* config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
[official-gcc.git] / gcc / ChangeLog
blobb9d4c358bef27f66c514d952411d849a45c72547
1 2013-09-26  Nick Clifton  <nickc@redhat.com>
3         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
4         time warning message.
5         (msp430_print_operand_raw): Delete unused letter parameter.
6         (TARGET_PRINT_OPERAND_ADDRESS): Define.
7         (msp430_print_operand_address): New function.
8         (msp430_print_operand): Move address printing code from here to
9         new function.
10         * config/msp430/msp430.md (movsipsi2): Add comment in generated
11         assembler.
12         (zero_extendpsisi2): Likewise.
13         (extendpsisi2): New pattern.
14         (andneghi3): New pattern.
16 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
18         * config/aarch64/aarch64.opt (mlra): New option.
19         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
20         (TARGET_LRA_P): Define.
22 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
24         * expr.c (expand_assignment): Remove obsolete comment.
26 2013-09-25  Jeff Law  <law@redhat.com>
28         * tree-flow.h (thread_through_all_blocks): Prototype moved into
29         tree-ssa-threadupdate.h.
30         (register_jump_thread): Similarly.
31         * tree-ssa-threadupdate.h: New header file.
32         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
33         * tree-vrp.c: Likewise.
34         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
35         (thread_around_empty_blocks): Change type of path vector argument to
36         an edge,type pair from just an edge.  Initialize both elements when
37         appending to a jump threading path.  Tweak references to elements
38         appropriately.
39         (thread_across_edge): Similarly.  Release memory for the elements
40         as needed.
41         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
42         (dump_jump_thread_path): New function broken out from
43         register_jump_thread.
44         (register_jump_thread): Use dump_jump_thread_path.  Change type of
45         path vector entries.  Search the path for NULL edges and dump
46         the path if one is found.  Tweak the conversion of path to 3-edge
47         form to use the block copy type information embedded in the path.
49 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
51         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
53 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
54             Vladimir Makarov  <vmakarov@redhat.com>
56         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
57         from the least significant bit.
58         (strip_address_mutations): Add bitfield operations handling.
59         (must_be_index_p): Add shifting and rotate operations handling.
60         (set_address_base): Use must_be_base_p predicate.
61         (set_address_index): Use must_be_index_p predicate.
63 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
64             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
65             Sergey Lega  <sergey.s.lega@intel.com>
66             Anna Tikhonova  <anna.tikhonova@intel.com>
67             Ilya Tocar  <ilya.tocar@intel.com>
68             Andrey Turetskiy  <andrey.turetskiy@intel.com>
69             Ilya Verbin  <ilya.verbin@intel.com>
70             Kirill Yukhin  <kirill.yukhin@intel.com>
71             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
73         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
74         Use new names.
75         (ix86_expand_vector_move_misalign): Support new unaligned load and
76         stores and use new names.
77         (CODE_FOR_sse2_storedqu): Rename to ...
78         (CODE_FOR_sse2_storedquv16qi): ... this.
79         (CODE_FOR_sse2_loaddqu): Rename to ...
80         (CODE_FOR_sse2_loaddquv16qi): ... this.
81         (CODE_FOR_avx_loaddqu256): Rename to ...
82         (CODE_FOR_avx_loaddquv32qi): ... this.
83         (CODE_FOR_avx_storedqu256): Rename to ...
84         (CODE_FOR_avx_storedquv32qi): ... this.
85         * config/i386/i386.md (fpint_logic): New.
86         * config/i386/sse.md (VMOVE): Extend for AVX512.
87         (VF): Ditto.
88         (VF_128_256): New.
89         (VF_512): Ditto.
90         (VI_UNALIGNED_LOADSTORE): Ditto.
91         (sse2_avx_avx512f): Ditto.
92         (sse2_avx2): Extend for AVX512.
93         (sse4_1_avx2): Ditto.
94         (avx2_avx512f): New.
95         (sse): Extend for AVX512.
96         (sse2): Ditto.
97         (sse4_1): Ditto.
98         (avxsizesuffix): Ditto.
99         (sseintvecmode): Ditto.
100         (ssePSmode): Ditto.
101         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
102         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
103         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
104         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
105         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
106         (<sse2_avx_avx512f>_storedqu<mode): ... this.
107         (<sse>_movnt<mode>): Replace constraint "x" with "v".
108         (STORENT_MODE): Extend for AVX512.
109         (*absneg<mode>2): Replace constraint "x" with "v".
110         (*mul<mode>3): Ditto.
111         (*ieee_smin<mode>3): Ditto.
112         (*ieee_smax<mode>3): Ditto.
113         (avx_cmp<mode>3): Replace VF with VF_128_256.
114         (*<sse>_maskcmp<mode>3_comm): Ditto.
115         (<sse>_maskcmp<mode>3): Ditto.
116         (<sse>_andnot<mode>3): Extend for AVX512.
117         (<code><mode>3, anylogic): Replace VF with VF_128_256.
118         (<code><mode>3, fpint_logic): New.
119         (*<code><mode>3): Extend for AVX512.
120         (avx512flogicsuff): New.
121         (avx512f_<logic><mode>): Ditto.
122         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
123         VF_128_256.
124         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
125         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
126         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
127         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
128         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
129         (xop_vpermil2<mode>3): Ditto.
130         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
131         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
132         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
133         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
135 2013-09-25  Tom Tromey  <tromey@redhat.com>
137         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
138         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
139         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
140         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
141         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
142         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
143         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
144         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
145         (GRAPHITE_HTAB_H): Remove.
147 2013-09-25  Tom Tromey  <tromey@redhat.com>
149         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
151 2013-09-25  Tom Tromey  <tromey@redhat.com>
153         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
155 2013-09-25  Tom Tromey  <tromey@redhat.com>
157         * config/i386/t-i386 (i386.o): Remove.
158         (i386-c.o): Use COMPILE and POSTCOMPILE.
160 2013-09-25  Tom Tromey  <tromey@redhat.com>
162         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
164 2013-09-25  Tom Tromey  <tromey@redhat.com>
166         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
167         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
168         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
169         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
170         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
171         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
172         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
173         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
174         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
175         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
176         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
177         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
178         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
179         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
180         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
181         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
182         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
183         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
184         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
185         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
186         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
187         (tree-ssa-pre.o, tree-ssa-sccvn.o)
188         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
189         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
190         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
191         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
192         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
193         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
194         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
195         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
196         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
197         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
198         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
199         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
200         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
201         (graphite-clast-to-gimple.o, graphite-dependences.o)
202         (graphite-interchange.o, graphite-poly.o)
203         (graphite-scop-detection.o, graphite-sese-to-poly.o)
204         (graphite-optimize-isl.o, tree-vect-loop.o)
205         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
206         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
207         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
208         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
209         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
210         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
211         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
212         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
213         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
214         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
215         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
216         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
217         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
218         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
219         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
220         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
221         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
222         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
223         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
224         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
225         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
226         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
227         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
228         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
229         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
230         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
231         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
232         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
233         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
234         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
235         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
236         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
237         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
238         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
239         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
240         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
241         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
242         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
243         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
244         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
245         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
246         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
247         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
248         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
249         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
250         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
251         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
252         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
253         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
254         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
255         (gcov-dump.o): Remove.
256         (default-c.o): Use COMPILE and POSTCOMPILE.
257         (CFLAGS-gcc.o): New variable.
258         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
260 2013-09-25  Tom Tromey  <tromey@redhat.com>
262         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
263         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
264         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
265         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
266         (c-family/c-pragma.o, c-family/c-pretty-print.o)
267         (c-family/c-semantics.o, c-family/c-ada-spec.o)
268         (c-family/array-notation-common.o, c-family/stub-objc.o)
269         (c-family/c-ubsan.o): Remove.
271 2013-09-25  Tom Tromey  <tromey@redhat.com>
273         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
275 2013-09-25  Tom Tromey  <tromey@redhat.com>
277         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
278         to add -DENABLE_SHARED_LIBGCC.
279         (gcc.o): Don't use subshell.
281 2013-09-25  Tom Tromey  <tromey@redhat.com>
283         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
284         * configure.ac: Don't invoke AM_PROG_CC_C_O.
285         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
286         * configure, config.in: Rebuild.
288 2013-09-25  Tom Tromey  <tromey@redhat.com>
290         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
291         (COMPILE, POSTCOMPILE): New variables.
292         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
293         (DEPFILES): New variable.
294         Include ".Po" files.
295         * configure.ac: Add checks for dependency checking.
296         * configure, aclocal.m4: Regenerate.
298 2013-09-25  Tom Tromey  <tromey@redhat.com>
300         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
301         ($(ALL_HOST_OBJS)): Move order-only dependency to end
302         of file.
304 2013-09-25  Tom Tromey  <tromey@redhat.com>
306         * Makefile.in (generated_files): Add options.h,
307         target-hooks-def.h, insn-opinit.h,
308         common/common-target-hooks-def.h, pass-instances.def,
309         c-family/c-target-hooks-def.h.
311 2013-09-25  Jeff Law  <law@redhat.com>
313         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
314         than foo[foo.length () - 1] to access last member in a vec.
315         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
317 2013-09-25  Richard Biener  <rguenther@suse.de>
319         PR middle-end/58521
320         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
322 2013-09-25  Jan Hubicka  <jh@suse.cz>
324         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
325         test.
327 2013-09-25  Marek Polacek  <polacek@redhat.com>
329         PR sanitizer/58420
330         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
331         when determining the type name.
333 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
335         * config/sh/sh.md: Fix formatting.
337 2013-09-24  Xinliang David Li <davidxl@google.com>
339         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): 
340         Check max peel iterations parameter.
341         * param.def: New parameter.
342         * doc/invoke.texi: Document New parameter.
344 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
346         * gimple-pretty-print.c: Various whitespace tweaks.
347         * tree-core.h: Likewise.
348         * tree-pretty-print.c: Likewise.
349         * tree-ssa-alias.c: Likewise.
350         * tree-ssa-copy.c: Likewise.
351         * tree-ssanames.c: Likewise.
352         * tree-ssanames.h: Likewise.
353         * tree-vrp.c: Likewise.
355 2013-09-24  Alan Modra  <amodra@gmail.com>
357         PR middle-end/57134
358         PR middle-end/57586
359         * stmt.c (expand_asm_operands): Call expand_expr with
360         EXPAND_MEMORY for output operands that disallow regs.  Don't
361         use EXPAND_WRITE on inout operands.
363 2013-09-24  Richard Biener  <rguenther@suse.de>
365         PR middle-end/58513
366         * tree.c (reference_alias_ptr_type): Move ...
367         * alias.c (reference_alias_ptr_type): ... here and implement
368         in terms of the new reference_alias_ptr_type_1.
369         (ref_all_alias_ptr_type_p): New helper.
370         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
371         use ref_all_alias_ptr_type_p.
372         (get_deref_alias_set): Add flag_strict_aliasing check here.
373         (reference_alias_ptr_type_1): New function, split out from ...
374         (get_alias_set): ... here.
375         (alias_ptr_types_compatible_p): New function.
376         * alias.h (reference_alias_ptr_type): Declare.
377         (alias_ptr_types_compatible_p): Likewise.
378         * tree.h (reference_alias_ptr_type): Remove.
379         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
380         to compare MEM_REF alias types.
382 2013-09-24  Richard Biener  <rguenther@suse.de>
384         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
386 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
388         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
389         reload helper function arrays into a single array reg_addr.
390         (reload_fpr_gpr): Likewise.
391         (reload_gpr_vsx): Likewise.
392         (reload_vsx_gpr): Likewise.
393         (struct rs6000_reg_addr): Likewise.
394         (reg_addr): Likewise.
395         (rs6000_debug_reg_global): Change rs6000_vector_reload,
396         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
397         (rs6000_init_hard_regno_mode_ok): Likewise.
398         (rs6000_secondary_reload_direct_move): Likewise.
399         (rs6000_secondary_reload): Likewise.
401         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
402         constraints: wu, ww, and wy.  Repurpose wv constraint added during
403         power8 changes.  Put wg constraint in alphabetical order.
405         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
406         for future work to add ISA 2.07 VSX single precision support.
407         (-mvsx-scalar-double): Change default from -1 to 1, update
408         documentation comment.
409         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
410         (-mupper-regs-df): New debug switch to control whether DF values
411         can go in the traditional Altivec registers.
412         (-mupper-regs-sf): New debug switch to control whether SF values
413         can go in the traditional Altivec registers.
415         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
416         and wy constraints.
417         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
418         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
419         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
420         (rs6000_debug_legitimate_address_p): Print if we are running
421         before, during, or after reload.
422         (rs6000_secondary_reload): Add a comment.
423         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
425         * config/rs6000/constraints.md (wa constraint): Sort w<x>
426         constraints.  Update documentation string.
427         (wd constraint): Likewise.
428         (wf constraint): Likewise.
429         (wg constraint): Likewise.
430         (wn constraint): Likewise.
