
9400         (iwmxxt_wshufh): Likewise.
9401         (eqv8qi3): Likewise.
9402         (eqv4hi3): Likewise.
9403         (eqv2si3): Likewise.
9404         (gtuv8qi3): Likewise.
9405         (gtuv4hi3): Likewise.
9406         (gtuv2si3): Likewise.
9407         (gtv8qi3): Likewise.
9408         (gtv4hi3): Likewise.
9409         (gtv2si3): Likewise.
9410         (smax<mode>3_iwmmxt): Likewise.
9411         (umax<mode>3_iwmmxt): Likewise.
9412         (smin<mode>3_iwmmxt): Likewise.
9413         (umin<mode>3_iwmmxt): Likewise.
9414         (iwmmxt_wpackhss): Likewise.
9415         (iwmmxt_wpackwss): Likewise.
9416         (iwmmxt_wpackdss): Likewise.
9417         (iwmmxt_wpackhus): Likewise.
9418         (iwmmxt_wpackwus): Likewise.
9419         (iwmmxt_wpackdus): Likewise.
9420         (iwmmxt_wunpckihb): Likewise.
9421         (iwmmxt_wunpckihh): Likewise.
9422         (iwmmxt_wunpckihw): Likewise.
9423         (iwmmxt_wunpckilb): Likewise.
9424         (iwmmxt_wunpckilh): Likewise.
9425         (iwmmxt_wunpckilw): Likewise.
9426         (iwmmxt_wunpckehub): Likewise.
9427         (iwmmxt_wunpckehuh): Likewise.
9428         (iwmmxt_wunpckehuw): Likewise.
9429         (iwmmxt_wunpckehsb): Likewise.
9430         (iwmmxt_wunpckehsh): Likewise.
9431         (iwmmxt_wunpckehsw): Likewise.
9432         (iwmmxt_wunpckelub): Likewise.
9433         (iwmmxt_wunpckeluh): Likewise.
9434         (iwmmxt_wunpckeluw): Likewise.
9435         (iwmmxt_wunpckelsb): Likewise.
9436         (iwmmxt_wunpckelsh): Likewise.
9437         (iwmmxt_wunpckelsw): Likewise.
9438         (ror<mode>3): Likewise.
9439         (ashr<mode>3_iwmmxt): Likewise.
9440         (lshr<mode>3_iwmmxt): Likewise.
9441         (ashl<mode>3_iwmmxt): Likewise.
9442         (ror<mode>3_di): Likewise.
9443         (ashr<mode>3_di): Likewise.
9444         (lshr<mode>3_di): Likewise.
9445         (ashl<mode>3_di): Likewise.
9446         (iwmmxt_wmadds): Likewise.
9447         (iwmmxt_wmaddu): Likewise.
9448         (iwmmxt_tmia): Likewise.
9449         (iwmmxt_tmiaph): Likewise.
9450         (iwmmxt_tmiabb): Likewise.
9451         (iwmmxt_tmiatb): Likewise.
9452         (iwmmxt_tmiabt): Likewise.
9453         (iwmmxt_tmiatt): Likewise.
9454         (iwmmxt_tmovmskb): Likewise.
9455         (iwmmxt_tmovmskh): Likewise.
9456         (iwmmxt_tmovmskw): Likewise.
9457         (iwmmxt_waccb): Likewise.
9458         (iwmmxt_wacch): Likewise.
9459         (iwmmxt_waccw): Likewise.
9460         (iwmmxt_waligni): Likewise.
9461         (iwmmxt_walignr): Likewise.
9462         (iwmmxt_walignr0): Likewise.
9463         (iwmmxt_walignr1): Likewise.
9464         (iwmmxt_walignr2): Likewise.
9465         (iwmmxt_walignr3): Likewise.
9466         (iwmmxt_wsadb): Likewise.
9467         (iwmmxt_wsadh): Likewise.
9468         (iwmmxt_wsadbz): Likewise.
9469         (iwmmxt_wsadhz): Likewise.
9470         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
9471         (iwmmxt_wabsdiffb): Likewise.
9472         (iwmmxt_wabsdiffh): Likewise.
9473         (iwmmxt_wabsdiffw): Likewise.
9474         (iwmmxt_waddsubhx): Likewise
9475         (iwmmxt_wsubaddhx): Likewise.
9476         (addc<mode>3): Likewise.
9477         (iwmmxt_avg4): Likewise.
9478         (iwmmxt_avg4r): Likewise.
9479         (iwmmxt_wmaddsx): Likewise.
9480         (iwmmxt_wmaddux): Likewise.
9481         (iwmmxt_wmaddsn): Likewise.
9482         (iwmmxt_wmaddun): Likewise.
9483         (iwmmxt_wmulwsm): Likewise.
9484         (iwmmxt_wmulwum): Likewise.
9485         (iwmmxt_wmulsmr): Likewise.
9486         (iwmmxt_wmulumr): Likewise.
9487         (iwmmxt_wmulwsmr): Likewise.
9488         (iwmmxt_wmulwumr): Likewise.
9489         (iwmmxt_wmulwl): Likewise.
9490         (iwmmxt_wqmulm): Likewise.
9491         (iwmmxt_wqmulwm): Likewise.
9492         (iwmmxt_wqmulmr): Likewise.
9493         (iwmmxt_wqmulwmr): Likewise.
9494         (iwmmxt_waddbhusm): Likewise.
9495         (iwmmxt_waddbhusl): Likewise.
9496         (iwmmxt_wqmiabb): Likewise.
9497         (iwmmxt_wqmiabt): Likewise.
9498         (iwmmxt_wqmiatb): Likewise.
9499         (iwmmxt_wqmiatt): Likewise.
9500         (iwmmxt_wqmiabbn): Likewise.
9501         (iwmmxt_wqmiabtn): Likewise.
9502         (iwmmxt_wqmiatbn): Likewise.
9503         (iwmmxt_wqmiattn): Likewise.
9504         (iwmmxt_wmiabb): Likewise.
9505         (iwmmxt_wmiabt): Likewise.
9506         (iwmmxt_wmiatb): Likewise.
9507         (iwmmxt_wmiatt): Likewise.
9508         (iwmmxt_wmiabbn): Likewise.
9509         (iwmmxt_wmiabtn): Likewise.
9510         (iwmmxt_wmiatbn): Likewise.
9511         (iwmmxt_wmiattn): Likewise.
9512         (iwmmxt_wmiawbb): Likewise.
9513         (iwmmxt_wmiawbt): Likewise.
9514         (iwmmxt_wmiawtb): Likewise.
9515         (iwmmxt_wmiawtt): Likewise.
9516         (iwmmxt_wmiawbbn): Likewise.
9517         (iwmmxt_wmiawbtn): Likewise.
9518         (iwmmxt_wmiawtbn): Likewise.
9519         (iwmmxt_wmiawttn): Likewise.
9520         (iwmmxt_wmerge): Likewise.
9521         (iwmmxt_tandc<mode>3): Likewise.
9522         (iwmmxt_torc<mode>3): Likewise.
9523         (iwmmxt_torvsc<mode>3): Likewise.
9524         (iwmmxt_textrc<mode>3): Likewise.
9525         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
9526         (wmmxt_pack): Likewise.
9527         (wmmxt_mult_c1): Likewise.
9528         (wmmxt_mult_c2): Likewise.
9529         (wmmxt_alu_c1): Likewise.
9530         (wmmxt_alu_c2): Likewise.
9531         (wmmxt_alu_c3): Likewise.
9532         (wmmxt_transfer_c1): Likewise.
9533         (wmmxt_transfer_c2): Likewise.
9534         (wmmxt_transfer_c3): Likewise.
9535         (marvell_f_iwmmxt_wstr): Likewise.
9536         (marvell_f_iwmmxt_wldr): Likewise.
9538 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
9540         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
9542 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
9544         Revert:
9545         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
9546         * lra-constraints.c (need_for_split_p): Check call used hard regs
9547         living through calls.
9549         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
9550         call used regs for call insn.
9552 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
9554         PR target/57736
9555         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
9556         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
9558 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
9560         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
9562 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
9564         * lra-constraints.c (need_for_split_p): Check call used hard regs
9565         living through calls.
9567 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
9569         PR target/57744
9570         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
9571         to tie with any other modes.  Eliminate Altivec vector mode tests,
9572         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
9573         code, to return 0 if testing MODE2 for a condition, if we've
9574         already tested MODE1 for the same condition.
9576 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9578         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
9579         layout.
9581 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9583         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
9584         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
9586 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9588         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
9589         Define.
9590         (aarch64_symbolic_constant_p): Remove.
9591         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
9592         static.  Fix line length and white space.
9593         (aarch64_symbolic_constant_p): Remove.
9594         * config/aarch64/predicates.md (aarch64_valid_symref):
9595         Use aarch64_classify_symbol_expression.
9597 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9599         * config/arm/constraints.md (Ts): New constraint.
9600         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
9601         16-bit encodings.
9602         (compare_scc): Use "Ts" constraint for operand 0.
9603         (ior_scc_scc): Likewise.
9604         (and_scc_scc): Likewise.
9605         (and_scc_scc_nodom): Likewise.
9606         (ior_scc_scc_cmp): Likewise for operand 7.
9607         (and_scc_scc_cmp): Likewise.
9608         * config/arm/thumb2.md (thumb2_movsi_insn):
9609         Add alternatives for 16-bit encodings.
9610         (thumb2_movhi_insn): Likewise.
9611         (thumb2_movsicc_insn): Likewise.
9612         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
9613         (thumb2_negscc): Use "Ts" constraint.
9614         Move mvn instruction outside cond_exec block.
9615         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
9616         for 16-bit encodings.
9618 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9620         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
9621         encoding.
9622         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
9623         (mulsi3subsi): Likewise.
9624         (mulsidi3adddi): Likewise.
9625         (mulsidi3_v6): Likewise.
9626         (umulsidi3_v6): Likewise.
9627         (umulsidi3adddi_v6): Likewise.
9628         (smulsi3_highpart_v6): Likewise.
9629         (umulsi3_highpart_v6): Likewise.
9630         (mulhisi3tb): Likewise.
9631         (mulhisi3bt): Likewise.
9632         (mulhisi3tt): Likewise.
9633         (maddhisi4): Likewise.
9634         (maddhisi4tb): Likewise.
9635         (maddhisi4tt): Likewise.
9636         (maddhidi4): Likewise.
9637         (maddhidi4tb): Likewise.
9638         (maddhidi4tt): Likewise.
9639         (zeroextractsi_compare0_scratch): Likewise.
9640         (insv_zero): Likewise.
9641         (insv_t2): Likewise.
9642         (anddi_notzesidi_di): Likewise.
9643         (anddi_notsesidi_di): Likewise.
9644         (andsi_notsi_si): Likewise.
9645         (iordi_zesidi_di): Likewise.
9646         (xordi_zesidi_di): Likewise.
9647         (andsi_iorsi3_notsi): Likewise.
9648         (smax_0): Likewise.
9649         (smax_m1): Likewise.
9650         (smin_0): Likewise.
9651         (not_shiftsi): Likewise.
9652         (unaligned_loadsi): Likewise.
9653         (unaligned_loadhis): Likewise.
9654         (unaligned_loadhiu): Likewise.
9655         (unaligned_storesi): Likewise.
9656         (unaligned_storehi): Likewise.
9657         (extv_reg): Likewise.
9658         (extzv_t2): Likewise.
9659         (divsi3): Likewise.
9660         (udivsi3): Likewise.
9661         (arm_zero_extendhisi2addsi): Likewise.
9662         (arm_zero_extendqisi2addsi): Likewise.
9663         (compareqi_eq0): Likewise.
9664         (arm_extendhisi2_v6): Likewise.
9665         (arm_extendqisi2addsi): Likewise.
9666         (arm_movt): Likewise.
9667         (thumb2_ldrd): Likewise.
9668         (thumb2_ldrd_base): Likewise.
9669         (thumb2_ldrd_base_neg): Likewise.
9670         (thumb2_strd): Likewise.
9671         (thumb2_strd_base): Likewise.
9672         (thumb2_strd_base_neg): Likewise.
9673         (arm_negsi2): Add alternative for 16-bit encoding.
9674         (arm_one_cmplsi2): Likewise.
9676 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9678         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
9679         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
9680         (movdfcc): Likewise.
9681         * config/arm/vfp.md (*thumb2_movsf_vfp):
9682         Disable predication for arm_restrict_it.
9683         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
9684         (*thumb2_movdfcc_vfp): Likewise.
9685         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
9686         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
9687         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
9688         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
9689         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
9690         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
9691         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
9692         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
9693         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
9694         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
9695         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
9696         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
9697         Disable predication for arm_restrict_it.
9699 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
9701         * config/i386/bmiintrin.h (_bextr_u32): New.
9702         (_bextr_u64): Ditto.
9704 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
9706         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
9707         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
9708         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
9709         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
9710         * config/mips/n32-elf.h: ...this new file.
9712 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
9714         PR target/57224
9715         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
9716         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
9718 2013-06-27  Catherine Moore  <clm@codesourcery.com>
9720         * config/mips/mips-tables.opt: Regenerate.
9721         * config/mips/mips-cpus.def: Add m14ke and m14kec.
9722         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
9723         * doc/invoke.texi: Add -m14kc.
9725 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
9727         PR target/57623
9728         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
9729         constraints of operand 1 and 2.
9731         PR target/57623
9732         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
9733         to match RTL canonicalization.  Swap predicates and
9734         constraints of operand 1 and 2.
9736 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
9738         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
9739         Process OP_INOUT regs for splitting too.
9741 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
9743         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
9744         decl before the loop, initialize to NULL.
9745         (vectorizable_load): Initialize ptr_incr to NULL.
9747 2013-06-27  Martin Jambor  <mjambor@suse.cz>
9749         PR lto/57208
9750         * ipa-ref.h (ipa_maybe_record_reference): Declare.
9751         * ipa-ref.c (ipa_maybe_record_reference): New function.
9752         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
9753         * ipa-cp.c (create_specialized_node): Record potential references from
9754         aggvals.
9755         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
9757 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
9759         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
9760         parameter 'mode' of type 'enum machine_mode mode'; change to pass
9761         'mode' to force_reg.
9762         (aarch64_add_offset): Update calls to aarch64_force_temporary.
9763         (aarch64_expand_mov_immediate): Likewise.
9765 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
9767         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
9768         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
9770 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9772         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
9773         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9774         (struct machine_function): Add tbegin_p.
9775         (s390_canonicalize_comparison): Fold CC mode compares to
9776         conditional jump if possible.
9777         (s390_emit_jump): Return the emitted jump.
9778         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
9779         Handle CCRAWmode compares.
9780         (s390_option_override): Default to -mhtm if available.
9781         (s390_reg_clobbered_rtx): Handle floating point regs as well.
9782         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
9783         FPRs instead of df_regs_ever_live_p.
9784         (s390_optimize_nonescaping_tx): New function.
9785         (s390_init_frame_layout): Extend clobbered_regs array to cover
9786         FPRs as well.
9787         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
9788         (s390_expand_tbegin): New function.
9789         (enum s390_builtin): New enum definition.
9790         (code_for_builtin): New array definition.
9791         (s390_init_builtins): New function.
9792         (s390_expand_builtin): New function.
9793         (TARGET_INIT_BUILTINS): Define.
9794         (TARGET_EXPAND_BUILTIN): Define.
9795         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
9796         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
9797         (s390_alc_comparison): Likewise.
9798         * config/s390/s390-modes.def: Add CCRAWmode.
9799         * config/s390/s390.h (processor_flags): Add PF_TX.
9800         (TARGET_CPU_HTM): Define macro.
9801         (TARGET_HTM): Define macro.
9802         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
9803         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
9804         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9805         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
9806         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
9807         (TBEGIN_MASK, TBEGINC_MASK): New constants.
9808         ("*cc_to_int"): Move up.
9809         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
9810         constants other than 0.
9811         ("*ccraw_to_int"): New insn and splitter definition.
9812         ("tbegin", "tbegin_nofloat", "tbegin_retry")
9813         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
9814         ("tx_assist"): New expander.
9815         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
9816         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
9817         * config/s390/s390.opt: Add -mhtm option.
9818         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
9819         * config/s390/htmxlintrin.h: New file.
9820         * config/s390/htmintrin.h: New file.
9821         * config/s390/s390intrin.h: New file.
9822         * doc/extend.texi: Document htm builtins.
9823         * config.gcc: Add the new header files to extra_headers.
9825 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
9827         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
9828         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
9830 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
9831             Pat Haugen  <pthaugen@us.ibm.com>
9832             Peter Bergner  <bergner@vnet.ibm.com>
9834         * config/rs6000/power8.md: New.
9835         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
9836         setting for power8 entry.
9837         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
9838         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
9839         test for Power4/Power5 only.
9840         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
9841         support.
9842         (force_new_group): Adjust comment.
9843         * config/rs6000/rs6000.md: Include power8.md.
9845 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
9847         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
9848         * config/arm/arm-protos.h (arm_max_conditional_execute): New
9849         declaration.
9850         (tune_params): Update comment.
9851         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
9852         (arm_max_conditional_execute): New function.
9853         (thumb2_final_prescan_insn): Use max_insn_skipped and
9854         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
9856 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
9858         PR tree-optimization/57705
9859         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
9860         SSA_NAME step, provided that it is not defined inside the loop.
9861         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
9862         (get_initial_def_for_induction): Handle SSA_NAME IV step.
9864 2013-06-25  Martin Jambor  <mjambor@suse.cz>
9866         PR middle-end/57670
9867         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
9868         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
9869         calls in the dump.
9870         (ipa_note_param_call): Initialize member_ptr flag.
9871         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
9872         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
9873         (ipa_write_indirect_edge_info): Stream member_ptr flag.
9874         (ipa_read_indirect_edge_info): Likewise.
9876 2013-06-25  Richard Biener  <rguenther@suse.de>
9878         PR middle-end/56977
9879         * passes.c (init_optimization_passes): Move pass_fold_builtins
9880         and pass_dce earlier with -Og.
9882 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
9884         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
9885         <BIT_FIELD_REF>: Remove trailing TAB.
9886         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
9887         remove blank line.
9889 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9891         PR tree-optimization/57358
9892         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
9893         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
9894         (ipa_analyze_params_uses): Generate pessimistic info when true.
9896 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9898         PR tree-optimization/57539
9899         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
9900         global.inlined_to of the new node to it.  All callers changed.
9901         * ipa-inline-transform.c (clone_inlined_nodes): New variable
9902         inlining_into, pass it to cgraph_clone_node.
9903         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
9904         ipa_free_edge_args_substructures.
9905         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
9906         rdesc linked list.  Do not assert rdesc edges have inlined caller.
9907         Assert we have found an rdesc in the rdesc list.
9909 2013-06-24  Richard Biener  <rguenther@suse.de>
9911         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
9912         (pointer_set_lookup): Declare.
9913         (class pointer_map): New template class implementing a
9914         generic pointer to T map.
9915         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
9916         pointer_map<T>::contains, pointer_map<T>::insert,
9917         pointer_map<T>::traverse): New functions.
9918         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
9919         (pointer_set_lookup): New function.
9920         (pointer_set_contains): Use pointer_set_lookup.
9921         (pointer_set_insert): Likewise.
9922         (insert_aux): Remove.
9923         (struct pointer_map_t): Embed a pointer_set_t.
9924         (pointer_map_create): Adjust.
9925         (pointer_map_destroy): Likewise.
9926         (pointer_map_contains): Likewise.
9927         (pointer_map_insert): Likewise.
9928         (pointer_map_traverse): Likewise.
9929         * tree-streamer.h (struct streamer_tree_cache_d): Use a
9930         pointer_map<unsigned> instead of a pointer_map_t.
9931         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
9932         (streamer_tree_cache_lookup): Likewise.
9933         (streamer_tree_cache_create): Likewise.
9934         (streamer_tree_cache_delete): Likewise.
9935         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
9936         pointer_map<unsigned> instead of a pointer_map_t.
9937         (lto_init_tree_ref_encoder): Adjust.
9938         (lto_destroy_tree_ref_encoder): Likewise.
9939         * lto-section-out.c (lto_output_decl_index): Likewise.
9940         (lto_record_function_out_decl_state): Likewise.
9941         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
9943 2013-06-24  Richard Biener  <rguenther@suse.de>
9945         PR tree-optimization/57488
9946         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
9948 2013-06-24  Alan Modra  <amodra@gmail.com>
9950         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
9951         (gen_easy_altivec_constant): Likewise.
9952         * config/rs6000/predicates.md (easy_vector_constant_add_self,
9953         easy_vector_constant_msb): Likewise.
9955 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
9957         PR target/57688
9958         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
9959         add missing return true.
9961 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
9963         PR target/52483
9964         * config/sh/predicates.md (general_extend_operand): Invoke
9965         general_movsrc_operand for memory operands.
9966         (general_movsrc_operand): Allow reg+reg addressing, do not use
9967         general_operand for memory operands.
9969 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
9971         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
9972         when current target options does not apply.
9973         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
9974         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
9975         * config/i386/bmiintrin.h: Pass appropriate target
9976         attributes to header.
9977         * config/i386/mmintrin.h: Ditto.
9978         * config/i386/nmmintrin.h: Ditto.
9979         * config/i386/avx2intrin.h: Ditto.
9980         * config/i386/fxsrintrin.h: Ditto.
9981         * config/i386/tbmintrin.h: Ditto.
9982         * config/i386/xsaveintrin.h: Ditto.
9983         * config/i386/f16cintrin.h: Ditto.
9984         * config/i386/xtestintrin.h: Ditto.
9985         * config/i386/xsaveoptintrin.h: Ditto.
9986         * config/i386/bmi2intrin.h: Ditto.
9987         * config/i386/lzcntintrin.h: Ditto.
9988         * config/i386/smmintrin.h: Ditto.
9989         * config/i386/wmmintrin.h: Ditto.
9990         * config/i386/x86intrin.h: Remove all header include guards.
9991         * config/i386/prfchwintrin.h: Ditto.
9992         * config/i386/pmmintrin.h: Ditto.
9993         * config/i386/tmmintrin.h: Ditto.
9994         * config/i386/xmmintrin.h: Ditto.
9995         * config/i386/popcntintrin.h: Ditto.
9996         * config/i386/rdseedintrin.h: Ditto.
9997         * config/i386/ammintrin.h: Ditto.
9998         * config/i386/emmintrin.h: Ditto.
9999         * config/i386/immintrin.h: Remove all header include guards.
10000         * config/i386/fma4intrin.h: Ditto.
10001         * config/i386/lwpintrin.h: Ditto.
10002         * config/i386/xopintrin.h: Ditto.
10003         * config/i386/ia32intrin.h: Ditto.
10004         * config/i386/avxintrin.h: Ditto.
10005         * config/i386/rtmintrin.h: Ditto.
10006         * config/i386/fmaintrin.h: Ditto.
10007         * config/i386/mm3dnow.h: Ditto.
10009 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
10011         * common/config/i386/i386-common.c: Handle LZCNT.
10013 2013-06-22  Andi Kleen  <ak@linux.intel.com>
10015         * doc/extend.texi: Use __atomic_store_n instead of
10016         __atomic_store in HLE example.
10018 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
10020         * config/sh/sh.c: Remove <cstdlib> workaround.
10022 2013-06-21  Andi Kleen  <ak@linux.intel.com>
10024         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
10026 2013-06-21  Andi Kleen  <ak@linux.intel.com>
10028         * doc/extend.texi: Document that __atomic_clear and
10029           __atomic_test_and_set should only be used with bool.
10031 2013-06-20  Jan Hubicka  <jh@suse.cz>
10033         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
10034         types_same_for_odr.
10035         * tree.c (decls_same_for_odr): New function.
10036         (same_for_edr): New function.
10037         (types_same_for_odr): New function.
10038         (get_binfo_at_offset): Use it.
10039         * tree.h (types_same_for_odr): Declare.
10041 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
10042             Jason Merrill  <jason@redhat.com>
10044         * system.h: Include <cstdlib> as well as <stdlib.h>.
10046 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
10048         PR target/57655
10049         * config/i386/i386.c (construct_container): Report error if
10050         long double is used with disabled x87 float returns.
10052 2013-06-20  Jan Hubicka  <jh@suse.cz>
10054         * lto-cgraph.c (input_symtab): Do not set cgraph state.
10056 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
10058         PR rtl-optimization/57425
10059         PR rtl-optimization/57569
10060         * alias.c (write_dependence_p): Remove parameters mem_mode and
10061         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
10062         Changed all callers.
10063         (canon_anti_dependence): Get comments and semantics in sync.
10064         Add parameter mem_canonicalized.  Changed all callers.
10065         * rtl.h (canon_anti_dependence): Update prototype.
10067 2013-06-20  Richard Biener  <rguenther@suse.de>
10069         * data-streamer-in.c (streamer_read_uhwi): Optimize single
10070         byte case, inline streamer_read_uchar and defer section
10071         overrun check.
10073 2013-06-20  Richard Biener  <rguenther@suse.de>
10075         PR tree-optimization/57584
10076         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
10077         SSA names into the expanded expression that take part in
10078         abnormal coalescing.
10080 2013-06-19  Sharad Singhai  <singhai@google.com>
10082         * gcov.c (print_usage): Handle new option.
10083         (process_args): Ditto.
10084         (get_gcov_intermediate_filename): New function.
10085         (output_intermediate_file): New function.
10086         (output_gcov_file): New function
10087         (generate_results): Handle new option.
10088         (release_function): Relase demangled name.
10089         (read_graph_file): Handle demangled name.
10090         (output_lines): Ditto.
10091         * doc/gcov.texi: Document gcov intermediate format.
10093 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
10095         PR bootstrap/57604
10096         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
10097         (lra_emit_add): Use the functions.  Add comment about Y as an
10098         address segment.
10100 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
10102         PR driver/57652
10103         * collect2.c (collect_atexit): New.
10104         (collect_exit): Delete.
10105         (main): Register collect_atexit with atexit.
10106         (collect_wait): Change collect_exit to exit.
10107         (do_wait): Same.
10108         * collect2.h (collect_exit): Delete.
10109         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
10111 2013-06-19  Wei Mi  <wmi@google.com>
10113         PR rtl-optimization/57518
10114         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
10115         if regno is used in paradoxical subreg.
10116         (update_equiv_regs): Check pdx_subregs[regno] before
10117         set a reg to be equivalent with a mem.
10119 2013-06-19  Matthias Klose  <doko@ubuntu.com>
10121         PR driver/57651
10122         * file-find.h (find_a_file): Add a mode parameter.
10123         * file-find.c (find_a_file): Likewise.
10124         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
10125         with X_OK for the executables.
10126         * collect2.c (main): Call find_a_file with X_OK.
10128 2013-06-19  Steve Ellcey  <sellcey@mips.com>
10130         PR target/56942
10131         * config/mips/mips.md (casesi_internal_mips16_<mode>):
10132         Use NEXT_INSN instead of next_real_insn.
10134 2013-06-19  Jan Hubicka  <jh@suse.cz>
10136         * cgraph.h (const_value_known_p): Replace by ...
10137         (ctor_for_folding): .. this one.
10138         * cgraphunit.c (process_function_and_variable_attributes): Use it.
10139         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
10140         * expr.c (expand_expr_real_1): Likewise.
10141         (string_constant): Likewise.
10142         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
10143         * ipa.c (process_references): Likewise.
10144         (symtab_remove_unreachable_nodes): Likewise.
10145         * ipa-inline-analysis.c (param_change_prob): Likewise.
10146         * gimple-fold.c (canonicalize_constructor_val): Likewise.
10147         (get_base_constructor): Likwise.
10148         * varpool.c (varpool_remove_node): Likewise.
10149         (varpool_remove_initializer): LIkewise.
10150         (dump_varpool_node): LIkwise.
10151         (const_value_known_p): Rewrite to ...
10152         (ctor_for_folding): ... this one.
10154 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
10156         PR driver/57651
10157         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
10158         PERSONALITY in $PATH derived prefixes.
10160 2013-06-19  Jeff Law  <law@redhat.com>
10162         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
10163         in comment.
10165         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
10166         (simplify_bitwise_binary): Use it to simpify certain binary ops on
10167         booleans.
10169 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
10171         * config/arm/vfp.md: Move VFP instruction classification documentation
10172         to ...
10173         * config/arm/arm.md: ... here.  Update instruction classification
10174         documentation.
10176 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
10178         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
10179         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
10180         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
10182 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
10184         PR target/57609
10185         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10186         with NEXT_INSN.  Use tablejump_p to check for jump table data
10187         insns.
10189 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
10191         PR c++/56544
10192         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
10193         that now in C++ the value is correct per the C++ standards.
10195 2013-06-19  Richard Biener  <rguenther@suse.de>
10197         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
10198         for global context.
10200 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10202         Revert:
10203         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10205         PR target/57609
10206         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10207         with next_active_insn.
10209 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
10211         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
10212         functions are inlined during failures to flag an error.
10213         * tree-inline.c (expand_call_inline): Allow the error to be flagged
10214         in early inline pass.
10216 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
10218         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
10219         in comments.
10221 2013-06-18  Julian Brown  <julian@codesourcery.com>
10223         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
10224         Permit virtual register pre-reload if !strict.
10225         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
10226         change.
10227         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
10228         prototype.
10229         * config/arm/neon.md (movmisalign<mode>): Use
10230         neon_perm_struct_or_reg_operand instead of
10231         neon_struct_or_register_operand.
10232         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
10233         neon_permissive_struct_operand instead of neon_struct_operand.
10234         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
10235         neon_vector_mem_operand.
10236         * config/arm/predicates.md (neon_struct_operand): Adjust call to
10237         neon_vector_mem_operand.
10238         (neon_permissive_struct_operand): New.
10239         (neon_struct_or_register_operand): Rename to...
10240         (neon_perm_struct_or_reg_operand): This. Adjust call to
10241         neon_vector_mem_operand.
10243 2013-06-18  Richard Biener  <rguenther@suse.de>
10245         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
10246         * lto-streamer.h: Include pointer-set.h.
10247         (struct lto_decl_slot): Remove.
10248         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
10249         Remove next_index entry.
10250         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
10251         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
10252         (lto_init_tree_ref_encoder): Adjust.
10253         (lto_destroy_tree_ref_encoder): Likewise.
10254         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
10255         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
10256         (lto_output_decl_index): Adjust.
10257         (lto_new_out_decl_state): Likewise.
10258         (lto_record_function_out_decl_state): Likewise.
10259         * lto-streamer-out.c (copy_function): Likewise.
10261 2013-06-18  Richard Biener  <rguenther@suse.de>
10263         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
10264         * cgraphunit.c: Include cfgloop.h.
10265         (init_lowered_empty_function): Initialize the loop tree.
10266         (assemble_thunk): Insert new BBs into loops.
10268 2013-06-18  Richard Biener  <rguenther@suse.de>
10270         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
10271         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
10272         the map from cache entry to cache index optional.
10273         (streamer_tree_cache_replace_tree): Adjust accordingly.
10274         (streamer_tree_cache_append): Likewise.
10275         (streamer_tree_cache_delete): Likewise.
10276         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
10277         streamer cache map from cache entry to cache index.
10278         * lto-streamer-out.c (create_output_block): Adjust.
10280 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
10282         * config/arm/arm.md (attribute "insn"): Move multiplication and
10283         division attributes to...
10284         (attribute "type"): ... here.  Remove mult.
10285         (attribute "mul32"): New attribute.
10286         (attribute "mul64"): Add umaal.
10287         (*arm_mulsi3): Update attributes.
10288         (*arm_mulsi3_v6): Likewise.
10289         (*thumb_mulsi3): Likewise.
10290         (*thumb_mulsi3_v6): Likewise.
10291         (*mulsi3_compare0): Likewise.
10292         (*mulsi3_compare0_v6): Likewise.
10293         (*mulsi_compare0_scratch): Likewise.
10294         (*mulsi_compare0_scratch_v6): Likewise.
10295         (*mulsi3addsi): Likewise.
10296         (*mulsi3addsi_v6): Likewise.
10297         (*mulsi3addsi_compare0): Likewise.
10298         (*mulsi3addsi_compare0_v6): Likewise.
10299         (*mulsi3addsi_compare0_scratch): Likewise.
10300         (*mulsi3addsi_compare0_scratch_v6): Likewise.
10301         (*mulsi3subsi): Likewise.
10302         (*mulsidi3adddi): Likewise.
10303         (*mulsi3addsi_v6): Likewise.
10304         (*mulsidi3adddi_v6): Likewise.
10305         (*mulsidi3_nov6): Likewise.
10306         (*mulsidi3_v6): Likewise.
10307         (*umulsidi3_nov6): Likewise.
10308         (*umulsidi3_v6): Likewise.
10309         (*umulsidi3adddi): Likewise.
10310         (*umulsidi3adddi_v6): Likewise.
10311         (*smulsi3_highpart_nov6): Likewise.
10312         (*smulsi3_highpart_v6): Likewise.
10313         (*umulsi3_highpart_nov6): Likewise.
10314         (*umulsi3_highpart_v6): Likewise.
10315         (mulhisi3): Likewise.
10316         (*mulhisi3tb): Likewise.
10317         (*mulhisi3bt): Likewise.
10318         (*mulhisi3tt): Likewise.
10319         (maddhisi4): Likewise.
10320         (*maddhisi4tb): Likewise.
10321         (*maddhisi4tt): Likewise.
10322         (maddhidi4): Likewise.
10323         (*maddhidi4tb): Likewise.
10324         (*maddhidi4tt): Likewise.
10325         (divsi3): Likewise.
10326         (udivsi3): Likewise.
10327         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
10328         (thumb2_mulsi_short_compare0): Likewise.
10329         (thumb2_mulsi_short_compare0_scratch): Likewise.
10330         * config/arm/arm1020e.md (1020mult1): Update attribute change.
10331         (1020mult2): Likewise.
10332         (1020mult3): Likewise.
10333         (1020mult4): Likewise.
10334         (1020mult5): Likewise.
10335         (1020mult6): Likewise.
10336         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
10337         change.
10338         (cortex_a15_mult64): Likewise.
10339         (cortex_a15_sdiv): Likewise.
10340         (cortex_a15_udiv): Likewise.
10341         * config/arm/arm1026ejs.md (mult1): Update attribute change.
10342         (mult2): Likewise.
10343         (mult3): Likewise.
10344         (mult4): Likewise.
10345         (mult5): Likewise.
10346         (mult6): Likewise.
10347         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
10348         (pj4_ir_div): Likewise.
10349         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
10350         (11_mult2): Likewise.
10351         (11_mult3): Likewise.
10352         (11_mult4): Likewise.
10353         (11_mult5): Likewise.
10354         (11_mult6): Likewise.
10355         (11_mult7): Likewise.
10356         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
10357         (cortex_a8_mla): Likewise.
10358         (cortex_a8_mull): Likewise.
10359         (cortex_a8_smulwy): Likewise.
10360         (cortex_a8_smlald): Likewise.
10361         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
10362         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
10363         (cortex_r4_mul_3): Likewise.
10364         (cortex_r4_mla_4): Likewise.
10365         (cortex_r4_mla_3): Likewise.
10366         (cortex_r4_smlald): Likewise.
10367         (cortex_r4_mull): Likewise.
10368         (cortex_r4_sdiv): Likewise.
10369         (cortex_r4_udiv): Likewise.
10370         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
10371         (cortex_a7_idiv): Likewise.
10372         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
10373         (9_mult2): Likewise.
10374         (9_mult3): Likewise.
10375         (9_mult4): Likewise.
10376         (9_mult5): Likewise.
10377         (9_mult6): Likewise.
10378         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
10379         (cortex_a53_sdiv): Likewise.
10380         (cortex_a53_udiv): Likewise.
10381         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
10382         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
10383         (mp626_mult2): Likewise.
10384         (mp626_mult3): Likewise.
10385         (mp626_mult4): Likewise.
10386         * config/arm/fa526.md (526_mult1): Update attribute change.
10387         (526_mult2): Likewise.
10388         * config/arm/arm-generic.md (mult): Update attribute change.
10389         (mult_ldsched_strongarm): Likewise.
10390         (mult_ldsched): Likewise.
10391         (multi_cycle): Likewise.
10392         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
10393         * config/arm/fa606te.md (606te_mult1): Update attribute change.
10394         (606te_mult2): Likewise.
10395         (606te_mult3): Likewise.
10396         (606te_mult4): Likewise.
10397         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
10398         (cortex_a9_mac16): Likewise.
10399         (cortex_a9_multiply): Likewise.
10400         (cortex_a9_mac): Likewise.
10401         (cortex_a9_multiply_long): Likewise.
10402         * config/arm/fa626te.md (626te_mult1): Update attribute change.
10403         (626te_mult2): Likewise.
10404         (626te_mult3): Likewise.
10405         (626te_mult4): Likewise.
10407 2013-06-18  Richard Biener  <rguenther@suse.de>
10409         PR lto/57334
10410         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
10412 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10414         PR target/57609
10415         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10416         with next_active_insn.
10418 2013-06-18  Alan Modra  <amodra@gmail.com>
10420         * config/rs6000/rs6000.h (enum data_align): New.
10421         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
10422         (DATA_ABI_ALIGNMENT): Define.
10423         (CONSTANT_ALIGNMENT): Correct comment.
10424         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
10425         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
10427 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
10429         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
10430         ATTRIBUTE_UNUSED marking.
10432 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
10434         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
10435         alternative and update.
10436         (aarch64_dup_lanedi): Delete.
10437         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
10438         * config/aarch64/aarch64-simd-builtins.def: Update.
10440 2013-06-17  Richard Biener  <rguenther@suse.de>
10442         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
10443         (lto_input_scc): Declare.
10444         (lto_input_tree_1): Likewise.
10445         (struct lto_stats_d): Add num_tree_bodies_output and
10446         num_pickle_refs_output.
10447         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
10448         (lto_read_tree_1): Split out from ...
10449         (lto_read_tree): ... this.
10450         (lto_input_scc): New function.
10451         (lto_input_tree_1): Split out from ...
10452         (lto_input_tree): ... this.  Handle LTO_tree_scc.
10453         (lto_data_in_create): Create the streamer cache without hashes.
10454         * lto-streamer-out.c (create_output_block): Create the streamer
10455         cache with hashes when not doing WPA.
10456         (lto_write_tree_1): Split out from ...
10457         (lto_write_tree): ... this.
10458         (get_symbol_initial_value): New function.
10459         (lto_output_tree_1): Split out from ...
10460         (lto_output_tree): ... this.  Write trees as series of SCCs
10461         using a DFS walk via DFS_write_tree.
10462         (struct sccs, struct scc_entry): New types.
10463         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
10464         (DFS_write_tree_body): New function.
10465         (DFS_write_tree): Likewise.
10466         (hash_tree): Likewise.
10467         (scc_entry_compare): Likewise.
10468         (hash_scc): Likewise.
10469         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
10470         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
10471         TREE_CHAIN as regular reference.
10472         (streamer_read_integer_cst): Remove.
10473         (streamer_get_pickled_tree): Adjust.
10474         * tree-streamer-out.c (streamer_write_chain): Disable streaming
10475         of DECL_EXTERNALs in BLOCK_VARS for now.
10476         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
10477         reference.
10478         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
10479         Add hash value argument and record that if hashes are recorded
10480         in the cache.
10481         (streamer_tree_cache_insert_1): Adjust.
10482         (streamer_tree_cache_insert): Likewise.
10483         (streamer_tree_cache_insert_at): Rename to ...
10484         (streamer_tree_cache_replace_tree): ... this and adjust.
10485         (streamer_tree_cache_append): Adjust.
10486         (record_common_node): Likewise.
10487         (streamer_tree_cache_create): Add argument whether to
10488         record hash values together with trees.
10489         (streamer_tree_cache_delete): Adjust.
10490         * tree-streamer.h (struct streamer_tree_cache_d): Add
10491         vector of hashes.
10492         (streamer_read_integer_cst): Remove.
10493         (streamer_tree_cache_insert): Adjust.
10494         (streamer_tree_cache_append): Likewise.
10495         (streamer_tree_cache_insert_at): Rename to ...
10496         (streamer_tree_cache_replace_tree): ... this and adjust.
10497         (streamer_tree_cache_create): Add argument whether to record hashes.
10498         (streamer_tree_cache_get): Rename to ...
10499         (streamer_tree_cache_get_tree): ... this.
10500         (streamer_tree_cache_get_hash): New function.
10501         * tree.c (cache_integer_cst): New function.
10502         * tree.h (cache_integer_cst): Declare.
10503         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
10504         * lto-symtab.c (lto_varpool_replace_node): Only release
10505         DECL_INITIAL of non-prevailing decls.
10506         * varpool.c (varpool_remove_initializer): Do not release
10507         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
10509 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
10511         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
10512         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
10513         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
10514         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
10515         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
10516         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
10517         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
10518         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
10519         instead of TARGET_64BIT.
10520         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
10521         Require ISA_HAS_<D>DIV.
10523 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
10525         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
10526         (mips*-*-linux*): Move default with_llsc setting to where other
10527         defaults are set.
10528         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
10529         with_arch block.
10530         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
10531         Likewise.  Remove default with_tune setting.  Move default float
10532         setting to its own block.  Handle with_llsc in the same block as above.
10534 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
10536         PR rtl-optimization/57425
10537         PR rtl-optimization/57569
10538         * alias.c (write_dependence_p): Add new parameters mem_mode,
10539         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
10540         Changed all callers.
10541         (canon_anti_dependence): New function.
10542         * cse.c (check_dependence): Use canon_anti_dependence.
10543         * cselib.c (cselib_invalidate_mem): Likewise.
10544         * rtl.h (canon_anti_dependence): Declare.
10546 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
10548         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
10549         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
10550         ".set mips3" for 64-bit targets.
10552 2013-06-15  Dehao Chen  <dehao@google.com>
10554         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
10555         * gimple-low.c (gimple_check_call_matching_types): Likewise.
10556         (gimple_check_call_args): Likewise.
10557         * value-prof.c (check_ic_target): Likewise.
10558         * ipa-inline.c (early_inliner): Likewise.
10559         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
10560         * cgraph.c (cgraph_create_edge_1): Likewise.
10561         (cgraph_make_edge_direct): Likewise.
10563 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
10565         PR target/57615
10566         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
10567         rs6000_output_move_128bit to handle emitting quad memory
10568         operations.  Set attribute length to 8 bytes.
10570 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
10572         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
10573         New pattern.
10574         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
10575         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
10576         (aarch64_<su>mlsl<mode>): Likewise.
10578 2013-06-14  Mike Stump  <mikestump@comcast.net>
10580         * Makefile.in (TARGET_H): Add insn-codes.h.
10582 2013-06-14  Alan Modra  <amodra@gmail.com>
10584         PR middle-end/57134
10585         PR middle-end/57586
10586         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
10587         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
10588         bitfield expansion when EXPAND_MEMORY.
10589         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
10591 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
10593         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
10594         test for clearing quad memory on 32-bit later.
10596 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
10598         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
10599         (fold_negate_expr): Likewise.
10600         (fold_real_zero_addition_p): Handle vectors.
10601         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
10603 2013-06-14  Alan Modra  <amodra@gmail.com>
10605         * varasm.c (force_const_mem): Revert 2013-06-07 change.
10607 2013-06-13  Jan Hubicka  <jh@suse.cz>
10609         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
10610         Local comdats are not externally visible.
10611         * symtab.c (dump_symtab_base): Dump externally visible.
10612         (verify_symtab_base): Verify back links in the symtab hash.
10614 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
10616         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
10617         CONVERT_EXPR as equal nodes.
10619 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
10621         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
10623 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
10625         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
10626         Generalize to complex and vector.
10627         * tree.c (build_all_ones_cst): New function.
10628         * tree.h (build_all_ones_cst): Declare it.
10630 2013-06-13  Alan Modra  <amodra@gmail.com>
10632         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
10633         * config/rs6000/rs6000.md (signbittf2): New insn.
10634         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
10635         (abstf2_internal, cmptf_internal2): Likewise.
10636         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
10638 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
10639             Pat Haugen  <pthaugen@us.ibm.com>
10640             Peter Bergner  <bergner@vnet.ibm.com>
10642         * config/rs6000/rs6000.c (emit_load_locked): Add support for
10643         power8 byte, half-word, and quad-word atomic instructions.
10644         (emit_store_conditional): Likewise.
10645         (rs6000_expand_atomic_compare_and_swap): Likewise.
10646         (rs6000_expand_atomic_op): Likewise.
10648         * config/rs6000/sync.md (larx): Add new modes for power8.
10649         (stcx): Likewise.
10650         (AINT): New mode iterator to include TImode as well as normal
10651         integer modes on power8.
10652         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
10653         that VSX registers are not considered.  Use AINT mode iterator
10654         instead of INT1 to allow inclusion of quad word atomic operations
10655         on power8.
10656         (load_locked<mode>): Likewise.
10657         (store_conditional<mode>): Likewise.
10658         (atomic_compare_and_swap<mode>): Likewise.
10659         (atomic_exchange<mode>): Likewise.
10660         (atomic_nand<mode>): Likewise.
10661         (atomic_fetch_<fetchop_name><mode>): Likewise.
10662         (atomic_nand_fetch<mode>): Likewise.
10663         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
10664         each type.
10665         (ATOMIC): On power8, add QImode, HImode modes.
10666         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
10667         modes that promote to SImode.
10668         (load_lockedti): Convert TImode arguments to PTImode, so that we
10669         get a guaranteed even/odd register pair.
10670         (load_lockedpti): Likewise.
10671         (store_conditionalti): Likewise.
10672         (store_conditionalpti): Likewise.
10674         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
10675         atomic load/store instructions.
10676         (HSI): Likewise.
10678 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
10680         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
10681         loads.
10682         (insn_count): New attribute, with most cases extracted from...
10683         (length): ...here.  Redefine most cases in terms of insn_count.
10684         (single_insn): Delete.
10685         (can_delay): Use insn_count to check for single instructions.
10686         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
10687         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
10688         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
10689         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
10690         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
10691         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
10692         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
10693         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
10694         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
10695         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
10696         rather than "length".
10697         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
10698         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
10699         Use "insn_count" rather than "length".
10700         * config/mips/mips-dsp.md
10701         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
10702         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
10703         length attributes.
10705 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
10707         PR tree-optimization/57361
10708         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
10710 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
10712         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
10713         to split.
10714         (aarch64_simd_combine<mode>): New instruction expansion.
10715         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
10716         function prototype.
10717         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
10718         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
10720 2013-06-12  Jan Hubicka  <jh@suse.cz>
10722         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
10723         decl has when in streaming stage.
10724         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
10725         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
10727 2013-06-12  Roland Stigge  <stigge@antcom.de>
10729         PR target/57578
10730         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
10732 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
10734         PR tree-optimization/57537
10735         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
10736         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
10738 2013-06-12  Richard Biener  <rguenther@suse.de>
10740         * data-streamer.h (streamer_write_char_stream): CSE
10741         obs->current_pointer.
10742         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
10743         streamer_write_char_stream manually and optimize the resulting loop.
10744         (streamer_write_hwi_stream): Likewise.
10746 2013-06-12  Jan Hubicka  <jh@suse.cz>
10748         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
10749         * cgraph.h (varpool_create_empty_node): Declare.
10750         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
10751         duplicated nodes.
10752         * symtab.c (symtab_unregister_node): Be lax about missin entries
10753         in node hash.
10754         (symtab_get_node): Update comment.
10755         * varpool.c (varpool_create_empty_node): Break out from ...
10756         (varpool_node_for_decl): ... here.
10757         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
10759 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
10761         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
10762         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
10763         part.  Use straight-line flow at the end.
10764         <COMPONENT_REF>: Remove superfluous else.
10765         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
10767 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
10769         PR target/56564
10770         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
10771         target hook even for !TREE_PUBLIC decls.  If no resolution info
10772         is available, return false for common and external decls.
10774 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10776         * config/rl78/constraints.md (U): New constraint.
10777         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
10778         valloc attribute.
10780 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
10782         PR target/57589
10783         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
10784         to allow returning address to AT_PLATFORM name.
10786 2013-06-11  Jan Hubicka  <jh@suse.cz>
10788         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
10789         * cgraph.h (symtab_node_base): Add weakref flag.
10790         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
10791         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
10792         (output_weakrefs): Use weakref flag.
10793         * fold-const.c (simple_operand_p): Handle WEAK.
10794         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
10795         * ipa.c (varpool_externally_visible_p): Drop weakref.
10796         (function_and_variable_visibility): Update comment; fix weakref
10797         sanity checks; do not clear DECL_WEAK on them.
10798         * lto-cgraph.c (lto_output_node): update.
10799         (lto_output_varpool_node): Update.
10800         (input_overwrite_node): Update.
10801         (input_node): Update.
10802         (input_varpool_node): Update.
10803         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
10804         (lto_symtab_merge_symbols): Add sanity check.
10805         (lto_symtab_prevailing_decl): Do not special case weakrefs.
10806         * passes.c (rest_of_decl_compilation): Set static flag, too.
10807         * symtab.c (dump_symtab_base): Dump weakref.
10808         (verify_symtab_base): Sanity check weakrefs.
10809         (symtab_make_decl_local): Remove duplicated code.
10810         (symtab_alias_ultimate_target): Simplify.
10811         * varpool.c (varpool_create_variable_alias): Set weakref flag.
10813 2013-06-11  Tom de Vries  <tom@codesourcery.com>
10815         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
10816         sequence_vect == NULL.
10818 2013-06-11  DJ Delorie  <dj@redhat.com>
10820         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
10821         (rl78_unwind_word_mode): New.
10823 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
10825         * final.c (debug_prefix_maps): Make static.
10827 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
10829         * function.c (initial_trampoline): Remove stray copy.
10831 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
10833         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
10835 2013-06-11  Martin Jambor  <mjambor@suse.cz>
10837         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
10838         within bounds at the beginning of the function.
10840 2013-06-11  Alan Modra  <amodra@gmail.com>
10842         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
10843         reporting.
10844         (get_named_section): Don't NULL !DECL_P decl.
10846 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
10848         * doc/invoke.texi (core-avx2): Document.
10849         (slm): Likewise.
10850         (atom): Updated with MOVBE.
10852 2013-06-11  Richard Biener  <rguenther@suse.de>
10854         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
10856 2013-06-11  Anton Blanchard  <anton@samba.org>
10858         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
10859         correct shift value in little-endian mode.
10861 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
10863         PR target/56564
10864         * varasm.c (get_variable_align): Move #endif to the right place.
10866 2013-06-10  Cary Coutant  <ccoutant@google.com>
10868         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
10869         for hash so that hash table traversal order is deterministic.
10871 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
10872             Pat Haugen  <pthaugen@us.ibm.com>
10873             Peter Bergner  <bergner@vnet.ibm.com>
10875         * config/rs6000/vector.md (GPR move splitter): Do not split moves
10876         of vectors in GPRS if they are direct moves or quad word load or
10877         store moves.
10879         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
10880         declaration.
10881         (direct_move_p): Likewise.
10882         (quad_load_store_p): Likewise.
10884         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
10885         classes into bins based on the physical register type.
10886         (reg_class_to_reg_type): Likewise.
10887         (IS_STD_REG_TYPE): Likewise.
10888         (IS_FP_VECT_REG_TYPE): Likewise.
10889         (reload_fpr_gpr): Arrays to determine what insn to use if we can
10890         use direct move instructions.
10891         (reload_gpr_vsx): Likewise.
10892         (reload_vsx_gpr): Likewise.
10893         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
10894         information that is a simplification of register classes.  Also
10895         precalculate direct move reload helpers.
10896         (direct_move_p): New function to return true if the operation can
10897         be done as a direct move instruciton.
10898         (quad_load_store_p): New function to return true if the operation
10899         is a quad memory operation.
10900         (rs6000_legitimize_address): If quad memory, only allow register
10901         indirect for TImode addresses.
10902         (rs6000_legitimate_address_p): Likewise.
10903         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
10904         (rs6000_reload_register_type): Likewise.
10905         (register_to_reg_type): Return register type.
10906         (rs6000_secondary_reload_simple_move): New helper function for
10907         secondary reload and secondary memory needed to identify anything
10908         that is a simple move, and does not need reloading.
10909         (rs6000_secondary_reload_direct_move): New helper function for
10910         secondary reload to identify cases that can be done with several
10911         instructions via the direct move instructions.
10912         (rs6000_secondary_reload_move): New helper function for secondary
10913         reload to identify moves between register types that can be done.
10914         (rs6000_secondary_reload): Add support for quad memory operations
10915         and for direct move.
10916         (rs6000_secondary_memory_needed): Likewise.
10917         (rs6000_debug_secondary_memory_needed): Change argument names.
10918         (rs6000_output_move_128bit): New function to return the move to
10919         use for 128-bit moves, including knowing about the various
10920         limitations of quad memory operations.
10922         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
10923         memory operations.  call rs6000_output_move_128bit for the actual
10924         instruciton(s) to generate.
10925         (vsx_movti_64bit): Likewise.
10927         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
10928         (UNSPEC_P8V_MTVSRWZ): Likewise.
10929         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
10930         (UNSPEC_P8V_MTVSRD): Likewise.
10931         (UNSPEC_P8V_XXPERMDI): Likewise.
10932         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
10933         (UNSPEC_FUSION_GPR): Likewise.
10934         (FMOVE128_GPR): New iterator for direct move.
10935         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
10936         (f32_sv): Likewise.
10937         (f32_dm): Likewise.
10938         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
10939         loads and direct move instructions.
10940         (zero_extendsidi2_lfiwzx): Likewise.
10941         (extendsidi2_lfiwax): Likewise.
10942         (extendsidi2_nocell): Likewise.
10943         (floatsi<mode>2_lfiwax): Likewise.
10944         (lfiwax): Likewise.
10945         (floatunssi<mode>2_lfiwzx): Likewise.
10946         (lfiwzx): Likewise.
10947         (fix_trunc<mode>_stfiwx): Likewise.
10948         (fixuns_trunc<mode>_stfiwx): Likewise.
10949         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
10950         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
10951         (parity<mode>2_cmpb): Set length/type attr.
10952         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
10953         for 'mr.' to fast_compare.
10954         (bpermd_<mode>): Change type attr to popcnt.
10955         (p8_fmrgow_<mode>): New insns for power8 direct move support.
10956         (p8_mtvsrwz_1): Likewise.
10957         (p8_mtvsrwz_2): Likewise.
10958         (reload_fpr_from_gpr<mode>): Likewise.
10959         (p8_mtvsrd_1): Likewise.
10960         (p8_mtvsrd_2): Likewise.
10961         (p8_xxpermdi_<mode>): Likewise.
10962         (reload_vsx_from_gpr<mode>): Likewise.
10963         (reload_vsx_from_gprsf): Likewise.
10964         (p8_mfvsrd_3_<mode>): LIkewise.
10965         (reload_gpr_from_vsx<mode>): Likewise.
10966         (reload_gpr_from_vsxsf): Likewise.
10967         (p8_mfvsrd_4_disf): Likewise.
10968         (multi-word GPR splits): Do not split direct moves or quad memory
10969         operations.
10971 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
10973         * tree-into-ssa.c (interesting_blocks): Make static.
10975 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
10977         PR target/56564
10978         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
10979         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
10980         Use DATA_ABI_ALIGNMENT for that case instead if defined.
10981         (get_variable_align): New function.
10982         (get_variable_section, emit_bss, emit_common,
10983         assemble_variable_contents, place_block_symbol): Use
10984         get_variable_align instead of DECL_ALIGN.
10985         (assemble_noswitch_variable): Add align argument, use it
10986         instead of DECL_ALIGN.
10987         (assemble_variable): Adjust caller.  Use get_variable_align
10988         instead of DECL_ALIGN.
10989         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
10990         caller.
10991         (DATA_ABI_ALIGNMENT): Define.
10992         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
10993         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
10994         opt is false, only return the psABI mandated alignment increase.
10995         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
10996         (DATA_ABI_ALIGNMENT): ... this.
10997         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
10998         (DATA_ABI_ALIGNMENT): ... this.
10999         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
11000         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
11001         (DATA_ABI_ALIGNMENT): ... this.
11002         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
11003         * doc/tm.texi: Regenerated.
11005 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
11007         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
11008         cmp_code to construct REG_EQUAL note.
11010 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
11012         PR target/57568
11013         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
11014         that operands[2] doesn't overlap with operands[0].
11016 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
11017             Jan Hubicka  <jh@suse.cz>
11019         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
11020         hack to mark symbols as used.
11022 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
11024         PR rtl-optimization/57559
11025         * lra-constraints.c (process_alt_operands): Don't discourage
11026         memory with known offset for offsetable memory constraint.
11027         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
11029 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
11031         * varasm.c (struct oc_local_state): Reorder fields.
11032         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
11033         and adjust accordingly.
11034         (output_constructor): Reorder initialization code and adjust call to
11035         output_constructor_bitfield.
11037 2013-06-07  Jan Hubicka  <jh@suse.cz>
11039         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
11041 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
11043         * tree-object-size.c (unknown): Make const.
11045 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11047         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
11048         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
11049         for last alternative in the cpu_facility attribute.
11051 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11053         PR target/56315
11054         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
11055         (xordi3): Change operand 2 constraint to arm_xordi_operand.
11056         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
11057         * config/arm/constraints.md (Dg): New constraint.
11058         * config/arm/neon.md (xordi3_neon): Remove.
11059         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
11060         * config/arm/predicates.md (arm_xordi_operand): New predicate.
11062 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11064         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
11065         Clean up alternatives.
11067 2013-06-07  Alan Modra  <amodra@gmail.com>
11069         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
11070         va_list_gpr_size.
11072 2013-06-07  Alan Modra  <amodra@gmail.com>
11074         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
11076 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11078         * config/arm/constraints.md (Df): New constraint.
11079         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
11080         Correct length attribute for last two alternatives.
11082 2013-06-07  Alan Modra  <amodra@gmail.com>
11084         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
11085         override user -mfp-in-toc.
11086         (offsettable_ok_by_alignment): Consider just the current access
11087         rather than the whole object, unless BLKmode.  Handle
11088         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
11089         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
11090         for -mcmodel=medium.
11091         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
11092         override user -mfp-in-toc or -msum-in-toc.  Default to
11093         -mno-fp-in-toc for -mcmodel=medium.
11095 2013-06-06  DJ Delorie  <dj@redhat.com>
11097         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
11098         TARGET_VALID_POINTER_MODE.
11100 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
11101             Pat Haugen  <pthaugen@us.ibm.com>
11102             Peter Bergner  <bergner@vnet.ibm.com>
11104         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
11105         Document new power8 builtins.
11107         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
11108         condition code register, to allow 128-bit logical operations to be
11109         done in the VSX or GPR registers.
11110         (nor<mode>3): Use the canonical form for nor.
11111         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
11112         vclz*, and vpopcnt* vector instructions.
11113         (nand<mode>3): Likewise.
11114         (orc<mode>3): Likewise.
11115         (clz<mode>2): LIkewise.
11116         (popcount<mode>2): Likewise.
11118         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
11119         that only the GPRs are recognized.
11121         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11122         support for new power8 builtins.
11124         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
11125         builtin functions.
11126         (xscvdpspn): Likewise.
11127         (vclz): Likewise.
11128         (vclzb): Likewise.
11129         (vclzh): Likewise.
11130         (vclzw): Likewise.
11131         (vclzd): Likewise.
11132         (vpopcnt): Likewise.
11133         (vpopcntb): Likewise.
11134         (vpopcnth): Likewise.
11135         (vpopcntw): Likewise.
11136         (vpopcntd): Likewise.
11137         (vgbbd): Likewise.
11138         (vmrgew): Likewise.
11139         (vmrgow): Likewise.
11140         (eqv): Likewise.
11141         (eqv_v16qi3): Likewise.
11142         (eqv_v8hi3): Likewise.
11143         (eqv_v4si3): Likewise.
11144         (eqv_v2di3): Likewise.
11145         (eqv_v4sf3): Likewise.
11146         (eqv_v2df3): Likewise.
11147         (nand): Likewise.
11148         (nand_v16qi3): Likewise.
11149         (nand_v8hi3): Likewise.
11150         (nand_v4si3): Likewise.
11151         (nand_v2di3): Likewise.
11152         (nand_v4sf3): Likewise.
11153         (nand_v2df3): Likewise.
11154         (orc): Likewise.
11155         (orc_v16qi3): Likewise.
11156         (orc_v8hi3): Likewise.
11157         (orc_v4si3): Likewise.
11158         (orc_v2di3): Likewise.
11159         (orc_v4sf3): Likewise.
11160         (orc_v2df3): Likewise.
11162         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
11163         allow power8 quad mode in 64-bit.
11164         (rs6000_builtin_vectorized_function): Add support to vectorize
11165         ISA 2.07 count leading zeros, population count builtins.
11166         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
11167         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
11168         (builtin_function_type): Add vgbbd builtin function which takes an
11169         unsigned argument.
11170         (altivec_expand_vec_perm_const): Add support for new power8 merge
11171         instructions.
11173         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
11174         that does not include TImdoe for use with 32-bit.
11175         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
11176         instructions.
11177         (UNSPEC_VSX_CVDPSPN): Likewise.
11178         (vsx_xscvdpspn): Likewise.
11179         (vsx_xscvspdpn): Likewise.
11180         (vsx_xscvdpspn_scalar): Likewise.
11181         (vsx_xscvspdpn_directmove): Likewise.
11182         (vsx_and<mode>3): Split logical operations into 32-bit and
11183         64-bit. Add support to do logical operations on TImode as well as
11184         VSX vector types.  Allow logical operations to be done in either
11185         VSX registers or in general purpose registers in 64-bit mode.  Add
11186         splitters if GPRs were used. For AND, add clobber of CCmode to
11187         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
11188         encoding.
11189         (vsx_and<mode>3_32bit): Likewise.
11190         (vsx_and<mode>3_64bit): Likewise.
11191         (vsx_ior<mode>3): Likewise.
11192         (vsx_ior<mode>3_32bit): Likewise.
11193         (vsx_ior<mode>3_64bit): Likewise.
11194         (vsx_xor<mode>3): Likewise.
11195         (vsx_xor<mode>3_32bit): Likewise.
11196         (vsx_xor<mode>3_64bit): Likewise.
11197         (vsx_one_cmpl<mode>2): Likewise.
11198         (vsx_one_cmpl<mode>2_32bit): Likewise.
11199         (vsx_one_cmpl<mode>2_64bit): Likewise.
11200         (vsx_nor<mode>3): Likewise.
11201         (vsx_nor<mode>3_32bit): Likewise.
11202         (vsx_nor<mode>3_64bit): Likewise.
11203         (vsx_andc<mode>3): Likewise.
11204         (vsx_andc<mode>3_32bit): Likewise.
11205         (vsx_andc<mode>3_64bit): Likewise.
11206         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
11207         and xxlorc instructions.
11208         (vsx_eqv<mode>3_64bit): Likewise.
11209         (vsx_nand<mode>3_32bit): Likewise.
11210         (vsx_nand<mode>3_64bit): Likewise.
11211         (vsx_orc<mode>3_32bit): Likewise.
11212         (vsx_orc<mode>3_64bit): Likewise.
11214         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
11216         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
11217         instruction.
11218         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
11219         (p8_vmrgow): Likewise.
11220         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
11221         GPRs to be split under VSX.
11222         (p8v_clz<mode>2): Add power8 count leading zero support.
11223         (p8v_popcount<mode>2): Add power8 population count support.
11224         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
11225         support.
11227         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
11228         instruction.
11230         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
11231         builtin functions.
11232         (vec_nand): Likewise.
11233         (vec_vclz): Likewise.
11234         (vec_vclzb): Likewise.
11235         (vec_vclzd): Likewise.
11236         (vec_vclzh): Likewise.
11237         (vec_vclzw): Likewise.
11238         (vec_vgbbd): Likewise.
11239         (vec_vmrgew): Likewise.
11240         (vec_vmrgow): Likewise.
11241         (vec_vpopcnt): Likewise.
11242         (vec_vpopcntb): Likewise.
11243         (vec_vpopcntd): Likewise.
11244         (vec_vpopcnth): Likewise.
11245         (vec_vpopcntw): Likewise.
11247 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
11249         PR rtl-optimization/57468
11250         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
11251         spilled pseudos.
11253 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
11255         PR rtl-optimization/57459
11256         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
11257         type when setting live regs.
11259 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
11261         * config/s390/s390.opt (mlra): New option.
11262         * config/s390/s390.c (s390_decompose_address): Check displacement
11263         for all registers for LRA.
11264         (s390_secondary_reload): Don't used secondary reloads for LRA.
11265         (s390_lra_p): New function.
11266         (TARGET_LRA_P): Define.
11267         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
11268         of attribute cpu_facility to zarch for the last alternative.
11269         (*cmpmem_short): Ditto.
11271 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
11273         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
11274         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
11275         (arm_expand_prologue): Likewise.
11277 2013-06-06  Teresa Johnson  <tejohnson@google.com>
11279         PR c++/53743
11280         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
11281         as this is now done by redirect_edge_and_branch_force.
11282         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
11283         barriers, and fix interaction with splitting.
11284         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
11285         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
11286         reflect changes made in the routine.
11287         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
11288         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
11289         since this is called in cfglayout mode, and replace partition fixup
11290         with assert as that is now done by force_nonfallthru_and_redirect.
11291         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
11292         already be marked with region crossing note.
11293         (insert_section_boundary_note): Make non-static, gate on flag
11294         has_bb_partition, rewrite to also check for multiple partitions.
11295         (rest_of_handle_reorder_blocks): Remove call to
11296         insert_section_boundary_note, now done later during free_cfg.
11297         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
11298         * bb-reorder.h (insert_section_boundary_note): Declare.
11299         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
11300         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
11301         invoke insert_section_boundary_note.
11302         (try_redirect_by_replacing_jump): Remove unnecessary
11303         check for region crossing note.
11304         (fixup_partition_crossing): New function.
11305         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
11306         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
11307         in non-cfglayout mode.
11308         (force_nonfallthru_and_redirect): Fixup partition boundaries,
11309         remove old code that tried to do this. Emit barrier correctly
11310         when we are in cfglayout mode.
11311         (last_bb_in_partition): New function.
11312         (rtl_split_edge): Correctly fixup partition boundaries.
11313         (commit_one_edge_insertion): Remove old code that tried to
11314         fixup region crossing edge since this is now handled in
11315         split_block, and set up insertion point correctly since
11316         block may now end in a jump.
11317         (verify_hot_cold_block_grouping): Guard against checking when not in
11318         linearized RTL mode.
11319         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
11320         notes.
11321         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
11322         rtl_verify_flow_info, so not called in cfglayout mode.
11323         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
11324         (fixup_reorder_chain): Remove old code that attempted to fixup region
11325         crossing note as this is now handled in force_nonfallthru_and_redirect.
11326         (duplicate_insn_chain): Don't duplicate switch section notes.
11327         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
11328         note.
11329         * basic-block.h (emit_barrier_after_bb): Declare.
11331 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11333         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
11334         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
11335         arm_usatsihi): Adjust alternatives for arm_restrict_it.
11337 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11339         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
11340         where appropriate.
11341         * config/arm/ldmstm.md: Regenerate.
11343 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11345         * config/arm/sync.md (atomic_loaddi_1):
11346         Disable predication for arm_restrict_it.
11347         (arm_load_exclusive<mode>): Likewise.
11348         (arm_load_exclusivesi): Likewise.
11349         (arm_load_exclusivedi): Likewise.
11350         (arm_load_acquire_exclusive<mode>): Likewise.
11351         (arm_load_acquire_exclusivesi): Likewise.
11352         (arm_load_acquire_exclusivedi): Likewise.
11353         (arm_store_exclusive<mode>): Likewise.
11354         (arm_store_exclusive<mode>): Likewise.
11355         (arm_store_release_exclusivedi): Likewise.
11356         (arm_store_release_exclusive<mode>): Likewise.
11358 2013-06-06  Richard Biener  <rguenther@suse.de>
11360         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
11361         after LTO_null.
11362         (lto_tag_is_tree_code_p): Adjust.
11363         (lto_tag_is_gimple_code_p): Likewise.
11364         (lto_gimple_code_to_tag): Likewise.
11365         (lto_tag_to_gimple_code): Likewise.
11366         (lto_tree_code_to_tag): Likewise.
11367         (lto_tag_to_tree_code): Likewise.
11368         * data-streamer.h (streamer_write_hwi_in_range): Use
11369         uhwi streaming to stream the normalized range.
11370         (streamer_read_hwi_in_range): Likewise.
11372 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11374         * config/arm/arm.md (enabled_for_depr_it): New attribute.
11375         (predicable_short_it): Likewise.
11376         (predicated): Likewise.
11377         (enabled): Handle above.
11378         (define_cond_exec): Set predicated attribute to yes.
11380 2013-06-05  Mike Stump  <mikestump@comcast.net>
11382         * gdbinit.in (__FUNCTION__): Add.
11384 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
11386         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
11387         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
11389 2013-06-05  Jan Hubicka  <jh@suse.cz>
11391         * varasm.c (mark_decl_referenced): Revert the removal until targets
11392         are fixed.
11394 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
11396         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
11397         instead of mark_decl_referenced.
11399 2013-06-05  Jan Hubicka  <jh@suse.cz>
11401         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
11402         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
11403         and symtab_used_from_object_file_p.
11404         (cgraph_make_node_local_1): Clear forced_by_abi.
11405         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
11406         * cgraph.h (symtab_node_base): Add forced_by_abi.
11407         (decide_is_variable_needed): Remove.
11408         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
11409         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
11410         (decide_is_symbol_needed): ... this one; handle symbols in general;
11411         always analyze virtuals; honnor forced_by_abi.
11412         (cgraph_finalize_function): Update.
11413         (varpool_finalize_decl): Update.
11414         (symbol_defined_and_needed): Remove.
11415         (analyze_functions): Update.
11416         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
11417         output_refs, input_overwrite_node): Handle forced_by_abi.
11418         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
11419         (address_taken_from_non_vtable_p): ... this one.
11420         (comdat_can_be_unshared_p_1): New function.
11421         (cgraph_comdat_can_be_unshared_p): Rename to ...
11422         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
11423         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
11424         (function_and_variable_visibility): Clear forced_by_abi as needed.
11425         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
11426         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
11427         * symtab.c (dump_symtab_base): Dump forced_by_abi.
11428         * varpool.c (decide_is_variable_needed): Remove.
11430 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11432         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
11433         (arm_option_override): Override arm_restrict_it where appropriate.
11434         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
11435         * config/arm/arm.opt (mrestrict-it): New command-line option.
11436         * doc/invoke.texi: Document -mrestrict-it.
11438 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
11440         * tsan.c (tsan_atomic_table): Make const.
11442 2013-06-05  Richard Biener  <rguenther@suse.de>
11444         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
11445         index associated with the tree we are supposed to replace.
11446         * tree-streamer-out.c (pack_ts_base_value_fields): Output
11447         TREE_ASM_WRITTEN as zero for everything but SSA names.
11449 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
11451         * tree-ssa-structalias.c (call_stmt_vars): Make static.
11453 2013-06-04  Jan Hubicka  <jh@suse.cz>
11455         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
11456         (input_node, input_varpool_node): Handle correctly external same
11457         body aliases.
11458         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
11459         nodes at ltrans stage.
11461 2013-06-04  Jan Hubicka  <jh@suse.cz>
11463         * ipa-inline.c (update_caller_keys): Fix availability test.
11464         (update_callee_keys): Likewise.
11465         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
11466         to follow ELF standard.
11468 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
11470         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
11471         (mips64r5900el-*-elf*): New configurations.
11472         * config/mips/mips-cpus.def (r5900): New processor.
11473         * config/mips/mips-tables.opt: Regenerate.
11474         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
11475         (mips_issue_rate): Handle PROCESSOR_R5900.
11476         (mips_reorg_process_insns): Force reorder mode for the R5900.
11477         * config/mips/mips.h (TARGET_MIPS5900): Define.
11478         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
11479         TARGET_MIPS5900.
11480         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
11481         TARGET_MIPS5900.
11482         * config/mips/mips.md (processor): Add r5900.
11483         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
11485 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11487         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
11488         into function to generate MOVI instruction.
11489         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
11490         (aarch64_preferred_simd_mode): Turn into wrapper.
11491         (aarch64_output_scalar_simd_mov_immediate): New function.
11492         * config/aarch64/aarch64-protos.h: Add prototype for above.
11494 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11496         * config/aarch64/aarch64.c (simd_immediate_info): Remove
11497         element_char member.
11498         (sizetochar): Return signed char.
11499         (aarch64_simd_valid_immediate): Remove elchar and other
11500         unnecessary variables.
11501         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
11502         Calculate element_char as required.
11503         * config/aarch64/aarch64-protos.h: Update and move prototype
11504         for aarch64_output_simd_mov_immediate.
11505         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11506         Update arguments.
11508 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11510         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
11511         information completed by aarch64_simd_valid_immediate.
11512         (aarch64_legitimate_constant_p): Update arguments.
11513         (aarch64_simd_valid_immediate): Work with struct rather than many
11514         pointers.
11515         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
11516         (aarch64_simd_make_constant): Update arguments.
11517         (aarch64_output_simd_mov_immediate): Work with struct rather than
11518         many pointers.  Output immediate directly rather than as operand.
11519         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
11520         Update prototype.
11521         * config/aarch64/constraints.md (Dn): Update arguments.
11523 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11525         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
11526         longer static.
11527         (aarch64_simd_immediate_valid_for_move): Remove.
11528         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
11529         (aarch64_simd_make_constant): Update call.
11530         (aarch64_output_simd_mov_immediate): Update call.
11531         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
11532         Add prototype.
11533         * config/aarch64/constraints.md (Dn): Update call.
11535 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11537         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
11538         return type to bool for prototype.
11539         (aarch64_legitimate_constant_p): Check for true instead of not -1.
11540         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
11541         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
11543 2013-06-04  Catherine Moore  <clm@codesourcery.com>
11545         * config/mips/mips.opt (meva): New.
11546         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
11547         (ASM_SPEC): Handle -meva.
11548         * doc/invoke.texi (meva):  Document.
11550 2013-06-04  Alan Modra  <amodra@gmail.com>
11552         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
11553         constant output.
11555 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11557         * rtl.def: Add extra fourth optional field to define_cond_exec.
11558         * gensupport.c (process_one_cond_exec): Process attributes from
11559         define_cond_exec.
11560         * doc/md.texi: Document fourth field in define_cond_exec.
11562 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
11564         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
11565         out the processing order as in store_bit_field_1.
11567 2013-06-04  Jan Hubicka  <jh@suse.cz>
11569         PR middle-end/57500
11570         * cgraphunit.c (cgraph_process_same_body_aliases): Create
11571         non-VAR_DECL node if it does not exist yet.
11573 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
11575         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
11576         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
11577         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
11578         target_cpu_default setting.
11580 2013-06-03  Teresa Johnson  <tejohnson@google.com>
11582         * dumpfile.c (opt_info_switch_p): Change -fopt-info
11583         default to -fopt-info=optimized instead of all.
11584         * doc/invoke.texi: Ditto.
11585         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
11586         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
11587         (execute_vect_slp): Emit BB vectorization success under
11588         MSG_OPTIMIZED_LOCATIONS.
11589         * tree-vect-slp.c (vect_slp_transform_bb): Change
11590         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
11591         * tree-vect-loop.c (vect_transform_loop): Ditto.
11593 2013-06-03  Jason Merrill  <jason@redhat.com>
11595         PR c++/57415
11596         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11597         Use TARGET_EXPR for C++.
11599 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
11601         PR rtl-optimization/57268
11602         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
11603         if DEBUG_INSN_P (insn).
11605         Reapply
11606         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
11608         PR rtl-optimization/57268
11609         * sched-deps.c (sched_analyze_2): Flush dependence lists if
11610         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11612 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11614         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
11615         (ix86_avoid_lea_for_addr): Likewise.
11616         (exact_dependency_1): Likewise.
11617         (ix86_adjust_cost): Likewise.
11618         (swap_top_of_ready_list): Fix formatting and !reload_completed check
11619         removed.
11620         (do_reorder_for_imul): Fix typo, formatting and
11621         !reload_completed check removed.
11622         (ix86_sched_reorder): Fix typo and formatting.
11623         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
11624         list.
11626 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
11628         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
11630 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
11632         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
11633         <STRING_CST>: Likewise.
11634         <VECTOR_CST>: Likewise.
11636 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
11637             Mikael Morin  <mikael@gcc.gnu.org>
11639         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
11640         * config.in: Regenerated.
11641         * configure: Regenerated.
11643 2013-06-01  Jan Hubicka  <jh@suse.cz>
11645         PR middle-end/57366
11646         * cgraphunit.c (compile): When weakref is not supported,
11647         set up transparent aliases before final output pass.
11648         * varasm.c (assemble_alias): Do not try to do it here.
11650 2013-06-01  Jan Hubicka  <jh@suse.cz>
11652         PR middle-end/57467
11653         * passes.c (for_per_function): Skip unanalyzed functions.
11655 2013-06-01  Jan Hubicka  <jh@suse.cz>
11657         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
11658         (lto_symtab_merge_symbols_1): ... this one.
11659         (lto_symtab_merge_cgraph_nodes): Rename to ...
11660         (lto_symtab_merge_symbols): ... this one; simplify.
11661         * cgraph.c (same_body_aliases_done): Rename to ...
11662         (cpp_implicit_aliases_done): ... this one.
11663         (cgraph_create_function_alias): Update.
11664         (cgraph_same_body_alias): Update.
11665         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
11666         (verify_edge_corresponds_to_fndecl): Simplify.
11667         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
11668         (cgraph_node): Remove same_body_alias.
11669         (varpool_node): Remove alias_of and extra_name_alias.
11670         (same_body_aliases_done): Rename to ..
11671         (cpp_implicit_aliases_done): ... this one.
11672         (symtab_alias_ultimate_target): Add default parameter.
11673         (symtab_resolve_alias): New function.
11674         (fixup_same_cpp_alias_visibility): Declare.
11675         (cgraph_function_node): Add default parameter.
11676         (cgraph_node_asm_name): Likewise.
11677         (cgraph_function_or_thunk_node): Add default parameter; do
11678         not ICE when it is NULL.
11679         (varpool_variable_node): Likewise.
11680         * tree-emutls.c (create_emultls_var): Update.
11681         (ipa_lower_emutls): Update.
11682         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
11683         (cgraph_reset_node): Reset alias info.
11684         (cgraph_finalize_function): Update.
11685         (fixup_same_cpp_alias_visibility): Move to symtab.c.
11686         (analyze_function): Simplify.
11687         (cgraph_process_same_body_aliases): Simplify.
11688         (analyze_functions): Fixup same body aliases.
11689         (handle_alias_pairs): Simplify.
11690         (assemble_thunk): Update.
11691         (assemble_thunks_and_aliases): Update.
11692         (output_weakrefs): Rewrite.
11693         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
11694         (lto_output_varpool_node): Likewise.
11695         (compute_ltrans_boundary): Remve assert.
11696         (get_alias_symbol): New functoin.
11697         (input_node): Rewrite alias handling.
11698         (input_varpool_node): Likewise.
11699         * ipa-pure-const.c (propagate_pure_const): Fix formating.
11700         * ipa.c (process_references): Handle weakrefs correctly.
11701         (symtab_remove_unreachable_nodes): Likewise.
11702         * trans-mem.c (get_cg_data): Update.
11703         (ipa_tm_create_version_alias): Update.
11704         (ipa_tm_execute): Update.
11705         * symtab.c (dump_symtab_base): Dump aliases.
11706         (verify_symtab_base): Verify aliases.
11707         (symtab_node_availability): New function.
11708         (symtab_alias_ultimate_target): Simplify.
11709         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
11710         handle all the fixup cases.
11711         (symtab_resolve_alias): New function.
11712         * passes.c (ipa_write_summaries): Handle weakrefs.
11713         * varpool.c (varpool_analyze_node): Simplify.
11714         (assemble_aliases): Update.
11715         (varpool_create_variable_alias): Simplify.
11716         (varpool_extra_name_alias): Simplify.
11717         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
11718         (lto_symtab_merge_symbols): ... this one.
11720 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
11722         Revert
11723         PR rtl-optimization/57268
11724         * sched-deps.c (sched_analyze_2): Flush dependence lists if
11725         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11727 2013-06-01  Tobias Burnus  <burnus@net-b.de>
11729         Partially reverted:
11730         2013-05-31  Tobias Burnus  <burnus@net-b.de>
11732         PR middle-end/57073
11733         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
11734         further up.
11736 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
11738         PR rtl-optimization/57268
11739         * sched-deps.c (sched_analyze_2): Flush dependence lists if
11740         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11742 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
11744         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
11745         unordered comparison operators when -fno-trapping-math is in effect
11746         on the e500.
11747         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
11748         and implement unordered comparison operators properly on the e500.
11750 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
11752         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
11753         for constant scalar integers.
11754         (simplify_relational_operation_1): Likewise.
11756 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
11758         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
11759         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
11760         Fix comment.
11762 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11763             Igor Zamyatin  <igor.zamyatin@intel.com>
11765         Silvermont (SLM) architecture performance tuning.
11766         * config/i386/i386.h (enum ix86_tune_indices): Add
11767         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
11768         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
11770         * config/i386/i386.c (initial_ix86_tune_features)
11771         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
11772         (ix86_lea_outperforms): Handle Silvermont tuning.
11773         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
11774         call.
11775         (ix86_use_lea_for_mov): Likewise.
11776         (ix86_avoid_lea_for_addr): Likewise.
11777         (ix86_lea_for_add_ok): Likewise.
11778         (exact_dependency_1): New function.
11779         (exact_store_load_dependency): Likewise.
11780         (ix86_adjust_cost): Handle Silvermont tuning.
11781         (do_reoder_for_imul): Likewise.
11782         (swap_top_of_ready_list): New function.
11783         (ix86_sched_reorder): Changed to handle Silvermont tuning.
11785         * config/i386/i386.md (peepholes that split memory operand in fp
11786         converts): New.
11788 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11790         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11791         Remove un-necessary braces.
11793 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11795         * config/aarch64/aarch64.c (aarch64_classify_symbol):
11796         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
11798 2013-05-31  Tobias Burnus  <burnus@net-b.de>
11800         PR middle-end/57073
11801         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
11803 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11805         PR target/56315
11806         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
11807         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
11808         * config/arm/neon.md (iordi3_neon): Remove.
11809         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
11810         * config/arm/predicates.md (imm_for_neon_logic_operand):
11811         Move to earlier in the file.
11812         (neon_logic_op2): Likewise.
11813         (arm_iordi_operand_neon): New predicate.
11815 2013-05-31  Richard Biener  <rguenther@suse.de>
11817         PR tree-optimization/57478
11818         PR tree-optimization/57453
11819         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
11820         are life as well.
11822 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11824         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
11825         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
11827 2013-05-30  Tobias Burnus  <burnus@net-b.de>
11828             Thomas Koenig  <tkoenig@gcc.gnu.org>
11830         PR middle-end/57073
11831         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
11832         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
11834 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
11836         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
11838 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
11840         * target.def (register_usage_leveling_p): New hook.
11841         * targhooks.c (default_register_usage_leveling_p): New.
11842         * targhooks.h (default_register_usage_leveling_p): New prototype.
11843         * lra-assigns.c (register_usage_leveling_p): Use the hook.
11844         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
11845         * doc/tm.texi: Update.
11846         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
11848 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
11850         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
11851         (*insv_reg<mode>): New define_insn.
11853 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
11855         PR rtl-optimization/57439
11856         * postreload.c (move2add_valid_value_p): Check that we have
11857         a zero subreg_regno_offset when accessing the register in
11858         the requested mode.
11860 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11861             Igor Zamyatin  <igor.zamyatin@intel.com>
11863         Silvermont (SLM) architecture pipeline model, tuning and
11864         insn selection.
11865         * config.gcc: Add slm config options and target.
11867         * config/i386/slm.md: New.
11869         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
11871         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
11872         PROCESSOR_SLM.
11873         (ix86_target_macros_internal): Likewise.
11875         * config/i386/i386.c (slm_cost): New cost.
11876         (m_SLM): New macro flag.
11877         (initial_ix86_tune_features): Set m_SLM.
11878         (x86_accumulate_outgoing_args): Likewise.
11879         (x86_arch_always_fancy_math_387): Likewise.
11880         (processor_target_table): Add slm cost.
11881         (cpu_names): Add slm cpu name.
11882         (x86_option_override_internal): Set SLM ISA.
11883         (ix86_issue_rate): New case PROCESSOR_SLM.
11884         (ia32_multipass_dfa_lookahead): Likewise.
11885         (fold_builtin_cpu): Add slm.
11887         * config/i386/i386.h (TARGET_SLM): New target macro.
11888         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
11889         (processor_type): Add PROCESSOR_SLM.
11891         * config/i386/i386.md (cpu): Add new value "slm".
11892         (slm.md): Include slm.md.
11894 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
11895             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11897         * config/arm/arm-protos.h: Add and update function protos.
11898         * config/arm/arm.c (use_simple_return_p): New added.
11899         (thumb2_expand_return): Check simple_return flag.
11900         * config/arm/arm.md: Add simple_return and conditional simple_return.
11901         * config/arm/iterators.md: Add iterator for return and simple_return.
11903 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11905         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
11906         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
11907         (arm_emit_vfp_multi_reg_pop): Likewise.
11908         (thumb2_emit_ldrd_pop): Likewise.
11909         (arm_expand_epilogue): Add misc REG_CFA notes.
11910         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
11912 2013-05-29  Lawrence Crowl  <crowl@google.com>
11914         * config/arm/t-arm: Update for below.
11916         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
11917         Change type to hash_table.  Update dependent calls and types.
11919         * config/i386/t-cygming: Update for below.
11921         * config/i386/t-interix: Update for below.
11923         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
11924         Change type to hash_table.  Update dependent calls and types.
11925         (i386_find_on_wrapper_list::wrappers): Likewise.
11927         * config/ia64/t-ia64: Update for below.
11929         * config/ia64/ia64.c (bundle_state_table):
11930         Change type to hash_table.  Update dependent calls and types.
11932         * config/mips/mips.c (mips_reorg_process_insns::htab):
11933         Change type to hash_table.  Update dependent calls and types.
11935         * config/sol2.c (solaris_comdat_htab):
11936         Change type to hash_table.  Update dependent calls and types.
11938         * config/t-sol2: Update for above.
11940 2013-05-29  Teresa Johnson  <tejohnson@google.com>
11942         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
11943         functions are not yet marked as defined.
11945 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
11946             Pat Haugen  <pthaugen@us.ibm.com>
11947             Peter Bergner  <bergner@vnet.ibm.com>
11949         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
11950         instructions.
11951         (VEC_A): Likewise.
11952         (VEC_C): Likewise.
11953         (vrotl<mode>3): Likewise.
11954         (vashl<mode>3): Likewise.
11955         (vlshr<mode>3): Likewise.
11956         (vashr<mode>3): Likewise.
11958         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11959         support for power8 V2DI builtins.
11961         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
11962         power8 V2DI builtins.
11963         (vupkhsw): Likewise.
11964         (vupklsw): Likewise.
11965         (vaddudm): Likewise.
11966         (vminsd): Likewise.
11967         (vmaxsd): Likewise.
11968         (vminud): Likewise.
11969         (vmaxud): Likewise.
11970         (vpkudum): Likewise.
11971         (vpksdss): Likewise.
11972         (vpkudus): Likewise.
11973         (vpksdus): Likewise.
11974         (vrld): Likewise.
11975         (vsld): Likewise.
11976         (vsrd): Likewise.
11977         (vsrad): Likewise.
11978         (vsubudm): Likewise.
11979         (vcmpequd): Likewise.
11980         (vcmpgtsd): Likewise.
11981         (vcmpgtud): Likewise.
11982         (vcmpequd_p): Likewise.
11983         (vcmpgtsd_p): Likewise.
11984         (vcmpgtud_p): Likewise.
11985         (vupkhsw): Likewise.
11986         (vupklsw): Likewise.
11987         (vaddudm): Likewise.
11988         (vmaxsd): Likewise.
11989         (vmaxud): Likewise.
11990         (vminsd): Likewise.
11991         (vminud): Likewise.
11992         (vpksdss): Likewise.
11993         (vpksdus): Likewise.
11994         (vpkudum): Likewise.
11995         (vpkudus): Likewise.
11996         (vrld): Likewise.
11997         (vsld): Likewise.
11998         (vsrad): Likewise.
11999         (vsrd): Likewise.
12000         (vsubudm): Likewise.
12002         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
12003         support for power8 V2DI instructions.
12005         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
12006         power8 V2DI instructions.  Combine pack and unpack insns to use an
12007         iterator for each mode.  Check whether a particular mode supports
12008         Altivec instructions instead of just checking TARGET_ALTIVEC.
12009         (UNSPEC_VPKUWUM): Likewise.
12010         (UNSPEC_VPKSHSS): Likewise.
12011         (UNSPEC_VPKSWSS): Likewise.
12012         (UNSPEC_VPKUHUS): Likewise.
12013         (UNSPEC_VPKSHUS): Likewise.
12014         (UNSPEC_VPKUWUS): Likewise.
12015         (UNSPEC_VPKSWUS): Likewise.
12016         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
12017         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
12018         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
12019         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
12020         (UNSPEC_VUPKHSB): Likewise.
12021         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
12022         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
12023         (UNSPEC_VUPKHSH): Likewise.
12024         (UNSPEC_VUPKLSB): Likewise.
12025         (UNSPEC_VUPKLSH): Likewise.
12026         (VI2): Likewise.
12027         (VI_char): Likewise.
12028         (VI_scalar): Likewise.
12029         (VI_unit): Likewise.
12030         (VP): Likewise.
12031         (VP_small): Likewise.
12032         (VP_small_lc): Likewise.
12033         (VU_char): Likewise.
12034         (add<mode>3): Likewise.
12035         (altivec_vaddcuw): Likewise.
12036         (altivec_vaddu<VI_char>s): Likewise.
12037         (altivec_vadds<VI_char>s): Likewise.
12038         (sub<mode>3): Likewise.
12039         (altivec_vsubcuw): Likewise.
12040         (altivec_vsubu<VI_char>s): Likewise.
12041         (altivec_vsubs<VI_char>s): Likewise.
12042         (altivec_vavgs<VI_char>): Likewise.
12043         (altivec_vcmpbfp): Likewise.
12044         (altivec_eq<mode>): Likewise.
12045         (altivec_gt<mode>): Likewise.
12046         (altivec_gtu<mode>): Likewise.
12047         (umax<mode>3): Likewise.
12048         (smax<mode>3): Likewise.
12049         (umin<mode>3): Likewise.
12050         (smin<mode>3): Likewise.
12051         (altivec_vpkuhum): Likewise.
12052         (altivec_vpkuwum): Likewise.
12053         (altivec_vpkshss): Likewise.
12054         (altivec_vpkswss): Likewise.
12055         (altivec_vpkuhus): Likewise.
12056         (altivec_vpkshus): Likewise.
12057         (altivec_vpkuwus): Likewise.
12058         (altivec_vpkswus): Likewise.
12059         (altivec_vpks<VI_char>ss): Likewise.
12060         (altivec_vpks<VI_char>us): Likewise.
12061         (altivec_vpku<VI_char>us): Likewise.
12062         (altivec_vpku<VI_char>um): Likewise.
12063         (altivec_vrl<VI_char>): Likewise.
12064         (altivec_vsl<VI_char>): Likewise.
12065         (altivec_vsr<VI_char>): Likewise.
12066         (altivec_vsra<VI_char>): Likewise.
12067         (altivec_vsldoi_<mode>): Likewise.
12068         (altivec_vupkhsb): Likewise.
12069         (altivec_vupkhs<VU_char>): Likewise.
12070         (altivec_vupkls<VU_char>): Likewise.
12071         (altivec_vupkhsh): Likewise.
12072         (altivec_vupklsb): Likewise.
12073         (altivec_vupklsh): Likewise.
12074         (altivec_vcmpequ<VI_char>_p): Likewise.
12075         (altivec_vcmpgts<VI_char>_p): Likewise.
12076         (altivec_vcmpgtu<VI_char>_p): Likewise.
12077         (abs<mode>2): Likewise.
12078         (vec_unpacks_hi_v16qi): Likewise.
12079         (vec_unpacks_hi_v8hi): Likewise.
12080         (vec_unpacks_lo_v16qi): Likewise.
12081         (vec_unpacks_hi_<VP_small_lc>): Likewise.
12082         (vec_unpacks_lo_v8hi): Likewise.
12083         (vec_unpacks_lo_<VP_small_lc>): Likewise.
12084         (vec_pack_trunc_v8h): Likewise.
12085         (vec_pack_trunc_v4si): Likewise.
12086         (vec_pack_trunc_<mode>): Likewise.
12088         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
12089         V2DI builtins.
12090         (vec_vmaxsd): Likewise.
12091         (vec_vmaxud): Likewise.
12092         (vec_vminsd): Likewise.
12093         (vec_vminud): Likewise.
12094         (vec_vpksdss): Likewise.
12095         (vec_vpksdus): Likewise.
12096         (vec_vpkudum): Likewise.
12097         (vec_vpkudus): Likewise.
12098         (vec_vrld): Likewise.
12099         (vec_vsld): Likewise.
12100         (vec_vsrad): Likewise.
12101         (vec_vsrd): Likewise.
12102         (vec_vsubudm): Likewise.
12103         (vec_vupkhsw): Likewise.
12104         (vec_vupklsw): Likewise.
12106 2013-05-29  Jan Hubicka  <jh@suse.cz>
12108         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
12109         flags; reorder rest of fields in more consistent way.
12110         (varpool_node): Remove analyzed, finalized and alias.
12111         (cgraph_ndoe): Likewise.
12112         (symtab_alias_ultimate_target): New function.
12113         (cgraph_function_node): Move offline.
12114         (cgraph_reset_node): Declare.
12115         (cgraph_comdat_can_be_unshared_p): Remove.
12116         (varpool_remove_initializer): Declare.
12117         (varpool_first_defined_variable, varpool_next_defined_variable
12118         cgraph_first_defined_function, cgraph_next_defined_function): Update.
12119         (cgraph_function_with_gimple_body_p): Update.
12120         (varpool_all_refs_explicit_p): Update.
12121         (symtab_alias_target): New function.
12122         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
12123         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
12124         (cgraph_function_or_thunk_node): Simplify using
12125         symtab_alias_ultimate_target.
12126         (varpool_variable_node): Likewise.
12127         * cgraph.c (cgraph_create_function_alias): Update.
12128         (cgraph_add_thunk): Update.
12129         (cgraph_remove_node): Update.
12130         (dump_cgraph_node): Do not dump removed flags.
12131         (cgraph_function_body_availability): Update.
12132         (cgraph_propagate_frequency): Update.
12133         (verify_cgraph_node): Check sanity of local flag.
12134         (cgraph_function_node): Move here from cgraph.h; revamp for
12135         cgraph_function_or_thunk_node.
12136         * lto-symtab.c (lto_varpool_replace_node): Update.
12137         (lto_symtab_resolve_can_prevail_p): Update.
12138         (lto_symtab_merge_cgraph_nodes): Update.
12139         * ipa-cp.c (determine_versionability, initialize_node_lattices,
12140         propagate_constants_accross_call, devirtualization_time_bonus,
12141         ipcp_propagate_stage): Update.
12142         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
12143         * ipa-inline-transform.c (clone_inlined_nodes,
12144         preserve_function_body_p): Update.
12145         * ipa-reference.c (propagate): Update.
12146         (write_node_summary_p): Update.
12147         * toplev.c (wrapup_global_declaration_2): Update.
12148         * cgraphunit.c (cgraph_analyze_function): Rename to ...
12149         (analyze_function) ... this one.
12150         (cgraph_process_new_functions): Update.
12151         (cgraph_reset_node): Export.
12152         (cgraph_finalize_function): Update.
12153         (cgraph_add_new_function): Update.
12154         (process_function_and_variable_attributes): Update.
12155         (varpool_finalize_decl): Update.
12156         (symbol_finalized): Remove.
12157         (symbol_finalized_and_needed): Rename to ...
12158         (symbol_defined_and_needed): ... update.
12159         (cgraph_analyze_functions): Update.
12160         (handle_alias_pairs): Update.
12161         (mark_functions_to_output): Update.
12162         (assemble_thunk): Update.
12163         (output_in_order): Update.
12164         (output_weakrefs): Update.
12165         (finalize_compilation_unit): Update.
12166         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
12167         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
12168         input_node, input_varpool_node): Update.
12169         * dbxout.c (dbxout_expand_expr): Update.
12170         * cgraphclones.c (cgraph_clone_node): Update.
12171         (cgraph_copy_node_for_versioning): Update.
12172         (cgraph_materialize_clone): Update.
12173         (cgraph_materialize_all_clones): Update.
12174         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
12175         propagate_pure_const, propagate_nothrow): Update.
12176         * lto-streamer-out.c (lto_output, write_symbol): Update.
12177         * ipa-utils.c (ipa_reverse_postorder): Update.
12178         * ipa-inline.c (can_inline_edge_p): Update.
12179         (update_caller_keys, ipa_inline): Update.
12180         * dwarf2out.c (reference_to_unused,
12181         premark_types_used_by_global_vars_helper): Update.
12182         * tree-eh.c (tree_could_trap_p): Update.
12183         * ipa-split.c (consider_split, execute_split_functions): Update.
12184         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
12185         has_addr_references_p): Update; move ahead in file for better
12186         readability.
12187         (process_references): Simplify.
12188         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
12189         bodies are removed.
12190         (cgraph_comdat_can_be_unshared_p): Make static.
12191         (cgraph_externally_visible_p): Update.
12192         (varpool_externally_visible_p): Update.
12193         (function_and_variable_visibility): Update.
12194         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
12195         ipa_tm_mark_force_output_node): Update.
12196         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
12197         estimate_edge_devirt_benefit, inline_generate_summary,
12198         inline_write_summary): Update.
12199         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
12200         * ipa-prop.c (ipa_compute_jump_functions): Update.
12201         (ipa_print_node_params, ipa_prop_read_section,
12202         ipa_update_after_lto_read, read_replacements_section): Update.
12203         * varasm.c (mark_decl_referenced): Update.
12204         (assemble_alias, dump_tm_clone_pairs): Update.
12205         * tree-inline.c (copy_bb): Update.
12206         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
12207         Update.
12208         * symtab.c (dump_symtab_base): Print new flags.
12209         (verify_symtab_base): Verify new flags.
12210         (symtab_alias_ultimate_target): New function.
12211         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
12212         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
12213         Update.
12214         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
12215         Update.
12216         * i386.c (ix86_get_function_versions_dispatcher,
12217         ix86_generate_version_dispatcher_body): Update.
12218         (fold_builtin_cpu): Use varpool_add_new_variable.
12219         * varpool.c (varpool_remove_initializer): Break out from ...
12220         (varpool_remove_node): ... this one.
12221         (dump_varpool_node, varpool_node_for_asm,
12222         cgraph_variable_initializer_availability, varpool_analyze_node,
12223         varpool_assemble_decl, varpool_remove_unreferenced_decls,
12224         varpool_finalize_named_section_flags, varpool_create_variable_alias):
12225         Update.
12227 2013-05-29  Jan Hubicka  <jh@suse.cz>
12229         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
12231 2013-05-29  Easwaran Raman  <eraman@google.com>
12233         PR tree-optimization/57442
12234         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
12235         when control exits the main loop.
12237 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
12239         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
12240         and RX600.
12241         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
12242         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
12243         * rx/t-rx: Add rx100 under multi library matches option for nofpu
12244         option.
12246 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12248         PR tree-optimization/57441
12249         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
12250         Don't limit size of incr_vec to number of candidates.
12252 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
12254         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
12255         and mips16 directories.
12256         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
12257         (MULTILIB_DIRNAMES): Ditto.
12258         (MULTILIB_EXCEPTIONS): Add new exceptions.
12259         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
12260         (MULTILIB_DIRNAMES): Ditto.
12261         (MULTILIB_EXCEPTIONS): Add new exceptions.
12263 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12264             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12266         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
12267         SYMBOL_TINY_ABSOLUTE.
12268         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
12269         SYMBOL_TINY_ABSOLUTE.
12270         (aarch64_expand_mov_immediate): Likewise.
12271         (aarch64_classify_symbol): Likewise.
12272         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
12273         Permit SYMBOL_TINY_ABSOLUTE.
12274         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
12276 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12277             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12279         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
12280         Refactor if/switch.  Replace gcc_assert with if.
12282 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12284         * config/i386/i386.c (initial_ix86_tune_features): Enable
12285         FP Reassociation for AMD bdver1 and bdver2.
12287 2013-05-29  Martin Jambor  <mjambor@suse.cz>
12289         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
12290         and IMAGPART_EXPR do not occur within other handled_components.
12292 2013-05-29  Richard Biener  <rguenther@suse.de>
12294         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
12295         access on whether the use is in the BB we currently try to
12296         vectorize.
12297         (vect_bb_vectorization_profitable_p): Pass the BB we currently
12298         vectorize to vect_bb_slp_scalar_cost.
12300 2013-05-29  Richard Biener  <rguenther@suse.de>
12302         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
12303         computing scalar cost offsetted by stmts that are kept live
12304         by scalar uses.
12305         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
12306         for computation of scalar cost.
12308 2013-05-28  Steve Ellcey  <sellcey@mips.com>
12310         * config/mips/mips-cpus.def (mips32r2): Change processor type.
12312 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
12314         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
12315         array notation built-in reduction functions.
12316         * doc/passes.texi (Passes): Added documentation about changes done
12317         for Cilk Plus.
12318         * doc/invoke.texi (C Dialect Options): Added documentation about
12319         the -fcilkplus flag.
12320         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
12321         (BUILTINS_DEF): Depend on cilkplus.def.
12322         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
12323         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
12324         * cilkplus.def: New file.
12326 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
12328         PR rtl-optimization/57439
12329         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
12331 2013-05-28  Easwaran Raman  <eraman@google.com>
12333         PR tree-optimization/57337
12334         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
12335         (find_insert_point): Correctly identify the insertion point
12336         when two statements with the same UID is compared.
12338 2013-05-28  Richard Biener  <rguenther@suse.de>
12340         PR tree-optimization/56787
12341         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
12342         from the list of data references.
12343         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
12344         clobbers.
12345         (vect_analyze_loop_operations): Likewise.
12346         (vect_transform_loop): Remove clobbers.
12348 2013-05-28  Martin Jambor  <mjambor@suse.cz>
12350         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
12351         and REALPART_EXPRs have scalar type.
12353 2013-05-28  Richard Biener  <rguenther@suse.de>
12355         PR tree-optimization/57411
12356         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
12357         virtual operands.
12358         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
12359         virtual operand propagation.
12361 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
12363         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
12364         destination register for bmasksi_vis.
12365         (vector_init_bshuffle): Likewise.
12366         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
12368 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
12370         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
12371         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
12372         mode if the instruction isn't available in the original mode.
12373         * config/sparc/sparc.opt (mfix-ut699): New option.
12374         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
12375         (divdf3): Turn into expander.
12376         (divdf3_nofix): New insn.
12377         (divdf3_fix): Likewise.
12378         (divsf3): Disable if -mfix-ut699.
12379         (sqrtdf2): Turn into expander.
12380         (sqrtdf2_nofix): New insn.
12381         (sqrtdf2_fix): Likewise.
12382         (sqrtsf2): Disable if -mfix-ut699.
12384 2013-05-27  Richard Biener  <rguenther@suse.de>
12386         PR middle-end/57412
12387         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
12388         block for the new loop.
12390 2013-05-27  Richard Biener  <rguenther@suse.de>
12392         PR tree-optimization/57343
12393         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
12394         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
12395         (number_of_iterations_cond): Do not build the folded tree.
12397 2013-05-27  Richard Biener  <rguenther@suse.de>
12399         Revert
12400         PR middle-end/57381
12401         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
12402         OEP_CONSTANT_ADDRESS_OF retained.
12404         PR tree-optimization/57417
12405         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
12406         for unchanged base.
12407         (set_ssa_val_to): Compare addresses using
12408         get_addr_base_and_unit_offset.
12410 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
12412         PR rtl-optimization/56833
12413         * postreload.c (move2add_record_mode): New function.
12414         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
12415         (move2add_use_add2_insn): Use move2add_record_sym_value.
12416         (move2add_use_add3_insn): Likewise.
12417         (reload_cse_move2add): Use move2add_valid_value_p and
12418         move2add_record_mode.  Invalidate call-clobbered and REG_INC
12419         affected regs by setting reg_mode to VOIDmode.
12420         (move2add_note_store): Don't pretend the inside of a SUBREG is
12421         the actual destination.  Invalidate single/leading registers by
12422         setting reg_mode to VOIDmode.
12423         Use move2add_record_sym_value, move2add_valid_value_p and
12424         move2add_record_mode.
12426 2013-05-27  Richard Biener  <rguenther@suse.de>
12428         PR tree-optimization/57396
12429         * tree-affine.c (double_int_constant_multiple_p): Properly
12430         return false for val == 0 and div != 0.
12432 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
12434         * config/mips/mips.h: Use #elif in preprocessor conditions.
12436 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
12438         PR target/53916
12439         * config/mips/constraints.md (kl): New constraint.
12440         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
12441         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
12442         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
12443         from LO for MIPS16.
12444         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
12446 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
12448         PR target/55777
12449         * config/mips/mips.c (mips_can_inline_p): New function.
12450         (TARGET_CAN_INLINE_P): Define.
12452 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
12454         * sched-int.h (ds_t, dw_t): Make unsigned int.
12455         Fix documentation that describes how all the ds_t bits are used.
12456         Reserve the last bit for delayed-branch scheduling.
12457         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
12458         (BITS_PER_DEP_WEAK): Fix definition and documentation.
12459         (gen_dep_weak_1): Remove prototype.
12460         * sched-deps.c (get_dep_weak_1): Make static.
12461         * target.def (speculate_insn, needs_block_p, gen_spec_check,
12462         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
12463         * doc/tm.texi: Regenerate.
12464         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
12466 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
12468         PR debug/56950
12469         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
12471 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
12472             Sandra Loosemore  <sandra@codesourcery.com>
12474         * config.gcc (powerpc-*): Allow native for with-cpu.
12476 2013-05-24  Jeff Law  <law@redhat.com>
12478         PR tree-optimization/57124
12479         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
12480         conversion feeding a condition if the range has an overflow
12481         if -fstrict-overflow.  Add warnings for when we do make the
12482         transformation.
12484 2013-05-24  Dehao Chen  <dehao@google.com>
12486         * tree-cfg.c (locus_discrim_map): Fix the typo.
12487         (locus_discrim_hasher): Likewise.
12488         (locus_discrim_hasher::hash): Likewise.
12489         (locus_discrim_hasher::equal): Likewise.
12491 2013-05-24  Martin Jambor  <mjambor@suse.cz>
12493         PR tree-optimization/57294
12494         * cgraph.h (ipa_record_stmt_references): Declare.
12495         * cgraphbuild.c (ipa_record_stmt_references): New function.
12496         (build_cgraph_edges): Use ipa_record_stmt_references.
12497         (rebuild_cgraph_edges): Likewise.
12498         (cgraph_rebuild_references): Likewise.
12499         * ipa-prop.c (ipa_modify_call_arguments): Discard references
12500         associated with the old statement and build references from the
12501         newly built statements.
12502         * ipa-ref.c (ipa_remove_stmt_references): New function.
12503         * ipa-ref.h (ipa_remove_stmt_references): Declare.
12505 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
12507         * lra-constraints.c (emit_spill_move): Use smaller mode for
12508         mem-mem moves.
12509         (check_and_process_move): Consider mem-reg moves for secondary
12510         too.
12511         (curr_insn_transform): Don't lose insns emitted before for
12512         secondary memory moves.
12513         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
12514         reg set up in the current insn.
12516 2013-05-24  Dehao Chen  <dehao@google.com>
12518         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
12519         hash function.
12520         (locus_descrim_hasher::equal): Likewise.
12521         (build_gimple_cfg): New discriminator assignment algorithm.
12522         (make_edges): Likewise.
12523         (next_discriminator_for_locus): Likewise.
12524         (same_line_p): Likewise.
12525         (assign_discriminators): Likewise.
12526         (make_cond_expr_edges): Likewise.
12527         (make_gimple_switch_edges): Likewise.
12528         (make_goto_expr_edges): Likewise.
12529         (make_gimple_asm_edges): Likewise.
12531 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
12533         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
12534         X format specifier to only display bottom 16 bits.
12535         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
12536         immediate to match for operand 2, since it will be masked.
12538 2013-05-24  Richard Biener  <rguenther@suse.de>
12540         PR tree-optimization/57287
12541         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
12542         all SSA names that occur in abnormal PHIs.
12544 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12546         PR tree-ssa/57385
12547         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
12548         that index is not negative.
12550 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
12552         PR rtl-optimization/55177
12553         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
12554         (simplify_byte_swapping_operation): New.
12555         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
12556         (simplify_relational_operation_1): Deal with BSWAP.
12558 2013-05-23  Richard Henderson  <rth@redhat.com>
12560         PR target/56742
12561         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
12562         (ix86_reorg): Call it.
12564 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
12566         PR target/57379
12567         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
12568         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
12569         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
12571 2013-05-23  Christian Bruel  <christian.bruel@st.com>
12573         PR debug/57351
12574         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
12576 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12577             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12579         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
12580         * config/aarch64/constraints.md (Usa): Remove.
12581         * doc/md.texi (AArch64 Usa): Remove.
12583 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12584             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12586         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
12587         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
12588         * config/aarch64/predicates.md (aarch64_const_address): Remove.
12589         (aarch64_mov_operand): Use aarch64_mov_operand_p.
12591 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
12593         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
12594         instruction (AdvSIMD).
12595         * config/aarch64/aarch64-builtins.c
12596         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
12597         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
12599 2013-05-23  Martin Jambor  <mjambor@suse.cz>
12601         PR middle-end/57347
12602         * tree.h (contains_bitfld_component_ref_p): Declare.
12603         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
12604         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
12605         caller.
12606         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
12607         not access a bit-field.  Assert all final offsets are byte-aligned.
12609 2013-05-23  Richard Biener  <rguenther@suse.de>
12611         PR tree-optimization/57380
12612         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
12613         least one invariant or re-used load.
12614         * passes.c (init_optimization_passes): Move pass_phiprop before
12615         pass_forwprop.
12617 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
12619         * config/aarch64/aarch64-simd.md
12620         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
12622 2013-05-23  Richard Biener  <rguenther@suse.de>
12624         PR middle-end/57381
12625         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
12626         OEP_CONSTANT_ADDRESS_OF retained.
12628 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
12630         PR middle-end/57344
12631         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
12632         don't lower unit.  Handle unit not being always BITS_PER_WORD.
12634 2013-05-23  Richard Biener  <rguenther@suse.de>
12636         PR rtl-optimization/57341
12637         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
12638         instead of true_dependence.
12640 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
12642         * bb-reorder.c (branch_threshold): Make const.
12643         (exec_threshold): Ditto.
12645 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
12646             Pat Haugen  <pthaugen@us.ibm.com>
12647             Peter Bergner  <bergner@vnet.ibm.com>
12649         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
12650         documentation for the power8 crypto builtins.
12652         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
12654         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
12655         macros for defining power8 builtin functions.
12656         (BU_P8V_AV_2): Likewise.
12657         (BU_P8V_AV_P): Likewise.
12658         (BU_P8V_VSX_1): Likewise.
12659         (BU_P8V_OVERLOAD_1): Likewise.
12660         (BU_P8V_OVERLOAD_2): Likewise.
12661         (BU_CRYPTO_1): Likewise.
12662         (BU_CRYPTO_2): Likewise.
12663         (BU_CRYPTO_3): Likewise.
12664         (BU_CRYPTO_OVERLOAD_1): Likewise.
12665         (BU_CRYPTO_OVERLOAD_2): Likewise.
12666         (XSCVSPDP): Fix typo, point to the correct instruction.
12667         (VCIPHER): Add power8 crypto builtins.