431         (ws constraint): Likewise.
432         (wt constraint): Likewise.
433         (wx constraint): Likewise.
434         (wz constraint): Likewise.
435         (wu constraint): New constraint for ISA 2.07 SFmode scalar
436         instructions.
437         (ww constraint): Likewise.
438         (wy constraint): Likewise.
439         (wv constraint): Repurpose ISA 2.07 constraint that we not used in
440         the previous submissions.
441         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
443 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
445         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
446         (REG_BR_PROB): Say that the probability is stored in an int_list.
447         * reg-notes.def: Update commentary to mention INT_LIST.
448         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
449         (INT_LIST): New rtx.
450         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
451         * rtlanal.c (int_reg_note_p): New function.
452         (alloc_reg_note): Assert that the note does not have an int argument.
453         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
454         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
455         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
456         rather than an INSN_LIST.  Handle INT_LIST.
457         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
458         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
459         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
460         Manipulate them as ints rather than rtxes.
461         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
462         * regmove.c (copy_src_to_dest): Likewise.
463         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
465         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
466         into the cases that need it.
467         * config/arm/arm.c (arm_unwind_emit): Likewise.
469         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
470         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
471         * loop-doloop.c (add_test, doloop_modify): Likewise.
472         * loop-unswitch.c (compare_and_jump_seq): Likewise.
473         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
474         * predict.c (combine_predictions_for_insn): Likewise.
475         * print-rtl.c (print_rtx): Handle INT_LIST.
476         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
477         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
478         * config/arm/arm.c (emit_unlikely_jump): Likewise.
479         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
480         (ix86_split_fp_branch, predict_jump): Likewise.
481         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
482         * config/sh/sh.c (expand_cbranchsi4): Likewise.
483         * config/spu/spu.c (ea_load_store_inline): Likewise.
485         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
486         value of a REG_BR_PROB note.
487         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
488         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
489         * emit-rtl.c (try_split): Likewise.
490         * predict.c (br_prob_note_reliable_p): Likewise.
491         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
492         * reorg.c (mostly_true_jump): Likewise.
493         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
494         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
495         * config/i386/i386.c (ix86_print_operand): Likewise.
496         * config/ia64/ia64.c (ia64_print_operand): Likewise.
497         * config/mmix/mmix.c (mmix_print_operand): Likewise.
498         * config/rs6000/rs6000.c (output_cbranch): Likewise.
499         * config/s390/s390.c (s390_expand_tbegin): Likewise.
500         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
501         * config/sparc/sparc.c (output_cbranch): Likewise.
502         * config/spu/spu.c (get_branch_target): Likewise.
503         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
504         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
506 2013-09-23  Jan Hubicka  <jh@suse.cz>
508         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
509         for ipa-devirt.
510         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
511         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
512         of external calls
513         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
514         (gimple_fold_call): Dump inconsistent devirtualizations; add
515         sanity check for type based devirtualizations.
516         * ipa-prop.c: Include ipa-utils.h
517         (ipa_intraprocedural_devirtualization): Add sanity check.
518         (try_make_edge_direct_virtual_call): Likewise.
520 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
522         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
523         assignment statements.
525 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
527         * gimple-pretty-print.c (dump_ssaname_info): New function.
528         (dump_gimple_phi): Call it.
529         (pp_gimple_stmt_1): Likewise.
530         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
531         (range_info_def): Declare.
532         * tree-pretty-print.c (pp_double_int): New function.
533         (dump_generic_node): Call it.
534         * tree-pretty-print.h (pp_double_int): Declare.
535         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
536         * tree-ssanames.h (range_info_def): New structure.
537         (value_range_type): Move definition here.
538         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
539         Declare.
540         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
541         initialization.
542         (set_range_info): New function.
543         (get_range_info): Likewise.
544         (duplicate_ssa_name_range_info): Likewise.
545         (duplicate_ssa_name_fn): Check pointer type and call
546         duplicate_ssa_name_range_info.
547         * tree-ssa-copy.c (fini_copy_prop): Likewise.
548         * tree-vrp.c (value_range_type): Remove definition, now in
549         tree-ssanames.h.
550         (vrp_finalize): Call set_range_info to update value range of
551         SSA_NAMEs.
552         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
553         (SSA_NAME_RANGE_INFO): New macro.
555 2013-09-23  Richard Biener  <rguenther@suse.de>
557         PR tree-optimization/58464
558         * tree-ssa-pre.c (phi_trans_lookup): Remove.
559         (phi_trans_add): Change to add conditionally on being not
560         present.
561         (phi_translate_1): Remove recursion detection here.
562         (phi_translate): Pre-seed the cache with NULL to catch
563         recursion here in a more generic way.
564         (bitmap_find_leader): Adjust comment.
565         (get_representative_for): Dump value-numbers.
566         (create_expression_by_pieces): Likewise.
567         (insert_into_preds_of_block): Likewise.
569 2013-09-23  Christian Bruel  <christian.bruel@st.com>
571         PR target/58475
572         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
573         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
575 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
577         Revert r202780:
578         2013-09-20  Renlin Li  <renlin.li@arm.com>
580         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
581         (aarch64_expand_epilogue): Likewise.
582         (aarch64_legitimize_reload_address): Likewise.
584 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
586         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
587         NULL_TREE before pushing them onto the vector.  Likewise for labels.
589 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
591         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
592         comparison operators when -fno-trapping-math is in effect.
593         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
594         comparison operators in TFmode and assert that unsupported operators
595         cannot reach here.
596         (ia64_print_operand): Likewise.
598 2013-09-21  Jan Hubicka  <jh@suse.cz>
600         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
601         (sse_typeless_stores): Enable for core
602         (sse_load0_by_pxor): Likewise.
603         (four_jump_limit): Disable for core.
604         (pad_returns): Likewise.
605         (avoid_vector_decode): Likewise.
606         (fuse_cmp_and_branch): Enable for cores.
607         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
609 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
611         PR middle-end/56791
612         * config/pa/pa.c (pa_option_override): Disable auto increment and
613         decrement instructions until reload is completed.
615         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
616         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
617         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
619 2013-09-20  DJ Delorie  <dj@redhat.com>
620             Nick Clifton  <nickc@redhat.com>
622         * config/rl78/rl78.c: Various whitespace and comment tweaks.
623         (need_to_save): Save bank 0 on interrupts.
624         (characterize_address): Strip far address wrappers.
625         (rl78_as_legitimate_address): Likewise.
626         (transcode_memory_rtx): Likewise.
627         (rl78_peep_movhi_p): Disable this peephole after devirt.
628         (rl78_propogate_register_origins): Forget all origins when a
629         CLOBBER is seen.
630         * config/rl78/rl78-virt.md: Various whitespace tweaks.
631         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
632         peephole2's.
633         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
634         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
635         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
636         relocating.
637         * config/rl78/constraints.md: Various whitespace and paren tweaks.
639 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
641         * config/pa/pa.md: In "scc" insn patterns, change output template to
642         handle const0_rtx in reg_or_0_operand operands.
644 2013-09-20  Martin Husemann  <martin@NetBSD.org>
646         PR target/56875
647         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
648         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
650 2013-09-20  Richard Biener  <rguenther@suse.de>
652         PR middle-end/58484
653         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
654         remembering SSA name version and block index.
655         (new_scev_info_str): Adjust.
656         (hash_scev_info): Likewise.  Also hash the block index.
657         (eq_scev_info): Adjust.
658         (find_var_scev_info): Likewise.
659         (struct instantiate_cache_entry): Remove.
660         (struct instantiate_cache_type): Use a htab to map name, block
661         to chrec.
662         (instantiate_cache_type::~instantiate_cache_type): Adjust.
663         (get_instantiated_value_entry): Likewise.
664         (hash_idx_scev_info, eq_idx_scev_info): New functions.
665         (instantiate_scev_name): Adjust.
667 2013-09-20  Jeff Law  <law@redhat.com>
669         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
671 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
673         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
674         Call aarch64_simd_expand_args to update op[argc].
676 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
678         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside 
679           plugin argument.
681 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
683        * gengtype.c (file_rules): Added rule for *.cc files.
684        (get_output_file_with_visibility): Give fatal message when no
685        rules found.
687 2013-09-20  Renlin Li  <renlin.li@arm.com>
689         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
690         (aarch64_expand_epilogue): Likewise.
691         (aarch64_legitimize_reload_address): Likewise.
693 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
695         PR middle-end/57748
696         * expr.c (expand_assignment): Remove misalignp code path.
698 2013-09-20  Marek Polacek  <polacek@redhat.com>
700         PR sanitizer/58413
701         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
702         TYPE_PRECISION.  Add asserts.
704 2013-09-20  Richard Biener  <rguenther@suse.de>
706         PR tree-optimization/58453
707         * tree-loop-distribution.c (distribute_loop): Apply the cost
708         model for -ftree-loop-distribute-patterns, too.
710 2013-09-20  Richard Biener  <rguenther@suse.de>
712         PR middle-end/58473
713         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
714         make type comparison less strict.
716 2013-09-20  Alan Modra  <amodra@gmail.com>
718         * configure: Regenerate.
719         * aclocal.m4: Regenerate.
721 2013-09-20  Marek Polacek  <polacek@redhat.com>
723         PR other/58467
724         * doc/extend.texi: Document that attribute used is meant to be used
725         on variables with static storage duration.
727 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
729         PR tree-optimization/58472
730         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
731         simd_lane_access set inv_p = false.
732         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
733         the simduid magic VAR_DECL.
735 2013-09-19  Jan Hubicka  <jh@suse.cz>
737         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
739 2013-09-17  Jeff Law  <law@redhat.com>
741         * tree-ssa-dom.c (record_temporary_equivalences): New function
742         split out of dom_opt_dom_walker::after_dom_children.
743         (dom_opt_dom_walker::thread_across_edge): Move common code
744         in here from dom_opt_dom_walker::after_dom_children.
745         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
747 2013-09-19  Jan Hubicka  <jh@suse.cz>
749         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
750         (TARGET_GENERIC): Use PROCESOR_GENERIC
751         (enum processor_type): Unify generic32 and 64.
752         * i386.md (cpu): Likewise.
753         * x86-tune.def (use_leave): Enable for generic32.
754         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
755         * athlon.md: Change generic64 to generic in all occurences.
756         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
757         (ix86_target_macros_internal): Likewise.
758         * driver-i386.c (host_detect_local_cpu): Likewise.
759         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename to ..
760         (generic_memcpy, generic_memset, generic_cost): This one.
761         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
762         (m_GENERIC32, m_GENERIC64): Remove.
763         (m_GENERIC): Turn into one flag.
764         (processor_target): Unify generic tunnings.
765         (ix86_option_override_internal): Replace generic32/64 by generic.
766         (ix86_issue_rate): Likewise.
767         (ix86_adjust_cost): Likewise.
769 2013-09-19  Jan Hubicka  <jh@suse.cz>
771         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
772         of speculative flag.
774 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
776         * omp-low.c (expand_omp_sections): Always pass len - 1 to
777         GOMP_sections_start, even if !exit_reachable.
779 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
781         * lra-constraints.c (need_for_all_save_p): Use macro
782         HARD_REGNO_CALL_PART_CLOBBERED.
783         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
784         set up pseudo conflict hard regs.
786 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
788         PR target/58452
789         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
790         operands.
792 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
794         PR rtl-optimization/58438
795         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
796         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
797         from previous subpasses.
799 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
801         * arm.c (arm_get_frame_offsets): Validate architecture supports
802         LDRD/STRD before accepting the tuning preference.
803         (arm_expand_prologue): Likewise.
804         (arm_expand_epilogue): Likewise.
806 2013-09-18  Richard Biener  <rguenther@suse.de>
808         PR tree-optimization/58417
809         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
810         have chrecs with symbols defined in the loop as operands.
811         (chrec_fold_multiply): Likewise.
812         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
813         parameters before folding binary operations.
814         (struct instantiate_cache_entry_hasher): Remove.
815         (struct instantiate_cache_type): Use a pointer-map.
816         (instantiate_cache_type::instantiate_cache_type): New function.
817         (instantiate_cache_type::get): Likewise.
818         (instantiate_cache_type::set): Likewise.
819         (instantiate_cache_type::~instantiate_cache_type): Adjust.
820         (get_instantiated_value_entry): Likewise.
821         (global_cache): New global.
822         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
823         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
824         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
825         (instantiate_scev_name): Adjust.
826         (instantiate_scev): Construct global instead of local cache.
827         (resolve_mixers): Likewise.
829 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
830             Paolo Carlini  <paolo.carlini@oracle.com>
832         PR c++/58458
833         * doc/implement-cxx.texi: Fix references to the C++ standards.
835 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
837         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
838         * tree-vect-data-refs.c (vect_analyze_data_refs): For
839         simd_lane_access drs, update also DR_ALIGNED_TO.
841 2013-09-18  Marek Polacek  <polacek@redhat.com>
843         PR sanitizer/58411
844         * doc/extend.texi: Document no_sanitize_undefined attribute.