12668         (VCIPHERLAST): Likewise.
12669         (VNCIPHER): Likewise.
12670         (VNCIPHERLAST): Likewise.
12671         (VPMSUMB): Likewise.
12672         (VPMSUMH): Likewise.
12673         (VPMSUMW): Likewise.
12674         (VPERMXOR_V2DI): Likewise.
12675         (VPERMXOR_V4SI: Likewise.
12676         (VPERMXOR_V8HI: Likewise.
12677         (VPERMXOR_V16QI: Likewise.
12678         (VSHASIGMAW): Likewise.
12679         (VSHASIGMAD): Likewise.
12680         (VPMSUM): Likewise.
12681         (VPERMXOR): Likewise.
12682         (VSHASIGMA): Likewise.
12684         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12685         __CRYPTO__ if the crypto instructions are available.
12686         (altivec_overloaded_builtins): Add support for overloaded power8
12687         builtins.
12689         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
12690         support for power8 crypto builtins.
12691         (builtin_function_type): Likewise.
12692         (altivec_init_builtins): Add support for builtins that take vector
12693         long long (V2DI) arguments.
12695         * config/rs6000/crypto.md: New file, define power8 crypto
12696         instructions.
12698 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
12699             Pat Haugen  <pthaugen@us.ibm.com>
12700             Peter Bergner  <bergner@vnet.ibm.com>
12702         * doc/invoke.texi (Option Summary): Add power8 options.
12703         (RS/6000 and PowerPC Options): Likewise.
12705         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
12706         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
12707         wm, wn, wr documentation.
12709         * config/rs6000/constraints.md (wm): New constraint for VSX
12710         registers if direct move instructions are enabled.
12711         (wn): New constraint for no registers.
12712         (wq): New constraint for quad word even GPR registers.
12713         (wr): New constraint if 64-bit instructions are enabled.
12714         (wv): New constraint if power8 vector instructions are enabled.
12715         (wQ): New constraint for quad word memory locations.
12717         * config/rs6000/predicates.md (const_0_to_15_operand): New
12718         constraint for 0..15 for crypto instructions.
12719         (gpc_reg_operand): If VSX allow registers in VSX registers as well
12720         as GPR and floating point registers.
12721         (int_reg_operand): New predicate to match only GPR registers.
12722         (base_reg_operand): New predicate to match base registers.
12723         (quad_int_reg_operand): New predicate to match even GPR registers
12724         for quad memory operations.
12725         (vsx_reg_or_cint_operand): New predicate to allow vector logical
12726         operations in both GPR and VSX registers.
12727         (quad_memory_operand): New predicate for quad memory operations.
12728         (reg_or_indexed_operand): New predicate for direct move support.
12730         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
12731         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
12732         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
12733         (POWERPC_MASKS): Add power8 options.
12734         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
12735         various options.
12737         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
12738         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
12740         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
12741         (-mpower8-fusion): New power8 options.
12742         (-mpower8-fusion-sign): Likewise.
12743         (-mpower8-vector): Likewise.
12744         (-mcrypto): Likewise.
12745         (-mdirect-move): Likewise.
12746         (-mquad-memory): Likewise.
12748         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
12749         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
12750         registers.
12751         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
12752         (rs6000_debug_vector_unit): Add p8_vector.
12753         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
12754         definitions.  Also print fusion state.
12755         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
12756         (rs6000_builtin_mask_calculate): Add power8 builtin support.
12757         (rs6000_option_override_internal): Add support for power8.
12758         (rs6000_common_init_builtins): Add debugging for skipped builtins
12759         if -mdebug=builtin.
12760         (rs6000_adjust_cost): Add power8 support.
12761         (rs6000_issue_rate): Likewise.
12762         (insn_must_be_first_in_group): Likewise.
12763         (insn_must_be_last_in_group): Likewise.
12764         (force_new_group): Likewise.
12765         (rs6000_register_move_cost): Likewise.
12766         (rs6000_opt_masks): Likewise.
12768         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
12769         power8 capable assembler, default to power7 options.
12770         (TARGET_DIRECT_MOVE): Likewise.
12771         (TARGET_CRYPTO): Likewise.
12772         (TARGET_P8_VECTOR): Likewise.
12773         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
12774         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
12775         (VECTOR_MEM_P8_VECTOR_P): Likewise.
12776         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
12777         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
12778         (TARGET_XSCVDPSPN): Likewise.
12779         (TARGET_XSCVSPDPN): Likewsie.
12780         (TARGET_SYNC_HI_QI): Likewise.
12781         (TARGET_SYNC_TI): Likewise.
12782         (MASK_CRYPTO): Likewise.
12783         (MASK_DIRECT_MOVE): Likewise.
12784         (MASK_P8_FUSION): Likewise.
12785         (MASK_P8_VECTOR): Likewise.
12786         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
12787         temporary used by some of the direct move instructions to get two FP
12788         temporary registers does not force creation of a stack frame.
12789         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
12790         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
12791         that any VSX registers are tieable, even if they are also an
12792         Altivec vector mode.
12793         (r6000_reg_class_enum): Add wm, wr, wv constraints.
12794         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
12795         (RS6000_BTM_CRYPTO): Likewise.
12796         (RS6000_BTM_COMMON): Likewise.
12798         * config/rs6000/rs6000.md (cpu attribute): Add power8.
12799         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
12800         (enum rs6000_vector): Add power8 vector support.
12802 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12804         PR target/19599
12805         PR target/57340
12806         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
12807         (any_sibcall_could_use_r3): this and handle indirect calls.
12808         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
12810 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12812         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
12814 2013-05-22  Richard Biener  <rguenther@suse.de>
12816         PR middle-end/57349
12817         * profile.c (branch_prob): Do not split blocks that are
12818         abnormally receiving from ECF_RETURNS_TWICE functions.
12820 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
12822         * recog.c (offsettable_address_addr_space_p): Fix calculation of
12823         address mode.  Move pointer mode initialization to the same place.
12825 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
12827         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
12828         while it has any effect.
12830 2013-05-21  Easwaran Raman  <eraman@google.com>
12832         PR tree-optimization/57322
12833         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
12834         UID of the statement added to the BB to be 1.
12836 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
12838         PR tree-optimization/57331
12839         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
12840         of conversion from pointer type to integral type with integer.
12842 2013-05-21  Martin Jambor  <mjambor@suse.cz>
12844         PR lto/57289
12845         * ipa-prop.c (ipa_read_node_info): Process param_used and
12846         controlled_uses in the same order as when writing.
12848 2013-05-21  Magnus Granberg  <baldrick@free.fr>
12850         PR plugins/56754
12851         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
12853 2013-05-21  Richard Biener  <rguenther@suse.de>
12855         PR tree-optimization/57318
12856         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
12857         estimate stmts with side-effects as likely eliminated.
12859 2013-05-21  Richard Biener  <rguenther@suse.de>
12861         PR tree-optimization/57330
12862         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
12863         preserve the call stmts fntype.
12865 2013-05-21  Richard Biener  <rguenther@suse.de>
12867         PR tree-optimization/57303
12868         * tree-ssa-sink.c (statement_sink_location): Improve killing
12869         stmt detection and properly handle self-assignments.
12871 2013-05-21  Christian Bruel  <christian.bruel@st.com>
12873         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
12874         spanning registers. LEAF_REG_REMAP is supported only for contiguous
12875         registers. Set register size out of the PARALLEL loop.
12877 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
12879         PR target/56547
12880         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
12881         (*fmasf4, *fmasf4_media): New insns.
12883 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12885         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
12886         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
12887         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
12888         (mips_idiv_insns): Update the comments to say that the returned
12889         instruction counts are in units of BASE_INSN_LENGTH.
12890         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
12891         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
12892         using 2 rather than 4 as the length of indirect MIPS16 and
12893         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
12894         length of a NOP.  Don't divide MIPS16 lengths by 2.
12895         (mips16_split_long_branches): Assume a branch is long if the
12896         length is greater than 4 rather than 8.
12897         * config/mips/mips.md (length): Give MIPS16 lengths directly,
12898         rather than multiplying them by 2.  Multiply instruction counts
12899         by BASE_INSN_LENGTH rather than 4.
12900         (*jump_mips16, tls_get_tp_mips16_<mode>)
12901         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
12903 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12905         * config/mips/mips.md (extended_mips16): Remove branch case.
12906         (length): Remove duplicated extended_mips16 test.
12908 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12910         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
12912 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
12914         * recog.h (Recog_data): Rename to...
12915         (recog_data_d): ...this.
12916         (recog_data): Update accordingly.
12917         * recog.c (recog_data): Likewise.
12918         * reload.c (save_recog_data): Likewise.
12919         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
12920         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
12922 2013-05-17  Julian Brown  <julian@codesourcery.com>
12924         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
12925         found in a REG_EQUAL note, invalidate it.
12927 2013-05-17  Easwaran Raman  <eraman@google.com>
12929         * tree-ssa-reassoc.c (find_insert_point): New function.
12930         (insert_stmt_after): Likewise.
12931         (get_def_stmt): Likewise.
12932         (ensure_ops_are_available): Likewise.
12933         (not_dominated_by): Likewise.
12934         (rewrite_expr_tree): Do not move statements beyond what is
12935         necessary. Remove call to swap_ops_for_binary_stmt...
12936         (reassociate_bb): ... and move it here.
12937         (build_and_add_sum): Assign UIDs for new statements.
12938         (linearize_expr): Likewise.
12939         (do_reassoc): Renumber gimple statement UIDs.
12941 2013-05-17  Jan Hubicka  <jh@suse.cz>
12943         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
12944         weakrefs.
12945         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
12946         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
12947         weakrefs.
12948         (output_weakrefs): Update.
12950 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
12951             Martin Jambor  <mjambor@suse.cz>
12953         PR middle-end/57276
12954         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
12955         value that corresponds to the given aggval is found in values vector.
12957 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
12959         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
12960         sse, sse2, sse3, ssse3 and sse4a flags to options.
12962 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
12964         * gengtype-state.c: (s_expr_writer): New class, to handle
12965         prettifying of output layout of s-expressions.
12966         (state_writer): New class, to write out gtype.state.
12967         (state_written_type_count): Move this variable into member data of
12968         state_writer.
12969         (s_expr_writer::s_expr_writer): New code: constructor for new class
12970         (state_writer::state_writer(): ditto
12971         (s_expr_writer::write_new_line): New function
12972         (s_expr_writer::write_any_indent): ditto
12973         (s_expr_writer::begin_s_expr): ditto
12974         (s_expr_writer::end_s_expr): ditto
12975         (write_state_fileloc): convert to method of state_writer...
12976         (state_writer:: write_state_fileloc): ...and use methods of
12977         s_expr_writer to write indentation into the gtype.state output file
12978         to visually represent the hierarchical structure of the list
12979         structures
12980         (write_state_fields): ditto, renaming to...
12981         (state_writer::write_state_fields)
12982         (write_state_a_string): ditto, renaming to...
12983         (state_writer::write_state_a_string)
12984         (write_state_string_option): ditto, renaming to...
12985         (state_writer::write_state_string_option)
12986         (write_state_type_option): ditto, renaming to...
12987         (state_writer::write_state_type_option)
12988         (write_state_nested_option): ditto, renaming to...
12989         (state_writer::write_state_nested_option)
12990         (write_state_option): ditto, renaming to...
12991         (state_writer::write_state_option)
12992         (write_state_options): ditto, renaming to...
12993         (state_writer::write_state_options)
12994         (write_state_lang_bitmap): ditto, renaming to...
12995         (state_writer::write_state_lang_bitmap)
12996         (write_state_version): ditto, renaming to...
12997         (state_writer::write_state_version)
12998         (write_state_scalar_type): ditto, renaming to...
12999         (state_writer::write_state_scalar_type)
13000         (write_state_string_type): ditto, renaming to...
13001         (state_writer::write_state_string_type)
13002         (write_state_undefined_type): ditto, renaming to...
13003         (state_writer::write_state_undefined_type)
13004         (write_state_struct_union_type): ditto, renaming to...
13005         (state_writer::write_state_struct_union_type)
13006         (write_state_struct_type): ditto, renaming to...
13007         (state_writer::write_state_struct_type)
13008         (write_state_user_struct_type): ditto, renaming to...
13009         (state_writer::write_state_user_struct_type)
13010         (write_state_lang_struct_type): ditto, renaming to...
13011         (state_writer::write_state_lang_struct_type)
13012         (write_state_param_struct_type): ditto, renaming to...
13013         (state_writer::write_state_param_struct_type)
13014         (write_state_pointer_type): ditto, renaming to...
13015         (state_writer::write_state_pointer_type)
13016         (write_state_array_type): ditto, renaming to...
13017         (state_writer::write_state_array_type)
13018         (write_state_gc_used): ditto, renaming to...
13019         (state_writer::write_state_gc_used)
13020         (write_state_common_type_content): ditto, renaming to...
13021         (state_writer::write_state_common_type_content)
13022         (write_state_type): ditto, renaming to...
13023         (state_writer::write_state_type)
13024         (write_state_pair_list): ditto, renaming to...
13025         (state_writer::write_state_pair_list)
13026         (write_state_pair): ditto, renaming to...
13027         (state_writer::write_state_pair)
13028         (write_state_typedefs): ditto, renaming to...
13029         (state_writer::write_state_typedefs)
13030         (write_state_structures): ditto, renaming to...
13031         (state_writer::write_state_structures)
13032         (write_state_param_structs): ditto, renaming to...
13033         (state_writer::write_state_param_structs)
13034         (write_state_variables): ditto, renaming to...
13035         (state_writer::write_state_variables)
13036         (write_state_srcdir): ditto, renaming to...
13037         (state_writer::write_state_srcdir)
13038         (write_state_files_list): ditto, renaming to...
13039         (state_writer::write_state_files_list)
13040         (write_state_languages): ditto, renaming to...
13041         (state_writer::write_state_languages)
13042         (write_state): create a state_writer instance and use it when
13043         writing out the state file
13045 2013-05-17  Mike Stump  <mikestump@comcast.net>
13047         PR rtl-optimization/57304
13048         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
13049         accessing DF_REF_REAL_LOC.
13051 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
13053         PR rtl-optimization/57281
13054         PR rtl-optimization/57300
13055         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
13056         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
13057         what the other splitter did if the registers are dead.
13059 2013-05-17  Richard Biener  <rguenther@suse.de>
13061         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
13062         MEM_REF offsets.
13064 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
13066         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
13067         linking.
13069 2013-05-17  Marek Polacek  <polacek@redhat.com>
13071         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
13072         length when doing non-zero store of storing '\0' to '\0'.
13074 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
13076         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
13077         vect_external_def oprnd1 with loop_vinfo, try to emit
13078         optional cast, negation and and stmts on the loop preheader
13079         edge instead of into the pattern def seq.
13081         PR tree-optimization/57051
13082         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
13083         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
13085 2013-05-16  Nick Clifton  <nickc@redhat.com>
13087         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
13088         (rl78_is_naked_func): New function.
13089         (rl78_expand_prologue): Skip prologue generation for naked functions.
13090         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
13091         * doc/extend.texi (naked): Add RL78 to the list of processors
13092         that supports this attribute.
13094 2013-05-16  Jeff Law  <law@redhat.com>
13096         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
13098 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
13100         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
13101         cache parameters using detect_caches_amd also for CYRIX,
13102         NSC and TM2 signatures.
13104 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
13105             Dzianis Kahanovich  <mahatma@eu.by>
13107         PR target/45359
13108         PR target/46396
13109         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13110         VIA/Centaur processors and determine their cache parameters
13111         using detect_caches_amd.
13113 2013-05-16  Teresa Johnson  <tejohnson@google.com>
13115         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
13116         (rtl_verify_edges): New function.
13117         (rtl_verify_bb_insns): Ditto.
13118         (rtl_verify_bb_pointers): Ditto.
13119         (rtl_verify_bb_insn_chain): Ditto.
13120         (rtl_verify_fallthru): Ditto.
13121         (rtl_verify_bb_layout): Ditto.
13122         (rtl_verify_flow_info_1): Outline checks into new functions.
13123         (rtl_verify_flow_info): Ditto.
13125 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
13127         * cfghooks.c (copy_bbs): Add update_dominance argument.
13128         * cfghooks.h (copy_bbs): Update prototype.
13129         * tree-cfg.c (gimple_duplicate_sese_region):
13130         Add update_dominance argument.
13131         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
13132         * tree-ssa-loop-ch.c (copy_loop_headers): Update
13133         gimple_duplicate_sese_region call.
13134         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
13135         Update copy_bbs call.
13136         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
13137         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
13139 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
13141         * tree-vectorizer.h (NUM_PATTERNS): Increment.
13142         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
13143         vect_recog_rotate_pattern.
13144         (vect_recog_rotate_pattern): New function.
13146 2013-05-16  Jason Merrill  <jason@redhat.com>
13148         * Makefile.in (LLINKER): New variable.
13149         (mostlyclean): Remove link mutex.
13150         * configure.ac: Handle --enable-link-mutex.
13151         * lock-and-run.sh: New script.
13153 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13155         PR target/19599
13156         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
13157         for NULL decl.
13159 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13161         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
13163 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
13165         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
13166         * config/arm/arm.c (next_consecutive_mem): New function.
13167         (gen_movmem_ldrd_strd): Likewise.
13168         * config/arm/arm.md (movmemqi): Update condition and code.
13169         (unaligned_loaddi, unaligned_storedi): New patterns.
13171 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13173         * config.gcc: Obsolete *-*-solaris2.9*.
13174         * doc/install.texi (Specific, *-*-solaris2*): Document it.
13176 2013-05-16  Richard Biener  <rguenther@suse.de>
13178         * passes.c (init_optimization_passes): Move pass_parallelize_loops
13179         earlier, after GRAPHITE transforms and IV canonicalization.
13181 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
13183         * omp-low.c (extract_omp_for_data): For collapsed loops,
13184         if at least one of the loops is known at compile time to
13185         iterate zero times, set count to 0.
13186         (expand_omp_regimplify_p): New function.
13187         (expand_omp_for_generic): For collapsed loops, if at least
13188         one of the loops isn't known to iterate at least once,
13189         add runtime check with setting count to 0.
13190         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
13191         For unsigned types if it isn't known at compile time that
13192         the loop will iterate at least once, add runtime check to bypass
13193         the whole loop if initial condition isn't true.
13195 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
13197         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
13199 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
13201         PR middle-end/57286
13202         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
13203         transformations to avoid an infinite loop.
13205 2013-05-16  Marek Polacek  <polacek@redhat.com>
13207         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
13209 2013-05-15  Leif Ekblad  <leif@rdos.net>
13211         * config/i386/i386.c (ix86_decompose_address): Use
13212         DEFAULT_TLS_SEG_REG to access TLS segment register.
13213         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
13214         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
13215         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
13217 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
13219         PR target/57260
13220         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
13221         sibling calls to functions that would normally be lazily bound,
13222         unless $gp is call-clobbered.
13224 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
13226         * config/i386/i386.c (ix86_option_override_internal): Update
13227         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
13228         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
13229         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
13230         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
13231         of TARGET_3DNOW.
13232         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
13234 2013-05-15  Andreas Schwab  <schwab@suse.de>
13236         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
13237         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
13238         third operand.
13240 2013-05-15  Teresa Johnson  <tejohnson@google.com>
13242         * loop-unroll.c (report_unroll_peel): Check decision before
13243         emitting unroll/peel message.
13245 2013-05-15  Teresa Johnson  <tejohnson@google.com>
13247         * function.h (has_bb_partition): New rtl_data flag.
13248         (bb_reorder_complete): Ditto.
13249         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
13250         instead of flag_reorder_blocks_and_partition.
13251         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
13252         with some enhancements.
13253         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
13254         * bb-reorder.c (connect_traces): Check for has_bb_partition
13255         instead of flag_reorder_blocks_and_partition.
13256         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
13257         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
13258         verify_hot_cold_block_grouping.
13259         (partition_hot_cold_basic_blocks): Set has_bb_partition.
13261 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13263         PR target/19599
13264         * config/arm/predicates.md (call_insn_operand): New predicate.
13265         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
13266         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
13267         if insn is not a tail call.
13268         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
13269         registers.
13270         * config/arm/arm.h (enum reg_class): New caller save register class.
13271         (REG_CLASS_NAMES): Likewise.
13272         (REG_CLASS_CONTENTS): Likewise.
13273         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
13274         without decls.
13276 2013-05-15  Richard Biener  <rguenther@suse.de>
13278         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
13279         of MSG_OPTIMIZED_LOCATIONS.
13280         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
13281         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
13282         message.
13283         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
13284         of MSG_OPTIMIZED_LOCATIONS.
13285         (execute_vect_slp): Likewise.
13286         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
13287         (vect_create_cond_for_alias_checks): Likewise.
13288         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
13289         (vect_recog_widen_mult_pattern): Likewise.
13290         (vect_recog_widen_sum_pattern): Likewise.
13291         (vect_recog_over_widening_pattern): Likewise.
13292         (vect_recog_widen_shift_pattern): Likewise.
13293         (vect_recog_vector_vector_shift_pattern): Likewise.
13294         (vect_recog_divmod_pattern): Likewise.
13295         (vect_recog_mixed_size_cond_pattern): Likewise.
13296         (vect_recog_bool_pattern): Likewise.
13297         (vect_pattern_recog_1): Likewise.
13299 2013-05-15  Martin Jambor  <mjambor@suse.cz>
13301         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
13302         non-functions to builtin_unreachable.
13303         * ipa-inline-transform.c (inline_call): Do not assert estimates were
13304         correct when new direct edges were discovered.
13306 2013-05-15  Martin Jambor  <mjambor@suse.cz>
13308         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
13309         header, print symbol order instead of node uid, print more information
13310         about indirect edge targets.
13311         (ipa_make_edge_direct_to_target): Print symbol order instead of node
13312         uids.
13313         (ipa_make_edge_direct_to_target): Likewise.
13314         (remove_described_reference): Likewise.
13315         (propagate_controlled_uses): Likewise.
13316         (ipa_print_node_params): Also print symbol order.
13317         (ipcp_transform_function): Print symbol order instead of node uids.
13318         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
13319         (cgraph_get_create_real_symbol_node): Likewise.
13320         * ipa-cp.c (print_lattice): Likewise.
13321         (print_all_lattices): Likewise.
13322         (determine_versionability): Likewise.
13323         (initialize_node_lattices): Likewise.
13324         (estimate_local_effects): Likewise.
13325         (update_profiling_info): Likewise.
13326         (create_specialized_node): Likewise.
13327         (perhaps_add_new_callers): Likewise.
13328         (decide_about_value): Likewise.
13329         (decide_whether_version_node): Likewise.
13330         (identify_dead_nodes): Likewise.
13331         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
13332         (dump_inline_summary): Likewise.
13333         (estimate_node_size_and_time): Likewise.
13334         (inline_analyze_function): Likewise.
13335         * ipa-inline.c (report_inline_failed_reason): Likewise.
13336         (want_early_inline_function_p): Likewise.
13337         (edge_badness): Likewise.
13338         (update_edge_key): Likewise.
13339         (inline_small_functions): Likewise.  Add dumping of order to two other
13340         dumps.
13341         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
13342         instead of node uids.
13343         (propagate_pure_const): Likewise.
13344         (propagate_pure_const): Likewise.
13345         * ipa-utils.c (dump_cgraph_node_set): Likewise.
13346         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
13347         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
13348         of node uids.
13349         * tree-pretty-print.c (dump_function_header): Likewise.
13350         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
13351         Print symbol order instead of node uids.
13353 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13355         * config/s390/s390.c (s390_register_move_cost): Don't impose the
13356         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
13358 2013-05-15  Richard Biener  <rguenther@suse.de>
13360         PR tree-optimization/57275
13361         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
13362         return value for fail to do runtime alias checks for gather loads.
13364 2013-05-15  Jan Hubicka  <jh@suse.cz>
13366         PR lto/57038
13367         PR lto/47375
13368         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
13369         weakrefs are not external.
13370         (lto_symtab_merge_decls): Fix thinko when dealing with
13371         non-lto_symtab decls.
13372         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
13373         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
13374         * varpool.c (dump_varpool_node): Dump more flags.
13376 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
13378         * config/i386/i386.c (processor_alias_table): Add instruction
13379         FSGSBASE for AMD bdver3 architecture.
13381 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
13383         * tree.c (warn_deprecated_use): Print file:line using locus color.
13384         * diagnostic.c (diagnostic_report_current_module): Print file:line
13385         and file:line:column using locus color.
13387 2013-05-14  Mike Stump  <mikestump@comcast.net>
13389         * gdbinit.in: Add __null.
13391 2013-05-14  Mike Stump  <mikestump@comcast.net>
13393         * recog.h: Rename struct recog_data to Recog_data.
13394         * recog.c: Likewise.
13395         * reload.c (can_reload_into): Likewise.
13396         * config/picochip/picochip.c: Likewise.
13398 2013-05-14  Mike Stump  <mikestump@comcast.net>
13400         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
13402 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
13404         * resource.h (struct resources): Remove unch_memory member.
13405         (CLEAR_RESOURCE): Don't clear unch_memory.
13406         * resource.c (mark_referenced_resources): Don't set it.
13407         (mark_set_resources): Likewise.
13408         (mark_target_live_regs): Don't clear it.
13409         (init_resource_info): Likewise.
13410         * reorg.c (resource_conflicts_p): Don't compare it.
13411         (redundant_insn): Don't set it.
13413         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
13414         Remove prototypes.
13415         * emit-rtl.c (next_label): Remove unused function.
13416         (skip_consecutive_labels, link_cc0_insns): Move to ...
13417         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
13418         only place where these functions are used, and make them static.
13420 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
13422         * fold-const.c (fold_negate_expr): Handle vectors.
13423         (fold_truth_not_expr): Make it static.
13424         (fold_invert_truthvalue): New static function.
13425         (invert_truthvalue_loc): Handle vectors. Do not call
13426         fold_truth_not_expr directly.
13427         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
13428         <TRUTH_NOT_EXPR>: Do not cast to boolean.
13429         (fold_comparison): Handle vector constants.
13430         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
13431         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
13432         * tree.h (fold_truth_not_expr): Remove declaration.
13434 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
13436         * config/aarch64/aarch64-simd.md
13437         (aarch64_vcond_internal<mode>): Rename to...
13438         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
13439         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
13440         float modes. Clarify all iterator modes.
13441         (vcond<mode><mode>): Use new name for vcond expanders.
13442         (vcond<v_cmp_result><mode>): Likewise.
13443         (vcondu<mode><mode>: Likewise.
13444         * config/aarch64/iterators.md (VDQF_COND): New.
13446 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
13448         PR bootstrap/57266
13449         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
13450         variable for the shift amount. Check that we shift by non-negative
13451         amounts.
13453 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
13455         PR target/42017
13456         * config/arm/arm.h (EPILOGUE_USES): Only return true
13457         for LR_REGNUM after epilogue_completed.
13459 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
13461         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
13462         is error_mark_node.
13464 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13466         PR target/57261
13467         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
13468         and Solaris 11+/x86 with gld.
13469         * configure: Regenerate.
13471 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
13473         * expmed.c (expand_shift_1): Canonicalize rotates by
13474         constant bitsize / 2 to bitsize - 1.
13475         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
13476         case ROTATERT>: Likewise.
13478         Revert:
13479         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
13481         * config/i386/i386.md (rotateinv): New code attr.
13482         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
13483         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
13484         roll $31, %eax, etc.
13486 2013-05-14  Richard Biener  <rguenther@suse.de>
13488         PR middle-end/57235
13489         * tree-eh.c (sink_clobbers): Give up for successors with
13490         multiple predecessors and no virtual uses.
13492 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
13494         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
13495         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
13497 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
13499         PR middle-end/57251
13500         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
13501         the case when both op0 and op1 have VOIDmode.
13503 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
13505         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
13506         in multiply-accumulate mode.
13508 2013-05-13  Guozhi Wei  <carrot@google.com>
13510         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
13512 2013-05-13  Kai Tietz  <ktietz@redhat.com>
13514         PR target/56975
13515         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
13516         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
13517         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
13518         * config/i386/i386.c (ix86_option_override_internal): Likewise.
13519         (ix86_expand_prologue): Likewise.
13520         (ix86_expand_split_stack_prologue): Likewise.
13521         (legitimate_pic_address_disp_p): Likewise.
13522         (legitimize_pic_address): Likewise.
13523         (legitimize_tls_address): Likewise.
13524         (legitimize_pe_coff_symbol): Likewise.
13525         (output_pic_addr_const): Likewise.
13526         (construct_plt_address): Likewise.
13527         (ix86_expand_call): Likewise.
13528         (x86_output_mi_thunk): Likewise.
13529         (x86_function_profiler): Likewise.
13531 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
13533         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
13534         similar switch cases.
13535         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
13536         (aarch64_simd_mov_to_<mode>low): Delete.
13537         (aarch64_simd_mov_to_<mode>high): Delete.
13538         (move_lo_quad_<mode>): Add w<-r alternative.
13539         (aarch64_simd_move_hi_quad_<mode>): Likewise.
13540         (aarch64_simd_mov_from_*): Update type attribute.
13541         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
13542         statement.
13544 2013-05-13  Jan Hubicka  <jh@suse.cz>
13546         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
13547         * config/i386/i386.c (ix86_compute_frame_layout,
13548         ix86_expand_epilogue, emit_i387_cw_initialization,
13549         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
13550         ix86_local_alignment): Fix use of size/speed predicates.
13552 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
13554         PR tree-optimization/45216
13555         PR tree-optimization/57157
13556         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
13557         the (-Y) & (B - 1) variant if OP is |.
13558         * expmed.c (expand_shift_1): For rotations by const0_rtx just
13559         return shifted.  Use (-op1) & (prec - 1) as other_amount
13560         instead of prec - op1.
13562 2013-05-13  Martin Jambor  <mjambor@suse.cz>
13564         PR middle-end/42371
13565         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
13566         (ipa_constant_data): New type.
13567         (ipa_jump_func): Use ipa_constant_data to hold information about
13568         constant jump functions.
13569         (ipa_get_jf_constant): Adjust to jump function type changes.
13570         (ipa_get_jf_constant_rdesc): New function.
13571         (ipa_param_descriptor): New field controlled_uses.
13572         (ipa_get_controlled_uses): New function.
13573         (ipa_set_controlled_uses): Likewise.
13574         * ipa-ref.h (ipa_find_reference): Declare.
13575         * ipa-prop.c (ipa_cst_ref_desc): New type.
13576         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
13577         changes.
13578         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
13579         New parameter cs.  Adjust all callers.
13580         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
13581         (remove_described_reference): New function.
13582         (jfunc_rdesc_usable): Likewise.
13583         (try_make_edge_direct_simple_call): Decrement controlled use count,
13584         attempt to remove reference if it hits zero.
13585         (combine_controlled_uses_counters): New function.
13586         (propagate_controlled_uses): Likewise.
13587         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
13588         (ipa_edge_duplication_hook): Duplicate reference descriptions.
13589         (ipa_print_node_params): Print described use counter.
13590         (ipa_write_jump_function): Adjust to jump function type changes.
13591         (ipa_read_jump_function): New parameter CS, pass it to
13592         ipa_set_jf_constant.  Adjust caller.
13593         (ipa_write_node_info): Stream controlled use count
13594         (ipa_read_node_info): Likewise.
13595         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
13596         asserting.
13597         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
13598         count.  Remove cloning-added reference if it reaches zero.
13599         * ipa-ref.c (ipa_find_reference): New function.
13601 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
13603         * config/i386/i386.c (processor_target_table): Modified default
13604         alignment values for AMD BD and BT architectures.
13606 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
13608         * tree-vect-generic.c (uniform_vector_p): Move ...
13609         * tree.c (uniform_vector_p): ... here.
13610         * tree.h (uniform_vector_p): Declare it.
13611         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
13612         into a scalar.
13614 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
13616         PR tree-optimization/57230
13617         * tree-ssa-strlen.c (handle_char_store): Record length for
13618         array store from STRING_CST.
13620         PR tree-optimization/57230
13621         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
13622         check.
13624 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
13626         * config/epiphany/epiphany.c (epiphany_init): Check size of
13627         NUM_MODES_FOR_MODE_SWITCHING.
13628         (epiphany_expand_prologue):
13629         Remove CONFIG_REGNUM initial value handling code.
13630         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
13631         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
13632         (emit_set_fp_mode, epiphany_mode_after): Likewise.
13633         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
13634         Don't return 1 for FP_MODE_NONE.
13635         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
13636         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
13637         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
13638         * config/epiphany/epiphany.md (save_config): New pattern.
13640 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
13642         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
13644 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
13646         * config/i386/i386.md (memory): Handle sseishft1.
13647         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
13648         (*vec_extractv2di_1): Ditto.
13650 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
13652         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
13653         saved registers.
13655 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13657         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
13658         Add mthumb/march=armv7-a multilib.
13659         Add mthumb/march=armv7-r multilib.
13660         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
13662 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
13664         * config/v850/t-rtems: Add more multilibs.
13666 2013-05-10  Richard Biener  <rguenther@suse.de>
13668         PR tree-optimization/57214
13669         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
13670         not propagate from SSA names that occur in abnormal PHI nodes.
13672 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
13674         * stor-layout.c (element_precision): New function.
13675         * machmode.h (element_precision): Declare it.
13676         * tree.c (build_minus_one_cst): New function.
13677         (element_precision): Likewise.
13678         * tree.h (build_minus_one_cst): Declare new function.
13679         (element_precision): Likewise.
13680         * fold-const.c (operand_equal_p): Use element_precision.
13681         (fold_binary_loc): Handle vector types.
13682         * convert.c (convert_to_integer): Use element_precision.
13683         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
13684         separately.
13686 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
13688         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
13689         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
13690         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
13691         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
13692         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
13693         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
13694         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
13695         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
13696         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
13697         (Uuw8): New constraints.
13698         (Usb4): Move into alphabetical order.
13699         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
13700         (sd8_operand, ub8_operand, uw8_operand): New predicates.
13701         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
13702         previously unnamed patterns.
13703         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
13704         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
13705         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
13706         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
13707         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
13708         of specific lengths.
13710 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
13712         * config/i386/i386.md (rotateinv): New code attr.
13713         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
13714         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
13715         roll $31, %eax, etc.
13717         PR tree-optimization/45216
13718         PR tree-optimization/57157
13719         * tree-ssa-forwprop.c (simplify_rotate): New function.
13720         (ssa_forward_propagate_and_combine): Call it.
13722 2013-05-10  Richard Biener  <rguenther@suse.de>
13724         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
13725         disable peeling when we version for aliasing.
13726         (vector_alignment_reachable_p): Honor explicit user alignment.
13727         (vect_supportable_dr_alignment): Likewise.
13728         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
13729         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
13730         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
13731         then peeling to arrange for the cost-model check to come first.
13733 2013-05-10  Alan Modra  <amodra@gmail.com>
13735         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
13736         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
13737         * configure: Regenerate.
13739 2013-05-10  Alan Modra  <amodra@gmail.com>
13741         PR target/55033
13742         * varasm.c (default_elf_select_section): Move !DECL_P check..
13743         (get_named_section): ..to here before calling get_section_name.
13744         Adjust assertion.
13745         (default_section_type_flags): Add DECL_P check.
13746         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
13747         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
13749 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
13751         * config/epiphany/epiphany.c (epiphany_expand_prologue):
13752         When using gen_stack_adjust_str with a register offset, add a
13753         REG_FRAME_RELATED_EXPR note.
13755 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
13757         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
13758         (*vec_extractv4si_zext_mem): Ditto.
13759         (*vec_extractv2di): Add 0->x and x->x alternatives.
13760         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
13761         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
13763 2013-05-09  Jason Merrill  <jason@redhat.com>
13765         N3639 C++1y VLA support
13766         * gimplify.c (gimplify_vla_decl): Don't touch an existing
13767         DECL_VALUE_EXPR.
13769         * tree.c (build_constructor_va): New.
13770         * tree.h: Declare it.
13772 2013-05-09  Martin Jambor  <mjambor@suse.cz>
13774         PR lto/57084
13775         * gimple-fold.c (canonicalize_constructor_val): Call
13776         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
13778 2013-05-09  Jan Hubicka  <jh@suse.cz>
13779             Richard Biener  <rguenther@suse.de>
13781         PR lto/54095
13782         * symtab.c (symtab_make_decl_local): Do not add private names.
13784 2013-05-09  Jan Hubicka  <jh@suse.cz>
13786         PR lto/54095
13787         * symtab.c (insert_to_assembler_name_hash): Handle clones.
13788         (unlink_from_assembler_name_hash): Likewise.
13789         (symtab_prevail_in_asm_name_hash, symtab_register_node,
13790         symtab_unregister_node, symtab_initialize_asm_name_hash,
13791         change_decl_assembler_name): Update.
13793 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
13795         * config/aarch64/aarch64.md: New movtf split.
13796         (*movtf_aarch64): Update.
13797         (aarch64_movdi_tilow): Handle TF modes and rename to
13798         aarch64_movdi_<mode>low.
13799         (aarch64_movdi_tihigh): Handle TF modes and rename to
13800         aarch64_movdi_<mode>high
13801         (aarch64_movtihigh_di): Handle TF modes and rename to
13802         aarch64_mov<mode>high_di
13803         (aarch64_movtilow_di): Handle TF modes and rename to
13804         aarch64_mov<mode>low_di
13805         (aarch64_movtilow_tilow): Remove spurious whitespace.
13806         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
13807         splits.
13808         (aarch64_print_operand): Update.
13810 2013-05-09  Alan Modra  <amodra@gmail.com>
13812         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
13813         powerpc64le.
13814         * configure: Regenerate.
13816 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
13818         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
13819         splitter preparation statements.
13820         * config/i386/sse.md (*vec_extract* splitters): Ditto.
13821         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
13822         adjust_address_nv.
13824 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13826         * gimple-ssa-strength-reduction.c (count_candidates): Change
13827         return value to int.
13828         (analyze_candidates_and_replace): Change type of length to int.
13830 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
13832         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
13833         (*vec_extract<mode>): Use VI12_128 mode iterator.
13834         (*vec_extract<mode>_mem): Ditto.
13835         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
13836         attribute.
13838 2013-05-08  Diego Novillo  <dnovillo@google.com>
13840         PR bootstrap/54659
13842         Revert:
13843         2012-08-17  Diego Novillo  <dnovillo@google.com>
13845         PR bootstrap/54281
13846         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
13847         * config.in: Regenerate.
13848         * configure: Regenerate.
13849         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
13851 2013-05-08  Jan Hubicka  <jh@suse.cz>
13853         PR lto/54095
13854         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
13855         * cgraph.h (symtab_node_base): Add unique_name.
13856         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
13857         input_overwrite_node, input_varpool_node): Stream unique_name.
13858         * cgraphclones.c (cgraph_create_virtual_clone,
13859         cgraph_function_versioning): Set unique_name.
13860         * ipa.c (function_and_variable_visibility): Set unique_name.
13862 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13864         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
13865         (alloc_cand_and_find_basis): Restrict conditional candidate
13866         processing to CAND_MULTs.
13868 2013-05-08  Jan Hubicka  <jh@suse.cz>
13870         PR lto/54095
13871         lto-symtab.c (lto_symtab_symbol_p): New function.
13872         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
13873         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
13874         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
13875         Skip static symbols.
13877 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
13879         PR tree-optimization/57200
13880         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
13881         Only call inform if the preceding warning_at returns true.
13883 2013-05-07  Han Shen  <shenhan@google.com>
13885         * cfgexpand.c (record_or_union_type_has_array_p): New function.
13886         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
13887         * common.opt (fstack-protector-strong): New option.
13888         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
13889         * doc/invoke.texi (Optimization Options): Document
13890         "-fstack-protector-strong".
13891         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
13893 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
13895         * config/mips/mips.c (mips_machine_reorg2): Return 0.
13897 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
13899         * ira.c (update_equiv_regs): Add insn having equiv memory even if
13900         it is not lhs of the insn.
13901         (setup_reg_equiv): Remove insn having equiv memory which it is not
13902         lhs of the insn.
13903         * lra-constraints.c (process_address): Try to improve generation
13904         code for address base + disp.
13905         (lra_constraints): Make correct the code for checking insn setting
13906         up backward equivalence.  Remove insn only if it is in the init
13907         insn list.
13908         * lra-eliminations.c (update_reg_eliminate): Change return value.
13909         (lra_eliminate): Use the result.
13911 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
13913         * config/i386/sse.md (ssescalarnummask): New mode attribute.
13914         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
13915         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
13916         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
13917         register target operands.
13918         (*vec_extractv8hi_sse2): New pattern.
13919         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
13920         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
13921         (*vec_extract<mode>_mem): New insn and split pattern.
13923 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
13925         * config/arm/arm.c (arm_asan_shadow_offset): New function.
13926         (TARGET_ASAN_SHADOW_OFFSET): Define.
13927         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
13928         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
13930 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13932         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
13933         (incr_vec_index): Return -1 if increment not found.
13934         (create_add_on_incoming_edge): Assert if increment not found.
13935         (record_increment): Limit number of increments recorded.
13936         (all_phi_incrs_profitable): Return false if an increment not found.
13937         (replace_profitable_candidates): Don't process increments that were
13938         not recorded.
13939         (analyze_candidates_and_replace): Limit size of incr_vec.
13941 2013-05-07  Richard Biener  <rguenther@suse.de>
13943         * calls.c (special_function_p): setjmp-like functions are leaf.
13944         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
13945         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
13947 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
13949         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
13950         (aarch64_simd_mov<mode>): New expander.
13951         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
13952         (aarch64_simd_mov_to_<mode>high): Likewise.
13953         (aarch64_simd_mov_from_<mode>low): Likewise.
13954         (aarch64_simd_mov_from_<mode>high): Likewise.
13955         (aarch64_dup_lane<mode>): Update.
13956         (aarch64_dup_lanedi): New instruction pattern.
13957         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
13958         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
13960 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13962         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
13963         (replace_mult_candidate): Remove unnecessary argument; remove
13964         unnecessary parameter from call to introduce_cast_before_cand.
13965         (replace_unconditional_candidate): Remove unnecessary parameter
13966         from call to replace_mult_candidate.
13967         (replace_conditional_candidate): Likewise.
13968         (insert_initializers): Use make_temp_ssa_name.
13969         (introduce_cast_before_cand): Remove unnecessary argument; use
13970         make_temp_ssa_name.
13971         (replace_one_candidate): Remove unnecessary argument; remove
13972         unnecessary parameter from calls to introduce_cast_before_cand.
13973         (replace_profitable_candidates): Remove unnecessary parameters
13974         from calls to replace_one_candidate.
13976 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13978         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13979         phi def as possibly hiding a basis for a CAND_ADD whose operands
13980         have been commuted in the analysis.
13981         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
13983 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13985         * config/aarch64/aarch64.md
13986         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
13987         shift value between 0-4.
13989 2013-05-07  Richard Biener  <rguenther@suse.de>
13991         * double-int.h (rshift): New overload.
13992         * double-int.c (rshift): New function.
13993         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
13994         (create_reference_ops_from_ref): Remove.
13995         (vn_reference_insert): Use shared ops for constructing the
13996         reference and copy it.
13998 2013-05-07  Richard Biener  <rguenther@suse.de>
14000         PR middle-end/57190
14001         * tree-eh.c (sink_clobbers): Properly propagate
14002         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
14004 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
14006         PR tree-optimization/57149
14007         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
14008         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
14009         collect_phi_def_edges, execute_late_warn_uninitialized): Use
14010         uninit_undefined_value_p instead of ssa_undefined_value_p.
14012         PR debug/57184
14013         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
14014         for modifier == EXPAND_INITIALIZER.
14016 2013-05-07  Anton Blanchard  <anton@samba.org>
14018         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
14019         for powerpc64 little endian.
14020         * configure: Regenerate.
14022 2013-05-06  Graham Stott  <grahams@btinternet.com>
14024         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
14025         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
14026         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
14027         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
14029 2013-05-06  Graham Stott  <grahams@btinternet.com>
14031         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
14032         codes which allow non-lvalues.
14034 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
14036         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
14037         components are all 1s.
14038         (integer_minus_onep): New function.
14039         * tree.h (integer_minus_onep): Declare it.
14040         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
14041         integer_minus_onep instead of integer_all_onesp.
14043 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
14045         PR target/52933
14046         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
14047         variations of these patterns.
14049 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
14051         * config/i386/i386.md (isa): Add x64_sse4 member.
14052         (enabled): Handle x64_sse4.
14053         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
14054         instruction for 64bit SSE4_1 targets.  Update insn attributes.
14055         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
14056         instruction for SSE4_1 targets.  Update insn attributes.
14057         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
14058         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
14059         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
14060         const_1 selector.
14061         (*vec_extractv4si): Rename from *sse4_1_pextrd.
14062         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
14063         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
14065 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
14067         PR target/57108
14068         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
14070 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
14072         * final.c (do_assembler_dialects): Don't handle curly braces and
14073         vertical bar escaped by % as dialect delimiters.
14074         (output_asm_insn): Print curly braces and vertical bar if escaped
14075         by % and ASSEMBLER_DIALECT defined.
14076         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
14077         * doc/tm.texi: Regenerated.
14079 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
14081         * config/mips/mips.c: Include tree-pass.h.
14082         (mips_reorg): Split in pre- and post-dbr_schedule parts.
14083         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
14084         (pass_mips_machine_reorg2): New machine specific pass.
14085         (insert_pass_mips_machine_reorg2): New pass plugin definition.
14086         (mips_option_override): Register the new pass.
14087         * rtl.h (cleanup_barriers): Remove prototype.
14088         (dbr_schedule): Likewise.
14089         * jump.c (cleanup_barriers): Make static.
14090         * reorg.c (dbr_schedule): Likewise.
14092 2013-05-06  Richard Biener  <rguenther@suse.de>
14094         PR tree-optimization/57185
14095         * tree-parloops.c (add_field_for_reduction): Handle anonymous
14096         SSA names properly.
14098 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
14100         PR target/57106
14101         * config/i386/i386.c (add_parameter_dependencies): Add dependence
14102         between "first_arg" and "insn", not "last" and "insn".
14104 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14106         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
14107         (find_candidates_in_block): Re-enable slsr_process_phi.
14108         (create_phi_basis): Fix double counting of candidate adjustment.
14110 2013-05-06  Richard Biener  <rguenther@suse.de>
14112         PR middle-end/57147
14113         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
14114         the edge is also fallthru, preserve it and just clear the
14115         abnormal flag.
14116         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
14117         also complex, preserve that and just clear the fallthru flag.
14118         * tree-inline.c (update_ssa_across_abnormal_edges): Also
14119         update virtual operands.
14121 2013-05-06  Alan Modra  <amodra@gmail.com>
14123         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
14124         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
14125         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
14126         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
14127         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
14128         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
14130 2013-05-06  Alan Modra  <amodra@gmail.com>
14132         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
14133         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
14134         (DEFAULT_ASM_ENDIAN): Define.
14135         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
14136         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
14137         Update -K PIC clause from sysv4.h.
14138         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
14139         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
14141 2013-05-06  Alan Modra  <amodra@gmail.com>
14143         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
14144         twice for little-endian.
14145         (ashrdi3_no_power, ashrdi3): Support little-endian.
14147 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
14149         PR target/55303
14150         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
14151         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
14152         related expanders.
14153         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
14154         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
14155         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
14156         New predicates.
14158 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
14159             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14161         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
14162         * config/pa/pa.opt: Make mbig-switch a no-op.
14163         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
14164         (CASE_VECTOR_MODE): Always return SImode.
14165         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
14166         for the !TARGET_BIG_SWITCH case.
14167         * config/pa/pa-linux.h: Likewise.
14168         * config/pa/pa-openbsd.h: Likewise.
14169         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
14170         * config/pa/pa.md (short_jump): Remove define_insn.
14171         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
14172         (casesi0): Remove define_insn.
14173         (type): Remove btable_branch.
14174         (pa_combine_type): Likewise.
14175         (in_nullified_branch_delay): Likewise.
14176         (in_call_delay): Likewise.
14177         (define_delay): Likewise.
14178         (define_insn_reservation "Z3"): Likewise.
14179         (define_insn_reservation "Z4"): Likewise.
14180         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
14181         (pa_adjust_insn_length): Remove adjustment for btable branches.
14182         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
14183         and mno-big-switch
14185 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
14187         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
14188         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
14189         Add m->r,x alternatives.
14190         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
14191         splitters using SWI48x mode iterator.
14192         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
14193         TARGET_64BIT.  Add m->x alternative.
14194         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
14195         Add o->x alternative.  Enable for TARGET_SSE.
14196         (sse_storeq): Remove expander.
14197         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
14198         with memory input operand.
14199         (*vec_extractv2di_1 splitter): New.
14200         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
14201         * config/i386/i386.md (ssevecmodelower): New mode attribute.
14203 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
14205         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
14206         (INT_LOWPART): Delete.
14207         (extract_MB): Adjust.
14208         (extract_ME): Adjust.
14209         (print_operand): Adjust.
14211 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
14213         * config/rs6000/predicates.md (reg_or_add_cint_operand,
14214         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
14215         (reg_or_logical_cint_operand, easy_fp_constant,
14216         logical_const_operand): Delete "CONST_DOUBLE" case.
14217         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
14218         "HOST_BITS_PER_WIDE_INT == 64" test.
14219         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
14220         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
14221         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
14222         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
14223         test.
14224         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
14225         CONST_DOUBLE DImode/VOIDmode case.
14226         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
14227         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
14228         CONST_DOUBLE VOIDmode case.
14229         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
14230         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
14231         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
14232         Delete CONST_DOUBLE case.
14233         (splitters for mov FMOVE64 const_double): Delete
14234         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
14235         "HOST_BITS_PER_WIDE_INT >= 64" test.
14236         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
14237         case.
14238         (mov DI const_double): Delete.
14240 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
14242         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
14243         on op shows all bits zero in mode of a lowpart subreg, return zero.
14245 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
14247         PR target/57150
14248         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
14249         to save TFmode registers and DImode to save TImode registers for
14250         caller save operations.
14251         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
14252         mark being partially clobbered since they only use the first
14253         double word.
14255         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
14256         and TDmode only use the upper 64-bits of each VSX register.
14258 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14260         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
14261         (find_candidates_in_block): Disable slsr_process_phi.
14263 2013-05-03  Guozhi Wei  <carrot@google.com>
14265         * coverage.c (coverage_obj_init): Move the construction of gcov
14266         constructor to ...
14267         (build_init_ctor): ... here.
14269 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14271         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
14272         (slsr_cand_d): Redefine def_phi.
14273         (stride_status, phi_adjust_status, count_phis_status): New enums.
14274         (find_phi_def): New.
14275         (find_basis_for_base_expr): New.
14276         (find_basis_for_candidate): Handle hidden bases.
14277         (alloc_cand_and_find_basis): Handle phi candidates.
14278         (slsr_process_phi): New.
14279         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
14280         (create_mul_imm_cand): Likewise.
14281         (create_add_ssa_cand): Exclude phi base candidates.
14282         (create_add_imm_cand): Likewise.
14283         (slsr_process_cast): Likewise.
14284         (slsr_process_copy): Likewise.
14285         (find_candidates_in_block): Handle phi candidates.
14286         (dump_candidate): Likewise.
14287         (unconditional_cands): Delete.
14288         (unconditional_cands_with_known_stride_p): Delete.
14289         (phi_dependent_cand_p): New.
14290         (cand_increment): Handle phi-dependent candidates.
14291         (replace_dependent): Delete.
14292         (replace_mult_candidate): New.
14293         (replace_unconditional_candidate): New.
14294         (incr_vec_index): Move to avoid forward reference.
14295         (create_add_on_incoming_edge): New.
14296         (create_phi_basis): New.
14297         (replace_dependents): Delete.
14298         (replace_conditional_candidate): New.
14299         (phi_add_costs): New.
14300         (replace_uncond_cands_and_profitable_phis): New.
14301         (record_increment): Handle phi adjustments.
14302         (record_phi_increments): New.
14303         (record_increments): Handle phi adjustments.
14304         (phi_incr_cost): New.
14305         (lowest_cost_path): Handle phis.
14306         (total_savings): Likewise.
14307         (analyze_increments): Likewise.
14308         (ncd_with_phi): New.
14309         (ncd_of_cand_and_phis): New.
14310         (nearest_common_dominator_for_cands): Handle phi increments.
14311         (all_phi_incrs_profitable): New.
14312         (replace_profitable_candidates): Handle phi-dependent candidates.
14313         (analyze_candidates_and_replace): Likewise.
14315 2013-05-03  Teresa Johnson  <tejohnson@google.com>
14317         PR bootstrap/57154
14318         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
14319         do not exceed REG_BR_PROB_BASE.
14321 2013-05-03  Jeff Law  <law@redhat.com>
14323         PR tree-optimization/57144
14324         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
14325         operand of the condition will bit into the new type when eliminating
14326         a cast feeding a condition.
14328 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
14330         PR rtl-optimization/57130
14331         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
14332         of COMPARE as in_code to the recursive call if needed.
14334 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
14336         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
14337         (enabled): Handle new members.
14338         * config/i386/sse.md (*vec_concatv2si): Merge from
14339         *vec_concatv2si_sse2 and vec_concatv2si_sse.
14340         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
14342 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
14344         PR tree-optimization/57027
14345         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
14346         for fnms opportunity, check we got the prerequisite kind
14347         of tree / gimple before using accessor functions.
14349 2013-05-03  Richard Biener  <rguenther@suse.de>
14351         * double-int.h (lshift): New overload without precision
14352         and arith argument.
14353         (operator *=, operator +=, operator -=): Move ...
14354         * double-int.c (operator *=, operator +=, operator -=): ... here
14355         and implement more efficiently.
14356         (mul_double_with_sign): Remove.
14357         (lshift_double): Adjust to take unsinged shift argument, push
14358         dispatching code to callers.
14359         (mul_double_wide_with_sign): Add early out for callers that
14360         are not interested in high parts or overflow.
14361         (lshift): New function.
14362         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
14363         dispatch code here.
14364         (lrotate, rrotate): Use logical shifts.
14365         * expr.c (get_inner_reference): Use lshift.
14366         * fixed-value.c (do_fixed_divide): Likewise.
14367         * tree-dfa.c (get_ref_base_and_extent): Likewise.
14368         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
14369         (indirect_refs_may_alias_p): Likewise.
14370         (stmt_kills_ref_p_1): Likewise.
14372 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
14374         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
14376 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
14378         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
14379         scalar form of FABD instruction.
14381 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
14383         * lra-constraints.c (process_alt_operands): Add checking alt
14384         number to choose the best alternative.
14386 2013-05-02  Richard Biener  <rguenther@suse.de>
14388         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
14389         bitmap and its handling.
14390         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
14392 2013-05-02  Richard Biener  <rguenther@suse.de>
14394         PR middle-end/57140
14395         * tree-inline.c (copy_loops): Properly handle removed loops.
14396         (copy_cfg_body): Mark destination loops for fixup if source
14397         loops needed fixup.
14399 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
14401         PR target/56732
14402         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
14403         generating simple_return for naked functions.
14405 2013-05-02  Martin Jambor  <mjambor@suse.cz>
14407         PR middle-end/56988
14408         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
14409         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
14410         flags match.
14411         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
14412         ipa_agg_replacement_value structures.
14413         (known_aggs_to_agg_replacement_list): Likewise.
14414         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
14415         (read_agg_replacement_chain): Likewise.
14416         (ipcp_transform_function): Also check that by_ref flags match.
14418 2013-05-02  Richard Biener  <rguenther@suse.de>
14420         * graphds.h (struct graph): Add obstack member.
14421         * graphds.c (new_graph): Initialize obstack and allocate
14422         vertices from it.
14423         (add_edge): Allocate edge from the obstack.
14424         (free_graph): Free the obstack instead of all edges and vertices.
14426 2013-05-02  Teresa Johnson  <tejohnson@google.com>
14428         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
14429         divides.
14430         * cfg.c (update_bb_profile_for_threading): Ditto.
14431         * tree-inline.c (copy_bb): Ditto.
14432         (copy_edges_for_bb): Ditto.
14433         (initialize_cfun): Ditto.