845         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
846         no_sanitize_undefined attribute.
848 2013-09-18  Nick Clifton  <nickc@redhat.com>
850         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
851         (ASM_DECLARE_FUNCTION_NAME): Define.
853 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
855         * compare-elim.c (find_comparison_dom_walker): New class
856         (find_comparisons_in_bb): Rename to
857         find_comparison_dom_walker::before_dom_children
858         (find_comparisons): Adjust
859         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
860         adjust.
861         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
862         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
863         (init_walk_dominator_tree): Remove declaration.
864         (fini_walk_dominator_tree): Remove declaration.
865         * fwprop.c (single_def_use_dom_walker): New class
866         (single_def_use_enter_block): Convert to
867         single_def_use_dom_walker::before_dom_children.
868         (single_def_use_leave_block): Convert to
869         single_def_use_dom_walker::after_dom_children.
870         (build_single_def_use_links): Adjust.
871         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
872         class.
873         (find_candidates_in_block): Convert to
874         find_candidates_dom_walker::before_dom_children.
875         (execute_strength_reduction): Adjust.
876         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
877         (sese_dom_walker): New class.
878         (sese_dom_walker::sese_dom_walker): New constructor.
879         (sese_dom_walker::~sese_dom_walker): New destructor.
880         (build_sese_conditions_before): Convert to
881         sese_dom_walker::before_dom_children.
882         (build_sese_conditions_after): Convert to
883         sese_dom_walker::after_dom_children.
884         (build_poly_scop): Adjust
885         * tree-into-ssa.c (rewrite_dom_walker): New class
886         (rewrite_enter_block): Convert to
887         rewrite_dom_walker::before_dom_children.
888         (rewrite_leave_block): Convert to
889         rewrite_dom_walker::after_dom_children.
890         (rewrite_update_dom_walker): New class.
891         (rewrite_update_enter_block): Convert to
892         rewrite_update_dom_walker::before_dom_children.
893         (rewrite_update_leave_block): Convert to
894         rewrite_update_dom_walker::after_dom_children.
895         (rewrite_blocks, rewrite_into_ssa): Adjust.
896         (mark_def_dom_walker): New class.
897         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
898         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
899         (mark_def_sites_blocks): Convert to
900         mark_def_dom_walker::before_dom_children.
901         (mark_def_site_blocks): Remove.
902         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
903         (tree_ssa_dominator_optimize): Adjust.
904         (dom_thread_across_edge): Convert to method
905         dom_opt_dom_walker::thread_across_edge.
906         (dom_opt_enter_block): Convert to member function
907         dom_opt_dom_walker::before_dom_children.
908         (dom_opt_leave_block): Convert to member function
909         dom_opt_dom_walker::after_dom_children.
910         * tree-ssa-dse.c (dse_dom_walker): New class.
911         (dse_enter_block): Convert to member function
912         dse_dom_walker::before_dom_children.
913         (tree_ssa_dse): Adjust.
914         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
915         (determine_invariantness_stmt): Convert to method
916         invariantness_dom_walker::before_dom_children.
917         (determine_invariantness): Remove
918         (move_computations_dom_walker): New class.
919         (move_computations_stmt): Convert to method
920         move_computations_dom_walker::before_dom_children.
921         (move_computations, tree_ssa_lim): Adjust.
922         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
923         (nt_init_block): Make method
924         notrappping_dom_walker::before_dom_children.
925         (nt_fini_block): Make
926         method nontrapping_dom_walker::after_dom_children.
927         (get_non_trapping): Adjust.
928         * tree-ssa-pre.c (eliminate_dom_walker): New class.
929         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
930         (eliminate_leave_block): Make method.
931         eliminate_dom_walker::after_dom_children.
932         (eliminate): Adjust
933         * tree-ssa-strlen.c (strlen_dom_walker): New class.
934         (strlen_enter_block): Make method
935         strlen_dom_walker::before_dom_children.
936         (strlen_leave_block): Make
937         method strlen_dom_walker::after_dom_children.
938         (tree_ssa_strlen): Adjust.
939         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
940         (tree_ssa_uncprop): Adjust.
941         (uncprop_leave_block): Make method
942         uncprop_dom_walker::after_dom_children.
943         (uncprop_leave_block): Make method
944         uncprop_dom_walker::before_dom_children.
946 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
948         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn
949         before branch in same basic block.  Check both src and dest of
950         the move insn.
952 2013-09-17  Nick Clifton  <nickc@redhat.com>
954         * config/rl78/rl78-real.md (bf): New pattern.
955         (bt): New pattern.
956         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
957         (rl78_print_operand): Do not put a # before a %B.
958         * config/rl78/rl78.opt: Tweak doc strings.
960 2013-09-17  DJ Delorie  <dj@redhat.com>
962         * config/rl78/constraints.md (Wcv): Allow up to $r31.
963         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
964         (rl78_option_override): Likewise, if -mallregs.
965         (is_virtual_register): Likewise.
966         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
967         (REGNO_OK_FOR_BASE_P): Likewise.
968         * config/rl78/rl78.opt (-mallregs): New.
970 2013-09-17  Nick Clifton  <nickc@redhat.com>
972         * config/rl78/rl78.c (need_to_save): Change return type to bool.
973         For interrupt functions: save all call clobbered registers if the
974         interrupt handler is not a leaf function.
975         (rl78_expand_prologue): Always recompute the frame information.
976         For interrupt functions: only select bank 0 if one of the bank 0
977         registers is going to be psuhed.
979 2013-09-17  DJ Delorie  <dj@redhat.com>
981         * config/rl78/constraints.md: For each W* constraint, rename to C*
982         and create a W* constraint that checks for an optional ES: prefix
983         pattern also.
984         * config/rl78/rl78.md (UNS_ES_ADDR): New.
985         (es_addr): New.  Used to wrap far addresses.
986         * config/rl78/rl78-protos.h (rl78_es_addr): New.
987         (rl78_es_base): New.
988         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
989         wrapped far addresses.
990         (rl78_print_operand_1): Unwrap far addresses before processing.
991         (rl78_lo16): Wrap far addresses in unspecs.
992         (rl78_es_addr): New.
993         (rl78_es_base): New.
994         (insn_ok_now): Check for not-yet-wrapped far addresses.
995         (transcode_memory_rtx): Properly re-wrap far addresses.
997 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
999         * config/sparc/t-rtems: Add leon3 multilibs.
1001 2013-09-17  Cong Hou  <congh@google.com>
1003         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
1004         when checking the dot production pattern. The type of rhs operand
1005         of multiply is now checked correctly.
1007 2013-09-17  Jeff Law  <law@redhat.com>
1009         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1010         edge implied equivalences into successor phis.
1011         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
1012         here from tree-ssa-threadedge.c.
1013         (mark_threaded_blocks): When threading through a joiner, if both
1014         successors of the joiner's clone reach the same block, verify the
1015         PHI arguments are equal.  If not, cancel the jump threading request.
1016         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
1017         tree-ssa-threadupdate.c
1018         (thread_across_edge): Don't check PHI argument equality when
1019         threading through joiner block here.
1020         
1021 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
1023         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
1024         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
1025         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
1026         make_pass_early_warn_uninitialized): Move to tree-ssa-uninit.c.
1027         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c
1028         (has_undefined_value_p): New.  Pass dependant parts of 
1029         ssa_undefined_value_p.
1030         (uninit_undefined_value_p): Use has_undefined_value_p.
1031         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
1032         make_pass_early_warn_uninitialized): Move from tree-ssa.c
1033         * tree-ssa.h: Adjust prototypes
1035 2013-09-17  Jan Hubicka  <jh@suse.cz>
1037         PR middle-end/58332
1038         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
1039         * ipa-inline.c (can_inline_edge_p): Do not downgrade
1040         FUNCTION_NOT_OPTIMIZED.
1041         * ipa-inline-analysis.c (compute_inline_parameters): Function
1042         not optimized is not inlinable unless it is alwaysinline.
1043         (inline_analyze_function): Force calls in not optimized
1044         function not inlinable.
1046 2013-09-17  Jan Hubicka  <jh@suse.cz>
1048         PR middle-end/58329
1049         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
1050         to return NULL.
1051         * ipa.c (function_and_variable_visibility): Likewise.
1052         * ipa-profile.c (ipa_profile): Likewise.
1054 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1056         PR ipa/58398
1057         * cgraph.c (cgraph_function_body_availability): Check for ifunc
1058         attribute, and don't inline the resolver in this case.
1060 2013-09-17  Teresa Johnson  <tejohnson@google.com>
1062         * coverage.c (get_coverage_counts): Add missing newline.
1064 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1066         PR tree-optimization/58088
1067         * fold-const.c (mask_with_trailing_zeros): New function.
1068         (fold_binary_loc): Make sure we don't recurse infinitely
1069         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
1070         Use mask_with_trailing_zeros where appropriate.
1072 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
1074         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
1075         of 'prev' var to get better distance estimation.
1077 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
1079         * tree-inline.h (struct copy_body_data): Add transform_parameter.
1080         * tree-inline.c (is_parameter_of): New predicate.
1081         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
1082         a parameter has been remapped.
1083         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
1084         (optimize_inline_calls): Initialize transform_parameter.
1085         (copy_gimple_seq_and_replace_locals): Likewise.
1086         (tree_function_versioning): Likewise.
1087         (maybe_inline_call_in_expr): Likewise.
1089 2013-09-17  Nick Clifton  <nickc@redhat.com>
1091         * config/msp430/msp430-protos.h: Add prototypes for new functions.
1092         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
1093         interrupt handlers.
1094         (is_attr_func): New function.
1095         (msp430_is_interrupt_func): New function.
1096         (is_naked_func): New function.
1097         (is_reentrant_func): New function.
1098         (is_critical_func): New function.
1099         (msp430_start_function): Add annotations for function attributes.
1100         (msp430_attr): New function.
1101         (msp430_attribute_table): New.
1102         (msp430_function_section): New function.
1103         (TARGET_ASM_FUNCTION_SECTION): Define.
1104         (msp430_builtin): New enum.
1105         (msp430_init_builtins): New function.
1106         (msp430_builtin_devl): New function.
1107         (msp430_expand_builtin): New function.
1108         (TARGET_INIT_BUILTINS): Define.
1109         (TARGET_EXPAND_BUILTINS): Define.
1110         (TARGET_BUILTIN_DECL): Define.
1111         (msp430_expand_prologue): Add support for naked, interrupt,
1112         critical and reentrant functions.
1113         (msp430_expand_epilogue): Likewise.
1114         (msp430_print_operand): Handle 'O' character.
1115         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
1116         NO_TRAMPOLINES.
1117         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
1118         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
1119         (pushm): Use a 'n' rather than an 'i' constraint.
1120         (msp_return): Add generation of the interrupt return instruction.
1121         (disable_interrupts): New pattern.
1122         (enable_interrupts): New pattern.
1123         (push_intr_state): New pattern.
1124         (pop_intr_state): New pattern.
1125         (bic_SR): New pattern.
1126         (bis_SR): New pattern.
1127         * doc/extend.texi: Document MSP430 function attributes and builtin
1128         functions.
1130 2013-09-17  Richard Biener  <rguenther@suse.de>
1132         PR tree-optimization/58432
1133         * tree-loop-distribution.c (tree_loop_distribution): Also
1134         scan PHIs for outside loop uses and seed a partition from them.
1136 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
1138         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
1139         (restructure_reference): Call backtrace_base_for_ref.
1141 2013-09-17  Alan Modra  <amodra@gmail.com>
1143         PR target/57589
1144         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11 patch.
1146 2013-09-16  DJ Delorie  <dj@redhat.com>
1148         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
1149         vregs location for RL78/G10.
1150         (rl78_expand_prologue): Avoid SEL on G10.
1151         (rl78_expand_epilogue): Likewise.
1152         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
1153         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
1154         __RL78_G10__ when appropriate.
1155         (ASM_SPEC): Pass -mg10 along to the assembler.
1156         * config/rl78/rl78.md (sel_rb): Disable for G10.
1157         * config/rl78/rl78.opt: Add -mg10 option.
1158         * config/rl78/t-rl78: Add -mg10 multilib.
1160 2013-09-16  Xinliang David Li  <davidxl@google.com>
1162         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
1163         * omp-low.c (omp_max_vf): Ditto.
1164         (expand_omp_simd): Ditto.
1165         * tree-vectorizer.c (vectorize_loops): Ditto.
1166         (gate_vect_slp): Ditto.
1167         (gate_increase_alignment): Ditto.
1168         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
1169         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
1170         (gate_tree_vectorize): Name change.
1171         (tree_vectorize): Ditto.
1172         (pass_vectorize::gate): Call new function.
1173         (pass_vectorize::execute): Ditto.
1174         opts.c: O3 default setting change.
1175         (finish_options): Check new flag.
1176         * doc/invoke.texi: Document new flags.