14434         (copy_cfg_body): Ditto.
14435         (expand_call_inline): Ditto.
14436         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
14437         (estimate_node_size_and_time): Ditto.
14438         (inline_merge_summary): Ditto.
14439         * cgraphclones.c (cgraph_clone_edge): Ditto.
14440         (cgraph_clone_node): Ditto.
14441         * sched-rgn.c (compute_dom_prob_ps): Ditto.
14442         (compute_trg_info): Ditto.
14444 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
14446         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
14447         S reg when fp attribute set.
14448         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
14450 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
14452         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
14453         New pattern.
14454         (*and_one_cmplsi3_compare0_uxtw): Likewise.
14455         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
14456         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
14458 2013-05-02  Richard Biener  <rguenther@suse.de>
14460         * tree-scalar-evolution.c (scev_info_hasher): Remove.
14461         (struct instantiate_cache_entry): New type.
14462         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
14463         (struct instantiate_cache_type): New type.
14464         (set_instantiated_value, get_instantiated_value): Remove.
14465         (get_instantiated_value_entry): New function.
14466         (instantiate_scev_name): Use the new cache and adjust.
14467         (instantiate_scev_poly): Adjust.
14468         (instantiate_scev_binary): Likewise.
14469         (instantiate_array_ref): Likewise.
14470         (instantiate_scev_convert): Likewise.
14471         (instantiate_scev_not): Likewise.
14472         (instantiate_scev_3): Likewise.
14473         (instantiate_scev_2): Likewise.
14474         (instantiate_scev_r): Likewise.
14475         (instantiate_scev): Likewise.
14476         (resolve_mixers): Likewise.
14478 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
14480         PR target/57091
14481         * lra-constraints.c (best_small_class_operands_num): Remove.
14482         (process_alt_operands): Remove small_class_operands_num.  Take
14483         small classes operands into losers and only if the operand is not
14484         matched.  Modify debugging output.
14485         (curr_insn_transform): Remove best_small_class_operands_num.
14486         Print insn name.
14488 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14490         * config/aarch64/aarch64-builtins.c
14491         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
14492         * config/aarch64/aarch64-simd-builtins.def
14493         (reduc_splus_): Add new modes.
14494         (reduc_uplus_): New.
14495         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
14496         (reduc_uplus_v4sf): Likewise.
14497         (reduc_splus_v4sf): Likewise.
14498         (aarch64_addv<mode>): Likewise.
14499         (reduc_uplus_<mode>): Likewise.
14500         (reduc_splus_<mode>): Likewise.
14501         (aarch64_addvv2di): Likewise.
14502         (reduc_uplus_v2di): Likewise.
14503         (reduc_splus_v2di): Likewise.
14504         (aarch64_addvv2si): Likewise.
14505         (reduc_uplus_v2si): Likewise.
14506         (reduc_splus_v2si): Likewise.
14507         (reduc_<sur>plus_<mode>): New.
14508         (reduc_<sur>plus_v2di): Likewise.
14509         (reduc_<sur>plus_v2si): Likewise.
14510         (reduc_<sur>plus_v4sf): Likewise.
14511         (aarch64_addpv4sf): Likewise.
14512         * config/aarch64/arm_neon.h
14513         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
14514         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
14515         add UNSPEC_SADDV, UNSPEC_UADDV.
14516         (SUADDV): New.
14517         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
14519 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14521         * config/aarch64/arm_neon.h
14522         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
14524 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14526         * config/aarch64/aarch64-builtins
14527         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
14529 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14531         * config/aarch64/aarch64-simd-builtins.def
14532         (reduc_smax_): New.
14533         (reduc_smin_): Likewise.
14534         (reduc_umax_): Likewise.
14535         (reduc_umin_): Likewise.
14536         (reduc_smax_nan_): Likewise.
14537         (reduc_smin_nan_): Likewise.
14538         (fmax): Remove.
14539         (fmin): Likewise.
14540         (smax): Update for V2SF, V4SF and V2DF modes.
14541         (smin): Likewise.
14542         (smax_nan): New.
14543         (smin_nan): Likewise.
14544         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
14545         (<su><maxmin><mode>3): ...This, refactor.
14546         (s<maxmin><mode>3): New.
14547         (<maxmin_uns><mode>3): Likewise.
14548         (reduc_<maxmin_uns>_<mode>): Refactor.
14549         (reduc_<maxmin_uns>_v4sf): Likewise.
14550         (reduc_<maxmin_uns>_v2si): Likewise.
14551         (aarch64_<fmaxmin><mode>: Remove.
14552         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
14553         new builtin names.
14554         (vmin<q>_f<32,64>): Likewise.
14555         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
14556         (FMAXMIN): New.
14557         (su): Add mappings for smax, smin, umax, umin.
14558         (maxmin): New.
14559         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
14560         (FMAXMIN): Rename as...
14561         (FMAXMIN_UNS): ...This.
14562         (maxminv): Remove.
14563         (fmaxminv): Likewise.
14564         (fmaxmin): Likewise.
14565         (maxmin_uns): New.
14566         (maxmin_uns_op): Likewise.
14568 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14570         * config/aarch64/arm_neon.h
14571         (vac<ge, gt><sd>_f<32, 64>): Rename to...
14572         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
14573         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
14575 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14577         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
14578         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
14580 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14582         * config/aarch64/aarch64-simd.md
14583         (vcond<mode>_internal): Handle special cases for constant masks.
14584         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
14585         (vcondu<mode><mode>): Likewise.
14586         (vcond<v_cmp_result><mode>): New.
14588 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14590         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
14591         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
14592         * config/aarch64/aarch64-simd-builtins.def
14593         (cmeq): Update to BUILTIN_VALLDI.
14594         (cmgt): Likewise.
14595         (cmge): Likewise.
14596         (cmle): Likewise.
14597         (cmlt): Likewise.
14598         * config/aarch64/arm_neon.h
14599         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
14600         to builtins or C as appropriate.
14602 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14604         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
14605         (cmgeu): ...This.
14606         (cmhi): Rename to...
14607         (cmgtu): ...This.
14608         * config/aarch64/aarch64-simd.md
14609         (simd_mode): Add SF.
14610         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
14611         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
14612         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
14613         (cstore<mode>_neg): ...This.
14614         * config/aarch64/iterators.md
14615         (VALLF): new.
14616         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
14617         (COMPARISONS): New.
14618         (UCOMPARISONS): Likewise.
14619         (optab): Add missing comparisons.
14620         (n_optab): New.
14621         (cmp_1): Likewise.
14622         (cmp_2): Likewise.
14623         (CMP): Likewise.
14624         (cmp): Remove.
14625         (VCMP_S): Likewise.
14626         (VCMP_U): Likewise.
14627         (V_cmp_result): Add DF, SF modes.
14628         (v_cmp_result): Likewise.
14629         (v): Likewise.
14630         (vmtype): Likewise.
14631         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
14633 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
14635         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
14636         define_insn to define_insn_and_split.
14637         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
14638         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
14639         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
14640         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
14641         (thumb2_negscc): Likewise.
14643 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
14645         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
14647 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
14649         * config/arm/thumb2.md: Remove trailing whitespaces.
14651 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
14653         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
14654         Use gen_int_mode rather than GEN_INT.
14656 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
14658         * value-prof.c (stream_in_histogram_value): Remove the strayed
14659         debug_gimple_stmt.
14661 2013-04-30  Richard Biener  <rguenther@suse.de>
14663         PR middle-end/57122
14664         * cfghooks.c (split_edge): Properly check for the loop latch edge.
14666 2013-04-30  Richard Biener  <rguenther@suse.de>
14668         PR middle-end/57107
14669         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
14671 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
14673         PR rtl-optimization/56957
14674         PR rtl-optimization/57105
14675         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
14676         variable.  Use just INSN_UID for determining whether an insn
14677         should be only disconnected from the insn stream.
14678         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
14680 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
14682         PR tree-optimization/57104
14683         * tsan.c (instrument_expr): Don't instrument accesses to
14684         DECL_HARD_REGISTER VAR_DECLs.
14686 2013-04-30  Richard Biener  <rguenther@suse.de>
14688         * function.h (loops_for_fn): New inline function.
14689         (set_loops_for_fn): Likewise.
14690         * cfgloop.h (place_new_loop): Add struct function parameter.
14691         (get_loop): Likewise.
14692         (get_loops): Likewise.
14693         (number_of_loops): Likewise.
14694         (fel_next): Adjust.
14695         (fel_init): Likewise.
14696         * cfg.c (get_loop_copy): Adjust.
14697         * cfgloop.c (flow_loops_dump): Likewise.
14698         (record_loop_exits): Likewise.
14699         (verify_loop_structure): Likewise.
14700         * cfgloopanal.c (mark_irreducible_loops): Likewise.
14701         (estimate_reg_pressure_cost): Likewise.
14702         (mark_loop_exit_edges): Likewise.
14703         * cfgloopmanip.c (place_new_loop): Likewise.
14704         (add_loop): Likewise.
14705         (duplicate_loop): Likewise.
14706         * graph.c (draw_cfg_nodes): Likewise.
14707         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
14708         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
14709         (extract_affine_chrec): Likewise.
14710         (build_scop_iteration_domain): Likewise.
14711         * graphite.c (graphite_initialize): Likewise.
14712         * ira-build.c (create_loop_tree_nodes): Likewise.
14713         (more_one_region_p): Likewise.
14714         (rebuild_regno_allocno_maps): Likewise.
14715         (mark_loops_for_removal): Likewise.
14716         (mark_all_loops_for_removal): Likewise.
14717         (remove_unnecessary_regions): Likewise.
14718         (ira_build): Likewise.
14719         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
14720         * loop-init.c (fix_loop_structure): Likewise.
14721         (gate_rtl_move_loop_invariants): Likewise.
14722         (gate_rtl_unswitch): Likewise.
14723         (gate_rtl_unroll_and_peel_loops): Likewise.
14724         (rtl_doloop): Likewise.
14725         * lto-streamer-in.c (input_cfg): Likewise.
14726         * lto-streamer-out.c (output_cfg): Likewise.
14727         * modulo-sched.c (sms_schedule): Likewise.
14728         * predict.c (tree_estimate_probability): Likewise.
14729         (tree_estimate_probability_driver): Likewise.
14730         (estimate_loops): Likewise.
14731         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
14732         (move_sese_region_to_fn): Likewise.
14733         (debug_loop_num): Likewise.
14734         * tree-chrec.c (chrec_evaluate): Likewise.
14735         (hide_evolution_in_other_loops_than_loop): Likewise.
14736         (chrec_component_in_loop_num): Likewise.
14737         (reset_evolution_in_loop): Likewise.
14738         (evolution_function_is_invariant_rec_p): Likewise.
14739         * tree-if-conv.c (main_tree_if_conversion): Likewise.
14740         * tree-inline.c (copy_loops): Likewise.
14741         (copy_cfg_body): Likewise.
14742         (tree_function_versioning): Likewise.
14743         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
14744         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
14745         Likewise.
14746         (add_to_evolution_1): Likewise.
14747         (scev_const_prop): Likewise.
14748         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
14749         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
14750         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
14751         (tree_ssa_lim_initialize): Likewise.
14752         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
14753         (verify_loop_closed_ssa): Likewise.
14754         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
14755         (tree_ssa_loop_im): Likewise.
14756         (tree_ssa_loop_unswitch): Likewise.
14757         (tree_vectorize): Likewise.
14758         (check_data_deps): Likewise.
14759         (tree_ssa_loop_ivcanon): Likewise.
14760         (tree_ssa_loop_bounds): Likewise.
14761         (tree_complete_unroll): Likewise.
14762         (tree_complete_unroll_inner): Likewise.
14763         (tree_parallelize_loops): Likewise.
14764         (tree_ssa_loop_prefetch): Likewise.
14765         (tree_ssa_loop_ivopts): Likewise.
14766         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
14767         * tree-vectorizer.c (vectorize_loops): Likewise.
14769 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
14771         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
14772         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
14773         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
14774         with EABI_LINK_SPEC.
14776 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14778         PR target/44578
14779         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
14780         alternative.
14782 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
14784         PR target/57097
14785         * lra-constraints.c (process_alt_operands): Discourage a bit more
14786         using memory for pseudos.  Print cost dump for alternatives.
14787         Modify cost values for conflicts with early clobbers.
14788         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
14790 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14792         PR target/57098
14793         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
14795 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
14797         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
14798         from/to S register.
14799         (movdi_aarch64): Support LDR/STR from/to D register.
14801 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
14803         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
14804         or higher by default.
14806 2013-04-29  Richard Biener  <rguenther@suse.de>
14808         PR middle-end/57075
14809         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
14810         even if not adding abnormal edges for calls that can make
14811         abnormal gotos.
14813 2013-04-29  Richard Biener  <rguenther@suse.de>
14815         PR middle-end/57103
14816         * tree-cfg.c (move_stmt_op): Fix condition under which to update
14817         TREE_BLOCK.
14818         (move_stmt_r): Remove redundant checking.
14820 2013-04-29  Teresa Johnson  <tejohnson@google.com>
14822         PR bootstrap/57077
14823         * basic-block.h (apply_scale): New function.
14824         (apply_probability): Use apply_scale.
14825         * gimple-streamer-in.c (input_bb): Ditto.
14826         * lto-streamer-in.c (input_cfg): Ditto.
14827         * lto-cgraph.c (merge_profile_summaries): Ditto.
14828         * tree-optimize.c (execute_fixup_cfg): Ditto.
14829         * tree-inline.c (copy_bb): Update comment to use apply_scale.
14830         (copy_edges_for_bb): Ditto.
14831         (copy_cfg_body): Ditto.
14833 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14835         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
14836         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
14837         (tail_merge_optimize): Handle current_loops == NULL.
14839 2013-04-26  Jeff Law  <law@redhat.com>
14841         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
14842         (simplify_cond_using_ranges): Generalize code to simplify
14843         COND_EXPRs where one argument is a constant and the other
14844         is an SSA_NAME created by an integral type conversion.
14846 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14848         * config/arm/arm.md (store_minmaxsi): Use only when
14849         optimize_insn_for_size_p.
14851 2013-04-29  Christian Bruel  <christian.bruel@st.com>
14853         PR target/57108
14854         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
14856 2013-04-29  Richard Biener  <rguenther@suse.de>
14858         PR middle-end/57089
14859         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
14860         loop tree make sure to schedule a fixup for the child as well.
14861         (expand_omp_for_generic): Properly add loops.
14862         (expand_omp_for_static_nochunk): Likewise.
14863         (expand_omp_for_static_chunk): Likewise.
14864         (expand_omp_for): For the degenerate case fixup loops.
14865         (expand_omp_sections): Fix default bb placement in loops.
14866         (expand_omp_atomic_pipeline): Properly add loops.
14868 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14870         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
14872 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14874         * tree-ssa-tail-merge.c: Update header comment.
14876 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14878         * config/aarch64/arm_neon.h
14879         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
14880         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
14881         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
14882         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
14883         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
14884         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
14885         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
14886         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
14888 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14890         * config/aarch64/aarch64-simd.md
14891         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
14892         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
14893         fix_trunc, fixuns_trunc.
14894         (ftrunc<VDQF:mode>2): New.
14895         * config/aarch64/iterators.md (optab): Add fix, fixuns.
14896         (fix_trunc_optab): New.
14898 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14900         * config/aarch64/aarch64-builtins.c
14901         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
14902         iceilf, lround, iroundf.
14904 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14906         PR target/54349
14907         * config/i386/i386.h (enum ix86_tune_indices)
14908         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
14909         New, split from X86_TUNE_INTER_UNIT_MOVES.
14910         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
14911         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
14912         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
14913         (TARGET_INTER_UNIT_MOVES): Remove.
14914         * config/i386/i386.c (initial_ix86_tune_features): Update.
14915         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
14916         (ix86_expand_convert_uns_didf_sse): Use
14917         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14918         (ix86_expand_vector_init_one_nonzero): Ditto.
14919         (ix86_expand_vector_init_interleave): Ditto.
14920         (inline_secondary_memory_needed): Return true for moves from SSE class
14921         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
14922         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14923         * config/i386/constraints.md (Yi, Ym): Depend on
14924         TARGET_INTER_UNIT_MOVES_TO_VEC.
14925         (Yj, Yn): New constraints.
14926         * config/i386/i386.md (*movdi_internal): Change constraints of
14927         operand 1 from Yi to Yj and from Ym to Yn.
14928         (*movsi_internal): Ditto.
14929         (*movdf_internal): Ditto.
14930         (*movsf_internal): Ditto.
14931         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
14932         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14933         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
14934         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
14935         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
14936         * config/i386/sse.md (movdi_to_sse): Ditto.
14937         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
14938         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
14939         TARGET_INTER_UNIT_MOVES.
14940         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
14941         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
14942         instead of TARGET_INTER_UNIT_MOVES.
14943         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
14944         operand 1 from Yi to Yj and from Ym to Yn.
14946 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14948         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
14949         (float_truncate_hi_): Likewise.
14950         (float_extend_lo_): Likewise.
14951         (float_truncate_lo_): Likewise.
14952         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
14953         (aarch64_float_extend_lo_v2df): Likewise.
14954         (vec_unpacks_hi_v4sf): Likewise.
14955         (aarch64_float_truncate_lo_v2sf): Likewise.
14956         (aarch64_float_truncate_hi_v4sf): Likewise.
14957         (vec_pack_trunc_v2df): Likewise.
14958         (vec_pack_trunc_df): Likewise.
14960 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14962         * config/aarch64/aarch64-builtins.c
14963         (aarch64_fold_builtin): Fold float conversions.
14964         * config/aarch64/aarch64-simd-builtins.def
14965         (floatv2si, floatv4si, floatv2di): New.
14966         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
14967         * config/aarch64/aarch64-simd.md
14968         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
14969         * config/aarch64/iterators.md (FLOATUORS): New.
14970         (optab): Add float, floatuns.
14971         (su_optab): Likewise.
14973 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14975         * config/aarch64/aarch64-builtins.c
14976         (aarch64_builtin_vectorized_function): Use new names for
14977         fcvt builtins.
14978         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
14979         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
14980         (fcvtzu): Split as...
14981         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
14982         (fcvtas): Split as...
14983         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
14984         (fcvtau): Split as...
14985         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
14986         (fcvtps): Split as...
14987         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
14988         (fcvtpu): Split as...
14989         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
14990         (fcvtms): Split as...
14991         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
14992         (fcvtmu): Split as...
14993         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
14994         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
14995         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
14996         (lfrintnusf, lfrintnudf): Likewise.
14997         * config/aarch64/aarch64-simd.md
14998         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
14999         define_insn.
15000         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
15001         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
15002         (fcvt_pattern): Likewise.
15004 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
15006         * config/aarch64/aarch64-simd.md
15007         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
15008         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
15010 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
15012         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
15013         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
15014         (vrnd<a,m,n,p>_f32): Implement using builtins.
15015         (vrnd<i,x><q>_f<32, 64>): New.
15017 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
15019         * config/aarch64/aarch64-builtins.c
15020         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
15021         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
15022         (frintz): Rename to...
15023         (btrunc): ...this.
15024         (frintp): Rename to...
15025         (ceil): ...this.
15026         (frintm): Rename to...
15027         (floor): ...this.
15028         (frinti): Rename to...
15029         (nearbyint): ...this.
15030         (frintx): Rename to...
15031         (rint): ...this.
15032         (frinta): Rename to...
15033         (round): ...this.
15034         * config/aarch64/aarch64-simd.md
15035         (aarch64_frint<frint_suffix><mode>): Delete.
15036         (<frint_pattern><mode>2): Convert to insn.
15037         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
15038         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
15039         (frint_pattern): Likewise.
15040         (frint_suffix): Likewise.
15042 2013-04-29  Richard Biener  <rguenther@suse.de>
15044         PR tree-optimization/57081
15045         * loop-init.c: Include tree-flow.h.
15046         (loop_optimizer_finalize): Free number of iteration estimates.
15047         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
15049 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
15051         PR tree-optimization/57083
15052         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
15053         non-singleton shift count range, zero extend low_bound for uns case.
15055         * config/i386/predicates.md (general_vector_operand): New predicate.
15056         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
15057         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
15058         if they aren't nonimmediate operands.  If their original values
15059         satisfy const_vector_equal_evenodd_p, don't shift them.
15060         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
15061         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
15062         if not nonimmediate_operand.
15063         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
15064         instead of register_operand.
15065         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
15067 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
15069         * stor-layout.c (finalize_size_functions): Allocate a structure and
15070         reset cfun before dumping the functions.
15072 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
15074         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
15076         PR target/56866
15077         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
15078         use xop_pmacsdqh if uns_p.
15079         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
15080         the immediate rotate count.
15082 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
15084         * rtl.h (struct rtx_def): Add comment for field jump.
15085         (LRA_SUBREG_P): New macro.
15086         * recog.c (register_operand): Check LRA_SUBREG_P.
15087         * lra.c (lra): Add note at the end of RTL code. Align non-empty
15088         stack frame.
15089         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
15090         (lra_final_code_change): Skip subreg change for operators.
15091         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
15092         if there are no operand changes.
15093         * lra-constraints.c (curr_insn_set): New.
15094         (match_reload): Set LRA_SUBREG_P.
15095         (emit_spill_move): Ditto.
15096         (check_and_process_move): Use curr_insn_set. Process only single
15097         set insns.  Don't initialize sec_mem_p and change_p.
15098         (simplify_operand_subreg): Use LRA_SUBREG_P.
15099         (reg_in_class_p): New function.
15100         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
15101         of #ifdef.  Add code to remove cycling.
15102         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
15103         non-null disp.  Reload inner instead of disp when base and index
15104         are null.  Try to put lo_sum into register.
15105         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
15106         (check_and_process_move): Move code for move cost check to
15107         simple_move_p.  Remove equiv_substitution.
15108         (simple_move_p): New function.
15109         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
15110         curr_insn_set.  Call check_and_process_move only for single set
15111         insns.  Use the new function.  Move call of check_and_process_move
15112         after operand equiv substitution and address process.
15114 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
15116         PR go/57045
15117         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
15118         with nonlocal goto receivers or returns twice calls, ignore
15119         unininitialized values from abnormal edges to nl goto receiver
15120         or returns twice call.
15122 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
15124         PR tree-optimization/57051
15125         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
15126         and VEC_RSHIFT_EXPR if shift count is a multiple of element
15127         bitsize.
15129 2013-04-26  Richard Biener  <rguenther@suse.de>
15131         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
15132         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
15133         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
15134         (fixup_loop_arrays_after_move): New function.
15135         (move_sese_region_to_fn): Properly outline the loop tree parts
15136         of the SESE region.
15138 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
15140         * config/i386/i386.md (type, unit): Fix long lines.
15142 2013-04-26  Richard Biener  <rguenther@suse.de>
15144         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
15145         (lto-streamer-out.o): Likewise.
15146         * cfgloop.c (init_loops_structure): Export, add struct function
15147         argument and adjust.
15148         (flow_loops_find): Adjust.
15149         * cfgloop.h (enum loop_estimation): Add EST_LAST.
15150         (init_loops_structure): Declare.
15151         * lto-streamer-in.c: Include cfgloop.h.
15152         (input_cfg): Input the loop tree.
15153         * lto-streamer-out.c: Include cfgloop.h.
15154         (output_cfg): Output the loop tree.
15155         (output_struct_function_base): Do not drop PROP_loops.
15157 2013-03-26  Richard Biener  <rguenther@suse.de>
15159         * tree-cfg.c (execute_build_cfg): Build the loop tree.
15160         (pass_build_cfg): Provide PROP_loops.
15161         (move_sese_region_to_fn): Remove loops that are outlined into fn
15162         for now.
15163         * tree-inline.c: Include cfgloop.h.
15164         (initialize_cfun): Do not drop PROP_loops.
15165         (copy_loops): New function.
15166         (copy_cfg_body): Copy loop structure.
15167         (tree_function_versioning): Initialize destination loop tree.
15168         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
15169         (pass_parallelize_loops): Do IL verification.
15170         * loop-init.c (loop_optimizer_init): Fixup loops if required.
15171         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
15172         the CFG make sure we fixup loops as well.
15173         * tree-ssa-tail-merge.c: Include cfgloop.h.
15174         (replace_block_by): When merging loop latches mark loops for fixup.
15175         * lto-streamer-out.c (output_struct_function_base): Drop
15176         PROP_loops for now.
15177         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
15178         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
15179         * ipa-split.c: Include cfgloop.h.
15180         (split_function): Add the new return block to the loop tree root.
15181         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
15182         whether we have removed the forwarder block.
15183         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
15184         * cfgloop.h (place_new_loop): Declare.
15185         * cfgloopmanip.c (place_new_loop): Export.
15186         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
15187         (tree-switch-conversion.o): Likewise.
15188         (tree-complex.o): Likewise.
15189         (tree-inline.o): Likewise.
15190         (tree-ssa-tailmerge.o): Likewise.
15191         (ipa-split.o): Likewise.
15192         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
15193         (tree-ssa-copy.o): Likewise.
15194         * tree-switch-conversion.c: Include cfgloop.h
15195         (process_switch): If we emit a bit-test cascade, schedule loops
15196         for fixup.
15197         * tree-complex.c: Include cfgloop.h.
15198         (expand_complex_div_wide): Properly add new basic-blocks to loops.
15199         * asan.c: Include cfgloop.h.
15200         (create_cond_insert_point): Properly add new basic-blocks to
15201         loops, schedule loop fixup.
15202         * cfgloop.c (verify_loop_structure): Check that looks are not
15203         marked for fixup.
15204         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
15205         to loops.
15206         (expand_omp_for_generic): Likewise.
15207         (expand_omp_sections): Likewise.
15208         (expand_omp_atomic_pipeline): Schedule loops for fixup.
15209         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
15210         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
15211         is initialized, not when loops are present.
15212         * tree-parloops.c (parallelize_loops): Remove checking here.
15213         * passes.c (init_optimization_passes): Schedule a copy-propagation
15214         pass before complete unrolling of inner loops.
15216 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
15218         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
15219         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
15220         (colorize_init): Add argument to _WIN32 version.
15221         * toplev.c: Include diagnostic-color.h.
15222         (process_options): Default to -fdiagnostics-color=auto if
15223         GCC_COLORS env var is in the environment.
15224         * common.opt (fdiagnostics-color=): Add Var and Init.
15225         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
15226         env var is in the environment, the default is auto rather than never.
15228         * diagnostic.h (file_name_as_prefix): Add context argument.
15229         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
15230         the string as locus.
15231         * langhooks.c (lhd_print_error_function): Adjust caller.
15233 2013-04-25  Lawrence Crowl  <crowl@google.com>
15235         * var-tracking.c (shared_hash_def::htab):
15236         Change type to hash_table.  Update dependent calls and types.
15238 2013-04-25  Lawrence Crowl  <crowl@google.com>
15240         * Makefile.in: Update as needed below.
15242         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
15243         Move declaration to after the type's method definitons.
15245         * attribs.c (htab_t scoped_attributes::attribute_hash):
15246         Change type to hash_table.  Update dependent calls and types.
15248         * bitmap.c (htab_t bitmap_desc_hash):
15249         Change type to hash_table.  Update dependent calls and types.
15251         * cselib.c (htab_t cselib_hash_table):
15252         Change type to hash_table.  Update dependent calls and types.
15254         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
15255         (hash_string_slot_node): Move implementation into lto-streamer.h
15256         struct string_slot_hasher.
15257         (eq_string_slot_node): Likewise.
15259         * data-streamer-out.c: Update output_block::string_hash_table
15260         dependent calls and types.
15262         * dwarf2cfi.c (htab_t trace_index):
15263         Change type to hash_table.  Update dependent calls and types.
15265         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
15266         Change type to hash_table.  Update dependent calls and types.
15267         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
15268         (htab_t optimize_external_refs::map): Likewise.
15269         (htab_t output_comp_unit::extern_map): Likewise.
15270         (htab_t output_comdat_type_unit::extern_map): Likewise.
15271         (htab_t output_macinfo::macinfo_htab): Likewise.
15272         (htab_t optimize_location_lists::htab): Likewise.
15273         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
15275         * except.c (htab_t ehspec_hash_type):
15276         Change type to hash_table.  Update dependent calls and types.
15277         (assign_filter_values::ttypes): Likewise.
15278         (assign_filter_values::ehspec): Likewise.
15279         (sjlj_assign_call_site_values::ar_hash): Likewise.
15280         (convert_to_eh_region_ranges::ar_hash): Likewise.
15282         * gcse.c (htab_t pre_ldst_table):
15283         Change type to hash_table.  Update dependent calls and types.
15285         * ggc-common.c (htab_t saving_htab):
15286         Change type to hash_table.  Update dependent calls and types.
15287         (htab_t loc_hash): Likewise.
15288         (htab_t ptr_hash): Likewise.
15289         (call_count): Rename ggc_call_count.
15290         (call_alloc): Rename ggc_call_alloc.
15291         (loc_descriptor): Rename make_loc_descriptor.
15292         (add_statistics): Rename ggc_add_statistics.
15294         * ggc-common.c (saving_htab):
15295         Change type to hash_table.  Update dependent calls and types.
15297         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
15298         (push_gimplify_context): Likewise.
15299         (pop_gimplify_context): Likewise.
15300         (struct gimple_temp_hash_elt): Added.
15301         (struct gimplify_hasher): Likewise.
15302         (struct gimplify_ctx.temp_htab):
15303         Change type to hash_table.  Update dependent calls and types.
15305         * gimple-fold.c: Include gimplify-ctx.h.
15307         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
15308         Change type to hash_table.  Update dependent calls and types.
15309         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
15310         avoid potential global name collision.
15312         * gimplify.c: Include gimplify-ctx.h.
15313         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
15314         (htab_t gimplify_ctx::temp_htab):
15315         Update dependent calls and types for new type hash_table.
15316         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
15317         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
15319         * gimplify-ctx.h: New.
15320         (struct gimple_temp_hash_elt): Move from gimplify.c.
15321         (class gimplify_hasher): New.
15322         (struct gimplify_ctx): Move from gimple.h.
15323         (htab_t gimplify_ctx::temp_htab):
15324         Change type to hash_table.  Update dependent calls and types.
15326         * graphite-clast-to-gimple.c: Include graphite-htab.h.
15327         (htab_t ivs_params::newivs_index):
15328         Change type to hash_table.  Update dependent calls and types.
15329         (htab_t ivs_params::params_index): Likewise.
15330         (htab_t print_generated_program::params_index): Likewise.
15331         (htab_t gloog::newivs_index): Likewise.
15332         (htab_t gloog::params_index): Likewise.
15334         * graphite.c: Include graphite-htab.h.
15335         4htab_t graphite_transform_loops::bb_pbb_mapping):
15336         Change type to hash_table.  Update dependent calls and types.
15338         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
15339         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
15340         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
15342         * graphite-dependences.c: Include graphite-htab.h.
15343         (loop_is_parallel_p): Change hash table type of parameter.
15345         * graphite-htab.h: New.
15346         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
15347         (extern find_pbb_via_hash): Move from graphite-poly.h.
15348         (extern loop_is_parallel_p): Move from graphite-poly.h.
15349         (extern get_loop_body_pbbs): Move from graphite-poly.h.
15351         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
15352         (extern loop_is_parallel_p): Move to graphite-htab.h.
15353         (extern get_loop_body_pbbs): Move to graphite-htab.h.
15355         * haifa-sched.c (htab_t delay_htab):
15356         Change type to hash_table.  Update dependent calls and types.
15357         (htab_t delay_htab_i2): Likewise.
15359         * ira-color.c (htab_t allocno_hard_regs_htab):
15360         Change type to hash_table.  Update dependent calls and types.
15362         * ira-costs.c (htab_t cost_classes_htab):
15363         Change type to hash_table.  Update dependent calls and types.
15365         * loop-invariant.c (htab_t merge_identical_invariants::eq):
15366         Change type to hash_table.  Update dependent calls and types.
15368         * loop-iv.c (htab_t bivs):
15369         Change type to hash_table.  Update dependent calls and types.
15371         * loop-unroll.c (htab_t opt_info::insns_to_split):
15372         Change type to hash_table.  Update dependent calls and types.
15373         (htab_t opt_info::insns_with_var_to_expand): Likewise.
15375         * lto-streamer.h (struct string_slot): Move from data-streamer.h
15376         (struct string_slot_hasher): New.
15377         (htab_t output_block::string_hash_table):
15378         Change type to hash_table.  Update dependent calls and types.
15380         * lto-streamer-in.c (freeing_string_slot_hasher): New.
15381         (htab_t file_name_hash_table):
15382         Change type to hash_table.  Update dependent calls and types.
15384         * lto-streamer-out.c: Update output_block::string_hash_table dependent
15385         calls and types.
15387         * lto-streamer.c (htab_t tree_htab):
15388         Change type to hash_table.  Update dependent calls and types.
15390         * omp-low.c: Include gimplify-ctx.h.
15392         * passes.c (htab_t name_to_pass_map):
15393         Change type to hash_table.  Update dependent calls and types.
15394         (pass_traverse): Rename to passes_pass_traverse.
15396         * plugin.c (htab_t event_tab):
15397         Change type to hash_table.  Update dependent calls and types.
15399         * postreload-gcse.c (htab_t expr_table):
15400         Change type to hash_table.  Update dependent calls and types.
15401         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
15403         * sese.c (debug_rename_map_1): Make extern.
15404         (htab_t copy_bb_and_scalar_dependences::rename_map):
15405         Change type to hash_table.  Update dependent calls and types.
15407         * sese.h (extern debug_rename_map): Move to .c file.
15409         * store-motion.c (htab_t store_motion_mems_table):
15410         Change type to hash_table.  Update dependent calls and types.
15412         * trans-mem.c (htab_t tm_new_mem_hash):
15413         Change type to hash_table.  Update dependent calls and types.
15415         * tree-browser.c (htab_t TB_up_ht):
15416         Change type to hash_table.  Update dependent calls and types.
15418         * tree-cfg.c (htab_t discriminator_per_locus):
15419         Change type to hash_table.  Update dependent calls and types.
15421         * tree-complex.c: Include tree-hasher.h
15422         (htab_t complex_variable_components):
15423         Change type to hash_table.  Update dependent calls and types.
15425         * tree-eh.c (htab_t finally_tree):
15426         Change type to hash_table.  Update dependent calls and types.
15428         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
15429         struct int_tree_hasher.
15430         (extern int_tree_map_eq): Likewise.
15431         (uid_decl_map_hash): Removed.
15432         (extern decl_tree_map_eq): Likewise.
15434         * tree-hasher.h: New.
15435         (struct int_tree_hasher): New.
15436         (typedef int_tree_htab_type): New.
15438         * tree-inline.c: Include gimplify-ctx.h.
15440         * tree-mudflap.c: Include gimplify-ctx.h.
15442         * tree-parloops.c: Include tree-hasher.h.
15443         (htab_t eliminate_local_variables_stmt::decl_address):
15444         Change type to hash_table.  Update dependent calls and types.
15445         (htab_t separate_decls_in_region::decl_copies): Likewise.
15447         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
15448         Change type to hash_table.  Update dependent calls and types.
15450         * tree-sra.c (candidates):
15451         Change type to hash_table.  Update dependent calls and types.
15453         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
15454         in tree-flow.h.
15455         (int_tree_map_hash): Likewise.
15457         * tree-ssa-dom.c (htab_t avail_exprs):
15458         Change type to hash_table.  Update dependent calls and types.
15460         * tree-ssa-live.c (var_map_base_init::tree_to_index):
15461         Change type to hash_table.  Update dependent calls and types.
15463         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
15464         Change type to hash_table.  Update dependent calls and types.
15466         * tree-ssa-phiopt.c (seen_ssa_names):
15467         Change type to hash_table.  Update dependent calls and types.
15469         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
15470         Change type to hash_table.  Update dependent calls and types.
15472         * tree-ssa-uncprop.c (equiv):
15473         Change type to hash_table.  Update dependent calls and types.
15475 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
15477         PR rtl-optimization/57003
15478         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
15479         call note_stores with kill_clobbered_value callback again after
15480         killing regs_invalidated_by_call.
15482 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15484         * config/aarch64/aarch64-simd.md
15485         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
15486         (aarch64_simd_bsl<mode>): Likewise.
15487         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
15489 2013-04-25  Marek Polacek  <polacek@redhat.com>
15491         PR tree-optimization/57066
15492         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
15494 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
15496         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
15498 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15500         * config/aarch64/aarch64-builtins.c
15501         (aarch64_fold_builtin): New.
15502         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
15503         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
15504         * config/aarch64/aarch64-simd-builtins.def (abs): New.
15505         * config/aarch64/arm_neon.h
15506         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
15508 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15509             Tejas Belagod  <tejas.belagod@arm.com>
15511         * config/aarch64/aarch64-builtins.c
15512         (aarch64_gimple_fold_builtin): New.
15513         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
15514         * config/aarch64/aarch64-simd-builtins.def (addv): New.
15515         * config/aarch64/aarch64-simd.md (addpv4sf): New.
15516         (addvv4sf): Update.
15517         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
15519 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15521         * config/aarch64/aarch64.md
15522         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
15524 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15526         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
15527         (*ngcsi_uxtw): New pattern.
15529 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15530             Julian Brown  <julian@codesourcery.com>
15532         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
15533         (TB_DREG): Add T_V4HF.
15534         (v4hf_UP): New macro.
15535         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15536         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15537         Handle initialisation of V4HF. Adjust initialisation of reinterpret
15538         built-ins.
15539         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15540         (arm_vector_mode_supported_p): Handle V4HF.
15541         (arm_mangle_map): Handle V4HFmode.
15542         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
15543         * config/arm/arm_neon_builtins.def: Add entries for
15544         vcvtv4hfv4sf, vcvtv4sfv4hf.
15545         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
15546         (neon_vcvtv4hfv4sf): Likewise.
15547         * config/arm/neon-gen.ml: Handle half-precision floating point
15548         features.
15549         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
15550         * config/arm/arm_neon.h: Regenerate.
15551         * config/arm/neon.ml (type elts): Add F16.
15552         (type vectype): Add T_float16x4, T_floatHF.
15553         (type vecmode): Add V4HF.
15554         (type features): Add Requires_FP_bit feature.
15555         (elt_width): Handle F16.
15556         (elt_class): Likewise.
15557         (elt_of_class_width): Likewise.
15558         (mode_of_elt): Refactor.
15559         (type_for_elt): Handle F16, fix error messages.
15560         (vectype_size): Handle T_float16x4.
15561         (vcvt_sh): New function.
15562         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
15563         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
15564         (string_of_mode): Handle V4HF.
15565         * doc/arm-neon-intrinsics.texi: Regenerate.
15567 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15569         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
15570         format specifier in 'X' case.
15572 2013-04-25  Alan Modra  <amodra@gmail.com>
15574         PR target/57052
15575         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
15576         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
15577         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
15578         Repeat for many other rotate/shift and mask patterns using subregs.
15579         Name lshiftrt insns.
15580         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
15581         on WORDS_BIG_ENDIAN.
15583 2013-04-25  Alan Modra  <amodra@gmail.com>
15585         * config.gcc: Support little-endian powerpc-linux targets.
15586         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
15587         (LINK_OS_LINUX_SPEC): Define.
15588         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
15589         Preserve MASK_LITTLE_ENDIAN.
15590         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
15591         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
15592         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
15593         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
15594         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
15595         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
15596         Correct fp word order for little-endian.  Don't shift toc entries
15597         smaller than a word for little-endian.
15598         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
15599         (bswapdi2 splits): Correct low-part subreg for little-endian.
15600         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
15601         low/high where such is correct only for be.
15602         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
15603         little-endian for -mcall-aixdesc.
15605 2013-04-25  Alan Modra  <amodra@gmail.com>
15607         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
15608         replace_equiv_address_nv.
15610 2013-04-25  Alan Modra  <amodra@gmail.com>
15612         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
15614 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15616         Revert:
15617         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15618         * rtl.h (struct rtx_def): ...
15620 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15622         PR rtl-optimizations/57046
15623         * lra-constraints (split_reg): Set up lra_risky_transformations_p
15624         for multi-reg splits.
15626 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
15628         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
15630 2013-04-24  Sterling Augustine  <saugustine@google.com>
15632         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
15633         (comp_dir_string, debug_str_dwo_section): New.
15634         (DEBUG_STR_DWO_SECTION): Rename to ...
15635         (DEBUG_DWO_STR_SECTION): ... this.
15636         (DEBUG_NORM_STR_SECTION): Delete.
15637         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
15638         (DEBUG_STR_DWO_SECTION_FLAGS): New.
15639         (find_AT_string): Move most logic to ...
15640         (find_AT_string_in_table): ... here.  New.
15641         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
15642         add_skeleton_AT_string.  Delete logic.
15643         (output_skeleton_debug_sections): Remove call to
15644         add_top_level_skeleton_die_attrs.
15645         (add_comp_dir_attribute): Move logic to comp_dir_string.
15646         (dwarf2out_init): Initialize debug_str_dwo_section.
15647         (output_indirect_string): Call find_string_form.
15648         (output_indirect_strings): Rewrite.
15649         (prune_unused_types): Empty skeleton_debug_str_hash.
15650         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
15651         (dwarf2out_finish):  Call output_indirect_strings.
15653 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
15655         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
15657 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15659         * rtl.h (struct rtx_def): Add comment for field jump.
15660         (LRA_SUBREG_P): New macro.
15661         * recog.c (register_operand): Check LRA_SUBREG_P.
15662         * lra.c (lra): Add note at the end of RTL code. Align non-empty
15663         stack frame.
15664         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
15665         (lra_final_code_change): Skip subreg change for operators.
15666         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
15667         if there are no operand changes.
15668         * lra-constraints.c (curr_insn_set): New.
15669         (match_reload): Set LRA_SUBREG_P.
15670         (emit_spill_move): Ditto.
15671         (check_and_process_move): Use curr_insn_set. Process only single
15672         set insns.  Don't initialize sec_mem_p and change_p.
15673         (simplify_operand_subreg): Use LRA_SUBREG_P.
15674         (reg_in_class_p): New function.
15675         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
15676         of #ifdef.  Add code to remove cycling.
15677         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
15678         non-null disp.  Reload inner instead of disp when base and index
15679         are null.  Try to put lo_sum into register.
15680         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
15681         (check_and_process_move): Move code for move cost check to
15682         simple_move_p.  Remove equiv_substitution.
15683         (simple_move_p): New function.
15684         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
15685         curr_insn_set.  Call check_and_process_move only for single set
15686         insns.  Use the new function.  Move call of check_and_process_move
15687         after operand equiv substitution and address process.
15689 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
15691         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
15692         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
15693         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
15695 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
15697         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
15699 2013-04-24  Marek Polacek  <polacek@redhat.com>
15701         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
15702         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
15703         (select_loops_exit_conditions): Likewise.
15704         (number_of_iterations_for_all_loops): Likewise.
15705         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
15706         (scev_analysis): Likewise.
15708 2013-04-02  Catherine Moore  <clm@codesourcery.com>
15709             Chao-ying Fu  <fu@mips.com>
15711         * config/mips/micromips.md (jraddiusp): New pattern.
15712         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
15713         instruction if possible.
15715 2013-04-24  Alan Modra  <amodra@gmail.com>
15717         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
15719 2013-04-24  Julian Brown  <julian@codesourcery.com>
15720             Chung-Lin Tang  <cltang@codesourcery.com>
15722         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
15723         dependency behavior in enumeration type DIE generation. Add TODO note
15724         to comments about future DW_FORM_sdata/udata re-work of related code.
15726 2013-04-23  Lawrence Crowl  <crowl@google.com>
15728         * Makefile.in: Update as needed below.
15730         * hash-table.h (class hash_table):
15731         Correct many methods with parameter types compare_type to the correct
15732         value_type.  (Correct code was unlikely to notice the change.)
15733         (hash_table::elements_with_deleted) New.
15734         (class hashtable::iterator): New.
15735         (hashtable::begin()): New.
15736         (hashtable::end()): New.
15737         (FOR_EACH_HASH_TABLE_ELEMENT): New.
15739         * statistics.c (statistics_hashes):
15740         Change type to hash_table.  Update dependent calls and types.
15742         * tree-into-ssa.c (var_infos):
15743         Change type to hash_table.  Update dependent calls and types.
15745         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
15746         Change type to hash_table.  Update dependent calls and types.
15748         * tree-ssa-loop-im.c (struct mem_ref.refs):
15749         Change type to hash_table.  Update dependent calls and types.
15751         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
15752         Change type to hash_table.  Update dependent calls and types.
15754         * tree-ssa-sccvn.c (vn_tables_s::nary):
15755         Change type to hash_table.  Update dependent calls and types.
15756         (vn_tables_s::phis): Likewise.
15757         (vn_tables_s::references): Likewise.
15759         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
15760         (vn_reference_eq): Update parameter and return types.
15762         * tree-ssa-structalias.c (pointer_equiv_class_table):
15763         Change type to hash_table.  Update dependent calls and types.
15764         (location_equiv_class_table): Likewise.
15766         * tree-vect-data-refs.c: Consequential changes for making
15767         peeling a hash_table.
15769         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
15770         (destroy_loop_vec_info): Dependent hash_table update.
15772         * tree-vectorizer.h (peeling_htab):
15773         Change type to hash_table.  Update dependent calls and types.
15775 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
15777         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
15778         to check the register content is equal or not.
15779         * lra-constraints.c (match_reload): Use lra_assign_reg_val
15780         to assign register content record.
15781         * lra-eliminations.c (update_reg_eliminate): Use
15782         lra_update_reg_val_offset to update register content offset.
15783         * lra-int.h (struct lra_reg): Add offset member.
15784         (lra_reg_val_equal_p): New static inline function.
15785         (lra_update_reg_val_offset): New static inline function.
15786         (lra_assign_reg_val): New static inline function.
15787         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
15788         to assign register content record.
15789         (initialize_lra_reg_info_element): Initial offset to zero.
15791 2013-04-23  Catherine Moore  <clm@codesourcery.com>
15793         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
15794         operands.  Record compression.
15796 2013-04-23  Xinliang David Li  <davidxl@google.com>
15798         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
15800 2013-04-23  Richard Biener  <rguenther@suse.de>
15802         PR middle-end/57036
15803         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
15804         parameter, only add abnormal goto edges from the copied body
15805         if the call could perform abnormal gotos.
15806         (copy_cfg_body): Adjust.
15808 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
15810         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
15812 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
15814         * coretypes.h (gimple_stmt_iterator): Add struct to make
15815         compatible with C.
15817 2013-04-23  Richard Biener  <rguenther@suse.de>
15819         PR tree-optimization/57026
15820         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
15821         from SSA names occuring in abnormal PHI nodes.
15823 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15825         * lto/lto.c (print_lto_report_1): Fix LTO report names.
15827 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15829         * lto/lto.c (print_lto_report_1): Declare early.
15830         (read_cgraph_and_symbols): Call print_lto_report_1 early.
15832 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15834         * common.opt (-flto-report-wpa): Add.
15835         * doc/invoke.texi (-flto-report-wpa): Add.
15836         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
15837         (lto_main): dito.
15839 2013-04-22  Xinliang David Li  <davidxl@google.com>
15841         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
15842         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
15843         * Makefile.in: New dependency
15845             David Daney <ddaney.cavm@gmail.com>
15847         * configure.ac (gcc_cv_as_micromips_support):  Use the
15848         --fatal-warnings option.
15849         * configure: Regenerate.
15851 2013-04-22  Marek Polacek  <polacek@redhat.com>
15853         PR sanitizer/56990
15854         * tsan.c (instrument_expr): Don't instrument expression
15855         in case its size is zero.
15857 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
15859         PR target/57032
15860         Revert:
15861         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15863         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15865 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15867         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
15868         (gimple_stmt_iterator): New typedef.
15869         * gimple.h (gimple_stmt_iterator): Rename to...
15870         (gimple_stmt_iterator_d): ... This.
15871         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
15872         trees be valid for GIMPLE and GENERIC.
15873         (TARGET_GIMPLE_FOLD_BUILTIN): New.
15874         * gimple-fold.c (gimple_fold_call): Call target hook
15875         gimple_fold_builtin.
15876         * hooks.c (hook_bool_gsiptr_false): New.
15877         * hooks.h (hook_bool_gsiptr_false): New.
15878         * target.def (fold_stmt): New.
15879         * doc/tm.texi: Regenerate.
15881 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
15883         PR target/57018
15884         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
15885         a set sp if no stack realignment.
15887 2013-04-22  Nick Clifton  <nickc@redhat.com>
15889         * config.gcc (tilegx-linux): Extend extra_objs rather than
15890         overwriting it.
15891         (tilepro-linux): Likewise.
15893 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15895         * config/aarch64/aarch64-builtins.c
15896         (CF): Remove.
15897         (CF0, CF1, CF2, CF3, CF4, CF10): New.
15898         (VAR<1-12>): Add MAP parameter.
15899         (BUILTIN_*): Likewise.
15900         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
15901         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
15902         (aarch64_ushl_n<mode>): Likewise.
15903         (aarch64_sshr_n<mode>): Likewise.
15904         (aarch64_ushr_n<mode>): Likewise.
15905         (aarch64_<maxmin><mode>): Likewise.
15906         (aarch64_sqrt<mode>): Likewise.
15907         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
15908         (vshr<q>_n_*): Likewise.
15910 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15912         * config/aarch64/aarch64-builtins.c
15913         (aarch64_simd_builtin_type_mode): Handle SF types.
15914         (sf_UP): Define.
15915         (BUILTIN_GPF): Define.
15916         (aarch64_init_simd_builtins): Handle SF types.
15917         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
15918         (frecps): Likewise.
15919         (frecpx): Likewise.
15920         * config/aarch64/aarch64-simd.md
15921         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
15922         (aarch64_frecpe<mode>): New.
15923         (aarch64_frecps<mode>): Likewise.
15924         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
15925         (v8type): Add frecp<esx>.
15926         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
15927         (aarch64_frecps<mode>): Likewise.
15928         * config/aarch64/iterators.md (FRECP): New.
15929         (frecp_suffix): Likewise.
15930         * config/aarch64/arm_neon.h
15931         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
15933 2013-04-22  Christian Bruel  <christian.bruel@st.com>
15935         PR target/56995
15936         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
15937         (REG_CLASS_NAMES): Idem.
15938         (REG_CLASS_CONTENTS): Idem.
15939         (REGCLASS_HAS_FP_REG): Idem.
15940         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
15941         (sh_conditional_register_usage): Idem.
15943 2013-04-21  Jeff Law  <law@redhat.com>
15945         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
15946         (ssa_forward_propagate_and_combine): Use it.
15948 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15950         * lra.c: Update the flow chart diagram.
15952 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15954         PR rtl-optimization/56847
15955         * lra-constraints.c (process_alt_operands): Discourage alternative
15956         with non-matche doffsettable memory constraint fro memory with
15957         known offset.
15959 2013-04-19  Richard Biener  <rguenther@suse.de>
15961         PR tree-optimization/56982
15962         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
15963         function.
15964         * gimplify.c (gimplify_call_expr): Notice special calls.
15965         (gimplify_modify_expr): Likewise.
15966         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
15967         abnormal control flow receivers.
15968         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
15969         in the same way as cfun->has_nonlocal_labels.
15970         (gimple_purge_dead_abnormal_call_edges): Likewise.
15971         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
15972         receivers start a basic-block.
15974 2013-04-19  Richard Biener  <rguenther@suse.de>
15976         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
15977         member ...
15978         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
15979         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
15980         (SLP_TREE_LOAD_PERMUTATION): Add.
15981         (vect_transform_slp_perm_load): Adjust prototype.
15982         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
15983         (vect_free_slp_instance): Likewise.
15984         (vect_create_new_slp_node): Likewise.
15985         (vect_supported_slp_permutation_p): Remove.
15986         (vect_slp_rearrange_stmts): Adjust.
15987         (vect_supported_load_permutation_p): Likewise.  Inline
15988         vect_supported_slp_permutation_p here.
15989         (vect_analyze_slp_instance): Compute load permutations per
15990         slp node instead of per instance.
15991         (vect_get_slp_defs): Adjust.
15992         (vect_transform_slp_perm_load): Likewise.
15993         (vect_schedule_slp_instance): Remove redundant code.
15994         (vect_schedule_slp): Remove hack for PR56270, add it ...
15995         * tree-vect-stmts.c (vectorizable_load): ... here, do not
15996         CSE loads for SLP.  Adjust.
15998 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
16000         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
16001         spelling in two comments.
16003 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
16005         PR target/56797
16006         * config/arm/arm.c (load_multiple_sequence): Require SP
16007         as base register for loads if SP is in the register list.
16009 2013-04-19  Martin Jambor  <mjambor@suse.cz>
16011         PR tree-optimization/56718
16012         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
16013         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
16014         and made public.  Adjusted all callers.
16015         (ipa_intraprocedural_devirtualization): New function.
16016         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
16017         (ipa_intraprocedural_devirtualization): Likewise.
16018         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
16020 2013-04-19  Richard Biener  <rguenther@suse.de>
16022         PR tree-optimization/57000
16023         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
16025 2013-04-19  Terry Guo  <terry.guo@arm.com>
16027         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
16028         Replace with ...
16029         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
16030         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
16031         (cortex_m4_fmacs): Use new reservations.
16032         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
16034 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
16036         PR rtl-optimization/56999
16037         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
16038         related code.
16039         (lra_coalesce): Remove split_origin_bitmap and related code.
16040         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
16041         ranges if necessary.
16043 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
16045         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
16046         New array.
16047         (ix86_expand_call): Remove clobbered_registers array and use
16048         x86_64_ms_sysv_extra_clobbered_registers instead.
16049         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
16050         Declare here.
16051         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
16052         predicate.
16053         * config/i386/i386.md (*call_rex64_ms_sysv): Use
16054         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
16055         (*call_value_rex64_ms_sysv): Ditto.
16057 2013-04-18  Cary Coutant  <ccoutant@google.com>
16059         * dwarf2out.c (output_pubnames): Check die_perennial_p of
16060         parent instead of die_mark.
16062 2013-04-18  Diego Novillo  <dnovillo@google.com>
16064         * gimple.c (create_gimple_tmp): New.
16065         (get_expr_type): New.
16066         (build_assign): New.
16067         (build_type_cast): New.
16068         * gimple.h (enum ssa_mode): Define.
16069         (gimple_seq_set_location): New.
16070         * asan.c (build_check_stmt): Change some gimple_build_* calls
16071         to use build_assign and build_type_cast.
16073 2013-04-18  Richard Biener  <rguenther@suse.de>
16075         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
16076         handle negative step.  Remove redundant checks.
16077         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
16078         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
16079         for negative step and grouped loads fail to vectorize.
16081 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
16083         * emit-rtl.c (reset_insn_used_flags): New function.
16084         (reset_all_used_flags): Use it.
16085         (verify_insn_sharing): New function.
16086         (verify_rtl_sharing): Fix verification for SEQUENCEs.
16088 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
16090         PR tree-optimization/56984
16091         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
16092         and (x >> M) >= N don't register any assertion if N << M is the
16093         minimum value.
16095 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
16097         * lower-subreg.c (resolve_simple_move): If called self-recursive,
16098         do not delete_insn insns that have not yet been emitted, only
16099         unlink them with remove_insn.
16100         * df-scan.c (df_insn_delete): Revert r197492.
16102 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
16104         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
16105         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
16107 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
16109         * config/arm/arm.md (movsicc_insn): Convert define_insn into
16110         define_insn_and_split.
16111         (and_scc,ior_scc,negscc): Likewise.
16112         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
16114 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
16116         * config/arm/arm.c (use_return_insn): Return 0 for targets that
16117         can benefit from using a sequence of LDRD instructions in epilogue
16118         instead of a single LDM instruction.
16120 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
16122         PR 45688
16123         * doc/extend.texi: Fix typo.
16125 2013-04-17  Richard Biener  <rguenther@suse.de>
16127         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
16128         (vect_build_slp_tree): ... here.
16129         (vect_build_slp_tree_1): Compute which stmts of the SLP group
16130         match.  Remove special-casing of mismatched complex loads.
16131         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
16132         re-try the match with swapped commutative operands.
16133         (vect_supported_load_permutation_p): Remove special-casing of
16134         mismatched complex loads.
16135         (vect_analyze_slp_instance): Adjust.
16137 2013-04-17  Richard Biener  <rguenther@suse.de>
16139         PR rtl-optimization/56921
16140         * cfgloop.h (struct loop): Add simple_loop_desc member.
16141         (struct niter_desc): Mark with GTY(()).
16142         (simple_loop_desc): Do not use aux field but simple_loop_desc.
16143         * loop-iv.c (get_simple_loop_desc): Likewise.
16144         (free_simple_loop_desc): Likewise.
16146         Revert
16147         2013-04-16  Richard Biener  <rguenther@suse.de>
16149         PR rtl-optimization/56921
16150         * loop-init.c (pass_rtl_move_loop_invariants): Add
16151         TODO_do_not_ggc_collect to todo_flags_finish.
16152         (pass_rtl_unswitch): Same.
16153         (pass_rtl_unroll_and_peel_loops): Same.
16154         (pass_rtl_doloop): Same.
16156 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
16158         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
16159         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
16160         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
16161         references.
16162         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
16163         * tree-streamer.c (record_common_node): Adjust reference in comment.
16165 2013-04-17  Terry Guo  <terry.guo@arm.com>
16167         * config/arm/cortex-m4.md: Add a new bypass.
16169 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16171         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
16172         New pattern.
16173         (*subs_<optab><mode>_multp2): New pattern.
16174         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
16175         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
16177 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16179         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
16180         (*subs_mul_imm_<mode>): New pattern.
16182 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
16184         PR target/56948
16185         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
16186         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
16187         (vsx_movti_32bit): Change j->wa to O->wa.
16189 2013-04-16  Richard Biener  <rguenther@suse.de>
16191         PR rtl-optimization/56921
16192         * loop-init.c (pass_rtl_move_loop_invariants): Add
16193         TODO_do_not_ggc_collect to todo_flags_finish.
16194         (pass_rtl_unswitch): Same.
16195         (pass_rtl_unroll_and_peel_loops): Same.
16196         (pass_rtl_doloop): Same.
16198 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16200         * config/arm/arm.c (emit_multi_reg_push): New declaration
16201         for an existing function.
16202         (arm_emit_strd_push): New function.
16203         (arm_expand_prologue): Used here.
16204         (arm_emit_ldrd_pop): New function.
16205         (arm_expand_epilogue): Used here.
16206         (arm_get_frame_offsets): Update condition.
16207         (arm_emit_multi_reg_pop): Add a special case for load of a single
16208         register with writeback.
16210 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
16212         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
16213         description.
16215 2013-04-16  Richard Biener  <rguenther@suse.de>
16217         PR tree-optimization/56756
16218         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
16219         (first_mem_ref_loc): New.
16220         (execute_sm): Place the load temporarily before a previous
16221         access instead of in the latch edge to ensure its SSA dependencies
16222         are defined at points dominating the load.
16224 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
16226         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
16227         correct fix by moving header and footer insn to the footer of
16228         the merged basic block.  Clear BB_END of the merged-away block.
16230         PR middle-end/43631
16231         * emit-rtl.c (make_note_raw): New function.
16232         (link_insn_into_chain): New static inline function.
16233         (add_insn): Use it.
16234         (add_insn_before, add_insn_after): Factor insn chain linking code...
16235         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
16236         using link_insn_into_chain.
16237         (note_outside_basic_block_p): New helper function for emit_note_after
16238         and emit_note_before.
16239         (emit_note_after): Use nobb variant of add_insn_after if the note
16240         should not be contained in a basic block.
16241         (emit_note_before): Use nobb variant of add_insn_before if the note
16242         should not be contained in a basic block.
16243         (emit_note_copy): Use make_note_raw.
16244         (emit_note): Likewise.
16245         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
16246         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
16247         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
16248         the moved barrier the tail of the basic block it follows.
16249         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
16251 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
16253         PR tree-optimization/56962
16254         * gimple-ssa-strength-reduction.c (record_increment): Only set
16255         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
16256         either rhs1 or rhs2 is equal to c->base_expr.
16258 2013-04-15  Richard Biener  <rguenther@suse.de>
16260         PR tree-optimization/56933
16261         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
16262         member.
16263         (GROUP_READ_WRITE_DEPENDENCE): Remove.
16264         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
16265         * tree-vect-data-refs.c (vect_analyze_group_access): Move
16266         dependence check ...
16267         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
16268         ... here.
16269         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
16270         GROUP_READ_WRITE_DEPENDENCE.
16272 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16274         * emit-rtl.c (reset_all_used_flags): New function.
16275         (verify_rtl_sharing): Call reset_all_used_flags before and after
16276         performing the checks.
16278 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16280         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
16281         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
16282         * config/arm/constraints.md (De): New constraint.
16283         * config/arm/neon.md (anddi3_neon): Delete.
16284         (neon_vand<mode>): Expand to standard anddi3 pattern.
16285         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
16286         Move earlier in the file.
16287         (neon_inv_logic_op2): Likewise.
16288         (arm_anddi_operand_neon): New predicate.
16290 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16292         * configure.ac (gcc_cv_ld_as_needed): Set
16293         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
16294         Use -z ignore, -z record on *-*-solaris2*.
16295         (HAVE_LD_AS_NEEDED): Update comment.
16296         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
16297         * configure: Regenerate.
16298         * config.in: Regenerate.
16299         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
16300         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
16301         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
16302         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
16303         equivalents.  Fix markup.
16304         * doc/tm.texi: Regenerate.
16306 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
16308         * config/i386/i386.opt: New option mstack-protector-guard=.
16309         * config/i386/i386-opts.h: Add enum stack_protector_guard.
16310         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
16311         TARGET_SSP_TLS_GUARD.
16312         * config/i386/i386.c (ix86_option_override_internal): Set
16313         ix86_stack_protector_guard.
16314         * config/i386/i386.md (stack_protect_set): Enable for
16315         TARGET_SSP_TLS_GUARD only.
16316         (stack_protect_set_<mode>): Ditto.
16317         (stack_protect_test): Ditto.
16318         (stack_protect_test_<mode>): Ditto.
16319         * doc/invoke.texi (i386 Option): Document.
16321 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
16323         PR target/56890
16324         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
16325         (S_MODES): Set H_MODE bit.
16326         (SF_MODES): Set only S_MODE and SF_MODE bits.
16327         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
16328         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
16329         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
16330         <MODE_FLOAT>: Likewise.
16332 2013-04-15  Joey Ye  <joey.ye@arm.com>
16334         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
16336 2013-04-15  Joey Ye  <joey.ye@arm.com>
16338         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
16339         for real far jump.
16340         (thumb_far_jump_used_p): Count instruction size and set
16341         far_jump_used.
16343 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
16345         * reorg.c (fill_simple_delay_slots): Reindent block of code.
16346         * resource.c (mark_target_live_regs): Reformat conditional block.
16348 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
16350         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
16351         notes, they are emitted only just before final.
16352         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
16354 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
16356         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
16357         * cfgrtl.c (delete_insn): Call it here instead.
16358         * lra-spills.c (lra_final_code_change): Use delete_insn.
16359         * haifa-sched.c (sched_remove_insn): Likewise.
16360         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
16361         returning to the nop pool.
16362         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
16363         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
16365 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
16367         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
16368         * doc/tm.texi: Regenerated.
16370 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
16372         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
16373         QImode checks.
16375 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
16377         * df-core.c (df_find_def): Compare register numbers.
16378         (df_find_use): Likewise.
16380 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
16382         PR target/56903
16383         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
16384         lra_in_progress for return.
16386 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
16388         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
16389         define_insn into define_insn_and_split and emit movsicc patterns.
16391 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
16393         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
16395 2013-04-12  Richard Biener  <rguenther@suse.de>
16397         * tree-pass.h (TODO_do_not_ggc_collect): New.
16398         * passes.c (execute_one_ipa_transform_pass): Honor
16399         TODO_do_not_ggc_collect.
16400         (execute_one_pass): Likewise.
16402         Revert
16403         2013-04-10  Richard Biener  <rguenther@suse.de>
16405         * passes.c (init_optimization_passes): Remove reload pass.
16406         * ira.c (do_reload): Merge into ...
16407         (ira): ... this.
16408         (rest_of_handle_reload): Remove.
16409         (pass_reload): Likewise.
16410         * config/i386/i386.c (ix86_option_override): Refer to ira instead
16411         of reload for vzeroupper pass placement.
16413 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
16415         PR tree-optimization/56918
16416         PR tree-optimization/56920
16417         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
16418         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
16419         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
16420         use wide_mul_with_sign method.
16422 2013-04-12  Richard Biener  <rguenther@suse.de>
16424         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
16425         not be considered a gimple constant.
16427 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
16429         * fold-const.c (const_binop): Handle vector shifts by a scalar.
16430         (fold_binary_loc): Call const_binop also for mixed vector-scalar
16431         operations.
16433 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
16434             Jakub Jelinek  <jakub@redhat.com>
16436         * opts.c: Include diagnostic-color.h.
16437         (common_handle_option): Handle OPT_fdiagnostics_color_.
16438         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
16439         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
16440         (diagnostic-color.o): New.
16441         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
16442         (diagnostic_color_rule): New enum.
16443         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
16444         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
16445         the location string.
16446         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
16447         either NULL, or color kind.
16448         * diagnostic-color.c: New file.
16449         * diagnostic-color.h: New file.
16450         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
16451         arguments.
16452         * doc/invoke.texi (-fdiagnostics-color): Document.
16453         * pretty-print.h (pp_show_color): Define.
16454         (struct pretty_print_info): Add show_color field.
16455         * diagnostic.c: Include diagnostic-color.h.
16456         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
16457         macros.  Colorize error:, warning: etc. strings and also the location
16458         string.
16459         (diagnostic_show_locus): Colorize the caret line.
16460         * pretty-print.c: Include diagnostic-color.h.
16461         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
16462         inside of %< %> quotes or quoted through q format modifier.
16464 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16466         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
16468 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16470         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
16471         code in CC_NZ mode.
16472         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
16473         pattern.
16475 2013-04-11  Marek Polacek  <polacek@redhat.com>
16477         PR tree-optimization/48184
16478         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
16480 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
16482         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
16483         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
16484         (skip_simple_arithmetic): Tidy up.
16485         * tree.h (skip_simple_constant_arithmetic): Declare.
16487 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16489         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
16491 2013-04-11  Richard Biener  <rguenther@suse.de>
16493         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16494         generate vector constants.
16496 2013-04-11  Richard Biener  <rguenther@suse.de>
16498         PR tree-optimization/56878
16499         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
16500         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
16501         New function.
16502         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16503         Prefer to align the DR with the most invariant base address.
16505 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
16507         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
16508         comment.
16510 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
16512         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
16513         floating-point vector comparisons against 0.
16515 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
16517         PR tree-optimization/56899
16518         * fold-const.c (extract_muldiv_1): Apply distributive law
16519         only if TYPE_OVERFLOW_WRAPS (ctype).
16521 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
16523         PR target/56124
16524         * ira-costs.c (scan_one_insn): Check whether the source rtx of
16525         loading has side effect.
16527 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
16529         * config/sparc/sparc.c: Include tree-pass.h.
16530         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
16531         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
16532         head of file.  Change return type.  Split off gate function.
16533         (sparc_gate_work_around_errata): New function.
16534         (pass_work_around_errata): New pass definition.
16535         (insert_pass_work_around_errata) New pass insert definition to
16536         insert pass_work_around_errata just after delayed-branch scheduling.
16537         (sparc_option_override): Insert the pass.
16538         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
16540 2013-04-10  David S. Miller  <davem@davemloft.net>
16542         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
16543         or -mcpu=hypersparc.
16545         * target.def (cstore_mode): New hook.
16546         * target.h: Include insn-codes.h
16547         * targhooks.c: Likewise.
16548         (default_cstore_mode): New function.
16549         * targhooks.h: Declare it.
16550         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
16551         * doc/tm.texi: Rebuild.
16552         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
16553         target hook, rather than inspecting the insn_data.
16554         * config/sparc/sparc.c (sparc_cstore_mode): New function.
16555         (TARGET_CSTORE_MODE): Redefine.
16556         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
16557         result patterns.
16558         * config/sparc/predicates.md (cstore_result_operand): New special
16559         predicate.
16560         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
16561         Use it for operand 0.
16562         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
16563         (*snesi_special): Likewise.
16564         (*snesi_zero): Likewise.
16565         (*seqsi_zero): Likewise.
16566         (*sltu_insn): Likewise.
16567         (*sgeu_insn): Likewise.
16568         (*seqdi_special): Make operand 0 and comparison operation be of
16569         DImode.
16570         (*snedi_special): Likewise.
16571         (*snedi_special_vis3): Likewise.
16572         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
16573         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
16574         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
16575         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
16576         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
16577         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
16578         (*sltu_extend_sp64): Likewise.
16579         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
16580         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
16581         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
16582         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
16583         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
16585 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
16587         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
16588         (aarch64_start_file): Use the new function.
16590 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
16591             Jason Merrill  <jason@redhat.com>
16593         * common.opt: Add -gdwarf.
16594         * opts.c (common_handle_option): Handle it.
16595         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
16597 2013-04-10  Richard Biener  <rguenther@suse.de>
16599         * passes.c (execute_todo): Do not call ggc_collect conditional here.
16600         (execute_one_ipa_transform_pass): But unconditionally here.
16601         (execute_one_pass): And here.
16602         (init_optimization_passes): Remove reload pass.
16603         * tree-pass.h (TODO_ggc_collect): Remove.
16604         (pass_reload): Likewise.
16605         * ira.c (do_reload): Merge into ...
16606         (ira): ... this.
16607         (rest_of_handle_reload): Remove.
16608         (pass_reload): Likewise.
16609         * config/i386/i386.c (ix86_option_override): Refer to ira instead
16610         of reload for vzeroupper pass placement.
16611         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
16612         and todo_flags_finish of all passes.
16614 2013-04-10  Richard Biener  <rguenther@suse.de>
16616         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
16617         first_const_oprnd field, rename first_def_type to first_op_type.
16618         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
16619         (vect_get_and_check_slp_defs): Always use the type of the
16620         operand.  Allow mixed vect_external_def, vect_constant_def types.
16621         (vect_get_constant_vectors): Handle mixed vect_external_def,
16622         vect_constant_def types.
16624 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
16626         PR tree-optimization/55524
16627         * tree-ssa-math-opts.c
16628         (convert_mult_to_fma): Don't use an fms construct
16629         when we don't have an fms operation, but fnma, and it looks
16630         likely that we'll be able to use the latter.
16632 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
16634         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
16635         function.
16636         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
16637         inline fail caused by overwritable functions.
16639 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
16641         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
16642         unnecessary bits in the constant power of two case.
16644 2013-04-10  Richard Biener  <rguenther@suse.de>
16646         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
16647         broken code swapping operands.
16648         (vect_build_slp_tree): Do not compute load permutations here.
16649         (vect_analyze_slp_instance): Compute load permutations here,
16650         after building the SLP tree.
16652 2013-04-09  Christian Bruel  <christian.bruel@st.com>
16654         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
16655         of next/prev_real_insn.
16657 2013-04-09  Jan Hubicka  <jh@suse.cz>
16659         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
16660         Drop aliased parameter.
16661         (function_and_variable_visibility): Do not handle alias pairs.
16662         * cgraph.c (varpool_externally_visible_p): Update prototype.
16663         * varpool.c (varpool_add_new_variable): Update.
16665 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16667         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
16669 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
16671         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
16673         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
16675 2013-04-09  Marek Polacek  <polacek@redhat.com>
16677         PR tree-optimization/48762
16678         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
16680 2013-04-09  Richard Biener  <rguenther@suse.de>
16682         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
16683         dealing with cost.
16684         (vect_build_slp_tree): Likewise.
16685         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
16686         calculating the cost of a SLP instance.
16687         (vect_analyze_slp_instance): Use it from here, after building
16688         the SLP tree.
16690 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
16692         PR middle-end/56883
16693         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
16694         expand_omp_for_static_chunk): Use simple_p = true in
16695         force_gimple_operand_gsi calls when assigning to addressable decls.
16697 2013-04-09  Jeff Law  <law@redhat.com>
16699         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
16700         when the boolean was created by converting a wider object which
16701         had a boolean range.
16703 2013-04-09  Richard Biener  <rguenther@suse.de>
16705         * tree-vectorizer.h (slp_void_p): Remove.
16706         (slp_tree): Typedef before _slp_tree declaration.
16707         (struct _slp_tree): Use a vector of slp_tree as children.
16708         (vect_get_place_in_interleaving_chain): Remove.
16709         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
16710         Move ...
16711         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
16712         and make static.
16713         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
16714         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
16715         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
16716         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
16717         Use slp_node instead of slp_void_p and adjust.
16719 2013-04-09  Richard Biener  <rguenther@suse.de>
16721         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
16722         work that is not necessary.
16724 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
16726         PR tree-optimization/56854
16727         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
16728         forward into clobber stmts if it would change MEM_REF lhs into
16729         non-MEM_REF.
16731 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
16733         * tree.c (type_hash_lookup, type_hash_add): Make static.
16734         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
16736 2013-04-09  Richard Biener  <rguenther@suse.de>
16738         * tree.h (unsave_expr_now): Remove.
16739         * tree-inline.c (mark_local_for_remap_r): Remove.
16740         (unsave_expr_1): Likewise.
16741         (unsave_r): Likewise.
16742         (unsave_expr_now): Likewise.
16743         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
16744         (propagate_tree_value): Likewise.
16746 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
16748         * doc/rtl.texi (sequence): Rewrite documentation to match the
16749         current use of SEQUENCE rtl objects.
16750         * rtl.def (SEQUENCE): Likewise.
16752         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
16753         Update documentation.
16754         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
16755         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
16757         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
16759 2013-04-08  Teresa Johnson  <tejohnson@google.com>
16761         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
16762         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
16763         methods.
16764         (estimate_edge_size_and_time): Add comment to suggest using rounding
16765         methods.
16766         (estimate_node_size_and_time): Ditto.
16767         (remap_edge_change_prob): Use helper rounding divide methods.
16768         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
16769         (gimple_mod_pow2_value_transform): Ditto.
16770         (gimple_mod_subtract_transform): Ditto.
16771         (gimple_ic_transform): Ditto.
16772         (gimple_stringops_transform): Ditto.
16773         * stmt.c (conditional_probability): Ditto.
16774         (emit_case_dispatch_table): Ditto.
16775         * lto-cgraph.c (merge_profile_summaries): Ditto.
16776         * tree-optimize.c (execute_fixup_cfg): Ditto.
16777         * cfgcleanup.c (try_forward_edges): Ditto.
16778         * cfgloopmanip.c (scale_loop_profile): Ditto.
16779         (loopify): Ditto.
16780         (duplicate_loop_to_header_edge): Ditto.
16781         (lv_adjust_loop_entry_edge): Ditto.
16782         * tree-vect-loop.c (vect_transform_loop): Ditto.
16783         * profile.c (compute_branch_probabilities): Ditto.
16784         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
16785         * lto-streamer-in.c (input_cfg): Ditto.
16786         * gimple-streamer-in.c (input_bb): Ditto.
16787         * ipa-cp.c (update_profiling_info): Ditto.
16788         (update_specialized_profile): Ditto.
16789         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
16790         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
16791         rounding methods.
16792         * sched-rgn.c (compute_dom_prob_ps): Ditto.
16793         (compute_trg_info): Ditto.
16794         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
16795         (purge_dead_edges): Ditto.
16796         * loop-unswitch.c (unswitch_loop): Ditto.
16797         * cgraphclones.c (cgraph_clone_edge): Ditto.
16798         (cgraph_clone_node): Ditto.
16799         * tree-inline.c (copy_bb): Ditto.
16800         (copy_edges_for_bb): Ditto.
16801         (initialize_cfun): Ditto.
16802         (copy_cfg_body): Ditto.
16803         (expand_call_inline): Ditto.
16805 2013-04-08  Kai Tietz  <ktietz@redhat.com>
16807         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
16808         TARGET_CYGWIN64 by TARGET_64BIT.
16810 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16812         * config/epiphany/epiphany.md (GPR_1): New constant.
16813         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
16814         * config/epiphany/epiphany.c (gen_compare_reg):
16815         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
16816         is already in place.
16817         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
16818         Don't require being called during rtl expansion; If y operlaps r0,
16819         return 0.
16820         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
16821         (epiphany_expand_epilogue): Likewise.
16823         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
16824         Don't use CC_FPmode for ORDERED / UNORDERED.
16825         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
16827         * config/epiphany/constraints.md (CnL): New constraint.
16828         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
16829         * config/epiphany/predicates.md (add_operand): Allow 1024.
16831         * config/epiphany/epiphany.md (logical_op): New code iterator.
16832         (op_mnc): New code attribute.
16833         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
16834         (mov_f+1, mov_f+2): New peephole2 patterns.
16836         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
16837         (cstoresi4): Also allow re-use of zero result when doing a NE
16838         comparison to a non-zero operand.
16839         Use (clobber (scratch)) for first insn if the gpr output is not needed.
16841         * config/epiphany/epiphany.md (<insn_opname>v2si3):
16842         Use gen_addsi3_i / gen_subsi3_i.
16844 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16846         PR c++/34949
16847         PR c++/50243
16848         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
16849         contain anything but clobbers, at most one __builtin_stack_restore,
16850         optionally debug stmts and final resx, and if it has at least one
16851         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
16852         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
16853         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
16854         which isn't defaut definition, remove them.
16855         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
16856         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
16857         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
16858         with MEM_REF LHS with SSA_NAME address.
16860 2013-04-08  Jeff Law  <law@redhat.com>
16862         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
16864 2013-04-08  Richard Biener  <rguenther@suse.de>
16866         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
16867         extra newline.
16868         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
16869         determined vector type.
16870         (vect_analyze_data_refs): Likewise.
16871         (vect_get_new_vect_var): Adjust.
16872         (vect_create_destination_var): Preserve SSA name versions.
16873         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
16874         not dump anything here.
16876 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16878         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
16879         Add member lr_slot_known.
16880         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
16881         if necessary.
16882         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
16883         Remove code that sets lr_slot_offset according to what a previous
16884         version of epiphany_emit_save_restore used to do.
16885         (epiphany_emit_save_restore): When doing an lr save or restore,
16886         set/verify lr_slot_known and lr_slot_offset.
16888 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
16890         PR target/54338
16891         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
16892         in ALL_REGS.
16894 2013-04-08  Richard Biener  <rguenther@suse.de>
16896         * alias.c (find_base_term): Fix thinko in previous change.
16898 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16900         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
16901         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
16902         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
16903         if possible to compute val.
16904         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
16905         For QImode integers don't require anything about precision.  Use
16906         const_with_all_bytes_same to find out if the constant doesn't have
16907         repeated bytes in it.
16909 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16911         * config/s390/s390.c (s390_expand_insv): Only accept insertions
16912         within mode size.
16914 2013-04-08  Marek Polacek  <polacek@redhat.com>
16916         PR rtl-optimization/48182
16917         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
16918         value to 1.
16920 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16922         PR target/55487
16923         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
16924         nuses, make sure we have a label.
16926 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16928         PR target/56843
16929         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
16930         (rs6000_emit_swdiv_low_precision): Remove.
16931         (rs6000_emit_swdiv): Rewrite to handle between one and four
16932         iterations of Newton-Raphson generally; modify required number of
16933         iterations for some cases.
16934         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
16936 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
16938         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
16939         set-but-unused variable.
16941         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
16942         basic blocks of released function bodies garbage-collectable.
16944         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
16945         (struct rtl_opt_pass): Add TODO_df_finish.
16947         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
16949 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16951         * config/arm/constraints.md (q): New constraint.
16952         * config/arm/ldrdstrd.md: New file.
16953         * config/arm/arm.md (ldrdstrd.md) New include.
16954         (arm_movdi): Use "q" instead of "r" constraint
16955         for double-word memory access.
16956         (movdf_soft_insn): Likewise.
16957         * config/arm/vfp.md (movdi_vfp): Likewise.
16958         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
16959         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
16960         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
16961         (mem_ok_for_ldrd_strd): Likewise.
16962         (output_move_double): Update assertion.
16964 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16966         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
16968 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16970         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
16971         define_insn_and_split.
16972         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
16974 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16976         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
16977         define_insn_and_split.
16978         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
16979         (shiftsi3_compare): New pattern.
16980         (rrx): New pattern.
16981         * config/arm/unspecs.md (UNSPEC_RRX): New.
16983 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16985         * config/arm/arm.md (negdi_extendsidi): New pattern.
16986         (negdi_zero_extendsidi): Likewise.
16988 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16990         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
16991         define_insn_and_split.
16992         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
16993         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
16995 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16997         * config/arm/arm.md (arm_subdi3): Convert define_insn into
16998         define_insn_and_split.
16999         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
17000         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
17002 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
17004         * config/arm/arm.md (subsi3_carryin): New pattern.
17005         (subsi3_carryin_const): Likewise.
17006         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
17007         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
17009 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
17011         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
17013 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
17015         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
17016         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
17018 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17020         * config/arm/arm.c (arm_expand_builtin): Change fcode
17021         type to unsigned int.
17023 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17025         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
17027 2013-04-04  Ian Lance Taylor  <iant@google.com>
17029         * doc/standards.texi (Standards): The Go frontend supports the Go 1
17030         language standard.
17032 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
17034         PR middle-end/56729
17035         * df-scan.c (df_insn_delete): Disable failing assert.
17037 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17039         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
17040         New function prototype.
17041         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
17042         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
17043         (arm_builtin_vectorized_function): New function.
17045 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17047         * config/arm/arm_neon_builtins.def: New file.
17048         * config/arm/arm.c (neon_builtin_data): Move contents to
17049         arm_neon_builtins.def.
17050         (enum arm_builtins): Include neon builtin definitions.
17051         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
17052         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
17054 2013-04-04  Marek Polacek  <polacek@redhat.com>
17056         PR tree-optimization/48186
17057         * predict.c (maybe_hot_frequency_p): Return false if
17058         HOT_BB_FREQUENCY_FRACTION is 0.
17059         (cgraph_maybe_hot_edge_p): Likewise.
17061 2013-04-04  Richard Biener  <rguenther@suse.de>
17063         PR tree-optimization/56826
17064         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
17065         more accurately.
17067 2013-04-04  Richard Biener  <rguenther@suse.de>
17069         PR tree-optimization/56213
17070         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
17071         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
17073 2013-04-04  Richard Biener  <rguenther@suse.de>
17075         PR tree-optimization/56837
17076         * tree-loop-distribution.c (classify_partition): For non-zero
17077         values require that the value has the same precision as its
17078         mode to be useful as memset value.
17080 2013-04-03  Nick Clifton  <nickc@redhat.com>
17082         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
17083         (fmssf4): Use fmsf.s on E3V5 architectures.
17084         (fnmasf4): Use fnmaf.s on E3V5 architectures.
17085         (fnmssf4): Use fnmsf.s on E3V5 architectures.
17087 2013-04-03  Jeff Law  <law@redhat.com>
17089         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
17090         (lra-eliminations.o): Likewise.
17092 2013-04-03  Teresa Johnson  <tejohnson@google.com>
17094         * gcov-io.c (compute_working_sets): Moved most of body of old
17095         compute_working_sets here from profile.c.
17096         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
17097         (gcov_working_set_t): Moved typedef here from basic-block.h
17098         (compute_working_set): Declare.
17099         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
17100         (get_working_sets): Renamed from compute_working_set,
17101         replace most of body with call to new compute_working_sets.
17102         (get_exec_counts): Replace call to compute_working_sets
17103         to get_working_sets.
17104         * profile.h (get_working_sets): Renamed from compute_working_set.
17105         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
17106         to get_working_sets.
17107         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
17108         * gcov-dump.c (dump_working_sets): New function.
17110 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
17112         * hwint.c (sext_hwi, zext_hwi): New functions.
17113         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
17114         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
17115         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
17116         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
17117         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
17118         (sext_hwi, zext_hwi): New functions.
17120 2013-04-03  Jeff Law  <law@redhat.com>
17122         PR tree-optimization/56799
17123         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
17124         back test for widening conversion erroneously dropped in prior change.
17126 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17128         PR target/56809
17129         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
17130         instead of next_real_insn.
17132 2013-04-03  Marek Polacek  <polacek@redhat.com>
17134         PR sanitizer/55702
17135         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
17137 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17139         PR target/56809
17140         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
17141         next_real_insn.
17142         (thumb1_output_casesi): Likewise.
17143         (thumb2_output_casesi): Likewise.
17145 2013-04-03  Richard Biener  <rguenther@suse.de>
17147         PR tree-optimization/56817
17148         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
17149         Split out ...
17150         (tree_unroll_loops_completely_1): ... new function to manually
17151         walk the loop tree, properly defering outer loops of unrolled
17152         loops to later iterations.
17154 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17156         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
17157         (vectorizable_load): Likewise.
17158         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
17159         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
17161 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17163         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
17164         BIT_FIELD_REF.
17166 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
17168         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
17170 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
17172         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
17174 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17176         PR tree-optimization/56790
17177         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
17178         folding.
17180 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17182         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
17183         Handle VEC_MERGE.
17184         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
17185         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
17186         equal arguments.
17188 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
17190         PR c/19449
17191         * tree.h (force_folding_builtin_constant_p): New decl.
17192         * builtins.c (force_folding_builtin_constant_p): New variable.
17193         (fold_builtin_constant_p): Fold immediately also if
17194         force_folding_builtin_constant_p.
17196 2013-04-03  Richard Biener  <rguenther@suse.de>
17198         PR tree-optimization/56812
17199         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
17200         DRs of the same interleaving chain are independent.
17202 2013-04-02  Jason Merrill  <jason@redhat.com>
17204         * gdbinit.in (pbb): Use debug fn.
17206 2013-04-02  Lawrence Crowl  <crowl@google.com>
17208         * sese.h (struct ivtype_map_elt_s): Remove unused.
17209         (extern debug_ivtype_map): Remove unused.
17210         (extern eq_ivtype_map_elts): Remove unused.
17211         * sese.c (debug_ivtype_map): Removed unused.
17212         (debug_ivtype_map_1): Removed unused.
17213         (debug_ivtype_elt): Remove unused.
17214         (eq_ivtype_map_elts): Remove unused.
17217 2013-04-02  Kai Tietz  <ktietz@redhat.com>
17219         PR target/52790
17220         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
17221         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
17222         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
17223         function.
17224         (legitimize_pe_coff_symbol): Likewise.
17225         (is_imported_p): New helper-function.
17226         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
17227         for Windows x64 targets.
17228         (ix86_expand_prologue): Optimize for pe-coff targets.
17229         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
17230         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
17231         medium/large code-model.
17232         (legitimize_pic_address): Likewise.
17233         (legitimize_tls_address): Likewise.
17234         (ix86_expand_call): Likewise.
17235         (x86_output_mi_thunk): Likewise.
17236         (get_dllimport_decl): Add new beimport argument.
17237         (construct_plt_address): Don't assert for x64 pe-coff targets.
17238         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
17239         targets.
17240         (SYMBOL_FLAG_STUBVAR): New macro.
17241         (SYMBOL_REF_STUBVAR_P): Likewise.
17242         * config/i386/winnt.c (stub_list): New structure.
17243         (stub_head): New local variable.
17244         (i386_pe_record_stub): New function.
17245         (i386_pe_file_end): Emit refptr-stubs.
17247 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
17249         PR rtl-optimization/56745
17250         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
17251         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
17253         PR c++/34949
17254         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
17255         and both of them are MEM_REFs, just compare first argument for
17256         equality and attempt to deal even with differing offsets.
17258         PR c++/34949
17259         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
17260         of gimple_clobber_p to be MEM_REF.
17261         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
17262         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
17263         after gimplification.
17264         * asan.c (get_mem_ref_of_assignment): Don't instrument
17265         gimple_clobber_p stmts.
17266         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
17267         gimple_clobber_p stmt if they have MEM_REF lhs and
17268         are dead because of another gimple_clobber_p stmt.
17269         * tree-ssa-live.c (clear_unused_block_pointer): Treat
17270         gimple_clobber_p stmts like debug stmts.
17271         (remove_unused_locals): Remove clobbers with MEM_REF lhs
17272         that refer to unused VAR_DECLs or uninitialized values.
17273         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
17274         gimple_clobber_p stmts if they refer to removed parameters.
17275         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
17276         formatting.
17278 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
17280         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
17281         using SWI48 mode attribute.
17283 2013-04-02  Wei Mi  <wmi@google.com>
17285         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
17286         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
17287         *<rotate_insn><mode>3_mask in i386.md.
17289 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17291         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
17293 2013-04-02  Richard Biener  <rguenther@suse.de>
17295         PR tree-optimization/56778
17296         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
17297         Runtime alias tests are not supported for gather loads.
17298         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
17299         stmts referenced from SSA operands before updating SSA form.
17301 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
17302             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17304         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
17305         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
17306         * config/arm/cortex-a53.md: New file.
17307         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
17308         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
17309         * config/arm/arm.c (arm_issue_rate): Likewise.
17310         * config/arm/arm-tune.md: Regenerate
17311         * config/arm/arm-tables.opt: Regenerate.
17312         * config/arm/arm-cores.def: Add cortex-a53.
17314 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
17316         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
17317         non-static link.
17319 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
17321         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
17322         scalar load/store operations using B/H registers.
17323         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
17325 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
17327         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
17328         scalar move.
17329         * config/aarch64/aarch64.c
17330         (aarch64_simd_scalar_immediate_valid_for_move): New.
17331         * config/aarch64/aarch64-protos.h
17332         (aarch64_simd_scalar_immediate_valid_for_move): New.
17333         * config/aarch64/constraints.md (Dh, Dq): New.
17334         * config/aarch64/iterators.md (hq): New.
17336 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
17338         * reorg.c (get_branch_condition): Deal with conditional returns.
17339         (fill_simple_delay_slots): Remove dead code dealing with jumps.
17341 2013-04-01  Wei Mi  <wmi@google.com>
17343         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
17344         Truncate operand 2 using %b asm operand modifier.
17345         (*<shift_insn><mode>3_mask): Ditto.
17346         (*<rotate_insn><mode>3_mask): Ditto.
17348 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
17350         PR middle-end/56798
17351         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
17353 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
17355         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
17356         of next_real_insn.
17357         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
17359 2013-03-30  Lawrence Crowl  <crowl@google.com>
17361         * dse.c (clear_alias_sets): Remove never set.
17362         (disqualified_clear_alias_sets): Remove never set.
17363         (clear_alias_mode_pool): Remove never set.
17364         (dse_step0): Remove condition that is never true.
17365         (canon_address): Remove condition that is never true.
17366         (dse_step7): Remove condition that is never true.
17367         (rest_of_handle_dse): Remove condition that is never true.
17368         (rest_of_handle_dse::did_global): Remove never read from above.
17369         (dse_step2_spill): Remove never called from above.
17370         (dse_step5_spill): Remove never called from above.
17372 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
17374         * doc/md.texi (Standard Names) <casesi>: Update documentation for
17375         JUMP_TABLE_DATA changes.
17376         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
17377         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
17378         (Insns) <jump_table_data>: New entry.
17379         * doc/tm.texi: Regenerate.
17381         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
17383         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
17384         for table jump at the end of a basic block using tablejump_p.
17385         * targhooks.c (default_invalid_within_doloop): Likewise.
17386         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
17387         target hook implementation that is identical to the default hook.
17388         (rs6000_invalid_within_doloop): Remove.
17390         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
17391         unused variable from tablejump_p call.
17393         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
17394         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
17395         (INSN_DELETED_P): Likewise.
17396         (emit_jump_table_data): New prototype.
17397         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
17398         after 4th as unused.
17399         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
17400         * sched-vis.c (print_insn): Likewise.
17401         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
17402         insn for compatibility with back ends that use next_active_insn to
17403         identify jump table data.
17404         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
17405         (remove_insn): Likewise.
17406         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
17407         to be emitted.
17408         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
17409         (emit_jump_table_data): New function.
17411         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
17412         basic block, a JUMP_TABLE_DATA never is.
17413         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
17414         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
17415         off from code handling real insns.
17416         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
17417         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
17418         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
17419         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
17420         is not a NONDEBUG_INSN_P.
17421         * ira-costs.c (scan_one_insn): Likewise.
17422         * jump.c (mark_all_labels): Likewise.
17423         (mark_jump_label_1): Likewise.
17424         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
17425         * lra.c (get_insn_freq): Expect all insns reaching here to be in
17426         a basic block.
17427         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
17428         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
17429         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
17430         JUMP_TABLE_DATA_P insns.
17431         (calculate_elim_costs_all_insns): Likewise.
17432         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
17433         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
17434         (delete_output_reload): Code style fixups.
17435         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
17436         insn flags on this non-insn.
17437         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
17438         as scheduling barriers, for pre-change compatibility.
17439         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
17440         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
17442         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
17443         redundant JUMP_TABLE_DATA_P test.
17444         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
17445         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
17446         (frv_for_each_packet): Likewise.
17447         * config/i386/i386.c (min_insn_size): Likewise.
17448         (ix86_avoid_jump_mispredicts): Likewise.
17449         * config/m32r/m32r.c (m32r_is_insn): Likewise.
17450         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
17451         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
17452         (mips16_insn_length): Robustify.
17453         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
17454         (mips16_split_long_branches): Likewise.
17455         * config/pa/pa.c (pa_combine_instructions): Likewise.
17456         * config/rs6000/rs6000.c (get_next_active_insn): Treat
17457         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
17458         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
17459         as contributing to pool range lengths.
17460         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
17461         Remove redundant JUMP_TABLE_DATA_P test.
17462         (sh_loop_align): Likewise.
17463         (split_branches): Likewise.
17464         (sh_insn_length_adjustment): Likewise.
17465         * config/spu/spu.c (get_branch_target): Likewise.
17467 2013-03-29  Jan Hubicka  <jh@suse.cz>
17469         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
17470         gcov streaming; stream hot bb threshold to ltrans.
17471         * predict.c (get_hot_bb_threshold): Break out from ....
17472         (maybe_hot_count_p): ... here.
17473         (set_hot_bb_threshold): New function.
17474         * lto-section-in.c (lto_section_name): Add profile.
17475         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
17476         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
17477         and data-streamer.h
17478         (histogram_entry): New structure.
17479         (histogram, histogram_pool): New global vars.
17480         (histogram_hash): New structure.
17481         (histogram_hash::hash): New method.
17482         (histogram_hash::equal): Likewise.
17483         (account_time_size): New function.
17484         (cmp_counts): New function.
17485         (dump_histogram): New function.
17486         (ipa_profile_generate_summary): New function.
17487         (ipa_profile_write_summary): New function.
17488         (ipa_profile_read_summary): New function.
17489         (ipa_profile): Decide on threshold.
17490         (pass_ipa_profile): Add ipa_profile_write_summary and
17491         ipa_profile_read_summary.
17492         * Makefile.in (ipa.o): Update dependencies.
17493         * lto-streamer.h (LTO_section_ipa_profile): New section.
17495 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
17497         * tree.h (VAR_P): New.
17499 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
17501         PR lto/56777
17502         * doc/invoke.texi ([-fwhole-program]): Fix typo.
17504 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
17506         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
17507         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
17508         (control_flow_insn_p): Likewise.
17509         * cfgrtl.c (duplicate_insn_chain): Likewise.
17510         * final.c (get_attr_length_1): Likewise.
17511         (shorten_branches): Likewise.
17512         (final_scan_insn): Likewise.
17513         * function.c (instantiate_virtual_regs): Likewise.
17514         * gcse.c (insert_insn_end_basic_block): Likewise.
17515         * ira-costs.c (scan_one_insn): Likewise.
17516         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
17517         * lra.c (check_rtl): Likewise.
17518         * reload1.c (elimination_costs_in_insn): Likewise.
17519         * reorg.c (follow_jumps): Likewise.
17521         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
17522         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
17523         (thumb_far_jump_used_p): Likewise.
17524         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
17525         (workaround_speculation): Likewise.
17526         (add_sched_insns_for_speculation): Likewise.
17527         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
17528         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
17529         (frv_for_each_packet): Likewise.
17530         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
17531         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
17532         (final_emit_insn_group_barriers): Likewise.
17533         * config/m32r/m32r.c (m32r_is_insn): Likewise.
17534         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
17535         (mips16_insn_length): Likewise.
17536         * config/pa/pa.c (pa_reorg): Likewise.
17537         (pa_combine_instructions): Likewise.
17538         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
17539         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
17540         (sh_reorg): Likewise.
17541         (split_branches): Likewise.
17542         * config/spu/spu.c (get_branch_target): Likewise.
17544         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
17545         JUMP_TABLE_DATA_P.
17547 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
17549         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
17550         Fix declaration name.
17552 2013-03-28  Lawrence Crowl  <crowl@google.com>
17554         * graphds.h (struct graph.indicies): Remove unused.
17555         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
17556         (SCOP_ORIGINAL_PDDRS): Remove unused.
17557         * sese.h (extern insert_loop_close_phis): Removed unused.
17558         (extern insert_guard_phis): Removed unused.
17559         (extern ivtype_map_elt_info): Removed unused.
17560         (new_ivtype_map_elt): Removed unused.
17561         * sese.c (ivtype_map_elt_info): Removed unused.
17563 2013-03-28  Lawrence Crowl  <crowl@google.com>
17565         * Makefile.in: Add several missing include dependences.
17566         (DUMPFILE_H): New.
17567         (test-dump.o): New.  This object is not added to any executable,
17568         but is present for ad-hoc testing.
17569         * bitmap.c
17570         (debug (const bitmap_head_def &)): New.
17571         (debug (const bitmap_head_def *)): New.
17572         * bitmap.h
17573         (extern debug (const bitmap_head_def &)): New.
17574         (extern debug (const bitmap_head_def *)): New.
17575         * cfg.c
17576         (debug (edge_def &)): New.
17577         (debug (edge_def *)): New.
17578         * cfghooks.c
17579         (debug (basic_block_def &)): New.
17580         (debug (basic_block_def *)): New.
17581         * dumpfile.h
17582         (dump_node (const_tree, int, FILE *)): Correct source file.
17583         * dwarf2out.c
17584         (debug (die_struct &)): New.
17585         (debug (die_struct *)): New.
17586         * dwarf2out.h
17587         (extern debug (die_struct &)): New.
17588         (extern debug (die_struct *)): New.
17589         * gimple-pretty-print.c
17590         (debug (gimple_statement_d &)): New.
17591         (debug (gimple_statement_d *)): New.
17592         * gimple-pretty-print.h
17593         (extern debug (gimple_statement_d &)): New.
17594         (extern debug (gimple_statement_d *)): New.
17595         * ira-build.c
17596         (debug (ira_allocno_copy &)): New.
17597         (debug (ira_allocno_copy *)): New.
17598         (debug (ira_allocno &)): New.
17599         (debug (ira_allocno *)): New.
17600         * ira-int.h
17601         (extern debug (ira_allocno_copy &)): New.
17602         (extern debug (ira_allocno_copy *)): New.
17603         (extern debug (ira_allocno &)): New.
17604         (extern debug (ira_allocno *)): New.
17605         * ira-lives.c
17606         (debug (live_range &)): New.
17607         (debug (live_range *)): New.
17608         * lra-int.h
17609         (debug (lra_live_range &)): New.
17610         (debug (lra_live_range *)): New.
17611         * lra-lives.c
17612         (debug (lra_live_range &)): New.
17613         (debug (lra_live_range *)): New.
17614         * omega.c
17615         (debug (omega_pb_d &)): New.
17616         (debug (omega_pb_d *)): New.
17617         * omega.h
17618         (extern debug (omega_pb_d &)): New.
17619         (extern debug (omega_pb_d *)): New.
17620         * print-rtl.c
17621         (debug (const rtx_def &)): New.
17622         (debug (const rtx_def *)): New.
17623         * print-tree.c
17624         (debug_tree (tree): Move within file.
17625         (debug_raw (const tree_node &)): New.
17626         (debug_raw (const tree_node *)): New.
17627         (dump_tree_via_hooks (const tree_node *, int)): New.
17628         (debug (const tree_node &)): New.
17629         (debug (const tree_node *)): New.
17630         (debug_verbose (const tree_node &)): New.
17631         (debug_verbose (const tree_node *)): New.
17632         (debug_head (const tree_node &)): New.
17633         (debug_head (const tree_node *)): New.
17634         (debug_body (const tree_node &)): New.
17635         (debug_body (const tree_node *)): New.
17636         (debug_vec_tree (tree): Move and reimplement in terms of dump.
17637         (debug (vec<tree, va_gc> &)): New.
17638         (debug (vec<tree, va_gc> *)): New.
17639         * rtl.h
17640         (extern debug (const rtx_def &)): New.
17641         (extern debug (const rtx_def *)): New.
17642         * sbitmap.c
17643         (debug_raw (simple_bitmap_def &)): New.
17644         (debug_raw (simple_bitmap_def *)): New.
17645         (debug (simple_bitmap_def &)): New.
17646         (debug (simple_bitmap_def *)): New.
17647         * sbitmap.h
17648         (extern debug (simple_bitmap_def &)): New.
17649         (extern debug (simple_bitmap_def *)): New.
17650         (extern debug_raw (simple_bitmap_def &)): New.
17651         (extern debug_raw (simple_bitmap_def *)): New.
17652         * sel-sched-dump.c
17653         (debug (vinsn_def &)): New.
17654         (debug (vinsn_def *)): New.
17655         (debug_verbose (vinsn_def &)): New.
17656         (debug_verbose (vinsn_def *)): New.
17657         (debug (expr_def &)): New.
17658         (debug (expr_def *)): New.
17659         (debug_verbose (expr_def &)): New.
17660         (debug_verbose (expr_def *)): New.
17661         (debug (vec<rtx> &)): New.
17662         (debug (vec<rtx> *)): New.
17663         * sel-sched-dump.h
17664         (extern debug (vinsn_def &)): New.
17665         (extern debug (vinsn_def *)): New.
17666         (extern debug_verbose (vinsn_def &)): New.
17667         (extern debug_verbose (vinsn_def *)): New.
17668         (extern debug (expr_def &)): New.
17669         (extern debug (expr_def *)): New.
17670         (extern debug_verbose (expr_def &)): New.
17671         (extern debug_verbose (expr_def *)): New.
17672         (extern debug (vec<rtx> &)): New.
17673         (extern debug (vec<rtx> *)): New.
17674         * sel-sched-ir.h
17675         (_list_iter_cond_expr): Make inline instead of static.
17676         * sreal.c
17677         (debug (sreal &)): New.
17678         (debug (sreal *)): New.
17679         * sreal.h
17680         (extern debug (sreal &)): New.
17681         (extern debug (sreal *)): New.
17682         * tree.h
17683         (extern debug_raw (const tree_node &)): New.
17684         (extern debug_raw (const tree_node *)): New.
17685         (extern debug (const tree_node &)): New.
17686         (extern debug (const tree_node *)): New.
17687         (extern debug_verbose (const tree_node &)): New.
17688         (extern debug_verbose (const tree_node *)): New.
17689         (extern debug_head (const tree_node &)): New.
17690         (extern debug_head (const tree_node *)): New.
17691         (extern debug_body (const tree_node &)): New.
17692         (extern debug_body (const tree_node *)): New.
17693         (extern debug (vec<tree, va_gc> &)): New.
17694         (extern debug (vec<tree, va_gc> *)): New.
17695         * tree-cfg.c
17696         (debug (struct loop &)): New.
17697         (debug (struct loop *)): New.
17698         (debug_verbose (struct loop &)): New.
17699         (debug_verbose (struct loop *)): New.
17700         * tree-dump.c: Add header dependence.
17701         * tree-flow.h
17702         (extern debug (struct loop &)): New.
17703         (extern debug (struct loop *)): New.
17704         (extern debug_verbose (struct loop &)): New.
17705         (extern debug_verbose (struct loop *)): New.
17706         * tree-data-ref.c
17707         (debug (data_reference &)): New.
17708         (debug (data_reference *)): New.
17709         (debug (vec<data_reference_p> &)): New.
17710         (debug (vec<data_reference_p> *)): New.
17711         (debug (vec<ddr_p> &)): New.
17712         (debug (vec<ddr_p> *)): New.
17713         * tree-data-ref.h
17714         (extern debug (data_reference &)): New.
17715         (extern debug (data_reference *)): New.
17716         (extern debug (vec<data_reference_p> &)): New.
17717         (extern debug (vec<data_reference_p> *)): New.
17718         (extern debug (vec<ddr_p> &)): New.
17719         (extern debug (vec<ddr_p> *)): New.
17720         * tree-ssa-alias.c
17721         (debug (pt_solution &)): New.
17722         (debug (pt_solution *)): New.
17723         * tree-ssa-alias.h
17724         (extern debug (pt_solution &)): New.
17725         (extern debug (pt_solution *)): New.
17726         * tree-ssa-alias.c
17727         (debug (_var_map &)): New.
17728         (debug (_var_map *)): New.
17729         (debug (tree_live_info_d &)): New.
17730         (debug (tree_live_info_d *)): New.
17731         * tree-ssa-alias.h
17732         (extern debug (_var_map &)): New.
17733         (extern debug (_var_map *)): New.
17734         (extern debug (tree_live_info_d &)): New.
17735         (extern debug (tree_live_info_d *)): New.
17737 2013-03-28  Jan Hubicka  <jh@suse.cz>
17739         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
17741 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
17743         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
17744         record only when desired or required.
17746 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
17748         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
17749         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
17751 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17753         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
17754         (*andsi3_compare0_uxtw): New pattern.
17755         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
17756         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
17758 2013-03-28  Jan Hubicka  <jh@suse.cz>
17760         * data-streamer-in.c (streamer_read_gcov_count): New function.
17761         * gimple-streamer-out.c: Include value-prof.h.
17762         (output_gimple_stmt): Output histogram.
17763         (output_bb): Use streamer_write_gcov_count.
17764         * value-prof.c: Include data-streamer.h
17765         (dump_histogram_value): Add HIST_TYPE_MAX.
17766         (stream_out_histogram_value): New function.
17767         (stream_in_histogram_value): New function.
17768         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
17769         (stream_out_histogram_value, stream_in_histogram_value): Declare.
17770         * data-streamer-out.c (streamer_write_gcov_count): New function.
17771         (streamer_write_gcov_count_stream): New function.
17772         * lto-cgraph.c (lto_output_edge): Update counter streaming.
17773         (lto_output_node): Likewise.
17774         (input_node, input_edge): Likewise.
17775         * lto-streamer-out.c (output_cfg): Update streaming.
17776         * lto-streamer-in.c (input_cfg): Likewise.
17777         * data-streamer.h (streamer_write_gcov_count,
17778         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
17779         * gimple-streamer-in.c: Include value-prof.h
17780         (input_gimple_stmt): Input histograms.
17781         (input_bb): Update profile streaming.
17783 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
17785         * genmodes.c (emit_max_int): New function.
17786         (emit_insn_modes_h): Added call to emit_max_function.
17787         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
17788         Added doc.
17789         * machmode.def: Fixed comment.
17791 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
17793         * combine.c (try_combine): Removed useless assert.
17794         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
17796 2013-03-28  Marek Polacek  <polacek@redhat.com>
17797             Richard Biener  <rguenther@suse.de>
17799         PR tree-optimization/56695
17800         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
17801         build signed result of a vector comparison.
17802         * tree-cfg.c (verify_gimple_comparison): Check that a result
17803         of a vector comparison has signed type.
17805 2013-03-28  Richard Biener  <rguenther@suse.de>
17807         PR tree-optimization/37021
17808         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
17809         do not restrict gaps between groups.
17810         * tree-vect-stmts.c (vectorizable_load): Properly account for
17811         a gap between groups.
17813 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
17815         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
17816         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
17817         is not enabled.
17819 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
17821         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
17822         * doc/extend.texi (Named Address Spaces): Ditto.
17823         (Variable Attributes): Ditto.
17825 2013-03-27  Kai Tietz  <ktietz@redhat.com>
17827         * config.build: Add support for cygwin x64 target.
17828         * config.gcc: Likewise.
17829         * config.host: Likewise.
17830         * configure.ac: Likewise
17831         * configure: Regenerated.
17833 2013-03-27  Kai Tietz  <ktietz@redhat.com>
17835         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
17836         * config/i386/t-cygwin-w64: New file.
17837         * config/i386/cygwin-w64.h: New file.
17838         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
17839         and add support for x64-cygwin target.
17840         (CPP_SPEC): Likewise.
17841         (CXX_WRAP_SPEC_LIST): Undefine before define.
17842         (LIBGCJ_SONAME): Use 15 as version.
17844 2013-03-27  Richard Biener  <rguenther@suse.de>
17846         PR tree-optimization/56716
17847         * tree-ssa-structalias.c (perform_var_substitution): Adjust
17848         dumping for ref nodes.
17850 2013-03-27  Martin Jambor  <mjambor@suse.cz>
17852         PR tree-optimization/55334
17853         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
17854         restricted pointers to arrays.
17856 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
17858         * Makefile.in (.SUFFIXES): Add .cc.
17859         (.c.o): Apply same recipe for implicit rule .cc.o.
17861 2013-03-27  Richard Biener  <rguenther@suse.de>
17863         PR tree-optimization/37021
17864         * tree-vect-data-refs.c (vect_check_strided_load): Allow
17865         REALPART/IMAGPART_EXPRs around the supported refs.
17866         * tree-ssa-structalias.c (find_func_aliases): Assume that
17867         floating-point values are not used to transfer pointers.
17869 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17871         * target.def (TARGET_HAS_IFUNC_P): New target hook.
17872         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
17873         * doc/tm.texi: Regenerate.
17874         * targhooks.h (default_has_ifunc_p): New.
17875         * targhooks.c (default_has_ifunc_p): Ditto.
17876         * config/linux-protos.h: New file.
17877         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
17878         hook for linux which disables support of indirect functions in android.
17879         * config/linux-android.c: New file.
17880         * config/t-linux-android.c: Ditto.
17881         * config.gcc: Added new object file linux-android.o.
17882         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
17883         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
17884         * varasm.c (do_assemble_alias): Likewise.
17885         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
17886         doesn't support indirect functions.
17887         * configure: Regenerate.
17889 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
17891         PR target/56102
17892         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
17893         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
17894         mult-word mode.
17896 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17898         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
17900 2013-03-27  Terry Guo  <terry.guo@arm.com>
17902         * config/arm/arm-cores.def: Added core cortex-r7.
17903         * config/arm/arm-tune.md: Regenerated.
17904         * config/arm/arm-tables.opt: Regenerated.
17905         * doc/invoke.texi: Added entry for core cortex-r7.
17907 2013-03-27  Walter Lee  <walt@tilera.com>
17909         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
17910         double-decrement of next_scratch_regno.
17912 2013-03-27  Walter Lee  <walt@tilera.com>
17914         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
17915         input operands.
17916         (insn_v1mulus): Ditto.
17917         (insn_v2muls): Ditto.
17919 2013-03-27  Walter Lee  <walt@tilera.com>
17921         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
17922         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
17924 2013-03-27  Walter Lee  <walt@tilera.com>
17926         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
17927         (*sibcall_value): Ditto.
17929 2013-03-27  Walter Lee  <walt@tilera.com>
17931         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
17932         (insn_mnz_v8qi): ... this ...
17933         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
17934         vector equivalent.
17935         (insn_v<n>mnz): Replaced by ...
17936         (insn_v1mnz): ... this ...
17937         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
17938         equivalent.
17939         (insn_mz_<mode>): Replaced by ...
17940         (insn_mz_v8qi): ... this ...
17941         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
17942         vector equivalent.
17943         (insn_v<n>mz): Replaced by ...
17944         (insn_v1mz): ... this ...
17945         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
17946         equivalent.
17948 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
17950         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
17952 2013-03-26  Roland McGrath  <mcgrathr@google.com>
17954         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
17955         than fprintf with a non-constant, non-format string.
17957 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17959         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
17960         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
17961         operand 0 predicate.
17962         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
17963         attribute.  Use general_x64nomem_operand as operand 1 predicate.