1177         * common.opt: New flags.
1179 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
1181         * doc/tm.texi.in (Cond Exec Macros): Remove node.
1182         (Condition Code): Don't reference it.
1183         * doc/tm.texi: Regenerate.
1185 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
1187         PR middle-end/58418
1188         * lra-constraints.c (undo_optional_reloads): Consider all optional
1189         reload even if it did not get a hard reg.
1191 2013-09-16  Teresa Johnson  <tejohnson@google.com>
1193         * dumpfile.c (dump_loc): Remove newline emission.
1194         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
1195         emission to dump_printf_loc calls where missing.
1196         (vect_mark_for_runtime_alias_test): Ditto.
1197         (vect_analyze_data_ref_dependence): Ditto.
1198         (vect_analyze_data_ref_dependences): Ditto.
1199         (vect_slp_analyze_data_ref_dependence): Ditto.
1200         (vect_slp_analyze_data_ref_dependences): Ditto.
1201         (vect_compute_data_ref_alignment): Ditto.
1202         (vect_update_misalignment_for_peel): Ditto.
1203         (vect_verify_datarefs_alignment): Ditto.
1204         (vector_alignment_reachable_p): Ditto.
1205         (vect_get_data_access_cost): Ditto.
1206         (vect_enhance_data_refs_alignment): Ditto.
1207         (vect_find_same_alignment_drs): Ditto.
1208         (vect_analyze_data_refs_alignment): Ditto.
1209         (vect_analyze_group_access): Ditto.
1210         (vect_analyze_data_ref_access): Ditto.
1211         (vect_analyze_data_ref_accesses): Ditto.
1212         (vect_prune_runtime_alias_test_list): Ditto.
1213         (vect_analyze_data_refs): Ditto.
1214         (vect_create_addr_base_for_vector_ref): Ditto.
1215         (vect_create_data_ref_ptr): Ditto.
1216         (vect_grouped_store_supported): Ditto.
1217         (vect_grouped_load_supported): Ditto.
1218         * value-prof.c (check_counter): Ditto.
1219         (check_ic_target): Ditto.
1220         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
1221         (vect_recog_widen_mult_pattern): Ditto.
1222         (vect_recog_widen_sum_pattern): Ditto.
1223         (vect_recog_over_widening_pattern): Ditto.
1224         (vect_recog_widen_shift_pattern): Ditto.
1225         (vect_recog_rotate_pattern): Ditto.
1226         (vect_recog_vector_vector_shift_pattern): Ditto.
1227         (vect_recog_divmod_pattern): Ditto.
1228         (vect_recog_mixed_size_cond_pattern): Ditto.
1229         (vect_recog_bool_pattern): Ditto.
1230         (vect_pattern_recog_1): Ditto.
1231         (vect_pattern_recog): Ditto.
1232         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
1233         (vect_is_simple_iv_evolution): Ditto.
1234         (vect_analyze_scalar_cycles_1): Ditto.
1235         (vect_get_loop_niters): Ditto.
1236         (vect_analyze_loop_1): Ditto.
1237         (vect_analyze_loop_form): Ditto.
1238         (vect_analyze_loop_operations): Ditto.
1239         (vect_analyze_loop_2): Ditto.
1240         (vect_analyze_loop): Ditto.
1241         (report_vect_op): Ditto.
1242         (vect_is_slp_reduction): Ditto.
1243         (vect_is_simple_reduction_1): Ditto.
1244         (vect_get_known_peeling_cost): Ditto.
1245         (vect_estimate_min_profitable_iters): Ditto.
1246         (vect_model_reduction_cost): Ditto.
1247         (vect_model_induction_cost): Ditto.
1248         (get_initial_def_for_induction): Ditto.
1249         (vect_create_epilog_for_reduction): Ditto.
1250         (vectorizable_reduction): Ditto.
1251         (vectorizable_induction): Ditto.
1252         (vectorizable_live_operation): Ditto.
1253         (vect_loop_kill_debug_uses): Ditto.
1254         (vect_transform_loop): Ditto.
1255         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
1256         (vect_stmt_relevant_p): Ditto.
1257         (process_use): Ditto.
1258         (vect_mark_stmts_to_be_vectorized): Ditto.
1259         (vect_model_simple_cost): Ditto.
1260         (vect_model_promotion_demotion_cost): Ditto.
1261         (vect_model_store_cost): Ditto.
1262         (vect_get_store_cost): Ditto.
1263         (vect_model_load_cost): Ditto.
1264         (vect_get_load_cost): Ditto.
1265         (vect_init_vector_1): Ditto.
1266         (vect_get_vec_def_for_operand): Ditto.
1267         (vect_finish_stmt_generation): Ditto.
1268         (vectorizable_call): Ditto.
1269         (vectorizable_conversion): Ditto.
1270         (vectorizable_assignment): Ditto.
1271         (vectorizable_shift): Ditto.
1272         (vectorizable_operation): Ditto.
1273         (vectorizable_store): Ditto.
1274         (vectorizable_load): Ditto.
1275         (vectorizable_condition): Ditto.
1276         (vect_analyze_stmt): Ditto.
1277         (vect_transform_stmt): Ditto.
1278         (vect_is_simple_use): Ditto.
1279         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
1280         (vect_can_advance_ivs_p): Ditto.
1281         (vect_update_ivs_after_vectorizer): Ditto.
1282         (vect_do_peeling_for_loop_bound): Ditto.
1283         (vect_gen_niters_for_prolog_loop): Ditto.
1284         (vect_update_inits_of_drs): Ditto.
1285         (vect_create_cond_for_alias_checks): Ditto.
1286         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
1287         (vect_build_slp_tree_1): Ditto.
1288         (vect_supported_load_permutation_p): Ditto.
1289         (vect_analyze_slp_instance): Ditto.
1290         (vect_analyze_slp): Ditto.
1291         (vect_make_slp_decision): Ditto.
1292         (vect_detect_hybrid_slp): Ditto.
1293         (vect_bb_vectorization_profitable_p): Ditto.
1294         (vect_slp_analyze_bb_1): Ditto.
1295         (vect_update_slp_costs_according_to_vf): Ditto.
1296         (vect_get_mask_element): Ditto.
1297         (vect_transform_slp_perm_load): Ditto.
1298         (vect_schedule_slp_instance): Ditto.
1299         (vect_schedule_slp): Ditto.
1300         (vect_slp_transform_bb): Ditto.
1301         * profile.c (read_profile_edge_counts): Ditto.
1302         (compute_branch_probabilities): Ditto.
1303         * coverage.c (get_coverage_counts): Ditto.
1305 2013-09-16  Diego Novillo  <dnovillo@google.com>
1307         * tree-core.h: Add missing comment lines from refactoring
1308         of tree.h.
1310 2013-09-16  Jan Hubicka  <jh@suse.cz>
1312         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
1313         abstract functions; for static functions check the presence
1314         of body.
1316 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
1318         * config/aarch64/aarch64-simd-builtins.def (fma): New.
1319         * config/aarch64/aarch64-simd.md
1320         (aarch64_mla_elt<mode>): New.
1321         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1322         (aarch64_mls_elt<mode>): Likewise.
1323         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1324         (aarch64_fma4_elt<mode>): Likewise.
1325         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1326         (aarch64_fma4_elt_to_128v2df): Likewise.
1327         (aarch64_fma4_elt_to_64df): Likewise.
1328         (fnma<mode>4): Likewise.
1329         (aarch64_fnma4_elt<mode>): Likewise.
1330         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1331         (aarch64_fnma4_elt_to_128v2df): Likewise.
1332         (aarch64_fnma4_elt_to_64df): Likewise.
1333         * config/aarch64/iterators.md (VDQSF): New.
1334         * config/aarch64/arm_neon.h
1335         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
1336         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
1338 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
1340         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
1341         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1342         (aarch64_mul3_elt_to_128df): Likewise.
1343         (aarch64_mul3_elt_to_64v2df): Likewise.
1344         * config/aarch64/iterators.md (VEL): Also handle DFmode.
1345         (VMUL): New.
1346         (VMUL_CHANGE_NLANES) Likewise.
1347         (h_con): Likewise.
1348         (f): Likewise.
1349         * config/aarch64/arm_neon.h
1350         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
1352 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
1354         * config/aarch64/arm_neon.h
1355         (vcvtx_high_f32_f64): Fix parameters.
1357 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1358             Uros Bizjak  <ubizjak@gmail.com>
1360         * config/alpha.c: Include tree-ssanames.h.
1362 2013-09-16  Richard Biener  <rguenther@suse.de>
1364         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
1365         (dot_rdg_1): Handle control_dd.
1366         (create_edge_for_control_dependence): New function.
1367         (create_rdg_edges): Add control dependences if asked for.
1368         (build_rdg): Likewise.
1369         (generate_loops_for_partition): If there are not necessary
1370         control stmts remove all their dependencies.
1371         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
1372         (distribute_loop): Pass on control dependences.
1373         (tree_loop_distribution): Compute control dependences and remove
1374         restriction on number of loop nodes.
1376 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
1378         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
1379         for internal calls.
1381 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
1383         * cse.c (try_const_anchors): Punt on CC modes.
1385 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1387         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
1389 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
1391         PR target/58382
1392         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
1393         calls to word_mode.
1395 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
1397         PR target/48094
1398         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
1399         seen.
1400         (darwin_objc1_section): Likewise.
1401         (darwin_file_end): Emit Image Info section when required.
1403 2013-09-14  Jan Hubicka  <jh@suse.cz>
1405         * tree-into-ssa.c (gate_into_ssa): New.
1406         (pass_data_build_ssa): Use it.
1407         * cgraph.h (expand_thunk): Update prototype.
1408         * cgraphunit.c (analyze_function): Expand thunks early.
1409         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
1410         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
1411         set lowered flag; do not add new function.
1412         (assemble_thunks_and_aliases): Update.
1413         * tree-ssa.c (gate_init_datastructures): New gate.
1414         (pass_data_init_datastructures): Use it.
1416 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
1418         PR target/58269
1419         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
1420         xmm register set described in the psABI.
1422 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
1424         * dwarf2out.c (should_emit_struct_debug): Add check
1425         for type_decl variable is not NULL.
1427 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
1429         * config.gcc: Use new winnt-c.c target hooks
1430         * config/t-winnt: New file
1431         * config/winnt-c.c: New file
1432         * doc/tm.texi.in: Document new hook
1433         * doc/tm.texi: Regenerated
1435 2013-09-13  Jan Hubicka  <jh@suse.cz>
1437         PR middle-end/58094
1438         * ipa-inline.c (check_callers): New function.
1439         (check_caller_edge): Remove.
1440         (want_inline_function_to_all_callers_p): Also permit alises that are
1441         called dirrectly.
1442         (inline_to_all_callers): Terminate the walk when devirtualization
1443         introduce new calls.
1445 2013-09-13  Jan Hubicka  <jh@suse.cz>
1447         * ipa-inline-analysis.c (struct growth_data): Add node.
1448         (do_estimate_growth_1): Fix detection of recursion.
1450 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
1452         PR tree-optimization/58392
1453         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
1454         to avoid shadowing of outer loop variable.  If
1455         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
1456         replace_by_duplicate_decl simduid of loops that have it set and
1457         set dest_cfun->has_simduid_loops and/or
1458         dest_cfun->has_force_vect_loops.
1459         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
1460         instead of maybe_lookup_decl.
1461         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
1462         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
1463         call.  Copy over force_vect and copy and remap simduid.  Set
1464         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
1465         (copy_cfg_body): Remove blocks_to_copy argument.  Use
1466         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
1467         caller.  Don't set cfun->has_simduid_loops and/or
1468         cfun->has_force_vect_loops here.
1469         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
1470         caller.
1471         (expand_call_inline, tree_function_versioning): Adjust copy_body
1472         callers.
1474 2013-09-13  Martin Jambor  <mjambor@suse.cz>
1476         PR bootstrap/58388
1477         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
1478         the assert if the edge was a speculative one.
1480 2013-09-13  Richard Biener  <rguenther@suse.de>
1482         * tree-data-ref.h (known_dependences_p): Move here ...
1483         * tree-loop-distribution.c (known_dependences_p): ... from here.
1484         (dump_rdg_component, debug_rdg_component): Remove.
1485         (dump_rdg): Adjust.
1486         (generate_loops_for_partition): Use gimple_uid instead of
1487         relying on matching stmt visit order.
1488         (rdg_build_partitions): Take starting stmt vector.
1489         (ldist_gen): Merge into ...
1490         (distribute_loop): ... this function.  Do not compute starting
1491         vertices vector.
1492         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
1494 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1496         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
1497         Set type attribute correctly. Set predicable_short_it attribute.
1498         (cmpsi_shiftsi): Remove %? from output template.
1500 2013-09-13  Richard Biener  <rguenther@suse.de>
1502         * tree-loop-distribution.c (struct rdg_component,
1503         rdg_defs_used_in_other_loops_p, free_rdg_components,
1504         rdg_build_components): Remove.
1505         (stmts_from_loop): Do not record virtual PHIs.
1506         (generate_loops_for_partition): Skip virtual PHIs.
1507         (build_rdg_partition_for_component): Rename to ...
1508         (build_rdg_partition_for_vertex): ... this and adjust.
1509         (rdg_build_partitions): Take a vector of starting vertices
1510         instead of components.  Remove unnecessary leftover handling.
1511         (ldist_gen): Do not build components or record other stores.
1512         (distribute_loop): Do not distribute loops containing stmts
1513         with side-effects.
1515 2013-09-13  Christian Bruel  <christian.bruel@st.com>
1517         PR target/58314
1518         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
1520 2013-09-13  Kai Tietz  <ktietz@redhat.com>
1522         * config.gcc: Separate cases for mingw and cygwin targets,
1523         and add 64-bit cygwin target case.
1525         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
1526         dll-export inline-functions.
1527         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
1530 2013-09-13  Jeff Law  <law@redhat.com>
1532         PR middle-end/58387
1533         Revert:
1534         2013-09-06  Jeff Law  <law@redhat.com>
1536         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1537         edge implied equivalences into successor phis.
1539 2013-09-12  DJ Delorie  <dj@redhat.com>
1541         * config/rl78/rl78-virt.md: Change from | to \; for asm line
1542         separators.
1544 2013-09-12  Brooks Moses  <bmoses@google.com>
1546         PR driver/42955
1547         * Makefile.in: Do not install driver binaries in $(target)/bin.
1549 2013-09-12  DJ Delorie  <dj@redhat.com>
1551         * config/rl78/rl78.opt (mrelax): New.
1552         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
1553         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
1555         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
1556         between SP and FP.
1557         (rl78_expand_epilogue): Likewise.
1559 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
1561         PR middle-end/58335
1562         * lra-eliminations.c (remove_reg_equal_offset_note): New.
1563         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
1564         pointer elimination with using remove_reg_equal_offset_note.
1566 2013-09-12  DJ Delorie  <dj@redhat.com>
1568         * config/msp430/: New port.
1569         * config.gcc (msp430): Added.
1570         * doc/invoke.texi: Document MSP430 options.
1571         * doc/install.texi: Document msp430-elf
1572         * doc/md.texi: Document msp430-elf
1573         * doc/contrib.texi: Document msp430-elf
1575         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
1576         PSImode.
1578 2013-09-12  Martin Jambor  <mjambor@suse.cz>
1580         PR ipa/58389
1581         * ipa-prop.c (remove_described_reference): Give up if the edge in the
1582         reference descriptor is NULL.
1583         (ipa_edge_removal_hook): If owning a reference descriptor, set its
1584         edge to NULL.
1586 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
1588         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
1589         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
1590         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
1591         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
1592         make_temp_ssa_name): move to tree-ssanames.h
1593         * tree-ssa-alias.h: Move prototype.
1594         * tree-ssa.h: Include tree-ssanames.h.
1595         * tree-ssanames.c (FREE_SSANAMES): Move to here.
1596         * tree-ssanames.h: New.  Move items from tree-flow*.h
1597         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
1599 2013-09-12  Richard Biener  <rguenther@suse.de>
1601         PR tree-optimization/58404
1602         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
1603         propagate non-invariant addresses into dereferences wrapped
1604         in component references.
1606 2013-09-12  Richard Biener  <rguenther@suse.de>
1608         PR tree-optimization/58402
1609         * passes.def: Move pass_late_warn_uninitialized later.
1611 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
1613         * tree-ssa.h: New.  Move content from tree-flow.h and
1614         tree-flow-inline.h.
1615         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
1616         Move prototypes belonging to tree-ssa.c.
1617         * tree-flow-inline.h (redirect_edge_var_map_def,
1618         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
1619         tree-ssa.h.
1620         * gimple.h: Adjust prototypes.
1621         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
1622         to...
1623         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
1624         * tree.h: Move prototype to tree-ssa.h.
1625         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
1626         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
1627         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
1628         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
1629         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
1630         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
1631         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
1632         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
1633         graphite-blocking.c, graphite-clast-to-gimple.c,
1634         graphite-dependences.c, graphite-interchange.c,
1635         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
1636         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
1637         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
1638         ipa-reference.c, ipa-split.c, ipa-utils.c,
1639         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
1640         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
1641         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
1642         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1643         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
1644         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
1645         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
1646         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
1647         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
1648         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
1649         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
1650         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
1651         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
1652         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
1653         value-prof.c, var-tracking.c,
1654         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
1656 2013-09-12  Richard Biener  <rguenther@suse.de>
1658         PR tree-optimization/58396
1659         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
1660         (build_rdg): Take a loop-nest parameter, fix memleaks.
1661         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
1663 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1665         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
1666         for SLM.
1668 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
1670         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
1671         The FMA instruction names should have a 'v' prefix.
1673 2013-09-12  Richard Biener  <rguenther@suse.de>
1675         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
1676         (dot_rdg): Use popen instead of system in optional code.
1677         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
1678         (already_processed_vertex_p): Adjust.
1679         (has_anti_or_output_dependence, predecessor_has_mem_write,
1680         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
1681         rdg_flag_uses): Remove.
1682         (rdg_flag_vertex): Simplify.
1683         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
1684         remove recursion.
1685         (build_rdg_partition_for_component): Process the first vertex
1686         of a component only.
1687         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
1689 2013-09-12  Alan Modra  <amodra@gmail.com>
1691         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
1693 2013-09-11  DJ Delorie  <dj@redhat.com>
1694             Nick Clifton  <nickc@redhat.com>
1696         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
1697         (rl78_stack_based_mem): New.
1698         * config/rl78/constraints.md (Iv08): New.
1699         (Iv16): New.
1700         (Iv24): New.
1701         (Is09): New.
1702         (Is17): New.
1703         (Is25): New.
1704         (ISsi): New.
1705         (IShi): New.
1706         (ISqi): New.
1707         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
1708         (movhi): Likewise.
1709         (movsi): Change from expand to insn-and-split.
1710         (ashrsi3): Clobber AX.
1711         (lshrsi3): New.
1712         (ashlsi3): New.
1713         (cbranchsi4): New.
1714         * config/rl78/rl78.md (CC_REG): Fix.
1715         (addsi3): Allow memory and immediate operands.
1716         (addsi3_internal): Split into...
1717         (addsi3_internal_virt): ...new, and ...
1718         (addsi3_internal_real): ...new.
1719         (subsi): New.
1720         (subsi3_internal_virt): New.
1721         (subsi3_internal_real): New.
1722         (mulsi3): Add memory operand.
1723         (mulsi3_rl78): Likewise.
1724         (mulsi3_g13): Likewise.
1725         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
1726         (cbranchqi4_real): Add more constraint options.
1727         (cbranchhi4_real): Expand pattern.
1728         (cbranchhi4_real_signed): New.
1729         (cbranchhi4_real_inverted): New.
1730         (cbranchsi4_real_lt): New.
1731         (cbranchsi4_real_ge): New.
1732         (cbranchsi4_real_signed): New.
1733         (cbranchsi4_real): New.
1734         (peephole2): New.
1735         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
1736         constant shifts.
1737         (lshrsi3_virt): Likewise.
1738         (ashlsi3_virt): Likewise.
1739         (cbranchqi4_virt_signed): New.
1740         (cbranchhi4_virt_signed): New.
1741         (cbranchsi4_virt): New.
1742         * config/rl78/rl78.c: Whitespace fixes throughout.
1743         (move_elim_pass): New.
1744         (pass_data_rl78_move_elim): New.
1745         (pass_rl78_move_elim): New.
1746         (make_pass_rl78_move_elim): New.
1747         (rl78_devirt_info): Run devirt earlier.
1748         (rl78_move_elim_info): New.
1749         (rl78_asm_file_start): Register it.
1750         (rl78_split_movsi): New.
1751         (rl78_as_legitimate_address): Allow virtual base registers when
1752         appropriate.
1753         (rl78_addr_space_convert): Remove spurious debug stuff.
1754         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
1755         (rl78_print_operand): More cases for not printing '#'.
1756         (rl78_expand_compare): Remove most of the logic.
1757         (content_memory): New.
1758         (clear_content_memory): New.
1759         (get_content_index): New.
1760         (get_content_name): New.
1761         (display_content_memory): New.
1762         (update_content): New.
1763         (record_content): New.
1764         (already_contains): New.
1765         (insn_ok_now): Re-recog insns with virtual registers.
1766         (add_postponed_content_update): New.
1767         (process_postponed_content_update): New.
1768         (gen_and_emit_move): New.
1769         (transcode_memory_rtx): Record new location content.
1770         Use gen_and_emit_move.
1771         (force_into_acc): New.
1772         (move_to_acc): Use gen_and_emit_move.
1773         (move_from_acc): Likewise.
1774         (move_acc_to_reg): Likewise.
1775         (move_to_x): Likewise.
1776         (move_to_hl): Likewise.
1777         (move_to_de): Likewise.
1778         (rl78_alloc_physical_registers_op1): Record location content.
1779         (has_constraint): New.
1780         (rl78_alloc_physical_registers_op2): Record location content.
1781         Optimize use of HL.
1782         (rl78_alloc_physical_registers_ro1): Likewise.
1783         (rl78_alloc_physical_registers_cmp): Likewise.
1784         (rl78_alloc_physical_registers_umul): Likewise.
1785         (rl78_alloc_address_registers_macax): New.
1786         (rl78_alloc_physical_registers): Initialize and set location
1787         content memory as needed.
1788         (rl78_reorg): Make sure split2 is called.
1789         (rl78_rtx_costs): New.
1791 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
1793         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
1794         for (not (neg ...)) and (neg (not ...)) cases.
1796 2013-09-11  Richard Biener  <rguenther@suse.de>
1798         PR middle-end/58377
1799         * passes.def: Split critical edges before late uninit warning passes.
1800         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
1802 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
1804         PR tree-optimization/58385
1805         * fold-const.c (build_range_check): If both low and high are NULL,
1806         use omit_one_operand_loc to preserve exp side-effects.
1808 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1810         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
1812 2013-09-11  Richard Biener  <rguenther@suse.de>
1814         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
1815         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
1816         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
1817         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
1818         stmts_from_loop, known_dependences_p, build_empty_rdg,
1819         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
1820         * tree-loop-distribution.c: ... here.
1821         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
1822         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
1823         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
1824         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
1825         * tree-loop-distribution.c: ... here.
1826         * tree-loop-distribution.c: Include gimple-pretty-print.h.
1827         (struct partition_s): Add loops member.
1828         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
1829         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
1830         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
1832 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1833             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1834             Sergey Lega  <sergey.s.lega@intel.com>
1835             Anna Tikhonova  <anna.tikhonova@intel.com>
1836             Ilya Tocar  <ilya.tocar@intel.com>
1837             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1838             Ilya Verbin  <ilya.verbin@intel.com>
1839             Kirill Yukhin  <kirill.yukhin@intel.com>
1840             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1842         * config/i386/constraints.md (k): New.
1843         (Yk): Ditto.
1844         * config/i386/i386.c (const regclass_map): Add new mask registers.
1845         (dbx_register_map): Ditto.
1846         (dbx64_register_map): Ditto.
1847         (svr4_dbx_register_map): Ditto.
1848         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
1849         disabled.
1850         (ix86_preferred_reload_class): Disable constants for mask registers.
1851         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
1852         (ix86_hard_regno_mode_ok): Support new mask registers.
1853         (x86_order_regs_for_local_alloc): Ditto.
1854         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
1855         (FIXED_REGISTERS): Add new mask registers.
1856         (CALL_USED_REGISTERS): Ditto.
1857         (REG_ALLOC_ORDER): Ditto.
1858         (VALID_MASK_REG_MODE): New.
1859         (FIRST_MASK_REG): Ditto.
1860         (LAST_MASK_REG): Ditto.
1861         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
1862         (MAYBE_MASK_CLASS_P): New.
1863         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
1864         (REG_CLASS_CONTENTS): Ditto.
1865         (MASK_REGNO_P): New.
1866         (ANY_MASK_REG_P): Ditto.
1867         (HI_REGISTER_NAMES): Add new mask registers.
1868         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
1869         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
1870         mask registers.
1871         (attribute "type"): Add mskmov, msklog.
1872         (attribute "length_immediate"): Support them.
1873         (attribute "memory"): Ditto.
1874         (attribute "prefix_0f"): Ditto.
1875         (*movhi_internal): Support new mask registers.
1876         (*movqi_internal): Ditto.
1877         (define_split): Split out clobber pattern is a logic
1878         insn on mask registers.
1879         (*k<logic><mode>): New.
1880         (*andhi_1): Extend to support mask regs.
1881         (*andqi_1): Extend to support mask regs.
1882         (kandn<mode>): New.
1883         (define_split): Split and-not to and and not if operands
1884         are not mask regs.