17964         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
17965         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17966         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
17967         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17968         (mov<mode>_insv_1): Remove expander.  Merge insn with
17969         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
17970         Use general_x64nomem_operand as operand 1 predicate.
17971         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
17972         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
17973         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17974         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
17975         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
17976         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
17977         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17978         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
17979         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
17980         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
17981         (general_x64nomem_operand): Ditto.
17983 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17985         * config/rtems.opt: Add -pthread option.
17987 2013-03-26  Richard Biener  <rguenther@suse.de>
17989         * alias.c (find_base_term): Avoid redundant and not used recursion.
17990         (base_alias_check): Get the initial base term from the caller.
17991         (true_dependence_1): Compute and pass base terms to base_alias_check.
17992         (write_dependence_p): Likewise.
17993         (may_alias_p): Likewise.
17995 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
17997         * config/aarch64/aarch64.c (aarch64_classify_address): Support
17998         PC-relative load in SI modes and above only.
18000 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
18002         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
18003         * config/arm/iwmmxt.md (WCGR0): Update.
18004         (WCGR1, WCGR2, WCGR3): Likewise.
18006 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
18008         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
18009         Use x64 and nox64 isa attributes.
18011 2013-03-26  Richard Biener  <rguenther@suse.de>
18013         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
18014         alignment computations and rely on get_object_alignment_1
18015         for the !TYPE_P case.
18016         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
18018 2013-03-26  Walter Lee  <walt@tilera.com>
18020         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
18021         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
18023 2013-03-25  Jeff Law  <law@redhat.com>
18025         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
18026         check for INTEGRAL_TYPE_P that was missing due to checking in
18027         wrong version of prior patch.
18029 2013-03-25  Walter Lee  <walt@tilera.com>
18031         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
18032         TILEGX_INSN_SHUFFLEBYTES1.
18033         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
18034         shufflebytes1.
18035         (tilegx_builtins): Ditto.
18036         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
18038 2013-03-25  Walter Lee  <walt@tilera.com>
18040         * config/tilegx/tilegx.md (floatsisf2): New pattern.
18041         (floatunssisf2): New pattern.
18042         (floatsidf2): New pattern.
18043         (floatunssidf2): New pattern.
18045 2013-03-25  Walter Lee  <walt@tilera.com>
18047         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
18048         tests for constraint J, K, N, P.
18050 2013-03-25  Walter Lee  <walt@tilera.com>
18052         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
18053         Use indirect/pcrel encoding.
18054         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
18055         Ditto.
18057 2013-03-25  Steve Ellcey  <sellcey@mips.com>
18059         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
18060         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
18061         * config/mips/mips.c (mips_option_override): Set IMADD default.
18062         * config/mips/mips.h (PTF_AVOID_IMADD): New.
18063         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
18064         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
18065         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
18066         * doc/invoke.texi (-mimadd/-mno-imadd): New.
18068 2013-03-25  Jeff Law  <law@redhat.com>
18070         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
18071         slightly to avoid creating and folding useless trees.  Simplify
18072         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
18074 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
18076         * config/i386/i386.md (*zero_extendsidi2): Merge with
18077         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
18078         * config/i386/predicates.md (x86_64_zext_operand): Rename from
18079         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
18080         targets.  Clarify comment.
18082 2013-03-25  Martin Jambor  <mjambor@suse.cz>
18084         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
18085         pass-through jump functions differently.
18086         (ipa_read_jump_function): Likewise.  Also use setter functions to set
18087         up jump functions.
18089 2013-03-25  Martin Jambor  <mjambor@suse.cz>
18091         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
18092         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
18093         process it.
18094         (ipa_get_indirect_edge_target): New function.
18095         (devirtualization_time_bonus): New parameter known_aggs, pass it to
18096         ipa_get_indirect_edge_target.  Update all callers.
18097         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
18098         ipa_get_indirect_edge_target_1 instead of calling
18099         ipa_get_indirect_edge_target.
18100         (create_specialized_node): Pass aggvlas to
18101         ipcp_discover_new_direct_edges.
18103 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18105         * config/arm/arm.md (f_sels, f_seld): New types.
18106         (*cmov<mode>): New pattern.
18107         * config/arm/predicates.md (arm_vsel_comparison_operator): New
18108         predicate.
18110 2013-03-25  Kai Tietz  <ktietz@redhat.com>
18112         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
18113         POSIX-printf for mingw-hosted  builds.
18115 2013-03-25  Richard Biener  <rguenther@suse.de>
18117         PR middle-end/56694
18118         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
18119         must-not-throw stmt location.
18121 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18123         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
18124         Emit load-acquire versions when acq is true.
18125         (arm_emit_store_exclusive): Add rel parameter.
18126         Emit store-release versions when rel is true.
18127         (arm_split_compare_and_swap): Use acquire-release instructions
18128         instead.
18129         of barriers when appropriate.
18130         (arm_split_atomic_op): Likewise.
18131         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
18132         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
18133         (VUNSPEC_SLX): Likewise.
18134         (VUNSPEC_LDA): Likewise.
18135         (VUNSPEC_STL): Likewise.
18136         * config/arm/sync.md (atomic_load<mode>): New pattern.
18137         (atomic_store<mode>): Likewise.
18138         (arm_load_acquire_exclusive<mode>): Likewise.
18139         (arm_load_acquire_exclusivesi): Likewise.
18140         (arm_load_acquire_exclusivedi): Likewise.
18141         (arm_store_release_exclusive<mode>): Likewise.
18143 2013-03-25  Catherine Moore  <clm@codesourcery.com>
18145         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
18146         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
18147         * config/mip/predicates.md (lwsp_swsp_operand,
18148         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
18149         sb16_operand, db4_operand, db7_operand, ib3_operand,
18150         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
18151         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
18152         andi16_operand): New predicates.
18153         * config/mips/mips.md (compression): New attribute.
18154         (enabled): New attribute.
18155         (length): Consider compression in computing length.
18156         (shift_compression): New code attribute.
18157         (*add<mode>3): New operands. Record compression.
18158         (sub<mode>3): Likewise.
18159         (one_cmpl<mode>2): Likewise.
18160         (*and<mode>3): Likewise.
18161         (*ior<mode>3): Likewise.
18162         (unnamed pattern for xor): Likewise.
18163         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
18164         (*<optab><mode>3): Likewise.
18165         (*mov<mode>_internal: Likewise.
18166         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
18167         (mips_unsigned_immediate_p): New.
18168         (umips_lwsp_swsp_address_p): New.
18169         (m16_based_address_p): New.
18170         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
18171         (mips_unsigned_immediate_p): New prototype.
18172         (lwsp_swsp_address_p): New prototype.
18173         (m16_based_address_p): New prototype.
18174         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
18175         (mips_signed_immediate_p): New function.
18176         (m16_based_address_p): New function.
18177         (lwsp_swsp_address_p): New function.
18178         (mips_print_operand_punctuation): Recognize short delay slot insns
18179         for microMIPS.add<mode>3"
18181 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18183         PR target/56720
18184         * config/arm/iterators.md (v_cmp_result): New mode attribute.
18185         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
18187 2013-03-25  Richard Biener  <rguenther@suse.de>
18189         PR tree-optimization/56689
18190         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
18191         any edge.
18193 2013-03-25  Richard Biener  <rguenther@suse.de>
18195         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
18196         of bitmap.
18197         (memory_references): Likewise.
18198         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
18199         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
18200         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
18201         (gather_mem_refs_in_loops): Fold into ...
18202         (analyze_memory_references): ... this.  Move initialization
18203         to tree_ssa_lim_initialize.
18204         (fill_always_executed_in): Rename to ...
18205         (fill_always_executed_in_1): ... this.
18206         (fill_always_executed_in): Move contains_call computation to
18207         this new function from ...
18208         (tree_ssa_lim_initialize): ... here.
18209         (tree_ssa_lim): Call fill_always_executed_in.
18211 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
18213         * postreload.c (reload_combine): Fix code detecting returns.
18215 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
18217         * function.c (emit_use_return_register_into_block): On cc0 targets,
18218         do not emit the sequence between cc0 setter and user.
18220 2013-03-25  Kai Tietz  <ktietz@redhat.com>
18222         * config/i386/predicates.md (local_symbolic_operand): Interpret
18223         dll-imported symbols as none-local.
18225 2013-03-25  Richard Biener  <rguenther@suse.de>
18227         * tree-ssa-loop-im.c (struct depend): Remove.
18228         (struct lim_aux_data): Make depends a vec of gimples.
18229         (free_lim_aux_data): Adjust.
18230         (add_dependency): Likewise.
18231         (set_level): Likewise.
18233 2013-03-25  Richard Biener  <rguenther@suse.de>
18235         PR middle-end/56434
18236         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
18237         the pointer returned by calls with ECF_MALLOC set.
18239 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
18241         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
18243 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
18245         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
18246         using MMXMODE mode iterator.
18247         (*move<mode>_internal): Merge with *movv2sf_internal and
18248         *movv2sf_internal_rex64 using MMXMODE mode iterator.
18250 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
18252         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
18253         (record_last_mem_set_info): Likewise.
18255         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
18256         of XNEWVEC followed by memset.
18257         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
18259 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
18261         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
18262         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
18263         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
18264         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
18265         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
18266         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
18267         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
18268         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
18269         BARRIER_P instead of GET_CODE.
18271 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
18273         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
18274         inaccuracy in the probing code.
18276         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
18277         (ctrapdi4): Likewise.
18279 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
18281         * calls.c (expand_call): Add missing guard to code handling return
18282         of non-BLKmode structures in MSB.
18283         * function.c (expand_function_end): Likewise.
18285 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
18287         * combine.c (try_combine): Adjust comment.  Do not add the set of
18288         insn #0 if the destination indirectly is set or dies in insn #2.
18289         Tidy up code to distribute a new note.
18291 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
18293         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
18294         also for alternatives 16 and 17.
18296 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
18298         * config/i386/sse.md (*mov<mode>_internal): Merge with
18299         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
18300         Emit insn template depending on type attribute.  Use
18301         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
18302         movd instead of movq mnemonic for interunit moves.  Rewrite mode
18303         attribute calculation.  Remove unit attribute calculation.
18304         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
18305         Set prefix_data16 attribute for DImode ssemov types.
18306         Use Ym instead of y for SSE-MMX conversion alternatives.
18307         Reorder operand constraints.
18309 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
18311         * df.h (df_insn_delete): Adjust prototype.
18312         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
18313         and let it decide whether mark the basic block dirty.
18314         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
18315         * df-scan.c (df_insn_info_delete): New helper function, split
18316         off from df_insn_delete.
18317         (df_scan_free_bb_info): Use it.
18318         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
18319         Likewise.
18320         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
18321         that the insn is actually an insn and it has a non-NULL basic block.
18322         Do not mark basic block dirty if only deleting a DEBUG_INSN.
18324 2013-03-22  Richard Biener  <rguenther@suse.de>
18326         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
18327         dep_ref members.
18328         (mem_ref_alloc): Do not allocate them.
18329         (refs_independent_p): Do not query or maintain a cache.
18331 2013-03-22  Richard Biener  <rguenther@suse.de>
18333         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
18334         (gather_mem_refs_in_loops): Do not compute it.
18335         (analyze_memory_references): Do not allocate it.
18336         (tree_ssa_lim_finalize): Do not free it.
18337         (for_all_locs_in_loop): Do not query all_refs_in_loop.
18339 2013-03-22  Richard Biener  <rguenther@suse.de>
18341         * is-a.h (as_a): Use gcc_checking_assert.
18343 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
18345         * config/aarch64/aarch64.c (aarch64_print_operand): New
18346         format specifier for printing a constant in hex.
18347         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
18348         format specifier for printing second operand.
18350 2013-03-22  Richard Biener  <rguenther@suse.de>
18352         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
18353         bitmaps.
18354         (gather_mem_refs_in_loops): Perform store accumulation here.
18355         (create_vop_ref_mapping_loop): Remove.
18356         (create_vop_ref_mapping): Likewise.
18357         (analyze_memory_references): Initialize refs_stored_in_loop.
18358         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
18359         (record_indep_loop): Remove.
18360         (record_dep_loop): New function.
18361         (ref_indep_loop_p_1): Adjust to only walk over references
18362         in the loop, not its subloops.
18363         (ref_indep_loop_p): Rename to ...
18364         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
18365         maintaining a more fine-grained cache.
18366         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
18367         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
18369 2013-03-22  Richard Biener  <rguenther@suse.de>
18371         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
18372         (struct mem_ref): Make accesses_in_loop a vec of a vec of
18373         aggregate mem_ref_loc.
18374         (free_mem_ref_locs): Inline into ...
18375         (memref_free): ... this and adjust.
18376         (mem_ref_alloc): Adjust.
18377         (mem_ref_locs_alloc): Remove.
18378         (record_mem_ref_loc): Adjust.
18379         (get_all_locs_in_loop): Rewrite into ...
18380         (for_all_locs_in_loop): ... this iterator.
18381         (rewrite_mem_ref_loc): New functor.
18382         (rewrite_mem_refs): Use for_all_locs_in_loop.
18383         (sm_set_flag_if_changed): New functor.
18384         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
18385         (ref_always_accessed): New functor.
18386         (ref_always_accessed_p): Use for_all_locs_in_loop.
18388 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
18390         * tree-pass.h (PROP_gimple_lvec): New.
18391         * passes.c (dump_properties): Handle PROP_gimple_lvec.
18392         (init_optimization_passes): Move pass_lower_vector.
18393         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
18394         PROP_gimple_lvec.
18395         (pass_lower_vector): Provide PROP_gimple_lvec.
18396         (pass_lower_vector_ssa): Likewise.
18397         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
18399 2013-03-21  Mark Wielaard  <mjw@redhat.com>
18401         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
18403 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
18405         * config/i386/i386.md (*movdi_internal): Disparage slightly
18406         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
18407         conversion alternatives.
18409 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
18411         PR middle-end/48087
18412         * diagnostic.def (DK_WERROR): New kind.
18413         * diagnostic.h (werrorcount): Define.
18414         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
18415         promoted to DK_ERROR, increment DK_WERROR counter instead of
18416         DK_ERROR counter.
18417         * toplev.c (toplev_main): Call print_ignored_options even if
18418         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
18419         even if just werrorcount is non-zero.
18421         PR debug/55608
18422         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
18423         on failure.
18424         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
18425         (string_cst_pool_decl): New function.
18426         (optimize_one_addr_into_implicit_ptr): New function.
18427         (resolve_addr_in_expr): Optimize DWARF location expression
18428         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
18429         which doesn't live in memory, but has DW_AT_location or
18430         DW_AT_const_value, or refers to a string literal, into
18431         DW_OP_GNU_implicit_pointer.
18432         (optimize_location_into_implicit_ptr): New function.
18433         (resolve_addr): If removing DW_AT_location of a variable because
18434         it was DW_OP_addr of address of the variable, but the variable doesn't
18435         live in memory, try to emit const value attribute for the initializer.
18437 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
18439         * tree.h (VECTOR_TYPE_P): New macro.
18440         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
18441         TYPE_MODE): Use it.
18442         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
18443         VEC_COND_EXPR cannot be lvalues.
18444         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
18446 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
18448         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
18449         Restrict the transformation to equal modes.
18451 2013-03-21  Richard Biener  <rguenther@suse.de>
18453         PR tree-optimization/39326
18454         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
18455         (MEM_ANALYZABLE): Adjust.
18456         (record_mem_ref_loc): Move bitmap ops ...
18457         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
18458         unanalyzable refs, do not record locations for it.
18459         (analyze_memory_references): Allocate ref zero as shared
18460         unanalyzable ref.
18461         (refs_independent_p): Do not test for unanalyzed mems here.
18462         (ref_indep_loop_p_1): Special-case disambiguation against
18463         the unanalyzed ref.
18464         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
18466 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
18468         * config/arm/arm-protos.h (tune_params): Add
18469         prefer_neon_for_64bits field.
18470         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
18471         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
18472         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
18473         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
18474         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
18475         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
18476         (arm_option_override): Handle -mneon-for-64bits new option.
18477         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
18478         (prefer_neon_for_64bits): Declare new variable.
18479         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
18480         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
18481         (arch_enabled): Handle new arch types. Remove support for onlya8
18482         and nota8.
18483         (one_cmpldi2): Use new arch names.
18484         (zero_extend<mode>di2, extend<mode>di2): Ditto.
18485         * config/arm/arm.opt (mneon-for-64bits): Add option.
18486         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
18487         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
18488         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
18489         of onlya8.
18490         * doc/invoke.texi (-mneon-for-64bits): Document.
18492 2013-03-21  Richard Biener  <rguenther@suse.de>
18494         PR tree-optimization/39326
18495         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
18496         (sort_bbs_in_loop_postorder_cmp): New function.
18497         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
18499 2013-03-21  Richard Biener  <rguenther@suse.de>
18501         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
18502         (vect_insert_into_interleaving_chain): Likewise.
18503         (vect_drs_dependent_in_basic_block): Inline ...
18504         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
18505         split out from ...
18506         (vect_analyze_data_ref_dependence): ... here.  Simplify.
18507         (vect_check_interleaving): Simplify.
18508         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
18509         (vect_slp_analyze_data_ref_dependences): ... this new function.
18510         (dr_group_sort_cmp): New function.
18511         (vect_analyze_data_ref_accesses): Compute data-reference groups
18512         here instead of in vect_analyze_data_ref_dependence.  Use
18513         a more efficient algorithm.
18514         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
18515         vect_slp_analyze_data_ref_dependences.  Call
18516         vect_analyze_data_ref_accesses earlier.
18517         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
18518         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
18519         (vect_slp_analyze_data_ref_dependences): New prototype.
18521 2013-03-21  Richard Biener  <rguenther@suse.de>
18523         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
18524         ref is stored in the loop.
18525         (find_refs_for_sm): Walk only over all stores.
18526         (store_motion_loop): Allocate from lim_bitmap_obstack.
18527         (store_motion): Likewise.
18529 2013-03-21  Richard Biener  <rguenther@suse.de>
18531         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
18532         Update virtual SSA form.
18534 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18536         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
18537         * configure: Regenerate.
18538         * config.in: Regenerate.
18539         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
18540         if !HAVE_LD_EH_FRAME_CIEV3.
18542 2013-03-21  Richard Biener  <rguenther@suse.de>
18544         * tree-cfg.c (verify_expr_no_block): New function.
18545         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
18546         nor DECL_VALUE_EXPR have locations with associated blocks.
18547         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
18548         (clear_unused_block_pointer): Remove code dealing with
18549         blocks in DECL_DEBUG_EXPR locations.
18551 2013-03-21  Richard Biener  <rguenther@suse.de>
18553         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
18554         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
18555         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
18556         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
18557         instead of DECL_DEBUG_EXPR_IS_FROM.
18558         * gimplify.c (gimplify_modify_expr): Likewise.
18559         * tree-cfg.c (verify_expr_location_1): Likewise.
18560         * tree-complex.c (create_one_component_var): Likewise.
18561         * tree-sra.c (create_access_replacement): Likewise.
18562         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
18563         (clear_unused_block_pointer): Likewise.
18564         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
18565         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
18566         * var-tracking.c (var_debug_decl): Likewise.
18567         (track_expr_p): Likewise.
18568         * tree-inline.c (add_local_variables): Likewise.  Set
18569         DECL_HAS_DEBUG_EXPR_P after copying it.
18570         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
18571         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
18573 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
18575         PR bootstrap/56656
18576         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
18577         * configure: Regenerate.
18578         * config.in: Regenerate.
18579         * config/i386/i386.md (*movdf_internal): Use
18580         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
18581         movd instead of movq mnemonic for interunit moves.
18582         (*movdi_internal): Ditto.
18584 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18586         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
18587         (abd<mode>_3): New pattern.
18588         (aba<mode>_3): New pattern.
18589         (fabd<mode>_3): New pattern.
18591 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18593         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
18594         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
18595         occurrence of REGISTER_PREFIX as its empty string.
18597 2013-03-20  Jeff Law  <law@redhat.com>
18599         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
18600         addititional equivalences for equality comparisons between an SSA_NAME
18601         and a constant where the SSA_NAME was set from a widening conversion.
18603 2013-03-20  Walter Lee  <walt@tilera.com>
18605         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
18607 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
18609         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
18610         depending on type attribute.
18611         (*movti_internal): Ditto.
18612         (*movtf_internal): Ditto.
18613         (*movxf_internal): Ditto.
18614         (*movdf_internal): Ditto.
18615         (*movsf_internal): Ditto.
18617 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
18619         * config/i386/i386.md (*movti_internal): Set prefix attribute to
18620         maybe_vex for sselog1 and ssemov types.
18621         (*movdi_internal): Reorder operand constraints.
18622         (*movsi_internal): Ditto.  Set prefix attribute to
18623         maybe_vex for sselog1 and ssemov types.
18624         (*movtf_internal): Set prefix attribute to maybe_vex
18625         for sselog1 and ssemov types.
18626         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
18627         DImode ssemov types.  Reorder operand constraints.
18628         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
18629         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
18630         attribute for SImode ssemov types.  Reorder operand constraints.
18632 2013-03-20  Martin Jambor  <mjambor@suse.cz>
18634         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
18635         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
18637 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
18639         * config/rs6000/predicates.md (indexed_address, update_address_mem
18640         update_indexed_address_mem): New predicates.
18641         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
18642         attribute for load/store instructions.
18643         * config/rs6000/dfp.md (movsd_store): Likewise.
18644         (movsd_load): Likewise.
18645         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
18646         (unnamed HI->DI extend define_insn): Likewise.
18647         (unnamed SI->DI extend define_insn): Likewise.
18648         (unnamed QI->SI extend define_insn): Likewise.
18649         (unnamed QI->HI extend define_insn): Likewise.
18650         (unnamed HI->SI extend define_insn): Likewise.
18651         (unnamed HI->SI extend define_insn): Likewise.
18652         (extendsfdf2_fpr): Likewise.
18653         (movsi_internal1): Likewise.
18654         (movsi_internal1_single): Likewise.
18655         (movhi_internal): Likewise.
18656         (movqi_internal): Likewise.
18657         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
18658         attribute for load/store instructions.
18659         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
18660         instructions.
18661         (mov<mode>_softfloat): Likewise.
18662         (mov<mode>_hardfloat32): Likewise.
18663         (mov<mode>_hardfloat64): Likewise.
18664         (mov<mode>_softfloat64): Likewise.
18665         (movdi_internal32): Likewise.
18666         (movdi_internal64): Likewise.
18667         (probe_stack_<mode>): Likewise.
18669 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
18671         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
18672         floating point, and decimal floating point to reload iterator.
18674         * config/rs6000/constraints.md (wl constraint): New constraints to
18675         return FLOAT_REGS if certain options are used to reduce the number
18676         of separate patterns that exist in the file.
18677         (wx constraint): Likewise.
18678         (wz constraint): Likewise.
18680         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18681         -mdebug=reg, print wg, wl, wx, and wz constraints.
18682         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
18683         Initialize the reload functions for 64-bit binary/decimal floating
18684         point types.
18685         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
18686         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
18687         create the buffer on the stack to overcome not having a 32-bit
18688         load and store.
18689         (rs6000_emit_move): Likewise.
18690         (rs6000_secondary_memory_needed_rtx): Likewise.
18691         (rs6000_alloc_sdmode_stack_slot): Likewise.
18692         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
18693         via xxlxor, just like DFmode 0.0.
18695         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
18696         define as 1 if we are running on a power7 or newer.
18697         (enum r6000_reg_class_enum): Add new constraints.
18699         * config/rs6000/dfp.md (movsd): Delete, combine with binary
18700         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
18701         with other moves by using conditional constraits (wg).  Use LFIWZX
18702         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
18703         (movsd splitter): Likewise.
18704         (movsd_hardfloat): Likewise.
18705         (movsd_softfloat): Likewise.
18707         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
18708         binary and decimal floating point moves.
18709         (fmove_ok): New attributes to combine binary and decimal floating
18710         point moves, and to combine power6x (mfpgpr) moves along normal
18711         floating moves.
18712         (real_value_to_target): Likewise.
18713         (f32_lr): Likewise.
18714         (f32_lm): Likewise.
18715         (f32_li): Likewise.
18716         (f32_sr): Likewise.
18717         (f32_sm): Likewise.
18718         (f32_si): Likewise.
18719         (movsf): Combine binary and decimal floating point moves.  Combine
18720         power6x (mfpgpr) moves with other moves by using conditional
18721         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
18722         (mov<mode> for SFmode/SDmode); Likewise.
18723         (SFmode/SDmode splitters): Likewise.
18724         (movsf_hardfloat): Likewise.
18725         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
18726         (movsf_softfloat): Likewise.
18727         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
18729         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
18730         wx and wz constraints.
18732         * config/rs6000/constraints.md (wg constraint): New constraint to
18733         return FLOAT_REGS if -mmfpgpr (power6x) was used.
18735         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
18736         constraint.
18738         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18739         -mdebug=reg, print wg, wl, wx, and wz constraints.
18740         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
18741         Initialize the reload functions for 64-bit binary/decimal floating
18742         point types.
18743         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
18744         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
18745         create the buffer on the stack to overcome not having a 32-bit
18746         load and store.
18747         (rs6000_emit_move): Likewise.
18748         (rs6000_secondary_memory_needed_rtx): Likewise.
18749         (rs6000_alloc_sdmode_stack_slot): Likewise.
18750         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
18751         via xxlxor, just like DFmode 0.0.
18753         * config/rs6000/dfp.md (movdd): Delete, combine with binary
18754         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
18755         with other moves by using conditional constraits (wg).  Use LFIWZX
18756         and STFIWX for loading SDmode on power7.
18757         (movdd splitters): Likewise.
18758         (movdd_hardfloat32): Likewise.
18759         (movdd_softfloat32): Likewise.
18760         (movdd_hardfloat64_mfpgpr): Likewise.
18761         (movdd_hardfloat64): Likewise.
18762         (movdd_softfloat64): Likewise.
18764         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
18765         64-bit binary and decimal floating point moves.
18766         (FMOVE64X): Likewise.
18767         (movdf): Combine 64-bit binary and decimal floating point moves.
18768         Combine power6x (mfpgpr) moves with other moves by using
18769         conditional constraits (wg).
18770         (mov<mode> for DFmode/DDmode): Likewise.
18771         (DFmode/DDmode splitters): Likewise.
18772         (movdf_hardfloat32): Likewise.
18773         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
18774         (movdf_softfloat32): Likewise.
18775         (movdf_hardfloat64_mfpgpr): Likewise.
18776         (movdf_hardfloat64): Likewise.
18777         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
18778         (movdf_softfloat64): Likewise.
18779         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
18780         (reload_<mode>_load): Move to later in the file so they aren't in
18781         the middle of the floating point move insns.
18782         (reload_<mode>_store): Likewise.
18784         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
18785         constraint.
18787         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
18788         constraint if -mdebug=reg.
18789         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
18790         Enable using dd reload support if needed.
18792         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
18793         binary and decimal floating point moves in rs6000.md.
18794         (movtd_internal): Likewise.
18796         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
18797         decimal floating point moves.
18798         (movtf): Likewise.
18799         (movtf_internal): Likewise.
18800         (mov<mode>_internal, TDmode/TFmode): Likewise.
18801         (movtf_softfloat): Likewise.
18802         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
18804         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
18805         movdi_internal64, using wg constraint for move direct operations.
18806         (movdi_internal64): Likewise.
18808         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
18809         MODES_TIEABLE_P for selected modes.  Print the numerical value of
18810         the various virtual registers. Use GPR/FPR first/last values,
18811         instead of hard coding the register numbers.  Print which modes
18812         have reload functions registered.
18813         (rs6000_option_override_internal): If -mdebug=reg, trace the options
18814         settings before/after setting cpu, target and subtarget settings.
18815         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
18816         and for secondary reload failures in rs6000_secondary_reload_inner.
18817         (rs6000_secondary_reload_fail): Likewise.
18818         (rs6000_secondary_reload_inner): Likewise.
18820         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
18821         macros for first/last GPR and FPR registers.
18822         (LAST_GPR_REGNO): Likewise.
18823         (FIRST_FPR_REGNO): Likewise.
18824         (LAST_FPR_REGNO): Likewise.
18826         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
18827         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
18828         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
18829         (vcond<mode><mode>): Likewise.
18830         (vcondu<mode><mode>): Likewise.
18831         (vector_gtu<mode>): Likewise.
18832         (vector_gte<mode>): Likewise.
18833         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
18834         to prevent the compiler from converting DImode operations to TImode.
18835         (ior<mode>3): Likewise.
18836         (and<mode>3): Likewise.
18837         (one_cmpl<mode>2): Likewise.
18838         (nor<mode>3): Likewise.
18839         (andc<mode>3): Likewise.
18841         * config/rs6000/constraints.md (wt constraint): New constraint
18842         that returns VSX_REGS if TImode is allowed in VSX registers.
18844         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
18845         constant under VSX.
18847         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
18848         similar to TImode, but it is restricted to being in the GPRs.
18850         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
18851         TImode to occupy a single VSX register.
18853         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
18854         -mvsx-timode for power7/power8.
18855         (power7 cpu): Likewise.
18856         (power8 cpu): Likewise.
18858         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
18859         sure that TFmode/TDmode take up two registers if they are ever
18860         allowed in the upper VSX registers.
18861         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
18862         registers.
18863         (rs6000_init_hard_regno_mode_ok): Likewise.
18864         (rs6000_debug_reg_global): Add debugging for PTImode and wt
18865         constraint.  Print if LRA is turned on.
18866         (rs6000_option_override_internal): Give an error if -mvsx-timode
18867         and VSX is not enabled.
18868         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
18869         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
18870         to reg+offset addressing.  Use PTImode when checking offset
18871         addresses for validity.
18872         (reg_offset_addressing_ok_p): Likewise.
18873         (rs6000_legitimate_offset_address_p): Likewise.
18874         (rs6000_legitimize_address): Likewise.
18875         (rs6000_legitimize_reload_address): Likewise.
18876         (rs6000_legitimate_address_p): Likewise.
18877         (rs6000_eliminate_indexed_memrefs): Likewise.
18878         (rs6000_emit_move): Likewise.
18879         (rs6000_secondary_reload): Likewise.
18880         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
18881         reloads to fpr registers to continue to use reg+offset addressing,
18882         but 64-bit reloads to altivec registers need reg+reg addressing.
18883         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
18884         it.  Treat LO_SUM like a PLUS operation.
18885         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
18886         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
18887         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
18888         registers to share a register with a smaller sized type, since VSX
18889         puts scalars in the upper 64-bits.
18890         (print_operand): Add support for PTImode.
18891         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
18892         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
18893         registers, but don't have arithmetic support.
18894         (rs6000_memory_move_cost): Add test for VSX.
18895         (rs6000_opt_masks): Add -mvsx-timode.
18897         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
18898         for TImode.
18899         (VSs): Likewise.
18900         (VSr): Use wt constraint for TImode.
18901         (VSv): Drop TImode support.
18902         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
18903         (vsx_movti_64bit): Likewise.
18904         (vsx_movti_32bit): Likewise.
18905         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
18906         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
18907         one '?' on the appropriate output constraint.  Do not allow TImode
18908         logical operations on 32-bit systems.
18909         (vsx_ior<mode>3): Likewise.
18910         (vsx_xor<mode>3): Likewise.
18911         (vsx_one_cmpl<mode>2): Likewise.
18912         (vsx_nor<mode>3): Likewise.
18913         (vsx_andc<mode>3): Likewise.
18914         (vsx_concat_<mode>): Likewise.
18915         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
18917         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
18918         OPTION_MASK_VSX_TIMODE.
18919         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
18920         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
18922         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
18923         (TI2 iterator): New iterator for TImode, PTImode.
18924         (wd mode attribute): Add values for vector types.
18925         (movti_string): Replace TI move operations with operations for TImode
18926         and PTImode.  Add support for TImode being allowed in VSX registers.
18927         (mov<mode>_string, TImode/PTImode): Likewise.
18928         (movti_ppc64): Likewise.
18929         (mov<mode>_ppc64, TImode/PTImode): Likewise.
18930         (TI mode splitters): Likewise.
18932         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
18933         constraint.
18935 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
18937         PR tree-optimization/56355
18938         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
18939         Also handle integers with undefined overflow.
18941 2013-03-20  Catherine Moore  <clm@codesourcery.com>
18942             Maciej W. Rozycki  <macro@codesourcery.com>
18943             Tom de Vries  <tom@codesourcery.com>
18944             Nathan Sidwell  <nathan@codesourcery.com>
18945             Iain Sandoe  <iain@codesourcery.com>
18946             Nathan Froyd  <froydnj@codesourcery.com>
18947             Chao-ying Fu  <fu@mips.com>
18949         * doc/extend.texi: (micromips, nomicromips, nocompression):
18950         Document new function attributes.
18951         * doc/invoke.texi (minterlink-compressed, mmicromips,
18952         m14k, m14ke, m14kec): Document new options.
18953         (minterlink-mips16): Update documentation.
18954         * doc/md.texi (ZC, ZD): Document new constraints.
18955         * configure.ac (gcc_cv_as_micromips): Check if linker
18956         supports the .set micromips directive.
18957         * configure: Regenerate.
18958         * config.in: Regenerate.
18959         * config/mips/mips-tables.opt: Regenerate.
18960         * config/mips/micromips.md: New file.
18961         * constraints.md (ZC, ZD): New constraints.
18962         * config/mips/predicates.md (movep_src_register): New predicate.
18963         (movep_src_operand): New predicate.
18964         (non_volatile_mem_operand): New predicate.
18965         * config/mips/mips.md (multimem): New type.
18966         (length): Differentiate between 17-bit and 18-bit branch offsets.
18967         (MOVEP1, MOVEP2): New mode iterator.
18968         (mov_<load>l): Use ZC constraint.
18969         (mov_<load>r): Likewise.
18970         (mov_<store>l): Likewise.
18971         (mov_<store>r): Likewise.
18972         (*branch_equality<mode>_inverted): Add microMIPS support.
18973         (*branch_equality<mode>): Likewise.
18974         (*jump_absolute): Likewise.
18975         (indirect_jump_<mode>): Likewise.
18976         (tablejump_<mode>): Likewise.
18977         (<optab>_internal): Likewise.
18978         (sibcall_internal): Likewise.
18979         (sibcall_value_internal): Likewise.
18980         (prefetch): Use constraint ZD.
18981         * config/mips/mips.opt (minterlink-compressed): New option.
18982         (minterlink-mips16): Now an alias for minterlink-compressed.
18983         (mmicromips): New option.
18984         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18985         (compare_and_swap_12): Likewise.
18986         (sync_add<mode>): Likewise.
18987         (sync_<optab>_12): Likewise.
18988         (sync_old_<optab>_12): Likewise.
18989         (sync_new_<optab>_12): Likewise.
18990         (sync_nand_12): Likewise.
18991         (sync_old_nand_12): Likewise.
18992         (sync_new_nand_12): Likewise.
18993         (sync_sub<mode>): Likewise.
18994         (sync_old_add<mode>): Likewise.
18995         (sync_old_sub<mode>): Likewise.
18996         (sync_new_add<mode>): Likewise.
18997         (sync_new_sub<mode>): Likewise.
18998         (sync_<optab><mode>): Likewise.
18999         (sync_old_<optab><mode>): Likewise.
19000         (sync_new_<optab><mode>): Likewise.
19001         (sync_nand<mode>): Likewise.
19002         (sync_old_nand<mode>): Likewise.
19003         (sync_new_nand<mode>): Likewise.
19004         (sync_lock_test_and_set<mode>): Likewise.
19005         (test_and_set_12): Likewise.
19006         (atomic_compare_and_swap<mode>): Likewise.
19007         (atomic_exchange<mode>_llsc): Likewise.
19008         (atomic_fetch_add<mode>_llsc): Likewise.
19009         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19010         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19011         (umips_save_restore_pattern_p): Likewise.
19012         (umips_load_store_pair_p): Likewise.
19013         (umips_output_load_store_pair): Likewise.
19014         (umips_movep_target_p): Likewise.
19015         (umips_12bit_offset_address_p): Likewise.
19016         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19017         (mips_base_mips16): Rename this...
19018         (mips_base_compression_flags): ...to this. Update all uses.
19019         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19020         (mips_mips16_decl_p): Delete.
19021         (mips_nomips16_decl_p): Delete.
19022         (mips_get_compress_on_flags): New function.
19023         (mips_get_compress_off_flags): New function.
19024         (mips_get_compress_mode): New function.
19025         (mips_get_compress_on_name): New function.
19026         (mips_get_compress_off_name): New function.
19027         (mips_insert_attributes): Support multiple compression types.
19028         (mips_merge_decl_attributes): Likewise.
19029         (umips_12bit_offset_address_p): New function.
19030         (mips_start_function_definition): Emit .set micromips directive.
19031         (mips_call_may_need_jalx_p): New function.
19032         (mips_function_ok_for_sibcall): Add microMIPS support.
19033         (mips_print_operand_punctuation): Support short delay slots and
19034         compact jumps.
19035         (umips_swm_mask, umips_swm_encoding): New.
19036         (umips_build_save_restore): New function.
19037         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19038         (was_mips16_p): Remove.
19039         (old_compression_mode): New.
19040         (mips_set_compression_mode): New function.
19041         (mips_set_current_function): Add microMIPS support.
19042         (mips_option_override): Likewise.
19043         (umips_save_restore_pattern_p): New function.
19044         (umips_output_save_restore): New function.
19045         (umips_load_store_pair_p_1): New function.
19046         (umips_load_store_pair_p): New function.
19047         (umips_output_load_store_pair_1): New function.
19048         (umips_output_load_store_pair): New function.
19049         (umips_movep_target_p) New function.
19050         (mips_prepare_pch_save): Add microMIPS support.
19051         * config/mips/mips.h (TARGET_COMPRESSION): New.
19052         (TARGET_CPU_CPP_BUILTINS): Update macro
19053         to use new compression flags and to support microMIPS.
19054         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19055         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19056         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19057         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19058         (ASM_SPEC): Support mmicromips and mno-micromips.
19059         (M16STORE_REG_P): New macro.
19060         (MIPS_CALL): Support TARGET_MICROMIPS.
19061         (MICROMIPS_J): New macro.
19062         (mips_base_mips16): Rename this...
19063         (mips_base_compression_flags): ...to this.
19064         (UMIPS_12BIT_OFFSET_P): New macro.
19065         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19066         (MULTILIB_DIRNAMES): Likewise.
19067 2013-03-20  Richard Biener  <rguenther@suse.de>
19069         PR tree-optimization/56661
19070         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
19071         the result does not have to be distinct.
19073 2013-03-20  Richard Biener  <rguenther@suse.de>
19075         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
19076         remap_gimple_op_r.
19078 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19079             Steven Bosscher  <steven@gcc.gnu.org>
19081         PR rtl-optimization/56605
19082         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
19084 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
19086         PR bootstrap/56656
19087         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
19088         that require movd instead of movq.
19090 2013-03-20  Richard Biener  <rguenther@suse.de>
19092         * tree-ssa-structalias.c (struct variable_info): Add pointer
19093         to the first field of an aggregate with sub-vars.  Make
19094         this and the pointer to the next subfield its ID.
19095         (vi_next): New function.
19096         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
19097         storedanything_id, integer_id): Increment by one.
19098         (new_var_info, get_call_vi, lookup_call_clobber_vi,
19099         get_call_clobber_vi): Adjust.
19100         (solution_set_expand): Simplify and speedup.
19101         (solution_set_add): Inline into ...
19102         (set_union_with_increment): ... this.  Adjust accordingly.
19103         (do_sd_constraint): Likewise.
19104         (do_ds_constraint): Likewise.
19105         (do_complex_constraint): Simplify.
19106         (build_pred_graph): Adjust.
19107         (solve_graph): Likewise.  Simplify and speedup.
19108         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
19109         get_constraint_for_component_ref, get_constraint_for_1,
19110         first_vi_for_offset, first_or_preceding_vi_for_offset,
19111         create_function_info_for, create_variable_info_for_1,
19112         create_variable_info_for, intra_create_variable_infos): Adjust.
19113         (init_base_vars): Push NULL for ID zero.
19114         (compute_points_to_sets): Adjust.
19116 2013-03-20  Richard Biener  <rguenther@suse.de>
19118         * cfgloop.c (verify_loop_structure): Streamline and avoid
19119         ICEing on corrupt loop tree.
19120         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
19121         loop tree.
19123 2013-03-20  Richard Biener  <rguenther@suse.de>
19125         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
19126         check whether an SSA update is needed.
19128 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
19130         * config/mips/constraints.md (T): Rename to...
19131         (Yf): ...this.
19132         (U): Rename to...
19133         (Yd): ...this.
19134         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
19135         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
19137 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
19139         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
19140         (*subsi3_carryin_uxtw): Likewise.
19142 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
19144         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
19145         (*rorsi3_insn_uxtw): Likewise.
19147 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
19149         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
19150         (*extrsi5_insn_uxtw): Likewise.
19152 2013-03-19  Richard Biener  <rguenther@suse.de>
19154         PR tree-optimization/56273
19155         * passes.c (init_optimization_passes): Move second VRP after DOM.
19157 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
19159         * config/i386/i386.md (*movti_internal): Merge from
19160         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
19161         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
19162         nox64 isa attributes.
19164 2013-03-18  Richard Biener  <rguenther@suse.de>
19166         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
19167         (unite): Likewise.
19168         (merge_node_constraints): Likewise.
19169         (build_succ_graph): Likewise.
19170         (valid_graph_edge): Inline into single caller.
19171         (unify_nodes): Likewise.  Use bitmap_set_bit return value
19172         and cache varinfo.
19173         (scc_visit): Fix formatting and variable use.
19174         (do_sd_constraint): Use gcc_checking_assert.
19175         (do_ds_constraint): Likewise.
19176         (do_complex_constraint): Likewise.
19177         (condense_visit): Likewise.  Cleanup.
19178         (dump_pred_graph): New function.
19179         (perform_var_substitution): Dump the pred-graph before
19180         variable substitution.
19181         (find_equivalent_node): Use gcc_checking_assert.
19182         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
19184 2013-03-18  Richard Biener  <rguenther@suse.de>
19186         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
19187         Remove cond_expr_stmt_list argument and do not gimplify the
19188         built expression.
19189         (vect_loop_versioning): Adjust.
19190         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
19191         Cleanup to use less temporaries.
19192         (vect_create_data_ref_ptr): Cleanup.
19194 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
19196         PR tree-optimization/56635
19197         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
19198         require types_compatible_p types.
19200 2013-03-18  Nick Clifton  <nickc@redhat.com>
19202         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
19203         spurious backslash.
19205         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
19206         Add missing line to comment describing function.
19208 2013-03-18  Richard Biener  <rguenther@suse.de>
19210         PR tree-optimization/56210
19211         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19212         Handle string / character search functions.
19213         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
19215 2013-03-18  Richard Biener  <rguenther@suse.de>
19217         PR middle-end/56483
19218         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
19219         and implement properly.
19220         * gimple.h (gimple_cond_single_var_p): Remove.
19222 2013-03-18  Richard Biener  <rguenther@suse.de>
19224         * tree-data-ref.h (find_data_references_in_loop): Declare.
19225         * tree-data-ref.c (get_references_in_stmt): Use a stack
19226         vector pre-allocated in the callers.
19227         (find_data_references_in_stmt): Adjust.
19228         (graphite_find_data_references_in_stmt): Likewise.
19229         (create_rdg_vertices): Likewise.
19230         (find_data_references_in_loop): Export.
19231         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
19232         Compute dependences here...
19233         (vect_analyze_data_refs): ...not here.  When we encounter
19234         a non-vectorizable data reference in basic-block vectorization
19235         truncate the data reference vector.  Do not bother to
19236         fixup data-dependence information for gather loads.
19237         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
19238         of data references, as reported.
19240 2013-03-18  Richard Biener  <rguenther@suse.de>
19242         PR tree-optimization/3713
19243         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
19244         has_constants and expr.
19245         (stmt_has_constants): Properly valueize SSA names when deciding
19246         whether the stmt has constants.
19248 2013-03-18  Richard Biener  <rguenther@suse.de>
19250         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
19251         whole function when there is nothing to do.
19252         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
19253         * tree-vectorizer.c (vectorize_loops): Update virtual and
19254         loop-closed SSA once.
19255         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
19257 2013-03-18  Richard Biener  <rguenther@suse.de>
19259         PR middle-end/56113
19260         * domwalk.c (bb_postorder): New global static.
19261         (cmp_bb_postorder): New function.
19262         (walk_dominator_tree): Replace scheme imposing an order for
19263         visiting dominator sons by one sorting them at the time they
19264         are pushed on the stack.
19266 2013-03-18  Richard Biener  <rguenther@suse.de>
19268         PR tree-optimization/39326
19269         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
19270         (struct mem_ref): Replace mem member with ao_ref typed member.
19271         (MEM_ANALYZABLE): Adjust.
19272         (memref_eq): Likewise.
19273         (mem_ref_alloc): Likewise.
19274         (gather_mem_refs_stmt): Likewise.
19275         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
19276         (execute_sm_if_changed_flag_set): Adjust.
19277         (execute_sm): Likewise.
19278         (ref_always_accessed_p): Likewise.
19279         (refs_independent_p): Likewise.
19280         (can_sm_ref_p): Likewise.
19282 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
19284         PR c/56566
19285         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
19286         return 1 even for !unsignedp.
19288 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19290         * config/i386/i386.md (isa): Add x64 and nox64.
19291         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
19292         (*pushtf): Enable *roF alternative for x64 isa only.
19293         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
19294         mode attribute of integer alternatives to DImode for TARGET_64BIT.
19295         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
19296         (*movtf_internal): Merge from *movtf_internal_rex64 and
19297         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
19298         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
19299         nox64 isa attributes.
19300         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
19301         nox64 isa attributes.
19302         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
19304 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19306         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19308 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
19310         PR target/56640
19311         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
19312         class names.  Remove trailing comma after "ALL_REGS".
19314 2013-03-16  Jan Hubicka  <jh@suse.cz>
19316         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
19317         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
19318         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
19319         of cgraph_get_create_node.
19320         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
19322 2013-03-16  Jason Merrill  <jason@redhat.com>
19324         PR debug/49090
19325         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
19326         with DW_AT_default_value.
19328 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
19330         * BASE-VER: Set to 4.9.0.
19332 2013-03-14  Andi Kleen  <ak@linux.intel.com>
19334         PR target/56619
19335         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
19336         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
19337         Document _x* TSX intrinsics.
19339 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
19340             David Holsgrove  <david.holsgrove@xilinx.com>
19342         * configure.ac: Add MicroBlaze TLS support detection.
19343         * configure: Regenerate.
19344         * config/microblaze/microblaze-protos.h
19345         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
19346         symbol_mentioned_p, label_mentioned_p): Add prototypes.
19347         * config/microblaze/microblaze.c (microblaze_address_type): Add
19348         ADDRESS_TLS and tls_reloc address types.
19349         (microblaze_address_info): Add tls_reloc.
19350         (TARGET_HAVE_TLS): Define.
19351         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
19352          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
19353          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
19354         load_tls_operand,  microblaze_call_tls_get_addr,
19355         microblaze_legitimize_tls_address): New functions.
19356         (microblaze_classify_unspec): Handle UNSPEC_TLS.
19357         (get_base_reg): Use microblaze_tls_symbol_p.
19358         (microblaze_classify_address): Handle TLS.
19359         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
19360         label_mentioned_p and microblaze_tls_referenced_p.
19361         (microblaze_legitimize_address): Handle TLS.
19362         (microblaze_address_insns): Handle ADDRESS_TLS.
19363         (pic_address_needs_scratch): Handle TLS.
19364         (print_operand_address): Handle TLS.
19365         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
19366         (microblaze_expand_move): Handle TLS.
19367         (microblaze_legitimate_constant_p): Check
19368         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
19369         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
19370         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
19371         (PIC_OFFSET_TABLE_REGNUM): Set.
19372         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
19373         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
19374         (addsi3, movsi_internal2, movdf_internal): Update constraints
19375         * config/microblaze/predicates.md (arith_plus_operand): Define
19376         (move_operand): Redefine as move_src_operand,
19377         check microblaze_tls_referenced_p.
19379 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
19381         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
19382         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
19384 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
19386         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
19387         CC mode for AND.
19389 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
19391         PR tree-optimization/53265
19392         * common.opt (Waggressive-loop-optimizations): New option.
19393         * tree-ssa-loop-niter.c: Include tree-pass.h.
19394         (do_warn_aggressive_loop_optimizations): New function.
19395         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
19396         if number_of_latch_executions returned constant.
19397         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
19398         early.  If number_of_latch_executions returned constant, set
19399         nb_iterations_upper_bound back to it.
19400         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
19401         field.
19402         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
19403         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
19405         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
19406         (MULTILIB_OSDIRNAMES): Set.
19407         * genmultilib: If defaultosdirname doesn't start with :: , set
19408         defaultosdirname2 instead, clear it and emit two . multilib_raw
19409         entries instead of just one.
19411 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
19413         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
19414         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
19415         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
19416         (SUBTARGET_OVERRIDE_OPTIONS): New.
19418 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
19420         PR target/49880
19421         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
19422         (musermode): Convert to Var(TARGET_USERMODE).
19423         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
19424         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
19425         * config/sh/sh.c (sh_option_override): Use
19426         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
19427         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
19428         condition.
19429         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
19430         TARGET_SH4.
19431         (udivsi3_i4_single, divsi3_i4_single): Use
19432         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
19434 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
19436         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
19437         default setting.
19439 2013-03-13  Richard Biener  <rguenther@suse.de>
19441         PR tree-optimization/56608
19442         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
19443         calls when vectorizing basic-blocks.
19445 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
19447         PR plugins/45078
19448         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
19449         tm_file.
19451 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
19453         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
19455 2013-03-11  Jan Hubicka  <jh@suse.cz>
19457         PR lto/56557
19458         * lto-streamer-out.c (output_symbol_p): Skip references from
19459         constructors of external variables.
19461 2013-03-11  Jan Hubicka  <jh@suse.cz>
19463         PR middle-end/56571
19464         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
19465         from pseudos.
19466         * emit-rtl.c (verify_rtx_sharing): Likewise.
19467         (copy_insn_1): Likewise.
19468         * rtl.c (copy_rtx): Likewise.
19470 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
19472         PR target/56591
19473         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
19474         output_operand_lossage message.
19476 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
19478         PR target/56470
19479         * arm.c (shift_op): Validate RTL pattern on the fly.
19480         (arm_print_operand, case 'S'): Don't use shift_operator to validate
19481         the RTL.
19483 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19485         PR target/56347
19486         * config/pa/pa.md (call_value): Check for calls to powf and direct to
19487         new call patterns that clobber %fr12.
19488         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
19489         split and postreload patterns.
19490         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
19491         registers %fr12 and %fr12R as call used.
19493 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
19495         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
19496         (canon_address, record_store, replace_read, check_mem_read_rtx,
19497         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
19498         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
19499         rest_of_handle_dse): Likewise.
19501 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
19503         PR middle-end/56524
19504         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
19505         Add base_optabs.
19506         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
19507         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
19508         (save_optabs_if_changed): Replace with...
19509         (init_tree_optimization_optabs): ...this.
19510         * optabs.c (save_optabs_if_changed): Rename to...
19511         (init_tree_optimization_optabs): ...this.  Take the optimization node
19512         as argument.  Do nothing if the base optabs are already correct.
19513         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
19514         to recompute optabs.
19515         * function.h (function): Remove optabs field.
19516         * function.c (invoke_set_current_function_hook): Call
19517         init_tree_optimization_optabs.  Use the result to initialize
19518         this_fn_optabs.
19520 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
19522         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
19523         if GTMA_HAS_NO_INSTRUMENTATION.
19524         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
19525         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
19526         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
19527         * gimple-pretty-print.c (dump_gimple_transaction): Handle
19528         GTMA_HAS_NO_INSTRUMENTATION.
19530 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
19532         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
19533         libasan_preinit.o.
19535 2013-03-08  Marek Polacek  <polacek@redhat.com>
19536             Jakub Jelinek  <jakub@redhat.com>
19538         PR tree-optimization/56478
19539         * predict.c (is_comparison_with_loop_invariant_p): Change the
19540         type of loop_step to tree.
19541         (predict_loops): Adjust.
19542         (predict_iv_comparison): Perform the computations on double_ints.
19544 2013-03-08  Richard Biener  <rguenther@suse.de>
19546         PR tree-optimization/56570
19547         * tree-cfg.c (verify_expr_location_1): Verify locations for
19548         DECL_DEBUG_EXPR.
19549         * tree-sra.c (create_access_replacement): Strip locations
19550         from DECL_DEBUG_EXPRs.
19552 2013-03-08  Richard Biener  <rguenther@suse.de>
19554         * tree-inline.c (expand_call_inline): Do not associate
19555         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
19556         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
19558 2013-03-08  Richard Biener  <rguenther@suse.de>
19560         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
19561         or block changes with -Og.  Fix for location / block encoding
19562         changes and PHI arguments with locations.
19564 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
19566         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
19567         for all counters.
19568         (struct output_info): Likewise.
19569         (register_overhead): Remove bad gcc_assert.
19570         (bitmap_find_bit): If there is only a single bitmap element, do not
19571         count a miss as a search.
19572         (print_statistics): Update for counter type changes.
19573         (dump_bitmap_statistics): Likewise.  Print headers such that they
19574         are properly lined up with the printed counters.
19576 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
19578         PR tree-optimization/56559
19579         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
19580         check that it has only a single use.
19582 2013-03-07  Richard Biener  <rguenther@suse.de>
19584         * doc/invoke.texi (fwhole-program): Discourage use in combination
19585         with -flto.
19587 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
19589         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
19591         PR tree-optimization/56539
19592         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
19593         instead of GSI_CONTINUE_LINKING as last argument to
19594         force_gimple_operand_gsi.  Adjust function comment.
19596         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
19597         aarch64-cores.def.
19599         PR middle-end/56548
19600         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
19601         promoted mode, convert the result back to the original mode.
19603 2013-03-06  Richard Biener  <rguenther@suse.de>
19605         PR middle-end/56294
19606         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
19607         (insert_updated_phi_nodes_compare_uids): New function.
19608         (update_ssa): Sort symbols_to_rename after UID before
19609         traversing it to insert PHI nodes.
19611 2013-03-06  Richard Biener  <rguenther@suse.de>
19613         PR middle-end/50494
19614         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
19615         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
19617         Revert
19618         2013-02-13  Richard Biener  <rguenther@suse.de>
19620         PR lto/50494
19621         * varasm.c (output_constant_def_1): Get the decl representing
19622         the constant as argument.
19623         (output_constant_def): Wrap output_constant_def_1.
19624         (make_decl_rtl): Use output_constant_def_1 with the decl
19625         representing the constant.
19626         (build_constant_desc): Optionally re-use a decl already
19627         representing the constant.
19628         (tree_output_constant_def): Adjust.
19630 2013-03-06  Joey Ye  <joey.ye@arm.com>
19632         PR lto/50293
19633         * gcc.c (convert_white_space): New function.
19634         (main): Handles white space in function name.
19636 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
19638         PR target/56529
19639         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
19640         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
19641         to SH_DIV_CALL_TABLE for TARGET_SH2.
19642         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
19643         list.
19644         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
19645         call-table options.
19647 2013-03-05  Sterling Augustine  <saugustine@google.com>
19648             Cary Coutant  <ccoutant@google.com>
19650         PR debug/55364
19651         * dwarf2out.c (resolve_addr): Don't call
19652         remove_loc_list_addr_table_entries a second time for the same
19653         expression.
19655 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19657         PR debug/56510
19658         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
19659         (avoid_complex_debug_insns): New function.
19660         (expand_debug_locations): Call it.
19662         PR rtl-optimization/56484
19663         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
19664         lifetimes of hard registers on small register class machines.
19666 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
19668         * config/microblaze/microblaze-protos.h: Rename
19669         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
19670         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
19671         fast_interrupt.
19672         (microblaze_fast_interrupt_function_p): New function.
19673         (microblaze_is_interrupt_handler): Rename to
19674         microblaze_is_interrupt_variant and add fast_interrupt check.
19675         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
19676         (save_restore_insns): Likewise.