1885         (*<code><mode>_1): Separate HI mode to new pattern...
1886         (*<code>hi_1): This.
1887         (*<code>qi_1): Extend to support mask regs.
1888         (kxnor<mode>): New.
1889         (kortestzhi): Ditto.
1890         (kortestchi): Ditto.
1891         (kunpckhi): Ditto.
1892         (*one_cmpl<mode>2_1): Remove HImode and handle it...
1893         (*one_cmplhi2_1): ...Here, now with mask registers support.
1894         (*one_cmplqi2_1): Support new mask registers.
1895         (HI/QImode arithmetics splitter): Don't split if mask registers
1896         are used.
1897         (HI/QImode not splitter): Ditto.
1898         * config/i386/predicated.md (mask_reg_operand): New.
1899         (general_reg_operand): Ditto.
1901 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1903         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
1904         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
1906 2013-09-10  Jeff Law  <law@redhat.com>
1908         PR tree-optimization/58380
1909         * tree-ssa-threadupdate.c (thread_block): Recognize another case
1910         of threading through a buried loop header.
1912         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
1913         return value for single successor case.
1915 2013-09-10  Jan Hubicka  <jh@suse.cz>
1917         * ipa-devirt.c (ipa_devirt): Enable with LTO.
1919 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
1921         PR target/58361
1922         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
1923         support conditional execution.
1924         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
1926 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
1928         * lra.c (lra): Clear lra_optional_reload_pseudos before every
1929         constraint pass.
1930         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
1931         Check destination too to check move insn.
1932         (undo_optional_reloads): Add check that the original peudo did not
1933         changed its allocation and the optional reload was inherited on last
1934         inheritance pass.  Break loop after deciding to keep optional reload.
1935         (lra_undo_inheritance): Add check that inherited pseudo still in
1936         memory.
1938 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
1940         * config/aarch64/aarch64.md (generic_sched): New.
1941         * config/aarch64/aarch64-generic.md (load): Make conditional
1942         on generic_sched attribute.
1943         (nonload): Likewise.
1945 2013-09-10  Jan Hubicka  <jh@suse.cz>
1947         * lto-cgraph.c: Include ipa-utils.h.
1948         (compute_ltrans_boundary): Also add possible targets into the boundary.
1950 2013-09-10  Jan Hubicka  <jh@suse.cz>
1952         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
1953         VAR_DECL of vtable rather than full expression.
1955 2013-09-10  Jan Hubicka  <jh@suse.cz>
1956             Paolo Carlini  <paolo.carlini@oracle.com>
1958         * cgraphunit.c (analyze_functions): Save input_location, set it
1959         to UNKNOWN_LOCATION and restore it at the end.
1961 2013-09-10  Martin Jambor  <mjambor@suse.cz>
1963         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
1964         represented by a thunk.
1966 2013-09-10  Jeff Law  <law@redhat.com>
1968         PR tree-optimization/58343
1969         * tree-ssa-threadupdate.c (thread_block): Identify and disable
1970         jump threading requests through loop headers buried in the middle
1971         of a jump threading path.
1973         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
1974         in return value/type.
1976 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
1978         PR rtl-optimization/58365
1979         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
1980         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
1981         it differs.
1983 2013-09-10  Richard Biener  <rguenther@suse.de>
1985         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
1986         * tree-data-ref.c (create_rdg_vertices): Collect all data
1987         references, signal failure to the caller, use data-ref API.
1988         (build_rdg): Compute data references only once.  Maintain lifetime
1989         of data references and data dependences from within RDG.
1990         (free_rdg): Free dependence relations.
1991         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
1992         inventing extra dependences.
1993         (distribute_loop): Update for RDG API changes.
1995 2013-09-10  Kai Tietz  <ktietz@redhat.com>
1997         * doc/invoke.texi (fms-extensions): Document changed
1998         behavior for ms-abi targets.
1999         * config/i386/i386.c (ix86_option_override_internal):
2000         Set default value of option -fms-extension for ms-abi targets.
2002 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
2004         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
2006 2013-09-10  Alan Modra  <amodra@gmail.com>
2008         PR target/58330
2009         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
2011 2013-09-10  Alan Modra  <amodra@gmail.com>
2013         * config/rs6000/predicates.md (add_cint_operand): New.
2014         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
2015         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
2016         using add_cint_operand.
2017         (largetoc_high_plus_aix): Likewise.
2019 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
2021         PR tree-optimization/58364
2022         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
2023         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
2024         the current range can't be an unconditional true or false.
2026 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
2028         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
2030 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
2032         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
2034 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
2036         PR c++/43452
2037         * doc/invoke.texi (-Wdelete-incomplete): Document it.
2039 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
2041         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
2042         NO_REGS for immediate that can't be moved directly into FP_REGS.
2044 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2046         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
2047         comparison with negated operand.
2048         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
2049         RTL form.
2051 2013-09-09  Richard Biener  <rguenther@suse.de>
2053         PR middle-end/58326
2054         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
2055         of a subloop record all its block as affecting loop-closed SSA form.
2057 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2059         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
2060         of an rtx/bitpos pair.
2061         (store_fixed_bit_field): Update accordingly.
2063 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2065         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
2066         GEN_INT.
2067         * builtins.c (expand_errno_check): Likewise.
2068         * dwarf2cfi.c (init_return_column_size): Likewise.
2069         * except.c (sjlj_mark_call_sites): Likewise.
2070         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
2071         * lra-constraints.c (emit_inc): Likewise.
2072         * ree.c (combine_set_extension): Likewise.
2073         * regmove.c (fixup_match_2): Likewise.
2074         * reload1.c (inc_for_reload): Likewise.
2076 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2078         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
2079         (force_to_mode, simplify_shift_const_1, simplify_comparison):
2080         Use gen_int_mode with the mode of the associated simplify_* call.
2081         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
2082         * expmed.c (expand_shift_1): Likewise.
2083         * function.c (instantiate_virtual_regs_in_insn): Likewise.
2084         * loop-iv.c (iv_number_of_iterations): Likewise.
2085         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
2086         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
2088 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2090         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
2091         of the associated expand_* call.
2092         (asan_emit_stack_protection): Likewise.
2093         * builtins.c (round_trampoline_addr): Likewise.
2094         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
2095         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
2096         (emit_store_flag): Likewise.
2097         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
2098         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
2099         Likewise.
2100         * function.c (instantiate_virtual_regs_in_insn): Likewise.
2101         * ifcvt.c (noce_try_store_flag_constants): Likewise.
2102         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
2103         * modulo-sched.c (generate_prolog_epilog): Likewise.
2104         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
2105         (expand_ctz, expand_ffs, expand_unop): Likewise.
2107 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2109         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
2110         of the associated gen_rtx_* call.
2111         * caller-save.c (init_caller_save): Likewise.
2112         * combine.c (find_split_point, make_extraction): Likewise.
2113         (make_compound_operation): Likewise.
2114         * dwarf2out.c (mem_loc_descriptor): Likewise.
2115         * explow.c (plus_constant, probe_stack_range): Likewise.
2116         * expmed.c (expand_mult_const): Likewise.
2117         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
2118         * reload1.c (init_reload): Likewise.
2119         * valtrack.c (cleanup_auto_inc_dec): Likewise.
2120         * var-tracking.c (adjust_mems): Likewise.
2121         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
2122         rather than gen_rtx_fmt_ee.
2124 2013-09-09  Jan Hubicka  <jh@suse.cz>
2126         PR middle-end/58294
2127         * value-prof.c (gimple_ic): Copy also abnormal edges.
2129 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
2131         * asan.c (asan_shadow_cst): Use gen_int_mode.
2133 2013-09-08  Jan Hubicka  <jh@suse.cz>
2135         * ipa-profile.c: Add toplevel comment.
2136         (ipa_propagate_frequency_1): Be more conservative when profile is read.
2137         (contains_hot_call_p): New function.
2138         (ipa_propagate_frequency): Set frequencies based on counts when
2139         profile is read.
2140         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
2141         profile; do not tamper with profile after inlining if it is read.
2143 2013-09-08  Jan Hubicka  <jh@suse.cz>
2145         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
2146         speculative edges.
2148 2013-09-08  Jan Hubicka  <jh@suse.cz>
2150         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
2151         summary generation.
2153 2013-09-08  Jeff Law  <law@redhat.com>
2155         PR bootstrap/58340
2156         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
2157         of 'found'.
2159 2013-09-08  Andi Kleen  <ak@linux.intel.com>
2161         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
2163 2013-09-08  Jan Hubicka  <jh@suse.cz>
2165         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
2167 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
2169         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
2170         for non-debug insns.
2171         * lra.c (new_insn_reg): Take the containing insn as a parameter.
2172         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
2173         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
2174         accordingly.
2176 2013-09-08  Jan Hubicka  <jh@suse.cz>
2178         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
2179         targets and devirtualize to BUILT_IN_UNREACHABLE.
2180         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
2181         * ipa.c (walk_polymorphic_call_targets): New function.
2182         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
2183         functions; use the new timevar.
2184         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
2185         was removed from the program.
2186         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
2187         not consider it in the walk when its vtable is dead.
2188         (possible_polymorphic_call_targets_1): Pass anonymous flag to
2189         record_binfo.
2190         (devirt_variable_node_removal_hook): New function.
2191         (possible_polymorphic_call_targets): Also register
2192         devirt_variable_node_removal_hook.
2193         (ipa_devirt): Do not do non-speculative devirtualization.
2194         (gate_ipa_devirt): One execute if devirtualizing speculatively.
2196 2013-09-08  Jan Hubicka  <jh@suse.cz>
2198         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
2199         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
2200         varpool_remove_variable_insertion_hook): Declare.
2201         * varpool.c (varpool_node_hook_list): New structure.
2202         (first_varpool_node_removal_hook,
2203         first_varpool_variable_insertion_hook): New variables.
2204         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
2205         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
2206         varpool_remove_variable_insertion_hook,
2207         varpool_call_variable_insertion_hooks): New functions.
2208         (varpool_remove_node): Use it.
2210 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
2212         PR c++/54941
2213         * diagnostic.c (diagnostic_build_prefix): When s.file is
2214         "<built-in>" don't output line and column numbers.
2216 2013-09-06  Jan Hubicka  <jh@suse.cz>
2218         * cgraphunit.c (expand_thunk): Get body before touching arguments.
2219         * lto-streamer-out.c: Stream thunks, too.
2220         * lto-streamer-in.c (input_function): Pop cfun here
2221         (lto_read_body): Instead of here.
2223 2013-09-06  Caroline Tice  <cmtice@google.com>
2225         * doc/install.texi: Add documentation for the --enable-vtable-verify
2226         and the --disable-libvtv configure options.
2228 2013-09-06  Jeff Law  <law@redhat.com>
2230         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
2231         edge implied equivalences into successor phis.
2233 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
2235         * resource.c (mark_referenced_resources): Handle COND_EXEC.
2237 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
2239         * resource.c (mark_target_live_regs): Compute resources taking
2240         into account if a call is predicated or not.
2242 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
2244         * toplev.c (output_stack_usage): Be prepared for suffixes created by
2245         the compiler in the function names.
2247 2013-09-06  Jan Hubicka  <jh@suse.cz>
2249         PR middle-end/58094
2250         * ipa-inline.c (has_caller_p): New function.
2251         (want_inline_function_to_all_callers_p): Use it.
2252         (sum_callers, inline_to_all_callers): Break out from ...
2253         (ipa_inline): ... here.
2255 2013-09-06  Jan Hubicka  <jh@suse.cz>
2257         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
2258         only when AVX is enabled.
2260 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2262         * config/aarch64/aarch64.md
2263         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
2264         is fpsimd_<load/store>2.
2265         (load_pair<mode>): Likewise.
2266         (store_pair<mode>): Likewise.
2268 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2270         * config/arm/types.md (type): Add "mrs" type.
2271         * config/aarch64/aarch64.md
2272         (aarch64_load_tp_hard): Make type "mrs".
2273         * config/arm/arm.md
2274         (load_tp_hard): Make type "mrs".
2275         * config/arm/cortex-a15.md: Update with new attributes.
2276         * config/arm/cortex-a5.md: Update with new attributes.
2277         * config/arm/cortex-a53.md: Update with new attributes.
2278         * config/arm/cortex-a7.md: Update with new attributes.
2279         * config/arm/cortex-a8.md: Update with new attributes.
2280         * config/arm/cortex-a9.md: Update with new attributes.
2281         * config/arm/cortex-m4.md: Update with new attributes.
2282         * config/arm/cortex-r4.md: Update with new attributes.
2283         * config/arm/fa526.md: Update with new attributes.
2284         * config/arm/fa606te.md: Update with new attributes.
2285         * config/arm/fa626te.md: Update with new attributes.
2286         * config/arm/fa726te.md: Update with new attributes.
2288 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2290         * config/aarch64/aarch64.md
2291         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
2292         (*movtf_aarch64): Likewise.