19677         (compute_frame_size): Likewise.
19678         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
19679         (microblaze_globalize_label): Likewise.
19680         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
19681         * config/microblaze/microblaze.md: Use wrapper
19682         microblaze_is_interrupt_variant.
19684 2013-03-05  Kai Tietz  <ktietz@redhat.com>
19686         * sdbout.c (sdbout_one_type): Switch to current function's section
19687         supporting cold/hot.
19689 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
19691         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
19692         -mxl-reorder.
19694 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19696         PR middle-end/56461
19697         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
19698         if VALGRIND_GET_VBITS is defined, temporarily make object
19699         memory all defined, and restore previous valgrind addressability
19700         and definability afterwards.  Free this_object at the end.
19702         PR middle-end/56461
19703         * lra.c (lra): Call lra_clear_live_ranges if live_p,
19704         right before calling lra_create_live_ranges, also call it
19705         when clearing live_p.  Only call lra_clear_live_ranges
19706         at the end if live_p.
19708         PR middle-end/56461
19709         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
19711 2013-03-05  Richard Biener  <rguenther@suse.de>
19713         PR tree-optimization/56521
19714         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
19715         value-id.
19717 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
19719         PR c++/55135
19720         * except.h (remove_unreachable_eh_regions): New prototype.
19721         * except.c (remove_eh_handler_splicer): New function, split out
19722         of remove_eh_handler.
19723         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
19724         warning about running it on many EH regions one at a time.
19725         (remove_unreachable_eh_regions_worker): New function, walk the
19726         EH tree in depth-first order and remove non-marked regions.
19727         (remove_unreachable_eh_regions): New function.
19728         * tree-eh.c (mark_reachable_handlers): New function, split out
19729         from remove_unreachable_handlers.
19730         (remove_unreachable_handlers): Use mark_reachable_handlers and
19731         remove_unreachable_eh_regions.
19732         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
19733         and remove_unreachable_eh_regions.
19735 2013-03-05  Richard Biener  <rguenther@suse.de>
19737         PR middle-end/56525
19738         * loop-init.c (fix_loop_structure): Remove loops in two stages,
19739         not freeing them until the end.
19741 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19743         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
19745 2013-03-05  Richard Biener  <rguenther@suse.de>
19747         PR tree-optimization/56270
19748         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
19749         of loads after scheduling an SLP instance.
19751 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19753         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
19754         tic6x.exp.
19755         (check_gcc_parallelize): Run guality.exp as a separate job from
19756         vect.exp with unsorted.exp and $(dg_target_exps) separately from
19757         struct-layout-1.exp with stackalign.exp.
19759         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
19761         PR middle-end/56461
19762         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
19763         load_index sbitmap even if some bit in it isn't set.
19765         PR middle-end/56461
19766         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
19767         (discover_iteration_bound_by_body_walk): Change queues to
19768         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
19769         spelling in comment.  Call safe_push on queues[bound_index] directly.
19770         Release queues[queue_index] in every iteration unconditionally.
19771         Release bounds vector.
19773         PR middle-end/56461
19774         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
19775         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
19776         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
19777         inner_phis vector.
19779 2013-03-05  Richard Biener  <rguenther@suse.de>
19781         PR lto/56515
19782         * tree-inline.c (remap_blocks_to_null): New function.
19783         (expand_call_inline): When expanding a call stmt without
19784         an associated block inline remap all callee blocks to NULL.
19786 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19788         PR rtl-optimization/56494
19789         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
19790         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
19791         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
19793         PR middle-end/56461
19794         * sel-sched-ir.c (free_sched_pools): Release
19795         succs_info_pool.stack[succs_info_pool.max_top] vectors too
19796         if succs_info_pool.max_top isn't -1.
19798         PR bootstrap/56509
19799         * opts.c (opts_obstack, opts_concat): Moved to...
19800         * opts-common.c (opts_obstack, opts_concat): ... here.
19802 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
19804         PR middle-end/56461
19805         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
19807 2013-03-04  Martin Jambor  <mjambor@suse.cz>
19809         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
19810         all appropriate places.
19812 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
19814         PR tree-optimization/56424
19815         * ipa-split.c (split_function): Do not set the RSO flag if result is
19816         not by reference and its type is a register type.
19818 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
19820         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
19821         (microblaze_legitimate_pic_operand): Likewise
19822         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
19823         new function microblaze_legitimate_pic_operand
19824         * config/microblaze/microblaze-protos.h
19825         (microblaze_legitimate_pic_operand): Declare.
19827 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19829         * config/microblaze/predicates.md (call_insn_simple_operand):
19830         New predicate for supported rtx code types.
19831         * config/microblaze/microblaze.md (call_internal1): Use
19832         call_insn_simple_operand predicate.
19834 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
19836         PR middle-end/56461
19837         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
19838         partitions.ordered_remove.
19840         PR middle-end/56461
19841         * tree-vect-stmts.c (vectorizable_conversion): Don't call
19842         vec_oprnds0.create (1) for modifier == NONE.
19844         PR middle-end/56461
19845         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
19846         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
19847         vec_oprnds1 right before pushing anything to it for
19848         scalar_shift_arg.
19850         PR middle-end/56461
19851         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
19852         set nbbs to 0 instead of having separate code path.
19853         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
19854         instead of false as last argument if returning NULL.
19856 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
19858         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
19859         the attribute is now called "target" instead of "option".
19860         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
19861         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
19862         attribute/pragma name for TARGET_OPTION_VALID_P and
19863         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
19864         * doc/tm.texi: Regenerated.
19866 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
19868         * config/microblaze/microblaze.c:
19869         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
19870         * config/microblaze/microblaze.h: Add -mxl-reorder to
19871         DRIVER_SELF_SPECS.
19872         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
19873         instructions emitted if TARGET_REORDER.
19874         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
19875         or 0 for -m/-mno case, but initialises as 2 to detect default use case
19876         separately.
19878 2013-03-01  Xinliang David Li  <davidxl@google.com>
19880         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
19881         walk length.
19883 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
19885         PR middle-end/56461
19886         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
19887         vector even when returning true.  Fix up function comment formatting.
19889         PR middle-end/56461
19890         * ira-build.c (ira_loop_nodes_count): New variable.
19891         (create_loop_tree_nodes): Initialize it.
19892         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
19894         PR middle-end/56461
19895         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
19896         method on dr_chain and result_chain.
19897         * tree-vect-stmts.c (vectorizable_store): Only call
19898         result_chain.create if j == 0.
19900         PR middle-end/56461
19901         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
19902         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
19903         before overwriting it.
19905 2013-03-01  Tobias Burnus  <burnus@net-b.de>
19907         * doc/extended.texi (C Extensions): Change order in @menu
19908         to match @node.
19909         (Other MIPS Built-in Functions): Move last MIPS entry before
19910         "picoChip Built-in Functions".
19911         (SH Built-in Functions): Move after RX Built-in Functions.
19912         * doc/gcc.texi (Introduction): Change order in @menu
19913         to match @node.
19914         * doc/md.texi (Constraints): Ditto.
19915         * gty.texi (Type Information): Ditto.
19916         (User-provided marking routines for template types): Make
19917         subsection.
19918         * doc/invoke.texi (AArch64 Options): Move before
19919         "Adapteva Epiphany Options".
19921 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
19922             Jakub Jelinek  <jakub@redhat.com>
19924         PR sanitizer/56454
19925         * asan.c (gate_asan): Lookup no_sanitize_address instead of
19926         no_address_safety_analysis attribute.
19927         * doc/extend.texi (no_address_safety_attribute): Rename to
19928         no_sanitize_address attribute, mention no_address_safety_analysis
19929         attribute as deprecated alias.
19931 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
19933         PR middle-end/56461
19934         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
19935         type to vec<vec<tree> > *.
19936         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
19937         to be vec<tree> instead of vec<tree> *, set vec_defs
19938         to vNULL and call vec_defs.create (number_of_vects), adjust other
19939         uses of vec_defs.
19940         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
19941         vectorizable_condition): Adjust vect_get_slp_defs callers.
19943 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19945         * config/aarch64/aarch64.c
19946         (aarch64_float_const_representable): Remove unused variable.
19948 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19950         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
19952 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19954         * config/aarch64/aarch64-builtins.c
19955         (aarch64_init_simd_builtins): Make static.
19957 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19959         * config/aarch64/aarch64.c
19960         (aarch64_simd_make_constant): Make static.
19962 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19964         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
19965         with no initialization to the RHS of debug statements.
19967 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19969         PR tree-optimization/56294
19970         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
19971         Adjust dumping.
19972         (get_access_replacement): Do not call create_access_replacement.
19973         Assert a replacement exists.
19974         (get_repl_default_def_ssa_name): Create the replacement declaration
19975         itself.
19977 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19979         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
19980         final_end_function.
19982 2013-02-28  Marek Polacek  <polacek@redhat.com>
19984         PR rtl-optimization/56466
19985         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
19986         if we're changing a loop.
19987         (peel_loops_completely): Likewise.
19989 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
19991         PR c++/55813
19992         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
19994 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19996         PR target/56445
19997         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
19998         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
19999         INTX_FTYPE_FX, FX_FTYPE_INTX.
20000         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
20002 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
20004         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
20005         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
20006         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
20007         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
20008         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
20009         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
20010         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
20011         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
20012         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
20013         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
20014         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
20015         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
20016         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
20017         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
20018         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
20019         (avrxmega6): Increase max flash segments from 5 to 6.
20020         * config/avr/t-multilib: Regenerate.
20021         * config/avr/avr-tables.opt: Regenerate.
20022         * doc/avr-mmcu.texi: Regenerate.
20024 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
20026         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
20027         (avr_device_to_arch): Rename to avr_device_to_ld.
20028         (avr_device_to_as): New prototype.
20029         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
20030         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
20031         * config/avr/driver-avr.c (avr_device_to_as): New.
20032         (avr_device_to_arch): Rename to avr_device_to_ld.
20034 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
20036         PR middle-end/56461
20037         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
20038         method on dr_chain and result_chain.
20040         PR middle-end/56461
20041         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
20042         pointer_set_destroy on not_executed_last_iteration.
20044         PR middle-end/56461
20045         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
20047         PR middle-end/56461
20048         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
20049         FOR_EACH_DEFINED_FUNCTION when freeing state.
20051         PR middle-end/56461
20052         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
20053         pool_free.
20054         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
20055         overwriting it.
20057         PR middle-end/56461
20058         * ipa-cp.c (decide_whether_version_node): Call vec_free on
20059         known_aggs[i].items and release known_aggs vector.
20061         PR middle-end/56461
20062         * ipa-reference.c (propagate): Free node_info even for alias nodes.
20064 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20066         * config/microblaze/microblaze.c (microblaze_emit_compare):
20067         Use xor for EQ/NE comparisions.
20068         * config/microblaze/microblaze.md (cstoresf4): Add constraints
20069         (cbranchsf4): Adjust operator to comparison_operator.
20071 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
20073         PR middle-end/56461
20074         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
20075         vector.
20076         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
20077         vec_safe_push, always update *slot.
20078         (redirect_edge_var_map_clear): Use vec_free.
20079         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
20080         (free_var_map_entry): Use vec_free.
20081         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
20082         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
20084 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
20086         PR middle-end/45472
20087         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
20088         when the may_trap_p bit of the exprs being merged differs.
20089         Reorder tests for speculativeness in the logical and operator.
20091 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
20093         * incpath.c (add_standard_paths): Use reconcat instead of concat
20094         where appropriate and avoid leaking memory.
20096         * opts.h: Include obstack.h.
20097         (opts_concat): New prototype.
20098         (opts_obstack): New declaration.
20099         * opts.c (opts_concat): New function.
20100         (opts_obstack): New variable.
20101         (init_options_struct): Call gcc_init_obstack on opts_obstack.
20102         (finish_options): Use opts_concat instead of concat
20103         and XOBNEWVEC instead of XNEWVEC.
20104         * opts-common.c (generate_canonical_option, decode_cmdline_option,
20105         generate_option): Likewise.
20106         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
20107         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
20109         PR target/56455
20110         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
20111         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
20113 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
20115         PR middle-end/56461
20116         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
20118 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
20120         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
20121         (arm_block_move_unaligned_straight): Likewise.
20122         (arm_adjust_block_mem): Likewise.
20124 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
20126         PR target/48901
20127         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
20128         temp, cond and label.
20129         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
20131         PR target/52500
20132         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
20133         * config/c6x/c6x.h (dbx_register_map): Update declaration.
20135         PR target/52501
20136         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
20137         of prologue/epilogue functions.
20139         PR target/52550
20140         * config/tilegx/tilegx.c (tilegx_expand_prologue):
20141         Remove unused variable cfa_offset.
20142         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
20144         PR target/54639
20145         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
20146         type promotion to unsigned.
20148         PR target/54640
20149         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
20150         for HOST_WIDE_INT of 32 bit / same size as int.
20151         (arm_block_move_unaligned_straight): Likewise.
20152         (arm_adjust_block_mem): Likewise.
20154         PR target/54662
20155         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
20156         ALL_CFLAGS.
20158 2013-02-26  Marek Polacek  <polacek@redhat.com>
20160         PR tree-optimization/56426
20161         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
20163 2013-02-26  Richard Biener  <rguenther@suse.de>
20165         PR target/56444
20166         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
20167         unused variable loops.
20169 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
20171         PR tree-optimization/56448
20172         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
20173         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
20174         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
20175         later operands of the references, or even first operand for
20176         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
20178         PR tree-optimization/56443
20179         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
20180         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
20181         to type_for_mode langhook.
20183 2013-02-25  Matt Turner  <mattst88@gmail.com>
20185         * doc/invoke.texi: Document r4700.
20187 2013-02-25  Richard Biener  <rguenther@suse.de>
20189         PR tree-optimization/56175
20190         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
20191         split out from ...
20192         (simplify_bitwise_binary): ... here.  Also guard the conversion
20193         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
20195 2013-02-25  Catherine Moore  <clm@codesourcery.com>
20197         Revert:
20198         2013-02-24  Catherine Moore  <clm@codesourcery.com>
20199             Maciej W. Rozycki  <macro@codesourcery.com>
20200             Tom de Vries  <tom@codesourcery.com>
20201             Nathan Sidwell  <nathan@codesourcery.com>
20202             Iain Sandoe  <iain@codesourcery.com>
20203             Nathan Froyd  <froydnj@codesourcery.com>
20204             Chao-ying Fu  <fu@mips.com>
20206         * doc/extend.texi: (micromips, nomicromips, nocompression):
20207         Document new function attributes.
20208         * doc/invoke.texi (minterlink-compressed, mmicromips,
20209         m14k, m14ke, m14kec): Document new options.
20210         (minterlink-mips16): Update documentation.
20211         * doc/md.texi (ZC, ZD): Document new constraints.
20212         * configure.ac (gcc_cv_as_micromips): Check if linker
20213         supports the .set micromips directive.
20214         * configure: Regenerate.
20215         * config.in: Regenerate.
20216         * config/mips/mips-tables.opt: Regenerate.
20217         * config/mips/micromips.md: New file.
20218         * constraints.md (ZC, AD): New constraints.
20219         * config/mips/predicates.md (movep_src_register): New predicate.
20220         (movep_src_operand): New predicate.
20221         (non_volatile_mem_operand): New predicate.
20222         * config/mips/mips.md (multimem): New type.
20223         (length): Differentiate between 17-bit and 18-bit branch offsets.
20224         (MOVEP1, MOVEP2): New mode iterator.
20225         (mov_<load>l): Use ZC constraint.
20226         (mov_<load>r): Likewise.
20227         (mov_<store>l): Likewise.
20228         (mov_<store>r): Likewise.
20229         (*branch_equality<mode>_inverted): Add microMIPS support.
20230         (*branch_equality<mode>): Likewise.
20231         (*jump_absolute): Likewise.
20232         (indirect_jump_<mode>): Likewise.
20233         (tablejump_<mode>): Likewise.
20234         (<optab>_internal): Likewise.
20235         (sibcall_internal): Likewise.
20236         (sibcall_value_internal): Likewise.
20237         (prefetch): Use constraint ZD.
20238         * config/mips/mips.opt (minterlink-compressed): New option.
20239         (minterlink-mips16): Now an alias for minterlink-compressed.
20240         (mmicromips): New option.
20241         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
20242         (compare_and_swap_12): Likewise.
20243         (sync_add<mode>): Likewise.
20244         (sync_<optab>_12): Likewise.
20245         (sync_old_<optab>_12): Likewise.
20246         (sync_new_<optab>_12): Likewise.
20247         (sync_nand_12): Likewise.
20248         (sync_old_nand_12): Likewise.
20249         (sync_new_nand_12): Likewise.
20250         (sync_sub<mode>): Likewise.
20251         (sync_old_add<mode>): Likewise.
20252         (sync_old_sub<mode>): Likewise.
20253         (sync_new_add<mode>): Likewise.
20254         (sync_new_sub<mode>): Likewise.
20255         (sync_<optab><mode>): Likewise.
20256         (sync_old_<optab><mode>): Likewise.
20257         (sync_new_<optab><mode>): Likewise.
20258         (sync_nand<mode>): Likewise.
20259         (sync_old_nand<mode>): Likewise.
20260         (sync_new_nand<mode>): Likewise.
20261         (sync_lock_test_and_set<mode>): Likewise.
20262         (test_and_set_12): Likewise.
20263         (atomic_compare_and_swap<mode>): Likewise.
20264         (atomic_exchange<mode>_llsc): Likewise.
20265         (atomic_fetch_add<mode>_llsc): Likewise.
20266         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
20267         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
20268         (umips_save_restore_pattern_p): Likewise.
20269         (umips_load_store_pair_p): Likewise.
20270         (umips_output_load_store_pair): Likewise.
20271         (umips_movep_target_p): Likewise.
20272         (umips_12bit_offset_address_p): Likewise.
20273         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
20274         (mips_base_mips16): Rename this...
20275         (mips_base_compression_flags): ...to this. Update all uses.
20276         (mips_attribute_table): Add micromips, nomicromips and nocompression.
20277         (mips_mips16_decl_p): Delete.
20278         (mips_nomips16_decl_p): Delete.
20279         (mips_get_compress_on_flags): New function.
20280         (mips_get_compress_off_flags): New function.
20281         (mips_get_compress_mode): New function.
20282         (mips_get_compress_on_name): New function.
20283         (mips_get_compress_off_name): New function.
20284         (mips_insert_attributes): Support multiple compression types.
20285         (mips_merge_decl_attributes): Likewise.
20286         (umips_12bit_offset_address_p): New function.
20287         (mips_start_function_definition): Emit .set micromips directive.
20288         (mips_call_may_need_jalx_p): New function.
20289         (mips_function_ok_for_sibcall): Add microMIPS support.
20290         (mips_print_operand_punctuation): Support short delay slots and
20291         compact jumps.
20292         (umips_swm_mask, umips_swm_encoding): New.
20293         (umips_build_save_restore): New function.
20294         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
20295         (was_mips16_p): Remove.
20296         (old_compression_mode): New.
20297         (mips_set_compression_mode): New function.
20298         (mips_set_current_function): Add microMIPS support.
20299         (mips_option_override): Likewise.
20300         (umips_save_restore_pattern_p): New function.
20301         (umips_output_save_restore): New function.
20302         (umips_load_store_pair_p_1): New function.
20303         (umips_load_store_pair_p): New function.
20304         (umips_output_load_store_pair_1): New function.
20305         (umips_output_load_store_pair): New function.
20306         (umips_movep_target_p) New function.
20307         (mips_prepare_pch_save): Add microMIPS support.
20308         * config/mips/mips.h (TARGET_COMPRESSION): New.
20309         (TARGET_CPU_CPP_BUILTINS): Update macro
20310         to use new compression flags and to support microMIPS.
20311         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
20312         (MIPS_ARCH_FLOAT_SPEC): Likewise.
20313         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
20314         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
20315         (ASM_SPEC): Support mmicromips and mno-micromips.
20316         (M16STORE_REG_P): New macro.
20317         (MIPS_CALL): Support TARGET_MICROMIPS.
20318         (MICROMIPS_J): New macro.
20319         (mips_base_mips16): Rename this...
20320         (mips_base_compression_flags): ...to this.
20321         (UMIPS_12BIT_OFFSET_P): New macro.
20322         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
20323         (MULTILIB_DIRNAMES): Likewise.
20325 2013-02-25  Tom de Vries  <tom@codesourcery.com>
20327         PR rtl-optimization/56131
20328         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
20329         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
20330         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
20332 2013-02-25  Tobias Burnus  <burnus@net-b.de>
20334         * doc/invoke.texi (-fsanitize=): Move from optimization
20335         to debugging options.
20337 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
20339         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
20341 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
20342             Alexander Monakov  <amonakov@ispras.ru>
20344         PR middle-end/56077
20345         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
20346         flush pending lists also on non-jumps.  Adjust comment.
20348 2013-02-24  Catherine Moore  <clm@codesourcery.com>
20349             Maciej W. Rozycki  <macro@codesourcery.com>
20350             Tom de Vries  <tom@codesourcery.com>
20351             Nathan Sidwell  <nathan@codesourcery.com>
20352             Iain Sandoe  <iain@codesourcery.com>
20353             Nathan Froyd  <froydnj@codesourcery.com>
20354             Chao-ying Fu  <fu@mips.com>
20356         * doc/extend.texi: (micromips, nomicromips, nocompression):
20357         Document new function attributes.
20358         * doc/invoke.texi (minterlink-compressed, mmicromips,
20359         m14k, m14ke, m14kec): Document new options.
20360         (minterlink-mips16): Update documentation.
20361         * doc/md.texi (ZC, ZD): Document new constraints.
20362         * configure.ac (gcc_cv_as_micromips): Check if linker
20363         supports the .set micromips directive.
20364         * configure: Regenerate.
20365         * config.in: Regenerate.
20366         * config/mips/mips-tables.opt: Regenerate.
20367         * config/mips/micromips.md: New file.
20368         * constraints.md (ZC, AD): New constraints.
20369         * config/mips/predicates.md (movep_src_register): New predicate.
20370         (movep_src_operand): New predicate.
20371         (non_volatile_mem_operand): New predicate.
20372         * config/mips/mips.md (multimem): New type.
20373         (length): Differentiate between 17-bit and 18-bit branch offsets.
20374         (MOVEP1, MOVEP2): New mode iterator.
20375         (mov_<load>l): Use ZC constraint.
20376         (mov_<load>r): Likewise.
20377         (mov_<store>l): Likewise.
20378         (mov_<store>r): Likewise.
20379         (*branch_equality<mode>_inverted): Add microMIPS support.
20380         (*branch_equality<mode>): Likewise.
20381         (*jump_absolute): Likewise.
20382         (indirect_jump_<mode>): Likewise.
20383         (tablejump_<mode>): Likewise.
20384         (<optab>_internal): Likewise.
20385         (sibcall_internal): Likewise.
20386         (sibcall_value_internal): Likewise.
20387         (prefetch): Use constraint ZD.
20388         * config/mips/mips.opt (minterlink-compressed): New option.
20389         (minterlink-mips16): Now an alias for minterlink-compressed.
20390         (mmicromips): New option.
20391         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
20392         (compare_and_swap_12): Likewise.
20393         (sync_add<mode>): Likewise.
20394         (sync_<optab>_12): Likewise.
20395         (sync_old_<optab>_12): Likewise.
20396         (sync_new_<optab>_12): Likewise.
20397         (sync_nand_12): Likewise.
20398         (sync_old_nand_12): Likewise.
20399         (sync_new_nand_12): Likewise.
20400         (sync_sub<mode>): Likewise.
20401         (sync_old_add<mode>): Likewise.
20402         (sync_old_sub<mode>): Likewise.
20403         (sync_new_add<mode>): Likewise.
20404         (sync_new_sub<mode>): Likewise.
20405         (sync_<optab><mode>): Likewise.
20406         (sync_old_<optab><mode>): Likewise.
20407         (sync_new_<optab><mode>): Likewise.
20408         (sync_nand<mode>): Likewise.
20409         (sync_old_nand<mode>): Likewise.
20410         (sync_new_nand<mode>): Likewise.
20411         (sync_lock_test_and_set<mode>): Likewise.
20412         (test_and_set_12): Likewise.
20413         (atomic_compare_and_swap<mode>): Likewise.
20414         (atomic_exchange<mode>_llsc): Likewise.
20415         (atomic_fetch_add<mode>_llsc): Likewise.
20416         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
20417         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
20418         (umips_save_restore_pattern_p): Likewise.
20419         (umips_load_store_pair_p): Likewise.
20420         (umips_output_load_store_pair): Likewise.
20421         (umips_movep_target_p): Likewise.
20422         (umips_12bit_offset_address_p): Likewise.
20423         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
20424         (mips_base_mips16): Rename this...
20425         (mips_base_compression_flags): ...to this. Update all uses.
20426         (mips_attribute_table): Add micromips, nomicromips and nocompression.
20427         (mips_mips16_decl_p): Delete.
20428         (mips_nomips16_decl_p): Delete.
20429         (mips_get_compress_on_flags): New function.
20430         (mips_get_compress_off_flags): New function.
20431         (mips_get_compress_mode): New function.
20432         (mips_get_compress_on_name): New function.
20433         (mips_get_compress_off_name): New function.
20434         (mips_insert_attributes): Support multiple compression types.
20435         (mips_merge_decl_attributes): Likewise.
20436         (umips_12bit_offset_address_p): New function.
20437         (mips_start_function_definition): Emit .set micromips directive.
20438         (mips_call_may_need_jalx_p): New function.
20439         (mips_function_ok_for_sibcall): Add microMIPS support.
20440         (mips_print_operand_punctuation): Support short delay slots and
20441         compact jumps.
20442         (umips_swm_mask, umips_swm_encoding): New.
20443         (umips_build_save_restore): New function.
20444         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
20445         (was_mips16_p): Remove.
20446         (old_compression_mode): New.
20447         (mips_set_compression_mode): New function.
20448         (mips_set_current_function): Add microMIPS support.
20449         (mips_option_override): Likewise.
20450         (umips_save_restore_pattern_p): New function.
20451         (umips_output_save_restore): New function.
20452         (umips_load_store_pair_p_1): New function.
20453         (umips_load_store_pair_p): New function.
20454         (umips_output_load_store_pair_1): New function.
20455         (umips_output_load_store_pair): New function.
20456         (umips_movep_target_p) New function.
20457         (mips_prepare_pch_save): Add microMIPS support.
20458         * config/mips/mips.h (TARGET_COMPRESSION): New.
20459         (TARGET_CPU_CPP_BUILTINS): Update macro
20460         to use new compression flags and to support microMIPS.
20461         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
20462         (MIPS_ARCH_FLOAT_SPEC): Likewise.
20463         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
20464         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
20465         (ASM_SPEC): Support mmicromips and mno-micromips.
20466         (M16STORE_REG_P): New macro.
20467         (MIPS_CALL): Support TARGET_MICROMIPS.
20468         (MICROMIPS_J): New macro.
20469         (mips_base_mips16): Rename this...
20470         (mips_base_compression_flags): ...to this.
20471         (UMIPS_12BIT_OFFSET_P): New macro.
20472         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
20473         (MULTILIB_DIRNAMES): Likewise.
20475 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
20477         PR target/52555
20478         * target-globals.c (save_target_globals): For init_reg_sets and
20479         target_reinit remporarily set this_fn_optabs to this_target_optabs.
20481 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
20483         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
20484         * config/aarch64/t-aarch64
20485         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
20487 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
20489         PR inline-asm/56148
20490         * lra-constraints.c (process_alt_operands): Reload operand
20491         conflicting with earlier clobber only if no more other conflicting
20492         operands.
20494 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
20496         PR sanitizer/56393
20497         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
20498         if not linking a shared library.
20500 2013-02-22  Seth LaForge  <sethml@google.com>
20502         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
20504 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
20506         * config/arm/arm.md (split for extendsidi): Update condition.
20507         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
20508         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
20509         (qhs_zextenddi_cstr): Likewise.
20511 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
20513         PR middle-end/56420
20514         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
20515         avoid signed wrapping.
20516         (expand_mult): Handle properly multiplication by
20517         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
20518         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
20519         in the compiler if coeff is HOST_WIDE_INT_MIN.
20520         (expand_divmod): Don't make ext_op1 static, change it's type to
20521         uhwi.  Avoid undefined behavior in -INTVAL (op1).
20523         PR rtl-optimization/50339
20524         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
20525         field.
20526         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
20527         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
20528         into splitting_ashiftrt field.
20529         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
20530         ASHIFTRT.
20531         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
20532         choices.
20534 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
20536         PR middle-end/56108
20537         * trans-mem.c (execute_tm_mark): Do not expand transactions that
20538         are sure to go irrevocable.
20540 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
20542         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
20543         scalars are valid operands.
20545 2013-02-21  Martin Jambor  <mjambor@suse.cz>
20547         PR tree-optimization/56310
20548         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
20549         only matching indices and non-negative final offsets.
20550         (intersect_aggregates_with_edge): Pass src_idx to
20551         agg_replacements_to_vector.  Pass src_idx insstead of index to
20552         intersect_with_agg_replacements.
20554 2013-02-21  Martin Jambor  <mjambor@suse.cz>
20556         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
20557         instead of hard-wired defaults.
20559 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
20561         * doc/invoke.texi (MIPS Options): Update documentation of the
20562         floating-point multiply-accumulate instruction restrictions.
20564 2013-02-21  Kostya Serebryany  <kcc@google.com>
20566         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
20567         asan_shadow_offset on x86_64 linux.
20569 2013-02-21  Richard Biener  <rguenther@suse.de>
20571         PR tree-optimization/56415
20572         Revert
20573         2013-02-11  Richard Biener  <rguenther@suse.de>
20575         PR tree-optimization/56273
20576         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20577         first VRP run.
20579 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
20581         PR bootstrap/56258
20582         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
20583         instead of @itemx.
20585         PR inline-asm/56405
20586         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
20587         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
20589 2013-02-20  Jan Hubicka  <jh@suse.cz>
20591         PR tree-optimization/56265
20592         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
20593         when target is referenced for first time.
20595 2013-02-20  Richard Biener  <rguenther@suse.de>
20597         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
20598         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
20599         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
20600         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
20601         not return anything.
20602         (rename_ssa_copies): Do not remove unused locals.
20603         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
20604         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
20605         * passes.c (execute_function_todo): Do not schedule unused locals
20606         removal if cleanup_tree_cfg did something.
20607         * tree-ssa-live.c (remove_unused_locals): Dump statistics
20608         about the number of removed locals.
20610 2013-02-20  Richard Biener  <rguenther@suse.de>
20612         PR tree-optimization/56398
20613         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
20615 2013-02-20  Martin Jambor  <mjambor@suse.cz>
20617         PR tree-optimization/55334
20618         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
20619         restricted pointers to arrays.
20621 2013-02-20  Richard Biener  <rguenther@suse.de>
20622             Jakub Jelinek  <jakub@redhat.com>
20624         PR tree-optimization/56396
20625         * tree-ssa-ccp.c (n_const_val): New static variable.
20626         (get_value): Return NULL for SSA names we don't have a lattice
20627         entry for.
20628         (ccp_initialize): Initialize n_const_val.
20629         * tree-ssa-copy.c (n_copy_of): New static variable.
20630         (init_copy_prop): Initialize n_copy_of.
20631         (get_value): Return NULL_TREE for SSA names we don't have a
20632         lattice entry for.
20634 2013-02-20  Martin Jambor  <mjambor@suse.cz>
20636         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
20638 2013-02-20  Richard Biener  <rguenther@suse.de>
20640         * genpreds.c (write_lookup_constraint): Do not compare first
20641         letter of the constraint again.
20643 2013-02-20  Richard Biener  <rguenther@suse.de>
20645         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
20646         and ceil_log2.
20647         (get_use_iv_cost): Terminate hashtable walk when coming across
20648         an empty entry.
20650 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
20652         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
20653         reassociation for avx2 targets.
20655 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20657         * config/microblaze/microblaze.c: microblaze_has_clz = 0
20658         Add version check for v8.10.a to enable microblaze_has_clz
20659         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
20660         version and TARGET_PATTERN_COMPARE check
20661         * config/microblaze/microblaze.md: New clzsi2 instruction
20663 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20665         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
20666         function before branching.
20668 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
20670         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
20671         DUMP_INSN_RTX_UID.
20672         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
20674 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
20676         PR middle-end/55889
20677         * sel-sched.c: Include ira.h.
20678         (implicit_clobber_conflict_p): New function.
20679         (moveup_expr): Use it.
20680         * Makefile.in (sel-sched.o): Depend on ira.h.
20682 2013-02-19  Richard Biener  <rguenther@suse.de>
20684         PR tree-optimization/56384
20685         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
20686         (vn_hash_type): Split out from ...
20687         (vn_hash_constant_with_type): ... here.
20688         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
20689         (vn_phi_eq): Compare types from vn_phi_s structure.
20690         (vn_phi_lookup): Populate vn_phi_s type.
20691         (vn_phi_insert): Likewise.
20693 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
20695         PR tree-optimization/56350
20696         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
20697         if haven't found reduction or nested cycle operand, rather than
20698         asserting we must find it.
20700         PR tree-optimization/56381
20701         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
20702         to fold_build3.
20704 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
20705             Jakub Jelinek  <jakub@redhat.com>
20707         PR target/52555
20708         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
20709         (swap_optab_enable): Same.
20710         (init_all_optabs): Use argument instead of global.
20711         * tree.h (struct tree_optimization_option): New field target_optabs.
20712         * expr.h (init_all_optabs): Add argument to prototype.
20713         (TREE_OPTIMIZATION_OPTABS): New.
20714         (save_optabs_if_changed): Protoize.
20715         * optabs.h: Declare this_fn_optabs.
20716         * optabs.c (save_optabs_if_changed): New.
20717         Declare this_fn_optabs.
20718         (init_optabs): Add argument to init_all_optabs() call.
20719         * function.c (invoke_set_current_function_hook): Handle per
20720         function optabs.
20721         * function.h (struct function): New field optabs.
20722         * config/mips/mips.c (mips_set_mips16_mode): Handle when
20723         optimization_current_node has changed.
20724         * target-globals.h (save_target_globals_default_opts): Protoize.
20725         * target-globals.c (save_target_globals_default_opts): New.
20727 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20729         PR target/56347
20730         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
20731         registers %fr12 and %fr12R as call used.
20733         PR target/56214
20734         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
20735         and HImode, require all displacements to be an integer multiple of
20736         their mode size.
20737         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
20738         only allow QImode and HImode when reload is in progress and strict is
20739         true.  Likewise for symbolic addresses.  Use base14_operand to check
20740         displacements in REG+BASE addresses.
20742 2013-02-18  Richard Biener  <rguenther@suse.de>
20744         PR tree-optimization/56366
20745         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20746         handle sign-conversion of outer-loop initial induction value.
20748 2013-02-18  Richard Biener  <rguenther@suse.de>
20750         PR middle-end/56349
20751         * cfghooks.c (merge_blocks): If we merge a latch into another
20752         block adjust references to it.
20753         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
20754         (verify_loop_structure): Verify that a recorded latch is in fact
20755         a latch.
20757 2013-02-18  Richard Biener  <rguenther@suse.de>
20759         PR tree-optimization/56321
20760         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
20761         order SSA name release and virtual operand unlinking.
20763 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20765         * config/microblaze/microblaze.md (save_stack_block): Define.
20766         (restore_stack_block): Likewise.
20768 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20770         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
20771         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
20772         * config/microblaze/microblaze.c (microblaze_option_override):
20773         Bail out early for PIC modes when target does not support PIC.
20775 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20777         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
20778         Replace with a microblaze version.
20779         (microblaze_trampoline_init): Adapt for microblaze.
20780         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
20781         microblaze.
20783 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
20784             Dodji Seketeli  <dodji@redhat.com>
20786         PR asan/56330
20787         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
20788         (instrument_mem_region_access): Do not forget to always put
20789         instrumentation of the of 'base' and 'base + len' in a "if (len !=
20790         0) statement, even for cases where either 'base' or 'base + len'
20791         are not instrumented -- because they have been previously
20792         instrumented.  Simplify the logic by putting all the statements
20793         instrument 'base + len' inside a sequence, and then insert that
20794         sequence right before the current insertion point.  Then, to
20795         instrument 'base + len', just get an iterator on that statement.
20796         And do not forget to update the pointer to iterator the function
20797         received as argument.
20799 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
20801         PR rtl-optimization/56348
20802         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
20804 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
20806         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
20807         (clean_graph_dump_file): Pass base to start_graph_dump.
20809 2013-02-14  Richard Henderson  <rth@redhat.com>
20811         PR target/55941
20812         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
20814 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
20816         * collect2-aix.h: Define F_LOADONLY.
20818 2013-02-14  Richard Biener  <rguenther@suse.de>
20820         PR lto/50494
20821         * varasm.c (output_constant_def_1): Get the decl representing
20822         the constant as argument.
20823         (output_constant_def): Wrap output_constant_def_1.
20824         (make_decl_rtl): Use output_constant_def_1 with the decl
20825         representing the constant.
20826         (build_constant_desc): Optionally re-use a decl already
20827         representing the constant.
20828         (tree_output_constant_def): Adjust.
20830 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
20832         Fix an asan crash
20833         * asan.c (instrument_builtin_call):  Really put the length of the
20834         second source argument into src1_len.
20836 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
20838         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
20839         argument.  If it is false, don't create edge from then_bb to
20840         fallthru_bb.
20841         (insert_if_then_before_iter): Pass true to it.
20842         (build_check_stmt): Pass false to it.
20843         (transform_statements): Flush hash table only on extended basic
20844         block boundaries, rather than at the beginning of every bb.
20845         Don't flush hash table on nonfreeing_call_p calls.
20846         * tree-flow.h (nonfreeing_call_p): New prototype.
20847         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
20849 2013-02-13  David S. Miller  <davem@davemloft.net>
20851         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
20853 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
20855         PR target/56184
20856         * ira.c (max_regno_before_ira): Move from ...
20857         (ira): ... here.
20858         (fix_reg_equiv_init): Use max_regno_before_ira instead of
20859         vec_safe_length.
20861 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
20863         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
20865 2013-02-13  Richard Biener  <rguenther@suse.de>
20867         PR lto/56295
20868         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
20869         globals in MEM_REFs.
20871 2013-02-13  Richard Biener  <rguenther@suse.de>
20873         * loop-init.c (loop_optimizer_init): Clear loop state when
20874         re-initializing preserved loops.
20875         * loop-unswitch.c (unswitch_single_loop): Return whether
20876         we unswitched the loop.  Do not verify loop state here.
20877         (unswitch_loops): When we unswitched a loop discover new loops.
20879 2013-02-13  Kostya Serebryany  <kcc@google.com>
20881         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
20882         on x86_64 linux.
20883         * sanitizer.def: Rename __asan_init to __asan_init_v1.
20885 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
20887         Avoid instrumenting duplicated memory access in the same basic block
20888         * Makefile.in (asan.o): Add new dependency on hash-table.h
20889         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
20890         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
20891         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
20892         (free_mem_ref_resources, has_mem_ref_been_instrumented)
20893         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
20894         (get_mem_ref_of_assignment): New functions.
20895         (get_mem_refs_of_builtin_call): Extract from
20896         instrument_builtin_call and tweak a little bit to make it fit with
20897         the new signature.
20898         (instrument_builtin_call): Use the new
20899         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
20900         of is_gimple_builtin_call.
20901         (instrument_derefs, instrument_mem_region_access): Insert the
20902         instrumented memory reference into the hash table.
20903         (maybe_instrument_assignment): Renamed instrument_assignment into
20904         this, and change it to advance the iterator when instrumentation
20905         actually happened and return true in that case.  This makes it
20906         homogeneous with maybe_instrument_assignment, and thus give a
20907         chance to callers to be more 'regular'.
20908         (transform_statements): Clear the memory reference hash table
20909         whenever we enter a new BB, when we cross a function call, or when
20910         we are done transforming statements.  Use
20911         maybe_instrument_assignment instead of instrumentation.  No more
20912         need to special case maybe_instrument_assignment and advance the
20913         iterator after calling it; it's now handled just like
20914         maybe_instrument_call.  Update comment.
20916 2013-02-13  Richard Biener  <rguenther@suse.de>
20918         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
20919         Fix loop discovery code.
20921 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
20923         PR inline-asm/56148
20924         * lra-constraints.c (process_alt_operands): Match early clobber
20925         operand with itself.  Check conflicts with earlyclobber only if
20926         the operand is not reloaded.  Prefer to reload conflicting operand
20927         if earlyclobber and matching operands are the same.
20929 2013-02-12  Richard Biener  <rguenther@suse.de>
20931         PR lto/56297
20932         * lto-streamer-out.c (write_symbol): Do not output symbols
20933         for hard register variables.
20935 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
20937         PR target/54222
20938         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
20939         (umulsidi3_insn, mulsidi3_insn): New insns.
20941 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
20943         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
20944         (struct tune_params): Add vec_costs field.
20945         * config/arm/arm.c (arm_builtin_vectorization_cost)
20946         (arm_add_stmt_cost): New functions.
20947         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
20948         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
20949         (arm_default_vec_cost): New struct of type cpu_vec_costs.
20950         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
20951         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
20952         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
20953         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
20955 2013-02-12  Richard Biener  <rguenther@suse.de>
20957         PR lto/56295
20958         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
20959         decls again if possible.
20961 2013-02-12  Richard Biener  <rguenther@suse.de>
20963         PR middle-end/56288
20964         * tree-ssa.c (verify_ssa_name): Fix check, move
20965         SSA_NAME_IN_FREE_LIST check up.
20967 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
20968             Steven Bosscher   <steven@gcc.gnu.org>
20970         PR rtl-optimization/56151
20971         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
20972         equal to op0 or op1, and last_insn pattern is CODE operation
20973         with MEM dest and one of the operands matches that MEM.
20975 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
20977         * doc/extend.texi: Document Function Multiversioning and "default"
20978         parameter string to target attribute.
20979         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
20980         target attribute parameter is "default".
20981         (ix86_compare_version_priority): Remove checks for target attribute.
20982         (ix86_mangle_function_version_assembler_name): Change error to sorry.
20983         Remove check for target attribute equal to NULL. Add assert.
20984         (ix86_generate_version_dispatcher_body): Change error to sorry.
20986 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
20987             Jack Howarth  <howarth@bromo.med.uc.edu>
20988             Patrick Marlier  <patrick.marlier@gmail.com>
20990         PR libitm/55693
20991         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
20992         define ENDFILE_SPEC as TM_DESTRUCTOR.
20993         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
20995 2013-02-11  Alexander Potapenko  <glider@google.com>
20996             Jack Howarth  <howarth@bromo.med.uc.edu>
20997             Jakub Jelinek  <jakub@redhat.com>
20999         PR sanitizer/55617
21000         * config/darwin.c (cdtor_record): Rename ctor_record.
21001         (sort_cdtor_records): Rename sort_ctor_records.
21002         (finalize_dtors): New routine to sort destructors by
21003         priority before use in assemble_integer.
21004         (machopic_asm_out_destructor): Use finalize_dtors if needed.
21006 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
21008         PR rtl-optimization/56275
21009         * simplify-rtx.c (avoid_constant_pool_reference): Check that
21010         offset is non-negative and less than cmode size before
21011         calling simplify_subreg.
21013 2013-02-11  Richard Biener  <rguenther@suse.de>
21015         PR tree-optimization/56264
21016         * cfgloop.h (fix_loop_structure): Adjust prototype.
21017         * loop-init.c (fix_loop_structure): Return the number of
21018         newly discovered loops.
21019         * tree-cfgcleanup.c (repair_loop_structures): When new loops
21020         are discovered, do a full loop-closed SSA rewrite.
21022 2013-02-11  Richard Biener  <rguenther@suse.de>
21024         PR tree-optimization/56273
21025         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
21026         first VRP run.
21027         (check_array_ref): Fix missing newline in dumps.
21028         (search_for_addr_array): Likewise.
21030 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
21032         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
21034 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
21036         PR target/56256
21037         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
21039 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
21041         PR rtl-optimization/56246
21042         * lra-constraints.c (simplify_operand_subreg): Try to reuse
21043         reload pseudo.
21044         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
21045         constraints are satisfied.
21047 2013-02-08  Jeff Law  <law@redhat.com>
21049         PR debug/53948
21050         * emit-rtl.c (reg_is_parm_p): New function.
21051         * regs.h (reg_is_parm_p): New prototype.
21052         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
21053         callee-clobbered registers.
21055 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
21057         PR target/56043
21058         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
21059         If there is no implicit builtin declaration, just return NULL.
21061 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
21063         * config/i386/sse.md (FMAMODEM): New mode iterator.
21064         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
21065         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
21067 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
21069         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
21070         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
21071         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
21073 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21075         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
21076         (microblaze*-*-elf): Likewise.
21077         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
21078         LINK_SPEC.
21079         * config/microblaze/microblaze-c.c: Add builtin defines for
21080         _LITTLE_ENDIAN and _BIG_ENDIAN.
21081         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
21082         add to TARGET_DEFAULT flags.
21083         Expand ASM_SPEC and LINK_SPEC.
21084         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
21085         * config/microblaze/microblaze.md: Update extendsidi2 and
21086         movdi_internal instructions to use low-order / high-order reg
21087         print_operands.
21088         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
21089         options and inversemask / mask of LITTLE_ENDIAN.
21090         * config/microblaze/t-microblaze: Expand multilib options to
21091         include mlittle-endian (le) and update exceptions patterns.
21093 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
21095         PR rtl-optimization/56195
21096         * lra-constraints.c (get_reload_reg): Don't reuse regs
21097         if they have smaller mode than requested, if they have
21098         wider mode than requested, try to return a SUBREG.
21100         PR tree-optimization/56250
21101         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
21102         if type is unsigned and code isn't MULT_EXPR.
21104 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
21106         PR tree-optimization/56064
21107         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
21108         bits according to mode.
21109         * fixed-value.h (fixed_from_double_int)
21110         (const_fixed_from_double_int): Adjust comments.
21112 2013-02-08  Richard Biener  <rguenther@suse.de>
21114         PR lto/56231
21115         * lto-streamer.h (struct data_in): Remove current_file, current_line
21116         and current_col members.
21117         * lto-streamer-out.c (lto_output_location): Stream changed bits
21118         en-block for efficiency.
21119         * lto-streamer-in.c (clear_line_info): Remove.
21120         (lto_input_location): Cache current file, line and column
21121         globally via local statics.  Read changed bits en-block.
21122         (input_function): Do not call clear_line_info.
21123         (lto_read_body): Likewise.
21124         (lto_input_toplevel_asms): Likewise.
21126 2013-02-08  Michael Matz  <matz@suse.de>
21128         PR tree-optimization/52448
21129         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
21130         (nt_call_phase): New static.
21131         (add_or_mark_expr): Only mark accesses with newer phase than any
21132         call seen.
21133         (nonfreeing_call_p): New.
21134         (nt_init_block): Update nt_call_phase, mark blocks as visited.
21135         (nt_fini_block): Keep blocks marked as visited.
21136         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
21138 2013-02-08  Richard Biener  <rguenther@suse.de>
21140         * ira.c (ira): Free broken dominator information.
21142 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
21144         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
21146 2013-02-08  Marek Polacek  <polacek@redhat.com>
21148         * cfgloop.c (verify_loop_structure): Add more checking of headers.
21150 2013-02-08  Richard Biener  <rguenther@suse.de>
21152         PR middle-end/56181
21153         * cfgloop.h (flow_loops_find): Adjust.
21154         (bb_loop_header_p): Declare.
21155         * cfgloop.c (bb_loop_header_p): New function split out from ...
21156         (flow_loops_find): ... here.  Adjust function signature,
21157         support incremental loop structure update.
21158         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
21159         * cfgloopmanip.c (fix_loop_structure): Move ...
21160         * loop-init.c (fix_loop_structure): ... here.
21161         (apply_loop_flags): Split out from ...
21162         (loop_optimizer_init): ... here.
21163         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
21164         in incremental mode, only remove dead loops here.
21166 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
21168         PR target/54222
21169         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
21170         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
21171         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
21172         (*round<mode>3.libgcc): New insns for fixed-modes.
21173         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
21174         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
21175         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
21176         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
21177         implementations.  Define to __builtin_avr_absFX,
21178         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
21179         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
21180         __builtin_avr_countlsFX, respectively.
21181         * config/avr/avr-c.c (target.h): Include it.
21182         (enum avr_builtin_id): New enum.
21183         (avr_resolve_overloaded_builtin): New static function.
21184         (avr_register_target_pragmas): Use it to set
21185         targetm.resolve_overloaded_builtin.
21186         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
21187         tree nodes used by DEF_BUILTIN.
21188         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
21189         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
21190         <AVR_BUILTIN_xxBITS>: Same.
21192 2013-02-08  Richard Biener  <rguenther@suse.de>
21194         * cfgloop.c (verify_loop_structure): Properly handle
21195         a loop exiting to another loop header.
21196         * ira-int.h (ira_loops): Remove.
21197         * ira.c (ira_loops): Remove.
21198         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
21199         (do_reload): Use loop_optimizer_finalize.
21200         * ira-build.c (create_loop_tree_nodes): Use get_loops and
21201         number_of_loops to access the loop tree.
21202         (more_one_region_p): Likewise.
21203         (finish_loop_tree_nodes): Likewise.
21204         (rebuild_regno_allocno_maps): Likewise.
21205         (mark_loops_for_removal): Likewise.
21206         (mark_all_loops_for_removal): Likewise.
21207         (remove_unnecessary_regions): Likewise.
21208         (ira_build): Likewise.
21209         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
21211 2013-02-08  Richard Biener  <rguenther@suse.de>
21213         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
21214         * ipa-pure-const.c (analyze_function): Avoid calling
21215         mark_irreducible_loops twice.
21216         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
21218 2013-02-07  David S. Miller  <davem@davemloft.net>
21220         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
21221         on 'reg'.
21222         * var-tracking.c (vt_add_function_parameter): Test the presence of
21223         HAVE_window_save properly and do not remap argument registers when
21224         we have a leaf function.
21226 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
21228         PR bootstrap/56227
21229         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
21230         instead of "ll".
21231         * config/i386/i386.c (ix86_print_operand): Ditto.
21233 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
21235         * lra-constraints.c (process_alt_operands): Fix recently added comment.
21237 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
21239         PR rtl-optimization/56225
21240         * lra-constraints.c (process_alt_operands): Check that reload hard
21241         reg can hold value for strict_low_part.
21243 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
21245         PR debug/56154
21246         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
21247         dwarf2out_end_function.
21248         (in_first_function_p, maybe_at_text_label_p,
21249         first_loclabel_num_not_at_text_label): New variables.
21250         (dwarf2out_var_location): In the first function find out
21251         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
21252         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
21253         functions.
21255 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
21257         PR rtl-optimization/56178
21258         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
21259         SUBREG of a register.  Tidy up related block of code.
21260         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
21261         note if the source is a register or a SUBREG of a register.
21263 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
21265         PR target/56228
21266         * config/rs6000/rs6000.md (ptrm): New mode attr.
21267         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
21268         call_value_indirect_aix<pttrsize>,
21269         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
21270         m in constraints.
21272 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
21274         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
21275         if -bnortl. Convert to strcmp and strncmp.
21277 2013-02-07  Alan Modra  <amodra@gmail.com>
21279         PR target/54009
21280         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
21281         addresses won't wrap when offsetting.
21282         (rs6000_secondary_reload): Provide secondary reloads needed for
21283         wrapping LO_SUM addresses.
21285 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
21287         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
21288         MACH, just __MACH__.
21290 2013-02-06  Richard Biener  <rguenther@suse.de>
21292         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
21293         instead of calling fix_loop_structure.
21295 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
21297         PR middle-end/56217
21298         * omp-low.c (use_pointer_for_field): Return false if
21299         lower_send_shared_vars doesn't generate any copy-out code.
21301 2013-02-06  Tom de Vries  <tom@codesourcery.com>
21303         PR rtl-optimization/56131
21304         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
21305         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
21306         of the label is NULL.  Add comment.
21308 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
21310         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
21312         PR sanitizer/55374
21313         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
21314         (STATIC_LIBTSAN_LIBS): Likewise.
21315         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
21316         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
21317         is defined, don't add anything else beyond that.
21318         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
21319         (LINK_COMMAND_SPEC): Use them.
21321         PR tree-optimization/56205
21322         * tree-stdarg.c (check_all_va_list_escapes): Return true if
21323         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
21324         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
21326 2013-02-05  Richard Biener  <rguenther@suse.de>
21328         PR tree-optimization/53342
21329         PR tree-optimization/53185
21330         * tree-vectorizer.h (vect_check_strided_load): Remove.
21331         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
21332         not disallow peeling for vectorized strided loads.
21333         (vect_check_strided_load): Make static and simplify.
21334         (vect_analyze_data_refs): Adjust.
21335         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
21336         correctly when vectorizing strided loads.
21338 2013-02-05  Richard Biener  <rguenther@suse.de>
21340         * doc/install.texi: Refer to ISL, not PPL.
21342 2013-02-05  Jan Hubicka  <jh@suse.cz>
21344         PR tree-optimization/55789
21345         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
21347 2013-02-05  Jan Hubicka  <jh@suse.cz>
21349         PR tree-optimization/55789
21350         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
21351         the dead call anyway.
21353 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
21355         PR sanitizer/55374
21356         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
21358 2013-02-04  Alexander Potapenko  <glider@google.com>
21359             Jack Howarth  <howarth@bromo.med.uc.edu>
21360             Jakub Jelinek  <jakub@redhat.com>
21362         PR sanitizer/55617
21363         * config/darwin.c (sort_ctor_records): Stabilized qsort
21364         on constructor priority by using original position.
21365         (finalize_ctors): New routine to sort constructors by
21366         priority before use in assemble_integer.
21367         (machopic_asm_out_constructor): Use finalize_ctors if needed.
21369 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
21371         PR libstdc++/54314
21372         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
21373         about visibility on artificial decls.
21374         * config/sol2.c (solaris_assemble_visibility): Likewise.
21376 2013-02-04  Kai Tietz  <ktietz@redhat.com>
21378         PR target/56186
21379         * config/i386/i386.c (function_value_ms_64): Add additional valtype
21380         argument and improve checking of return-argument types for 16-byte
21381         modes.
21382         (ix86_function_value_1): Add additional valtype argument on call
21383         of function_value_64.
21384         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
21385         handling infunction_value_64 function.
21387 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
21389         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
21391 2013-02-04  Richard Biener  <rguenther@suse.de>
21393         PR tree-optimization/56188
21394         * tree-ssa-structalias.c (label_visit): Consider case with
21395         initially non-empty points-to set.
21396         (perform_var_substitution): Dump node mapping and clean up.
21398 2013-02-04  Richard Guenther  <rguenther@suse.de>
21400         PR lto/56168
21401         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
21402         node prevail as last resort.
21403         (lto_symtab_merge_decls): Remove guard on LTRANS here.
21404         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
21406 2013-02-04  Richard Biener  <rguenther@suse.de>
21408         PR tree-optimization/56113
21409         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
21410         Merge into ...
21411         (equiv_class_lookup_or_add): ... this.
21412         (label_visit): Adjust and fix error in previous patch.
21413         (perform_var_substitution): Adjust.
21415 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
21417         * config/sh/divtab.c: Fix formatting and comments throughout the file.
21418         * config/sh/sh4-300.md: Likewise.
21419         * config/sh/sh4a.md: Likewise.
21420         * config/sh/constraints.md: Likewise.
21421         * config/sh/sh.md: Likewise.
21422         * config/sh/netbsd-elf.h: Likewise.
21423         * config/sh/predicates.md: Likewise.
21424         * config/sh/sh-protos.h: Likewise.
21425         * config/sh/ushmedia.h: Likewise.
21426         * config/sh/linux.h: Likewise.
21427         * config/sh/sh.c: Likewise.
21428         * config/sh/superh.h: Likewise.
21429         * config/sh/elf.h: Likewise.
21430         * config/sh/sh4.md: Likewise.
21431         * config/sh/sh.h: Likewise.