2293         * config/arm/arm.md
2294         (thumb1_movdi_insn): Use "multiple" for type where more than one
2295         instruction is used for a move.
2296         (*arm32_movhf): Likewise.
2297         (*thumb_movdf_insn): Likewise.
2299 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2301         * config/arm/types.md (type): Rename fcpys to fmov.
2302         * config/arm/vfp.md
2303         (*arm_movsi_vfp): Rename type fcpys as fmov.
2304         (*thumb2_movsi_vfp): Likewise
2305         (*movhf_vfp_neon): Likewise
2306         (*movhf_vfp): Likewise
2307         (*movsf_vfp): Likewise
2308         (*thumb2_movsf_vfp): Likewise
2309         (*movsfcc_vfp): Likewise
2310         (*thumb2_movsfcc_vfp): Likewise
2311         * config/aarch64/aarch64-simd.md
2312         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
2313         * config/aarch64/aarch64.md
2314         (*movsi_aarch64): Replace type mov_reg with fmovs.
2315         (*movdi_aarch64): Likewise
2316         (*movsf_aarch64): Likewise
2317         (*movdf_aarch64): Likewise
2318         * config/arm/arm.c
2319         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
2320         * config/arm/iwmmxt.md
2321         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
2322         * config/arm/arm1020e.md: Update with new attributes.
2323         * config/arm/cortex-a15-neon.md: Update with new attributes.
2324         * config/arm/cortex-a5.md: Update with new attributes.
2325         * config/arm/cortex-a53.md: Update with new attributes.
2326         * config/arm/cortex-a7.md: Update with new attributes.
2327         * config/arm/cortex-a8-neon.md: Update with new attributes.
2328         * config/arm/cortex-a9.md: Update with new attributes.
2329         * config/arm/cortex-m4-fpu.md: Update with new attributes.
2330         * config/arm/cortex-r4f.md: Update with new attributes.
2331         * config/arm/marvell-pj4.md: Update with new attributes.
2332         * config/arm/vfp11.md: Update with new attributes.
2334 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2336         * config/aarch64/aarch64.md
2337         (*madd<mode>): Fix type attribute.
2338         (*maddsi_uxtw): Likewise.
2339         (*msub<mode>): Likewise.
2340         (*msubsi_uxtw): Likewise.
2341         (<su_optab>maddsidi4): Likewise.
2342         (<su_optab>msubsidi4): Likewise.
2344 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2346         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
2347         * config/arm/arm.md (core_cycles): Remove fdiv.
2348         * config/arm/vfp.md:
2349         (*sqrtsf2_vfp): Update for attribute changes.
2350         (*sqrtdf2_vfp): Likewise.
2351         * config/aarch64/aarch64.md:
2352         (sqrt<mode>2): Update for attribute changes.
2353         * config/arm/arm1020e.md: Update with new attributes.
2354         * config/arm/cortex-a15-neon.md: Update with new attributes.
2355         * config/arm/cortex-a5.md: Update with new attributes.
2356         * config/arm/cortex-a53.md: Update with new attributes.
2357         * config/arm/cortex-a7.md: Update with new attributes.
2358         * config/arm/cortex-a8-neon.md: Update with new attributes.
2359         * config/arm/cortex-a9.md: Update with new attributes.
2360         * config/arm/cortex-m4-fpu.md: Update with new attributes.
2361         * config/arm/cortex-r4f.md: Update with new attributes.
2362         * config/arm/marvell-pj4.md: Update with new attributes.
2363         * config/arm/vfp11.md: Update with new attributes.
2365 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2367         * config/arm/types.md
2368         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
2369         * config/aarch64/aarch64.md
2370         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
2371         new attributes.
2372         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
2373         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
2374         (float<GPI:mode><GPF:mode>2): Likewise.
2375         * config/arm/vfp.md
2376         (*truncsisf2_vfp): Update with new attributes.
2377         (*truncsidf2_vfp): Likewise.
2378         (fixuns_truncsfsi2): Likewise.
2379         (fixuns_truncdfsi2): Likewise.
2380         (*floatsisf2_vfp): Likewise.
2381         (*floatsidf2_vfp): Likewise.
2382         (floatunssisf2): Likewise.
2383         (floatunssidf2): Likewise.
2384         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
2385         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
2386         * config/arm/arm1020e.md: Update with new attributes.
2387         * config/arm/cortex-a15-neon.md: Update with new attributes.
2388         * config/arm/cortex-a5.md: Update with new attributes.
2389         * config/arm/cortex-a53.md: Update with new attributes.
2390         * config/arm/cortex-a7.md: Update with new attributes.
2391         * config/arm/cortex-a8-neon.md: Update with new attributes.
2392         * config/arm/cortex-a9.md: Update with new attributes.
2393         * config/arm/cortex-m4-fpu.md: Update with new attributes.
2394         * config/arm/cortex-r4f.md: Update with new attributes.
2395         * config/arm/marvell-pj4.md: Update with new attributes.
2396         * config/arm/vfp11.md: Update with new attributes.
2398 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2400         * config/aarch64/arm_neon.h
2401         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
2402         (vqtbx<1,2,3,4><q>_s8): Likewise.
2404 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2406         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
2407         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
2408         (add<mode>3): Add type attribute.
2409         (add<mode>3): Likewise.
2410         (usadd<mode>3): Likewise.
2411         (ssadd<mode>3): Likewise.
2412         (sub<mode>3): Likewise.
2413         (sub<mode>3): Likewise.
2414         (ussub<mode>3): Likewise.
2415         (sssub<mode>3): Likewise.
2416         (ssmulsa3): Likewise.
2417         (usmulusa3): Likewise.
2418         (arm_usatsihi): Likewise.
2419         * config/arm/vfp.md
2420         (*movdi_vfp): Add types for all instructions.
2421         (*movdi_vfp_cortexa8): Likewise.
2422         (*movhf_vfp_neon): Likewise.
2423         (*movhf_vfp): Likewise.
2424         (*movdf_vfp): Likewise.
2425         (*thumb2_movdf_vfp): Likewise.
2426         (*thumb2_movdfcc_vfp): Likewise.
2427         * config/arm/arm.md: Add type attribute to all insn patterns.
2428         (*thumb1_adddi3): Add type attribute.
2429         (*arm_adddi3): Likewise.
2430         (*adddi_sesidi_di): Likewise.
2431         (*adddi_zesidi_di): Likewise.
2432         (*thumb1_addsi3): Likewise.
2433         (addsi3_compare0): Likewise.
2434         (*addsi3_compare0_scratch): Likewise.
2435         (*compare_negsi_si): Likewise.
2436         (cmpsi2_addneg): Likewise.
2437         (*addsi3_carryin_<optab>): Likewise.
2438         (*addsi3_carryin_alt2_<optab>): Likewise.
2439         (*addsi3_carryin_clobercc_<optab>): Likewise.
2440         (*subsi3_carryin): Likewise.
2441         (*subsi3_carryin_const): Likewise.
2442         (*subsi3_carryin_compare): Likewise.
2443         (*subsi3_carryin_compare_const): Likewise.
2444         (*arm_subdi3): Likewise.
2445         (*thumb_subdi3): Likewise.
2446         (*subdi_di_zesidi): Likewise.
2447         (*subdi_di_sesidi): Likewise.
2448         (*subdi_zesidi_di): Likewise.
2449         (*subdi_sesidi_di): Likewise.
2450         (*subdi_zesidi_ze): Likewise.
2451         (thumb1_subsi3_insn): Likewise.
2452         (*arm_subsi3_insn): Likewise.
2453         (*anddi3_insn): Likewise.
2454         (*anddi_zesidi_di): Likewise.
2455         (*anddi_sesdi_di): Likewise.
2456         (*ne_zeroextracts): Likewise.
2457         (*ne_zeroextracts): Likewise.
2458         (*ite_ne_zeroextr): Likewise.
2459         (*ite_ne_zeroextr): Likewise.
2460         (*anddi_notdi_di): Likewise.
2461         (*anddi_notzesidi): Likewise.
2462         (*anddi_notsesidi): Likewise.
2463         (andsi_notsi_si): Likewise.
2464         (thumb1_bicsi3): Likewise.
2465         (*iordi3_insn): Likewise.
2466         (*iordi_zesidi_di): Likewise.
2467         (*iordi_sesidi_di): Likewise.
2468         (*thumb1_iorsi3_insn): Likewise.
2469         (*xordi3_insn): Likewise.
2470         (*xordi_zesidi_di): Likewise.
2471         (*xordi_sesidi_di): Likewise.
2472         (*arm_xorsi3): Likewise.
2473         (*andsi_iorsi3_no): Likewise.
2474         (*smax_0): Likewise.
2475         (*smax_m1): Likewise.
2476         (*arm_smax_insn): Likewise.
2477         (*smin_0): Likewise.
2478         (*arm_smin_insn): Likewise.
2479         (*arm_umaxsi3): Likewise.
2480         (*arm_uminsi3): Likewise.
2481         (*minmax_arithsi): Likewise.
2482         (*minmax_arithsi_): Likewise.
2483         (*satsi_<SAT:code>): Likewise.
2484         (arm_ashldi3_1bit): Likewise.
2485         (arm_ashrdi3_1bit): Likewise.
2486         (arm_lshrdi3_1bit): Likewise.
2487         (*arm_negdi2): Likewise.
2488         (*thumb1_negdi2): Likewise.
2489         (*arm_negsi2): Likewise.
2490         (*thumb1_negsi2): Likewise.
2491         (*negdi_extendsid): Likewise.
2492         (*negdi_zero_extend): Likewise.
2493         (*arm_abssi2): Likewise.
2494         (*thumb1_abssi2): Likewise.
2495         (*arm_neg_abssi2): Likewise.
2496         (*thumb1_neg_abss): Likewise.
2497         (one_cmpldi2): Likewise.
2498         (extend<mode>di2): Likewise.
2499         (*compareqi_eq0): Likewise.
2500         (*arm_extendhisi2addsi): Likewise.
2501         (*arm_movdi): Likewise.
2502         (*thumb1_movdi_insn): Likewise.
2503         (*arm_movt): Likewise.
2504         (*thumb1_movsi_insn): Likewise.
2505         (pic_add_dot_plus_four): Likewise.
2506         (pic_add_dot_plus_eight): Likewise.
2507         (tls_load_dot_plus_eight): Likewise.
2508         (*thumb1_movhi_insn): Likewise.
2509         (*thumb1_movsf_insn): Likewise.
2510         (*movdf_soft_insn): Likewise.
2511         (*thumb_movdf_insn): Likewise.
2512         (cbranchsi4_insn): Likewise.
2513         (cbranchsi4_scratch): Likewise.
2514         (*negated_cbranchsi4): Likewise.
2515         (*tbit_cbranch): Likewise.
2516         (*tlobits_cbranch): Likewise.
2517         (*tstsi3_cbranch): Likewise.
2518         (*cbranchne_decr1): Likewise.
2519         (*addsi3_cbranch): Likewise.
2520         (*addsi3_cbranch_scratch): Likewise.
2521         (*arm_cmpdi_insn): Likewise.
2522         (*arm_cmpdi_unsig): Likewise.
2523         (*arm_cmpdi_zero): Likewise.
2524         (*thumb_cmpdi_zero): Likewise.
2525         (*deleted_compare): Likewise.
2526         (*mov_scc): Likewise.
2527         (*mov_negscc): Likewise.
2528         (*mov_notscc): Likewise.
2529         (*cstoresi_eq0_thumb1_insn): Likewise.
2530         (cstoresi_nltu_thumb1): Likewise.
2531         (cstoresi_ltu_thu): Likewise.
2532         (thumb1_addsi3_addgeu): Likewise.
2533         (*arm_jump): Likewise.
2534         (*thumb_jump): Likewise.
2535         (*check_arch2): Likewise.
2536         (arm_casesi_internal): Likewise.
2537         (thumb1_casesi_dispatch): Likewise.
2538         (*arm_indirect_jump): Likewise.
2539         (*thumb1_indirect_jump): Likewise.
2540         (nop): Likewise.
2541         (*and_scc): Likewise.
2542         (*ior_scc): Likewise.
2543         (*compare_scc): Likewise.
2544         (*cond_move): Likewise.
2545         (*cond_arith): Likewise.
2546         (*cond_sub): Likewise.
2547         (*cmp_ite0): Likewise.
2548         (*cmp_ite1): Likewise.
2549         (*cmp_and): Likewise.
2550         (*cmp_ior): Likewise.
2551         (*ior_scc_scc): Likewise.
2552         (*ior_scc_scc_cmp): Likewise.
2553         (*and_scc_scc): Likewise.
2554         (*and_scc_scc_cmp): Likewise.
2555         (*and_scc_scc_nod): Likewise.
2556         (*negscc): Likewise.
2557         (movcond_addsi): Likewise.
2558         (movcond): Likewise.
2559         (*ifcompare_plus_move): Likewise.