21433 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21435         * config/pa/constraints.md: Adjust unused letters.  Change "T"
21436         constraint to match_test floating_point_store_memory_operand().
21437         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
21438         (base14_operand): New.
21439         (floating_point_store_memory_operand): New.
21440         (integer_store_memory_operand): Revise to use base14_operand and
21441         reg_plus_base_memory_operand.
21442         (move_dest_operand): Allow symbolic_memory_operands.
21443         (symbolic_memory_operand): Check for LO_SOM.
21444         (symbolic_operand): Change default case to break.
21445         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
21446         CONST_DOUBLE values to be reloaded by putting them into memory when
21447         the destination is a floating point register.
21448         (movdf): Remove code to handle CONST_DOUBLE.
21449         (movsf): Likewise.
21450         (reload_indf_r1): New.
21451         (reload_insf_r1): New.
21452         Consistently use "Q" and "T" constraints with integer and floating
21453         point move instructions, respectively.
21454         (movdi): Remove FAIL.
21455         Change predicate for source operand unamed DImode move from
21456         general_operand to move_src_operand.
21457         (umulsidi3): Change predicate for destination operand to
21458         register_operand.
21459         Likewise for similar unamed patterns.
21460         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
21461         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
21462         (hppa_legitimize_address): Simplify mask calculation.
21463         (pa_emit_move_sequence): Revised handling of secondary reloads from
21464         REG+D addresses for floating point loads and stores.  Directly handle
21465         loading CONST0_RTX (mode) to a floating point register.
21466         (pa_secondary_reload): Handle reloading DF and SFmode constant values
21467         to floating point registers.  Don't restrict secondary reloads to
21468         floating point registers to integer modes.  Revise some comments and
21469         cleanup some code.
21470         (TARGET_LEGITIMATE_ADDRESS_P): Define.
21471         (pa_legitimate_address_p): New.
21472         (pa_legitimize_reload_address): New.
21473         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
21474         (STRICT_REG_OK_FOR_BASE_P): New.
21475         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
21476         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
21478 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
21479             Andrew Dixie  <andrewd@gentrack.com>
21481         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
21482         flag set.
21484 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
21486         * expmed.c (extract_bit_field_1): Pass the full width of the
21487         structure to get_best_reg_extraction_insn.
21489 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
21491         PR target/54601
21492         * configure.ac (use_cxa_atexit): Add AIX.
21493         * configure: Regenerate.
21495         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
21497 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
21499         PR debug/54793
21500         * final.c (need_profile_function): New variable.
21501         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
21502         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
21503         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
21504         notes, targetm.asm_out.function_prologue doesn't emit anything,
21505         HAVE_prologue and profiler should be emitted before prologue,
21506         set need_profile_function instead of emitting it.
21507         (final_scan_insn): If need_profile_function, emit
21508         profile_function on the first NOTE_INSN_BASIC_BLOCK or
21509         NOTE_INSN_FUNCTION_BEG note.
21511 2013-02-01  Richard Henderson  <rth@redhat.com>
21513         * config/rs6000/rs6000.md (smulditi3): New.
21514         (umulditi3): New.
21516         * config/alpha/alpha.md (umulditi3): New.
21518 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
21520         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
21521         (ASM_OUTPUT_ALIGNED_LOCAL): New.
21523 2013-02-01  Richard Biener  <rguenther@suse.de>
21525         PR tree-optimization/56113
21526         * tree-ssa-structalias.c (label_visit): Reduce work for
21527         single-predecessor nodes.
21529 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
21531         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
21532         range isn't testing for zero.
21534 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
21536         PR middle-end/56113
21537         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
21539 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
21540             Nick Clifton  <nickc@redhat.com>
21542         * config/v850/constraints.md (Q): Define as a memory constraint.
21543         * config/v850/predicates.md (label_ref_operand): New predicate.
21544         (e3v5_shift_operand): New predicate.
21545         (ior_operator): New predicate.
21546         * config/v850/t-v850: Add e3v5 multilib.
21547         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
21548         (v850_gen_movdi): Prototype.
21549         * config/v850/v850.c: Add support for e3v5 architecture.
21550         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
21551         TARGET_V850E_UP.
21552         (construct_save_jarl): Add e3v5 long JARL support.
21553         (v850_adjust_insn_length): New function.  Adjust length of call
21554         insns when using e3v5 instructions.
21555         (v850_gen_movdi): New function: Generate instructions to move a
21556         DImode value.
21557         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
21558         (CPP_SPEC): Define __v850e3v5__ as appropriate.
21559         (TARGET_USE_FPU): Enable for e3v5.
21560         (CONST_OK_FOR_W): New macro.
21561         (ADJUST_INSN_LENGTH): Define.
21562         * config/v850/v850.md (UNSPEC_LOOP): Define.
21563         (attr cpu): Add v850e3v5.
21564         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
21565         (movdi): New pattern.
21566         (movdi_internal): New pattern.
21567         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
21568         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
21569         (cstoresf4): Likewise.
21570         (cstoredf4): Likewise.
21571         (insv): New pattern.
21572         (rotlso3_a): New pattern.
21573         (rotlsi3_b): New pattern
21574         (rotlsi3_v850e3v5): New pattern.
21575         (doloop_begin): New pattern.
21576         (fix_loop_counter): New pattern.
21577         (doloop_end): New pattern.
21578         (branch_normal): Add e3v5 long branch support.
21579         (branch_invert): Likewise.
21580         (branch_z_normal): Likewise.
21581         (branch_z_invert): Likewise.
21582         (branch_nz_normal): Likewise.
21583         (branch_nz_invert): Likewise.
21584         (call_internal_short): Add e3v5 register-indirect JARL support.
21585         (call_internal_long): Likewise.
21586         (call_value_internal_short): Likewise.
21587         (call_value_internal_long): Likewise.
21588         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
21589         (mloop): New option.
21590         * config.gcc: Add support for configuring v840e3v5 target.
21591         * doc/invoke.texi: Document new v850 specific command line options.
21593 2013-01-31  Paul Koning  <ni1d@arrl.net>
21595         PR debug/55059
21596         PR debug/54508
21597         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
21598         children if parent is a class.
21599         (prune_unused_types_prune): Don't add DW_AT_declaration.
21601 2013-01-31  Richard Biener  <rguenther@suse.de>
21603         PR tree-optimization/56157
21604         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
21605         match up operand with SLP child.
21607 2013-01-31  Jason Merrill  <jason@redhat.com>
21609         PR debug/54410
21610         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
21611         parameters the first time.
21612         (gen_scheduled_generic_parms_dies): Check completeness here.
21614 2013-01-31  Richard Biener  <rguenther@suse.de>
21616         PR middle-end/53073
21617         * common.opt (faggressive-loop-optimizations): New flag,
21618         enabled by default.
21619         * doc/invoke.texi (faggressive-loop-optimizations): Document.
21620         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
21621         infer_loop_bounds_from_undefined by it.
21623 2013-01-31  Richard Biener  <rguenther@suse.de>
21625         PR tree-optimization/56150
21626         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
21627         visit virtual operands.
21628         (find_uses_to_rename_bb): Likewise.
21630 2013-01-31  Richard Biener  <rguenther@suse.de>
21632         PR tree-optimization/56150
21633         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
21634         mixed store non-store stmts.
21636 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
21638         PR sanitizer/55374
21639         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
21640         LIBASAN_EARLY_SPEC is defined.
21641         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
21642         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
21643         before %o.
21644         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
21646         PR c++/55742
21647         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
21648         invalid args instead of ICEing on it.
21649         (ix86_valid_target_attribute_tree): Return error_mark_node if
21650         ix86_valid_target_attribute_inner_p failed.
21651         (ix86_valid_target_attribute_p): Return false only if
21652         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
21653         target("default") attribute.
21654         (sorted_attr_string): Change argument from const char * to tree,
21655         merge in all target attribute arguments rather than just one.
21656         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
21657         instead of free.  Avoid using strcat.
21658         (ix86_mangle_function_version_assembler_name): Mangle
21659         target("default") as if no target attribute is present.  Adjust
21660         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
21661         instead of xmalloc and XDELETEVEC instead of free.
21662         (ix86_function_versions): Don't return true if one of the decls
21663         doesn't have target attribute.  If they don't and one of the decls
21664         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
21665         sorted_attr_string caller.  Use XDELETEVEC instead of free.
21666         (ix86_supports_function_versions): Remove.
21667         (make_name): Fix up formatting.
21668         (make_dispatcher_decl): Remove resolver_name and its initialization.
21669         Avoid leaking memory.
21670         (is_function_default_version): Return true if there is
21671         target("default") attribute rather than no target attribute at all.
21672         (make_resolver_func): Avoid leaking memory.
21673         (ix86_generate_version_dispatcher_body): Likewise.
21674         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
21675         * target.def (supports_function_versions): Remove.
21676         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
21677         * doc/tm.texi: Regenerated.
21679 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
21681         PR rtl-optimization/56144
21682         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
21683         for values with side effects.
21685 2013-01-30  Richard Biener  <rguenther@suse.de>
21687         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
21688         (sparseset_pop): Likewise.
21689         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
21690         to be able to use quick_push in the worker loop.
21692 2013-01-30  Marek Polacek  <polacek@redhat.com>
21694         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
21696 2013-01-30  Richard Biener  <rguenther@suse.de>
21698         PR lto/56147
21699         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
21701 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
21703         PR tree-optimization/56064
21704         * fixed-value.c (fixed_from_double_int): New function.
21705         * fixed-value.h (fixed_from_double_int): New prototype.
21706         (const_fixed_from_double_int): New static inline function.
21707         * fold-const.c (native_interpret_fixed): New static function.
21708         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
21709         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
21710         (native_encode_fixed): New static function.
21711         (native_encode_expr) <FIXED_CST>: Use it.
21712         (native_interpret_int): Move double_int worker code to...
21713         * double-int.c (double_int::from_buffer): ...this new static method.
21714         * double-int.h (double_int::from_buffer): Prototype it.
21716 2013-01-30  Richard Biener  <rguenther@suse.de>
21718         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
21719         New pointer-map and obstack.
21720         (init_alias_vars): Allocate pointer-map and obstack.
21721         (delete_points_to_sets): Free them.
21722         (find_what_var_points_to): Cache result.
21723         (find_what_p_points_to): Adjust for changed interface of
21724         find_what_var_points_to.
21725         (compute_points_to_sets): Likewise.
21726         (ipa_pta_execute): Likewise.
21728 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
21730         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
21731         * configure: Regenerate.
21732         * config.in: Regenerate.
21733         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
21734         #nobits/#progbits if supported.
21736 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
21738         PR target/56121
21739         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
21740         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
21741         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
21743 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21745         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
21746         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
21748 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21750         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
21751         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
21753 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21755         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
21756         declaration.
21757         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
21758         * config/arm/cortex-a7.md: New bypasses using
21759         arm_mac_accumulator_is_result.
21761 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21763         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
21764         (cortex_a7_neon_mla): Likewise.
21765         (cortex_a7_fpfmad): New reservation.
21766         (cortex_a7_fpmacs): Use ffmas and update required units.
21767         (cortex_a7_fpmuld): Update required units and latency.
21768         (cortex_a7_fpmacd): Likewise.
21769         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
21770         (cortex_a7_neon). Likewise.
21771         (bypass) Update participating units.
21773 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21775         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
21776         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
21777         from fmac to ffma.
21778         * config/arm/vfp11.md (vfp_farith): Use ffmas.
21779         (vfp_fmul): Use ffmad.
21780         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
21781         (cortex_r4_fmacd): Use ffmad.
21782         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
21783         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
21784         (cortex_a9_fmacd): Use ffmad.
21785         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
21786         (cortex_a8_vfp_macd): Use ffmad.
21787         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
21788         (cortex_a5_fpmacd): Use ffmad.
21789         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
21790         (cortex_a15_vfp_macd): Use ffmad.
21791         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
21793 2013-01-29  Jason Merrill  <jason@redhat.com>
21795         PR libstdc++/54314
21796         * varasm.c (default_assemble_visibility): Don't warn about
21797         visibility on artificial decls.
21799 2013-01-29  Richard Biener  <rguenther@suse.de>
21801         PR tree-optimization/56113
21802         * tree-ssa-structalias.c (equiv_class_lookup): Also return
21803         the bitmap leader.
21804         (label_visit): Free duplicate bitmaps and record the leader instead.
21805         (perform_var_substitution): Adjust.
21807 2013-01-29  Richard Biener  <rguenther@suse.de>
21809         PR tree-optimization/55270
21810         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
21811         the CFG, schedule loops for fixup.
21813 2013-01-29  Nick Clifton  <nickc@redhat.com>
21815         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
21816         SP_REG.
21818 2013-01-28  Leif Ekblad  <leif@rdos.net>
21820         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
21821         * config/i386/i386.h (TARGET_RDOS): New macro.
21822         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
21823         * config/i386/i386.c (ix86_option_override_internal): For 64bit
21824         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
21825         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
21826         DEFAULT_LARGE_SECTION_THRESHOLD.
21827         * config/i386/i386.md (R14_REG, R15_REG): New constants.
21828         * config/i386/rdos.h: New file.
21829         * config/i386/rdos64.h: New file.
21831 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
21833         PR other/54814
21834         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
21835         TEST_HARD_REG_BIT.
21837 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21839         PR rtl-optimization/56117
21840         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
21841         call cselib_lookup_from_insn on the MEM before calling
21842         add_insn_mem_dependence.
21844 2013-01-28  Richard Biener  <rguenther@suse.de>
21846         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
21847         to a stmt that didn't have one.
21848         (copy_phis_for_bb): Likewise for PHI arguments.
21849         (copy_debug_stmt): Likewise for debug stmts.
21851 2013-01-28  Richard Biener  <rguenther@suse.de>
21853         PR tree-optimization/56034
21854         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
21855         (partition_builtin_p): Adjust.
21856         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
21857         it is the last partition.
21858         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
21859         up the vertex for the definition.
21860         (classify_partition): Classify whether a partition is a
21861         PKIND_REDUCTION, thus has uses outside of the loop.
21862         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
21863         Merge all PKIND_REDUCTION partitions into the last partition.
21864         (tree_loop_distribution): Seed partitions from reductions as well.
21866 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21868         PR tree-optimization/56125
21869         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
21870         pow(x,c) into sqrt(x) * powi(x, n/2) or
21871         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
21872         optimizing for size.
21873         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
21874         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
21875         integer.
21877         PR tree-optimization/56094
21878         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
21879         to UNKNOWN_LOCATION while gimplifying expr.
21881 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
21883         PR target/56114
21884         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
21885         operand 0 in movabs insn template for -masm=intel asm alternative.
21886         (*movabs<mode>_2): Ditto for operand 1.
21888 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
21890         PR target/54663
21891         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
21892         of microblaze-c.o
21894 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21896         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
21897         tm_file.
21899 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21901         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
21902         Undef to avoid warning.
21904 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
21906         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
21907         * configure: Regenerate.
21909 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
21911         PR tree-optimization/56098
21912         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
21913         for stmts with volatile ops.
21914         (cond_store_replacement): Don't optimize if assign has volatile ops.
21915         (cond_if_else_store_replacement_1): Don't optimize if either
21916         then_assign or else_assign have volatile ops.
21917         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
21918         volatile ops.
21920 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21922         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
21924 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21926         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
21927         missing ':' in asm example.
21929 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
21931         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
21932         entries into lane and laneq entries.
21933         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
21934         Remove AdvSIMD scalar modes.
21935         (aarch64_sq<r>dmulh_laneq<mode>): New.
21936         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
21937         modes.
21938         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
21939         builtin implementations to relfect changes in RTL in aarch64-simd.md.
21940         * config/aarch64/iterators.md (VCOND): New.
21941         (VCONQ): New.
21943 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21945         PR target/54222
21946         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
21947         Add NULL LIBNAME argument to existing definitions.
21948         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
21949         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
21950         * config/avr/avr.c (DEF_BUILTIN): Same.
21951         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
21952         (avr_expand_builtin): Expand to a vanilla call if a libgcc
21953         implementation is available (DECL_ASSEMBLER_NAME is set).
21954         (avr_fold_absfx): New static function.
21955         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
21956         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
21957         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
21958         AVR_BUILTIN_ABSLLK.
21959         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
21960         (abshk, absk, abslk, absllk): Provide as static inline functions.
21962 2013-01-25  Marek Polacek  <polacek@redhat.com>
21964         PR tree-optimization/56035
21965         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
21967 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
21969         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
21970         (*movtf_internal_rex64): Add (!o,C) alternative
21971         (*movxf_internal_rex64): Ditto.
21972         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
21974 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
21976         * doc/invoke.texi: fix typo.
21977         * doc/objc.texi: fix typo.
21979 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
21981         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
21982         for the first two alternatives.
21984 2013-01-24  Diego Novillo  <dnovillo@google.com>
21986         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
21987         (ggc-zone.o): Remove.
21988         * configure.ac: Remove option --with-gc.
21989         * configure: Re-generate.
21990         * doc/install.texi: Remove documentation for --with-gc.
21991         * gengtype.c (write_enum_defn): Remove.  Update all users.
21992         (write_Types_process_field): Remove generation of gt_e_* argument.
21993         (output_type_enum): Remove.  Update all users.
21994         (write_enum_defn): Remove.  Update all users.
21995         (enum alloc_zone): Remove.  Update all users.
21996         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
21997         * ggc-common.c (ggc_splay_alloc): Remove first argument.
21998         Update all callers.
21999         (struct ptr_data): Remove field TYPE.  Update all users.
22000         (gt_pch_note_object): Remove argument TYPE.  Update all users.
22001         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
22002         Update all users.
22003         * ggc-none.c (ggc_alloc_typed_stat): Remove.
22004         (struct alloc_zone): Remove.
22005         (ggc_internal_alloc_zone_stat): Remove.
22006         (ggc_internal_cleared_alloc_zone_stat): Remove.
22007         * ggc-page.c (ggc_alloc_typed_stat): Remove.
22008         (ggc_pch_count_object): Remove last argument.  Update all users.
22009         (ggc_pch_alloc_object): Remove last argument.  Update all users.
22010         (struct alloc_zone): Remove.
22011         * ggc-zone.c: Remove.
22012         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
22013         (struct alloc_zone): Remove.
22014         (ggc_alloc_typed_stat): Remove.
22015         (ggc_alloc_typed): Remove.
22016         (ggc_splay_alloc): Remove first argument.
22017         (rtl_zone): Remove.  Update all users.
22018         (tree_zone): Remove.  Update all users.
22019         (tree_id_zone): Remove.  Update all users.
22020         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
22021         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
22022         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
22023         * tree-ssanames.c: Remove references to zone allocator in comments.
22025 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
22027         * config/avr/avr.c (avr_out_fract): Make register numbers that
22028         might be outside of source operand signed.
22030 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
22032         * config/i386/constraints.md (Yf): New constraint.
22033         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
22034         of f constraint to conditionaly disable x87 register preferences.
22035         (*movdf_internal): Ditto.
22036         (*movsf_internal): Ditto.
22038 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
22040         PR inline-asm/55934
22041         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
22042         that have operands with impossible constraints.
22043         Add a FIXME for a speed-up opportunity.
22044         * lra-constraints.c (process_alt_operands): Verify that a class
22045         selected from constraints on asms is valid for the operand mode.
22046         (curr_insn_transform): Remove incorrect comment.
22048 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
22050         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
22051         TOC operand is a valid symbol ref in the constant pool.
22053 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
22055         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
22057 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
22059         PR target/54222
22060         * config/avr/stdfix.h: New file.
22061         * t-avr (stdfix-gcc.h): New rule to build it.
22062         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
22064 2013-01-23  Kostya Serebryany  <kcc@google.com>
22066         * config/darwin.h: remove dependency on
22067         CoreFoundation (asan on Mac OS).
22069 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
22071         PR target/49069
22072         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
22073         instead of cmpdi_operand for first comparison operand.
22074         Don't assert that comparison operands aren't both constants.
22076 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
22078         * doc/install.texi (Downloading the Source): Update references to
22079         downloading separate components.
22081 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
22083         * doc/extend.texi (__int128): Improve grammar.
22085 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
22087         PR target/56028
22088         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
22089         alternative to (o,r).
22090         (*movdi_internal_rex64): Remove (!o,n) alternative.
22091         (DImode immediate->memory splitter): Remove.
22092         (DImode immediate->memory peephole2): Remove.
22093         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
22094         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
22095         alternative to (!o,*r).
22096         (*movtf_internal_sse): New pattern.
22097         (*movxf_internal_rex64): New pattern.
22098         (*movxf_internal): Disable for TARGET_64BIT.
22099         (*movdf_internal_rex64): Remove (!o,F) alternative.
22101 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
22103         PR middle-end/56074
22104         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
22105         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
22106         * tree-vect-loop-manip.c (find_loop_location): Also ignore
22107         stmt locations where LOCATION_LOCUS of the stmt location is
22108         UNKNOWN_LOCATION or BUILTINS_LOCATION.
22110         PR target/55686
22111         * config/i386/i386.md (UNSPEC_STOS): New.
22112         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
22113         *strsetqi_1): Add UNSPEC_STOS.
22115 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
22117         PR c++/56067
22118         * doc/invoke.texi: Remove left over -Wsynth example.
22120 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
22122         PR tree-optimization/56051
22123         * fold-const.c (fold_binary_loc): Don't fold
22124         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
22125         a narrowing conversion, or widening conversion from signed
22126         to unsigned.
22128 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
22130         PR rtl-optimization/56023
22131         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
22132         dependent on debug instruction.
22134 2013-01-21  Martin Jambor  <mjambor@suse.cz>
22136         PR middle-end/56022
22137         * function.c (allocate_struct_function): Call
22138         invoke_set_current_function_hook earlier.
22140 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
22142         * reload1.c (init_reload): Only initialize reload_obstack
22143         during the first call.
22145 2013-01-21  Marek Polacek  <polacek@redhat.com>
22147         * cfgloop.c (verify_loop_structure): Fix up grammar.
22149 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
22151         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
22152         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
22154 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
22156         PR target/56058
22157         * config/arm/marvell-pj4.md: Update copyright year.
22158         Fix up use of alu to alu_reg and simple_alu_imm.
22160 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
22162         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
22164 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
22166         PR target/55433
22167         * lra-constraints.c (curr_insn_transform): Don't reuse original
22168         insn for secondary memory move when memory mode should be different.
22170 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22172         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
22173         atomic_storedi_1): New patterns.
22175 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
22177         btver2 pipeline descriptions.
22178         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
22179         descriptions.
22180         * config/i386/i386.md (btver2_decode): New type attributes.
22181         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
22182         type attributes.
22183         * config/i386/btver2.md: New file describing btver2 pipelines.
22185 2013-01-19  Andrew Pinski  <apinski@cavium.com>
22187         PR tree-optimization/52631
22188         * tree-ssa-sccvn (visit_use): Before looking up the original
22189         statement, try looking up the simplified expression.
22191 2013-01-19  Anthony Green  <green@moxielogic.com>
22193         * config/moxie/moxie.c (moxie_expand_prologue): Set
22194         current_function_static_stack_size.
22196 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
22198         PR tree-optimization/56029
22199         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
22200         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
22202 2013-01-18  Sharad Singhai  <singhai@google.com>
22204         PR tree-optimization/55995
22205         * dumpfile.c (dump_loc): Print location only if available.
22206         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
22208 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
22210         PR target/55433
22211         * lra-constraints.c (curr_insn_transform): Reuse original insn for
22212         secondary memory move.
22213         (inherit_reload_reg): Use rclass instead of cl for
22214         check_secondary_memory_needed_p.
22216 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
22218         PR middle-end/56015
22219         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
22220         the case where writing real complex part of target modifies op1.
22222 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
22224         * config/aarch64/aarch64-simd.md
22225         (aarch64_vcond_internal<mode>): Handle unordered cases.
22226         * config/aarch64/iterators.md (v_cmp_result): New.
22228 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
22229             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
22231         * config/arm/marvell-pj4.md: New file.
22232         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
22233         * config/arm/arm.md (generic_sched): Add marvell_pj4.
22234         (generic_vfp): Likewise.
22235         * config/arm/arm-cores.def: Add marvell-pj4.
22236         * config/arm/arm-tune.md: Regenerate.
22237         * config/arm/arm-tables.opt: Regenerate.
22238         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
22239         * doc/invoke.texi: Document marvell-pj4.
22241 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
22243         * config/aarch64/arm_neon.h: Map scalar types to standard types.
22245 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
22247         PR debug/54114
22248         PR debug/54402
22249         PR debug/49888
22250         * var-tracking.c (negative_power_of_two_p): New.
22251         (global_get_addr_cache, local_get_addr_cache): New.
22252         (get_addr_from_global_cache, get_addr_from_local_cache): New.
22253         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
22254         heading comment.
22255         (vt_stack_offset_p): Remove.
22256         (vt_canon_true_dep): Always canonicalize loc's address.
22257         (clobber_overlapping_mems): Make sure we have a MEM.
22258         (local_get_addr_clear_given_value): New.
22259         (val_reset): Clear local cached entries.
22260         (compute_bb_dataflow): Create and release the local cache.
22261         Disable duplicate MEMs clobbering.
22262         (emit_notes_in_bb): Clobber MEMs likewise.
22263         (vt_emit_notes): Create and release the local cache.
22264         (vt_initialize, vt_finalize): Create and release the global
22265         cache, respectively.
22266         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
22268 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
22270         PR libmudflap/53359
22271         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
22272         not found in the symtab.
22274 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
22276         PR debug/56006
22277         PR rtl-optimization/55547
22278         PR rtl-optimization/53827
22279         PR debug/53671
22280         PR debug/49888
22281         * alias.c (offset_overlap_p): New, factored out of...
22282         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
22283         the conservative special case for symbolic constants.  Don't
22284         adjust zero sizes on alignment.
22286 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
22288         PR rtl-optimization/52573
22289         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
22290         REG_UNUSED for the same register.
22292 2013-01-17  Richard Biener  <rguenther@suse.de>
22293             Marek Polacek  <polacek@redhat.com>
22295         PR rtl-optimization/55833
22296         * loop-unswitch.c (unswitch_loops): Move loop verification...
22297         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
22298         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
22299         Set it to true when we're removing a loop from hierarchy tree in
22300         an irreducible region.
22301         (fix_bb_placements): Adjust caller.
22302         (fix_loop_placements): Likewise.
22304 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
22306         * config/avr/builtins.def (DEF_BUILTIN): Factor out
22307         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
22308         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
22309         Remove ID.  Adjust comments.
22310         * config/avr/avr-c.c (avr_builtin_name): Remove.
22311         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
22312         * config/avr/avr.c (avr_tolower): New static function.
22313         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
22314         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
22315         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
22316         default expansion.
22318 2013-01-17  Jan Hubicka  <jh@suse.cz>
22320         PR tree-optimization/55273
22321         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
22323 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
22325         PR target/55981
22326         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
22327         store through atomic_store<mode>_1.
22328         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
22330 2013-01-17  Martin Jambor  <mjambor@suse.cz>
22332         PR tree-optimizations/55264
22333         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
22334         for virtual methods.
22335         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
22336         virtual methods before inlining is over.
22337         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
22338         virtual functions.
22339         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
22340         non-virtual.
22342 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
22344         PR rtl-optimization/56005
22345         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
22346         pending reads for prefetch.
22348 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
22350         * config/aarch64/aarch64.md
22351         (*cstoresi_neg_uxtw): New pattern.
22352         (*cmovsi_insn_uxtw): New pattern.
22353         (*<optab>si3_uxtw): New pattern.
22354         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
22355         (*<optab>si3_insn_uxtw): New pattern.
22356         (*bswapsi2_uxtw): New pattern.
22358 2013-01-16  Richard Biener  <rguenther@suse.de>
22360         * tree-inline.c (tree_function_versioning): Remove set but
22361         never used variable.
22363 2013-01-16  Richard Biener  <rguenther@suse.de>
22365         PR tree-optimization/55964
22366         * tree-flow.h (rename_variables_in_loop): Remove.
22367         (rename_variables_in_bb): Likewise.
22368         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
22369         (copy_loop_before): Adjust and delete update-ssa status.
22370         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
22371         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
22372         (rename_variables_in_loop): Remove.
22373         (slpeel_update_phis_for_duplicate_loop): Likewise.
22374         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
22375         use available cfg machinery instead of duplicating it.
22376         Update PHI nodes and perform poor-mans SSA update here.
22377         (slpeel_tree_peel_loop_to_edge): Adjust.
22379 2013-01-16  Richard Biener  <rguenther@suse.de>
22381         PR tree-optimization/54767
22382         PR tree-optimization/53465
22383         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
22384         (vrp_visit_phi_node): For PHI arguments coming via backedges
22385         drop all symbolical range information.
22386         (execute_vrp): Compute backedges.
22388 2013-01-16  Richard Biener  <rguenther@suse.de>
22390         * doc/install.texi: Update CLooG and ISL requirements to
22391         0.18.0 and 0.11.1.
22393 2013-01-16  Christian Bruel  <christian.bruel@st.com>
22395         PR target/55301
22396         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
22397         (broken_move): Handle UNSPECV_SP_SWITCH_B.
22398         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
22400 2013-01-16  DJ Delorie  <dj@redhat.com>
22402         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
22403         (UNSPECV_SP_SWITCH_E): New.
22404         (sp_switch_1): Change to an unspec.
22405         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
22406         replace $r15.
22408 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
22410         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
22411         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
22412         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
22413         (expand_mem_thread_fence): Ditto.
22414         (expand_mem_signal_fence): Ditto.
22415         (expand_atomic_load): Ditto.
22416         (expand_atomic_store): Ditto.
22418 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
22420         PR rtl-optimization/55547
22421         PR rtl-optimization/53827
22422         PR debug/53671
22423         PR debug/49888
22424         * alias.c (memrefs_conflict_p): Set sizes to negative after
22425         AND adjustments.
22427 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
22429         PR target/55940
22430         * function.c (thread_prologue_and_epilogue_insns): Always
22431         add crtl->drap_reg to set_up_by_prologue.set, even if
22432         stack_realign_drap is false.
22434 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
22436         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
22437         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
22438         *call): Fix indention.
22440 2013-01-15  Tom de Vries  <tom@codesourcery.com>
22442         PR target/55876
22443         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
22444         Update comment.
22446 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
22448         PR rtl-optimization/55153
22449         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
22451 2013-01-15  Martin Jambor  <mjambor@suse.cz>
22453         PR tree-optimization/55920
22454         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
22455         accesses as grp_to_be_debug_replaced.
22457 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
22459         PR tree-optimization/55920
22460         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
22461         there is non-useless type conversion needed from debug rhs to lhs,
22462         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
22464 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
22465             Mikael Pettersson  <mikpe@it.uu.se>
22467         PR target/43961
22468         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
22469         Thumb.
22470         (ASM_OUTPUT_CASE_LABEL): Remove.
22471         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
22472         * final.c (shorten_branches): Update alignment of labels before
22473         jump tables if CASE_VECTOR_SHORTEN_MODE.
22475 2013-01-15  Richard Biener  <rguenther@suse.de>
22477         PR bootstrap/55961
22478         * system.h: Do not include gmp.h for building host tools.
22480 2013-01-15  Richard Biener  <rguenther@suse.de>
22482         PR middle-end/55882
22483         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
22484         account for bitpos when computing alignment.
22486 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
22488         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
22489         (ix86_target_macros_internal): Likewise.
22491         * config/i386/i386.c (m_CORE2I7): Removed.
22492         (m_CORE_HASWELL): New macro.
22493         (m_CORE_ALL): Likewise.
22494         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
22495         (initial_ix86_arch_features): Likewise.
22496         (processor_target_table): Initializations for Core avx2.
22497         (cpu_names): New names "core-avx2".
22498         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
22499         PROCESSOR_CORE_HASWELL.
22500         (ix86_issue_rate): New case.
22501         (ia32_multipass_dfa_lookahead): Likewise.
22502         (ix86_sched_init_global): Likewise.
22504         * config/i386/i386.h (TARGET_HASWELL): New macro.
22505         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
22506         (processor_type): New PROCESSOR_HASWELL.
22508 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
22510         PR tree-optimization/55955
22511         * tree-vect-loop.c (vectorizable_reduction): Give up early on
22512         *SHIFT_EXPR and *ROTATE_EXPR codes.
22514         PR tree-optimization/48766
22515         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
22516         -ftrapv disable -fwrapv.
22518 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22520         PR target/55974
22521         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
22522         etc. to 1 and not to __flash.
22523         Use LL suffix for __INT24_MAX__ with -mint8.
22524         Use ULL suffix for __UINT24_MAX__ with -mint8.
22526 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22528         * config/avr/avr-arch.h
22529         (struct base_arch_s): Use typedef avr_arch_t instead.
22530         (struct arch_info_s): Use typedef avr_arch_info_t instead.
22531         (struct mcu_type_s): Use typedef avr_mcu_t instead.
22532         * config/avr/avr.c: Same.
22533         * config/avr/avr-devices.c: Same.
22534         * config/avr/driver-avr.c: Same.
22535         * config/avr/gen-avr-mmcu-texi.c: Same.
22536         * config/avr/avr-mcus.def: Adjust comment.
22538 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
22540         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
22541         * config/aarch64/iterators.md (VALLDI): New.
22543 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
22544             Andi Kleen  <ak@linux.intel.com>
22546         PR target/55948
22547         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
22548         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
22549         memmodel flag.
22551 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22553         * config/avr/avr-stdint.h: Remove trailing blanks.
22554         * config/avr/avr-log.h: Same.
22555         * config/avr/avr-arch.h: Same.
22556         * config/avr/avr-devices.c: Same.
22557         * config/avr/avr-dimode.md: Same.
22558         * config/avr/predicates.md: Same.
22559         * config/avr/avr-c.c: Same.  And fix typo.
22561         * config/avr/avr-protos.h: Same.  And:
22562         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
22563         (init_cumulative_args): Rename to avr_init_cumulative_args.
22564         (expand_prologue): Rename to avr_expand_prologue.
22565         (expand_epilogue): Rename to avr_expand_epilogue.
22566         (adjust_insn_length): Rename to avr_adjust_insn_length.
22567         (notice_update_cc): Rename to avr_notice_update_cc.
22568         (final_prescan_insn): Rename to avr_final_prescan_insn.
22569         * config/avr/avr.c: Same.
22570         * config/avr/avr.h: Same.
22571         * config/avr/avr.md: Remove trailing blanks.
22572         (prologue): Use avr_expand_prologue.
22573         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
22575 2013-01-14  Richard Biener  <rguenther@suse.de>
22577         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
22578         verify_location, collect_subblocks): New functions.
22579         (verify_gimple_in_cfg): Verify that locations only reference
22580         BLOCKs in the functions BLOCK tree.
22582 2013-01-14  Richard Biener  <rguenther@suse.de>
22584         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
22585         PHI argument.
22586         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
22587         unshare reference.
22588         (insert_out_of_ssa_copy_on_edge): Likewise.
22589         (rewrite_close_phi_out_of_ssa): Likewise.
22590         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
22591         debug expressions.
22592         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
22593         propagated constants.
22594         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
22595         can not be shared.
22597 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22599         * config/avr/avr-modes.def: Add GPL copyright notice.
22601 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
22603         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
22604         MEMMODEL_MASK to determine memory model.
22605         (atomic_store<mode>): Ditto from operands[2].
22606         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
22608 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
22610         PR fortran/55935
22611         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
22612         (fold_gimple_assign): Don't call unshare_expr here.
22613         (fold_ctor_reference): Call unshare_expr.
22615 2013-01-13  Terry Guo  <terry.guo@arm.com>
22617         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
22618         * doc/fragments.texi: Document MULTILIB_REUSE.
22619         * gcc.c (multilib_reuse): New internal spec.
22620         (set_multilib_dir): Also search multilib from multilib_reuse.
22621         * genmultilib (tmpmultilib3): Refactor code.
22622         (tmpmultilib4): Ditto.
22623         (multilib_reuse): New multilib argument.
22625 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
22627         * Makefile.in: Update copyright.
22629 2013-01-12  Tom de Vries  <tom@codesourcery.com>
22631         PR middle-end/55890
22632         * calls.c (expand_call): Check if arg_nr is valid.
22634 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
22636         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
22637         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
22638         documentation.  Add missing '__' in front of
22639         __builtin_ia32_packssdw256.
22641 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22643         PR target/55719
22644         * config/s390/s390.c (s390_preferred_reload_class): Do not return
22645         NO_REGS for larl operands.
22646         (s390_reload_larl_operand): Use s390_load_address instead of
22647         emit_move_insn.
22649 2013-01-11  Richard Biener  <rguenther@suse.de>
22651         * tree-cfg.c (verify_node_sharing_1): Split out from ...
22652         (verify_node_sharing): ... here.
22653         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
22655 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
22657         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
22658         Substitute TREECHECKING.
22659         * configure: Regenerate.
22660         * Makefile.in (TREECHECKING): New.
22662 2013-01-11  Richard Guenther  <rguenther@suse.de>
22664         PR tree-optimization/44061
22665         * tree-vrp.c (extract_range_basic): Compute zero as
22666         value-range for __builtin_constant_p of function parameters.
22668 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
22670         Update copyright years.
22672 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
22674         PR rtl-optimization/55672
22675         * lra-eliminations.c (mark_not_eliminable): Permit addition with
22676         const to be eliminable.
22678 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
22680         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
22681         * configure: Regenerate.
22683 2013-01-10  Richard Biener  <rguenther@suse.de>
22685         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
22687 2013-01-10  Richard Biener  <rguenther@suse.de>
22689         PR bootstrap/55792
22690         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
22691         locations for virtual PHI arguments.
22692         (rewrite_update_phi_arguments): Likewise.
22694 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
22696         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
22697         on to assembler.
22699 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
22701         PR tree-optimization/55921
22702         * tree-complex.c (expand_complex_asm): New function.
22703         (expand_complex_operations_1): Call it for GIMPLE_ASM.
22705 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22707         PR target/55718
22708         * config/s390/s390.c (s390_symref_operand_p)
22709         (s390_loadrelative_operand_p): Merge the two functions.
22710         (s390_check_qrst_address, print_operand_address): Add parameters
22711         to s390_loadrelative_operand_p invokation.
22712         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
22713         (s390_reload_larl_operand, s390_secondary_reload): Use
22714         s390_loadrelative_operand_p instead of s390_symref_operand_p.
22715         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
22717 2013-01-09  Mike Stump  <mikestump@comcast.net>
22719         * dse.c (record_store): Remove unnecessary assert.
22721 2013-01-09  Jan Hubicka  <jh@suse.cz>
22723         PR tree-optimization/55569
22724         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
22725         * cfgloop.h (scale_loop_profile): Likewise.
22727 2013-01-09  Jan Hubicka  <jh@suse.cz>
22729         PR lto/45375
22730         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
22731         functions.
22732         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
22734 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
22736         PR middle-end/55114
22737         * expr.h (maybe_emit_group_store): Declare.
22738         * expr.c (maybe_emit_group_store): New function.
22739         * builtins.c (expand_builtin_int_roundingfn): Call it.
22740         (expand_builtin_int_roundingfn_2): Likewise.
22742 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
22744         PR rtl-optimization/55829
22745         * lra-constraints.c (match_reload): Add code for absent output.
22746         (curr_insn_transform): Add code for reloads of matched inputs
22747         without output.
22749 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
22751         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
22752         attribute of movddup insn to DF.
22753         (*vec_interleave_lowv2df): Ditto.
22754         (vec_dupv2df): Ditto.
22756 2013-01-09  Jan Hubicka  <jh@suse.cz>
22758         PR tree-optimiation/55875
22759         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
22760         EVERY_ITERATION parameter.
22761         (number_of_iterations_exit): Check if exit is executed every iteration.
22762         (idx_infer_loop_bounds): Similarly here.
22763         (n_of_executions_at_most): Simplify
22764         to only test for cases where statement is dominated by the
22765         particular bound; handle correctly the "postdominance" test.
22766         (scev_probably_wraps_p): Use max loop iterations info
22767         as a global bound first.
22769 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
22770             Nick Clifton  <nickc@redhat.com>
22772         * config/v850/v850.md (cbranchsf4): New pattern.
22773         (cstoresf4): New pattern.
22774         (cbranchdf4): New pattern.
22775         (cstoredf4): New pattern.
22776         (movsicc): Disallow floating point comparisons.
22777         (cmpsf_le_insn): Fix order of operators.
22778         (cmpsf_lt_insn): Likewise.
22779         (cmpsf_eq_insn): Likewise.
22780         (cmpdf_le_insn): Likewise.
22781         (cmpdf_lt_insn): Likewise.
22782         (cmpdf_eq_insn): Likewise.
22783         (cmpsf_ge_insn): Use LE comparison.
22784         (cmpdf_ge_insn): Likewise.
22785         (cmpsf_gt_insn): Use LT comparison.
22786         (cmpdf_gt_insn): Likewise.
22787         (cmpsf_ne_insn): Delete pattern.
22788         (cmpdf_ne_insn): Delete pattern.
22789         * config/v850/v850.c (v850_gen_float_compare): Use
22790         gen_cmpdf_eq_insn for NE comparison.
22791         (v850_float_z_comparison_operator)
22792         (v850_float_nz_comparison_operator): Move from here ...
22793         * config/v850/predicates.md: ... to here.  Move GT and GE
22794         comparisons into v850_float_z_comparison_operator.
22795         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
22796         Delete prototype.
22797         (v850_float_nz_comparison_operator): Likewise.
22799 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22801         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
22802         with calls to gen_insvsi/gen_insvdi.
22804 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
22806         * config/i386/i386.c (initial_ix86_tune_features): Set up
22807         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
22809 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
22810             Jakub Jelinek  <jakub@redhat.com>
22812         PR tree-optimization/48189
22813         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
22814         If nitercst is 0, don't predict the exit edge.
22816 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22818         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
22819         in asm_fprintf with reg_names.
22820         (aarch64_print_operand_address): Likewise.
22821         (aarch64_return_addr): Likewise.
22822         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
22824 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22826         * config/pa/pa.h (VAL_U6_BITS_P): Define.
22827         (INT_U6_BITS): Likewise.
22828         * config/pa/predicates.md (uint6_operand): New predicate.
22829         (shift5_operand, shift6_operand): Likewise.
22830         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
22831         arith32_operand.
22832         (lshrdi3): Use shift6_operand.
22833         (shrpsi4, shrpdi4): New insn patterns.
22834         (extzv): Delete expander.
22835         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
22836         predicates in unamed zero extract patterns.  Tighten common constraint.
22837         (extv): Delete expander.
22838         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
22839         predicates in unamed sign extract patterns.  Tighten common constraint.
22840         (insv): Delete expander.
22841         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
22842         predicates in unamed insert patterns.  Tighten common constraint.
22843         Change uint32_operand predicate to uint6_operand predicate in unamed
22844         DImode pattern to insert constant values of type 1...1xxxx.
22846 2013-01-04  Jan Hubicka  <jh@suse.cz>
22848         PR tree-optimization/55823
22849         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
22850         issue.
22852 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22853             Uros Bizjak  <ubizjak@gmail.com>
22855         PR rtl-optimization/55845
22856         * df-problems.c (can_move_insns_across): Stop scanning at
22857         volatile_insn_p source instruction or give up if
22858         across_from .. across_to range contains any volatile_insn_p
22859         instructions.
22861 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22863         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
22864         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
22865         Declare.
22866         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
22867         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
22869 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22871         PR fortran/55341
22872         * asan.c (asan_clear_shadow): New function.
22873         (asan_emit_stack_protection): Use it.
22875 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22877         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
22878         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
22879         with tab instead of space.
22881 2013-01-08  Nick Clifton  <nickc@redhat.com>
22883         * config/rl78/rl78.c (rl78_expand_prologue): Always select
22884         register bank 0 at the start of an interrupt handler.
22885         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
22886         MDBH registers.
22888 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22890         * config/aarch64/aarch64-simd.md
22891         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
22892         (aarch64_simd_bsl): Likewise.
22893         (aarch64_vcond_internal<mode>): Likewise.
22894         (vcond<mode><mode>): Likewise.
22895         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
22896         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
22898 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22900         * config/aarch64/aarch64-builtins.c
22901         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
22903 2013-01-08  Martin Jambor  <mjambor@suse.cz>
22905         PR debug/55579
22906         * tree-sra.c (analyze_access_subtree): Return true also after
22907         potentially creating a debug-only replacement.
22909 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22911         PR middle-end/55890
22912         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
22914         PR tree-optimization/54120
22915         * tree-vrp.c (range_fits_type_p): Don't allow
22916         src_precision < precision from signed vr to unsigned_p
22917         if vr->min or vr->max is negative.
22918         (simplify_float_conversion_using_ranges): Test can_float_p
22919         against CODE_FOR_nothing.
22921 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22922             Richard Biener  <rguenther@suse.de>
22924         PR middle-end/55851
22925         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
22926         types instead of just INTEGER_TYPE types.
22928 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
22930         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
22931         TF_SIZE): Define.
22933 2013-01-07  Steve Ellcey  <sellcey@mips.com>
22935         PR target/42661
22936         * config/mips/mips.opt: Change mad to mmad to match documentation.
22938 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22940         PR target/55897
22941         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
22942         .progmemx.data now.
22944 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22946         PR target/55897
22947         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
22948         (avr_addrspace_t): Add .section_name field.
22949         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
22950         array size.
22951         (avr_addrspace): Same.  Initialize .section_name.  Remove last
22952         NULL entry.  Put __memx into .progmemx.data.
22953         (progmem_section_prefix): Remove.
22954         (avr_asm_init_sections): No need to initialize progmem_section.
22955         (avr_asm_named_section): Use avr_addrspace[].section_name to get
22956         section name prefix.
22957         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
22958         retrieve the progmem section.
22959         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
22960         boundary to run over avr_addrspace[].
22961         (avr_register_target_pragmas): Ditto.
22963 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
22965         * varasm.c (output_constant_def_contents): For asan_protect_global
22966         protected strings, adjust DECL_ALIGN if needed, before testing for
22967         anchored symbols.
22968         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
22969         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
22970         normal decls.
22971         (output_object_block): For asan protected decls, emit asan padding
22972         after their contents.
22973         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
22974         (asan_finish_file): Test it here instead.
22976 2013-01-07  Nick Clifton  <nickc@redhat.com>
22977             Matthias Klose  <doko@debian.org>
22978             Doug Kwan  <dougkwan@google.com>
22979             H.J. Lu  <hongjiu.lu@intel.com>
22981         PR driver/55470
22982         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
22984         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
22986         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
22988         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
22990         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
22992 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22994         PR target/54461
22995         * doc/install.texi (Cross-Compiler-Specific Options): Document
22996         --with-avrlibc.
22998 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
23000         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
23001         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
23002         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
23003         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
23004         vqmovun_high_s64): Fix source operand number and update copyright.
23006 2013-01-07  Richard Biener  <rguenther@suse.de>
23008         PR middle-end/55890
23009         * gimple.h (gimple_call_builtin_p): New overload.
23010         * gimple.c (validate_call): New function.
23011         (gimple_call_builtin_p): Likewise.
23012         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23013         Use gimple_call_builtin_p.
23014         (find_func_clobbers): Likewise.
23015         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
23016         (strlen_optimize_stmt): Likewise.
23018 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
23020         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
23021         (vld1q_dup_*): Likewise.
23022         (vld1_*): Likewise.
23023         (vld1q_*): Likewise.
23024         (vld1_lane_*): Likewise.
23025         (vld1q_lane_*): Likewise.
23027 2013-01-07  Richard Biener  <rguenther@suse.de>
23029         * lto-streamer.h (LTO_minor_version): Bump to 2.
23031 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
23033         * config/aarch64/aarch64-protos.h
23034         (aarch64_const_double_zero_rtx_p): Rename to...
23035         (aarch64_float_const_zero_rtx_p): ...this.
23036         (aarch64_float_const_representable_p): New.
23037         (aarch64_output_simd_mov_immediate): Likewise.
23038         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
23039         move immediate case.
23040         * config/aarch64/aarch64.c
23041         (aarch64_const_double_zero_rtx_p): Rename to...
23042         (aarch64_float_const_zero_rtx_p): ...this.
23043         (aarch64_print_operand): Allow printing of new constants.
23044         (aarch64_valid_floating_const): New.
23045         (aarch64_legitimate_constant_p): Check for valid floating-point
23046         constants.
23047         (aarch64_simd_valid_immediate): Likewise.
23048         (aarch64_vect_float_const_representable_p): New.
23049         (aarch64_float_const_representable_p): Likewise.
23050         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
23051         (aarch64_output_simd_mov_immediate): New.
23052         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
23053         (*movdf_aarch64): Likewise.
23054         * config/aarch64/constraints.md (Ufc): New.
23055         (Y): call aarch64_float_const_zero_rtx.
23056         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
23058 2013-01-07  Richard Biener  <rguenther@suse.de>
23060         PR tree-optimization/55888
23061         PR tree-optimization/55862
23062         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
23063         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
23064         not if it is contained therein.
23066 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
23068         * config/avr/t-avr: Typo.
23070 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
23072         PR55243
23073         * config/avr/t-avr: Don't automatically rebuild
23074         $(srcdir)/config/avr/t-multilib
23075         $(srcdir)/config/avr/avr-tables.opt
23076         $(srcdir)/doc/avr-mmcu.texi
23077         (avr-mcus): New phony target to build them on request.
23078         (s-avr-mlib, s-avr-mmcu-texi): Remove.
23079         * avr/avr-mcus.def: Adjust comments.
23081 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
23083         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
23085 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
23087         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
23089 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
23091         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
23093 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
23095         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
23096         to generate profiling.
23097         * config/rs6000/aix64.h (LIB_SPEC): Same.
23099 2013-01-04  Andrew Pinski  <apinski@cavium.com>
23101         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
23102         New function.
23103         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
23105 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
23107         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
23108         unconditionally.
23109         (ix86_expand_move): Ditto.
23110         (ix86_zero_extend_to_Pmode): Ditto.
23111         (ix86_expand_call): Ditto.
23112         (ix86_expand_special_args_builtin): Ditto.
23113         (ix86_expand_builtin): Ditto.
23115 2013-01-04  Richard Biener  <rguenther@suse.de>
23117         PR tree-optimization/55862
23118         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
23119         translating them through PHI nodes.
23121 2013-01-04  Martin Jambor  <mjambor@suse.cz>
23123         PR tree-optimization/55755
23124         * tree-sra.c (sra_modify_assign): Do not check that an access has no
23125         children when trying to avoid producing a VIEW_CONVERT_EXPR.
23127 2013-01-04  Marek Polacek  <polacek@redhat.com>
23129         PR middle-end/55859
23130         * opts.c (default_options_optimization): Clarify error message.
23132 2013-01-04  Richard Biener  <rguenther@suse.de>
23134         PR middle-end/55863
23135         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
23136         reassociation.
23138 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23140         PR target/53789
23141         * config/pa/pa.md (movsi): Revert previous change.
23142         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
23143         references.
23145 2013-01-03  Richard Henderson  <rth@redhat.com>
23147         * config/i386/i386.c (ix86_expand_move): Always assign to op1
23148         after eliminating TLS symbols.
23150 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
23152         PR bootstrap/50167
23153         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
23154         * graphite-poly.c (debug_gmp_value): Likewise.
23156 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
23158         PR target/55712
23159         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
23160         selected code model, define __code_mode_small__, __code_model_medium__,
23161         __code_model_large__, __code_model_32__ or __code_model_kernel__.
23162         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
23163         xchg temporary register with %k.  Declare temporary register as
23164         early clobbered.
23165         [__x86_64__]: For medium and large code models, preserve %rbx register.
23167 2013-01-03  Richard Biener  <rguenther@suse.de>
23169         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
23170         (dump_subscript): Adjust.
23171         (finalize_ddr_dependent): Do not dump redundant info.
23172         (analyze_siv_subscript): Adjust.
23173         (subscript_dependence_tester): Likewise.
23174         (compute_affine_dependence): Likewise.
23176 2013-01-03  Richard Biener  <rguenther@suse.de>
23178         Revert
23179         2013-01-03  Richard Biener  <rguenther@suse.de>
23181         PR tree-optimization/55857
23182         * tree-vect-stmts.c (vectorizable_load): Do not setup
23183         re-alignment for invariant loads.
23185         2013-01-02  Richard Biener  <rguenther@suse.de>
23187         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
23188         invariant load do not generate a vector load from the scalar location.
23190 2013-01-03  Richard Biener  <rguenther@suse.de>
23192         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
23193         for not vectorizing.
23194         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
23195         not build INDIRECT_REFs, call get_name once only.
23196         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
23197         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
23199 2013-01-03  Richard Biener  <rguenther@suse.de>
23201         PR tree-optimization/55857
23202         * tree-vect-stmts.c (vectorizable_load): Do not setup
23203         re-alignment for invariant loads.
23205 2013-01-03  Richard Biener  <rguenther@suse.de>
23207         PR lto/55848
23208         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
23209         prefer a built-in decl.
23211 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
23213         * gcc.c (process_command): Update copyright notice dates.
23214         * gcov.c (print_version): Likewise.
23215         * gcov-dump.c (print_version): Likewise.
23217         PR rtl-optimization/55838
23218         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
23219         iv0.step, iv1.step and step.
23221 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
23222             Marc Glisse  <marc.glisse@inria.fr>
23224         PR tree-optimization/55832
23225         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
23226         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
23227         integer_{one,zero}_node.
23229 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
23231         PR debug/54402
23232         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
23233         * var-tracking.c (reverse_op): Don't add reverse ops to
23234         VALUEs that have already
23235         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
23237 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
23239         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
23241 2013-01-02  Teresa Johnson  <tejohnson@google.com>
23243         * dumpfile.c (dump_loc): Print filename with location.
23244         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
23245         new location_t parameter to emit complete unroll message with
23246         new dump framework.
23247         (canonicalize_loop_induction_variables): Compute loops location
23248         and pass to try_unroll_loop_completely.
23249         * loop-unroll.c (report_unroll_peel): New function.
23250         (peel_loops_completely): Use new dump format with location
23251         for main dumpfile message, and invoke report_unroll_peel on success.
23252         (decide_unrolling_and_peeling): Ditto.
23253         (decide_peel_once_rolling): Remove old dumpfile message subsumed
23254         by report_unroll_peel.
23255         (decide_peel_completely): Ditto.
23256         (decide_unroll_constant_iterations): Ditto.
23257         (decide_unroll_runtime_iterations): Ditto.
23258         (decide_peel_simple): Ditto.
23259         (decide_unroll_stupid): Ditto.
23260         * cfgloop.c (get_loop_location): New function.
23261         * cfgloop.h (get_loop_location): Declare.
23263 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
23265         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
23266         NULL.
23268 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23270         PR middle-end/55198
23271         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
23272         BLKmode objects when EXPAND_MEMORY is specified.
23274 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
23276         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
23277         in loop predicate.
23278         (fold_builtin_cpu): Do not share cpu model decls across statements.
23280 2013-01-02  Jason Merrill  <jason@redhat.com>
23282         PR c++/55804
23283         * tree.c (build_array_type_1): Revert earlier change.
23285 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
23287         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
23288         "cortex-a57".
23289         * config/aarch64/aarch64-tune.md: Re-generate.
23291 2013-01-02  Richard Biener  <rguenther@suse.de>
23293         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
23294         invariant load do not generate a vector load from the scalar location.
23296 2013-01-02  Richard Biener  <rguenther@suse.de>
23298         PR bootstrap/55784
23299         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
23300         * configure: Regenerate.
23302 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
23304         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
23305         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
23306         (expand_builtin_int_roundingfn_2): Keep the original target around
23307         for the fallback case.
23309 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
23311         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
23312         to be clear for sign changes.
23314 2013-01-01  Jan Hubicka  <jh@suse.cz>
23316         * ipa-inline-analysis.c: Fix formatting.
23318 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
23320         PR tree-optimization/55831
23321         * tree-vect-loop.c (get_initial_def_for_induction): Use
23322         gsi_after_labels instead of gsi_start_bb.
23324 Copyright (C) 2013 Free Software Foundation, Inc.
23326 Copying and distribution of this file, with or without modification,
23327 are permitted in any medium without royalty provided the copyright
23328 notice and this notice are preserved.