2560         (*if_plus_move): Likewise.
2561         (*ifcompare_move_plus): Likewise.
2562         (*if_move_plus): Likewise.
2563         (*ifcompare_arith_arith): Likewise.
2564         (*if_arith_arith): Likewise.
2565         (*ifcompare_arith_move): Likewise.
2566         (*if_arith_move): Likewise.
2567         (*ifcompare_move_arith): Likewise.
2568         (*if_move_arith): Likewise.
2569         (*ifcompare_move_not): Likewise.
2570         (*if_move_not): Likewise.
2571         (*ifcompare_not_move): Likewise.
2572         (*if_not_move): Likewise.
2573         (*ifcompare_shift_move): Likewise.
2574         (*if_shift_move): Likewise.
2575         (*ifcompare_move_shift): Likewise.
2576         (*if_move_shift): Likewise.
2577         (*ifcompare_shift_shift): Likewise.
2578         (*ifcompare_not_arith): Likewise.
2579         (*ifcompare_arith_not): Likewise.
2580         (*if_arith_not): Likewise.
2581         (*ifcompare_neg_move): Likewise.
2582         (*if_neg_move): Likewise.
2583         (*ifcompare_move_neg): Likewise.
2584         (*if_move_neg): Likewise.
2585         (prologue_thumb1_interwork): Likewise.
2586         (*cond_move_not): Likewise.
2587         (*sign_extract_onebit): Likewise.
2588         (*not_signextract_onebit): Likewise.
2589         (stack_tie): Likewise.
2590         (align_4): Likewise.
2591         (align_8): Likewise.
2592         (consttable_end): Likewise.
2593         (consttable_1): Likewise.
2594         (consttable_2): Likewise.
2595         (consttable_4): Likewise.
2596         (consttable_8): Likewise.
2597         (consttable_16): Likewise.
2598         (*thumb1_tablejump): Likewise.
2599         (prefetch): Likewise.
2600         (force_register_use): Likewise.
2601         (thumb_eh_return): Likewise.
2602         (load_tp_hard): Likewise.
2603         (load_tp_soft): Likewise.
2604         (tlscall): Likewise.
2605         (*arm_movtas_ze): Likewise.
2606         (*arm_rev): Likewise.
2607         (*arm_revsh): Likewise.
2608         (*arm_rev16): Likewise.
2609         * config/arm/thumb2.md
2610         (*thumb2_smaxsi3): Likewise.
2611         (*thumb2_sminsi3): Likewise.
2612         (*thumb32_umaxsi3): Likewise.
2613         (*thumb2_uminsi3): Likewise.
2614         (*thumb2_negdi2): Likewise.
2615         (*thumb2_abssi2): Likewise.
2616         (*thumb2_neg_abss): Likewise.
2617         (*thumb2_movsi_insn): Likewise.
2618         (tls_load_dot_plus_four): Likewise.
2619         (*thumb2_movhi_insn): Likewise.
2620         (*thumb2_mov_scc): Likewise.
2621         (*thumb2_mov_negs): Likewise.
2622         (*thumb2_mov_negs): Likewise.
2623         (*thumb2_mov_nots): Likewise.
2624         (*thumb2_mov_nots): Likewise.
2625         (*thumb2_movsicc_): Likewise.
2626         (*thumb2_movsfcc_soft_insn): Likewise.
2627         (*thumb2_indirect_jump): Likewise.
2628         (*thumb2_and_scc): Likewise.
2629         (*thumb2_ior_scc): Likewise.
2630         (*thumb2_ior_scc_strict_it): Likewise.
2631         (*thumb2_cond_move): Likewise.
2632         (*thumb2_cond_arith): Likewise.
2633         (*thumb2_cond_ari): Likewise.
2634         (*thumb2_cond_sub): Likewise.
2635         (*thumb2_negscc): Likewise.
2636         (*thumb2_movcond): Likewise.
2637         (thumb2_casesi_internal): Likewise.
2638         (thumb2_casesi_internal_pic): Likewise.
2639         (*thumb2_alusi3_short): Likewise.
2640         (*thumb2_mov<mode>_shortim): Likewise.
2641         (*thumb2_addsi_short): Likewise.
2642         (*thumb2_subsi_short): Likewise.
2643         (thumb2_addsi3_compare0): Likewise.
2644         (*thumb2_cbz): Likewise.
2645         (*thumb2_cbnz): Likewise.
2646         (*thumb2_one_cmplsi2_short): Likewise.
2647         (*thumb2_negsi2_short): Likewise.
2648         (*orsi_notsi_si): Likewise.
2649         * config/arm/arm1020e.md: Update with new attributes.
2650         * config/arm/arm1026ejs.md: Update with new attributes.
2651         * config/arm/arm1136jfs.md: Update with new attributes.
2652         * config/arm/arm926ejs.md: Update with new attributes.
2653         * config/arm/cortex-a15.md: Update with new attributes.
2654         * config/arm/cortex-a5.md: Update with new attributes.
2655         * config/arm/cortex-a53.md: Update with new attributes.
2656         * config/arm/cortex-a7.md: Update with new attributes.
2657         * config/arm/cortex-a8.md: Update with new attributes.
2658         * config/arm/cortex-a9.md: Update with new attributes.
2659         * config/arm/cortex-m4.md: Update with new attributes.
2660         * config/arm/cortex-r4.md: Update with new attributes.
2661         * config/arm/fa526.md: Update with new attributes.
2662         * config/arm/fa606te.md: Update with new attributes.
2663         * config/arm/fa626te.md: Update with new attributes.
2664         * config/arm/fa726te.md: Update with new attributes.
2666 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
2668         * config/aarch64/aarch64-simd.md
2669         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
2670         <vwx> iterator to ensure correct register choice.
2671         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
2672         (aarch64_sqdmull_n<mode>): Likewise.
2673         (aarch64_sqdmull2_n<mode>_internal): Likewise.
2674         * config/aarch64/arm_neon.h
2675         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
2676         (vml<as><q>_n_<su>16): Likewise.
2677         (vml<as>l_high_lane<q>_<su>16): Likewise.
2678         (vml<as>l_high_n_<su>16): Likewise.
2679         (vml<as>l_lane<q>_<su>16): Likewise.
2680         (vml<as>l_n_<su>16): Likewise.
2681         (vmul<q>_lane<q>_<su>16): Likewise.
2682         (vmul<q>_n_<su>16): Likewise.
2683         (vmull_lane<q>_<su>16): Likewise.
2684         (vmull_n_<su>16): Likewise.
2685         (vmull_high_lane<q>_<su>16): Likewise.
2686         (vmull_high_n_<su>16): Likewise.
2687         (vqrdmulh<q>_n_s16): Likewise.
2689 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
2691         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
2692         have the correct lane parameter.
2694 2013-09-06  Richard Biener <rguenther@suse.de>
2696         * cfganal.c (control_dependences::~control_dependences):
2697         Properly free all of the vector.
2699 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
2701         PR target/58269
2702         * config/i386/i386.c (ix86_conditional_register_usage):
2703         Proper initialize extended SSE registers.
2705 2013-09-06  Jan Hubicka  <jh@suse.cz>
2707         PR tree-optimization/58311
2708         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
2710 2013-09-06  Jan Hubicka  <jh@suse.cz>
2712         * Makefile.in (tree-sra.o): Update dependencies.
2713         * tree-sra.c: Include ipa-utils.h
2714         (scan_function): Use recursive_call_p.
2715         (has_caller_p): New function.
2716         (cgraph_for_node_and_aliases): Count also callers of aliases.
2718 2013-09-06  Jan Hubicka  <jh@suse.cz>
2720         PR middle-end/58094
2721         * cgraph.h (symtab_semantically_equivalent_p): Declare.
2722         * tree-tailcall.c: Include ipa-utils.h.
2723         (find_tail_calls): Use it.
2724         * ipa-pure-const.c (check_call): Likewise.
2725         * ipa-utils.c (recursive_call_p): New function.
2726         * ipa-utils.h (recursive_call_p): Dclare.
2727         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
2728         (symtab_semantically_equivalent_p): New function.
2729         * Makefile.in (tree-tailcall.o): Update dependencies.
2731 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
2733         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
2734         non-inlinable part.
2736 2013-09-06  Richard Biener  <rguenther@suse.de>
2738         * lto-streamer.h (lto_global_var_decls): Remove.
2739         * Makefile.in (OBJS): Remove lto-symtab.o.
2740         (lto-symtab.o): Remove.
2741         (GTFILES): Remove lto-symtab.c
2742         * lto-symtab.c: Move to lto/
2744 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2746         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
2747         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
2748         (UNSPEC_FPINT_RINT): New constant definitions.
2749         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
2750         definition with 2 attributes.
2751         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
2752         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
2753         definitions.
2755 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2757         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
2758         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
2759         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
2760         attribute to "z196_alone".
2761         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
2762         "zEC12_simple".
2764 2013-09-06  Richard Biener  <rguenther@suse.de>
2766         * basic-block.h (class control_dependences): New.
2767         * tree-ssa-dce.c (control_dependence_map): Remove.
2768         (cd): New global.
2769         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
2770         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
2771         find_pdom, find_control_dependence, find_all_control_dependences):
2772         Move to cfganal.c.
2773         (mark_control_dependent_edges_necessary,
2774         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
2775         tree_dce_done, perform_tree_ssa_dce): Adjust.
2776         * cfganal.c (set_control_dependence_map_bit,
2777         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
2778         find_all_control_dependences): Move from tree-ssa-dce.c and
2779         implement as methods of control_dependences class.
2780         (control_dependences::control_dependences): New.
2781         (control_dependences::~control_dependences): Likewise.
2782         (control_dependences::get_edges_dependent_on): Likewise.
2783         (control_dependences::get_edge): Likewise.
2785 2013-09-04  Jan Hubicka  <jh@suse.cz>
2787         * tree.c (types_same_for_odr): Drop overactive check.
2788         * ipa-devirt.c (hash_type_name): Likewise.
2790 2013-09-04  Jan Hubicka  <jh@suse.cz>
2792         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
2793         (analyze_functions): ... here.
2795 2013-09-04  Jan Hubicka  <jh@suse.cz>
2797         PR middle-end/58201
2798         * cgraphunit.c (analyze_functions): Clear AUX fields
2799         after processing; initialize assembler name has.
2801 2013-09-05  Jeff Law  <law@redhat.com>
2803         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
2804         from thread_around_empty_block.  Record threading path into PATH.
2805         Recurse if threading through the initial block is successful.
2806         (thread_across_edge): Corresponding changes to slightly simplify.
2808 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2810         * config/aarch64/aarch64.md
2811         (type): Remove frecpe, frecps, frecpx.
2812         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
2813         fix to be a TARGET_SIMD instruction.
2814         (aarch64_frecps): Remove.
2815         * config/aarch64/aarch64-simd.md
2816         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
2817         (aarch64_frecps<mode>): Handle all float/vector of float modes.
2819 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2820             Sofiane Naci  <sofiane.naci@arm.com>
2822         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
2823         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
2824         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
2825         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
2826         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
2827         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
2828         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
2829         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
2830         into "clz, "rbit".  Rename "shift" to "shift_imm".
2831         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
2832         changes.  Update for attribute changes all occurrences of arlo_* and
2833         shift* types.
2834         * config/arm/arm-fixed.md: Update for attribute changes
2835         all occurrences of arlo_* types.
2836         * config/arm/thumb2.md: Update for attribute changes all occurrences
2837         of arlo_* types.
2838         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
2839         (cortexa7_older_only): Likewise.
2840         (cortexa7_younger):  Likewise.
2841         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
2842         (1020alu_shift_op): Likewise.
2843         (1020alu_shift_reg_op): Likewise.
2844         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
2845         (alu_shift_op): Likewise.
2846         (alu_shift_reg_op): Likewise.
2847         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
2848         (11_alu_shift_op): Likewise.
2849         (11_alu_shift_reg_op): Likewise.
2850         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
2851         (9_alu_shift_reg_op): Likewise.
2852         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
2853         attribute changes.
2854         (cortex_a15_alu_shift): Likewise.
2855         (cortex_a15_alu_shift_reg): Likewise.
2856         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
2857         attribute changes.
2858         (cortex_a5_alu_shift): Likewise.
2859         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
2860         attribute changes.
2861         (cortex_a53_alu_shift): Likewise.
2862         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
2863         attribute changes.
2864         (cortex_a7_alu_reg): Likewise.
2865         (cortex_a7_alu_shift): Likewise.
2866         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
2867         attribute changes.
2868         (cortex_a8_alu_shift): Likewise.
2869         (cortex_a8_alu_shift_reg): Likewise.
2870         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
2871         (cortex_a9_dp_shift): Likewise.
2872         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
2873         attribute changes.
2874         * config/arm/cortex-r4.md
2875         (cortex_r4_alu): Update for attribute changes.
2876         (cortex_r4_mov): Likewise.
2877         (cortex_r4_alu_shift_reg): Likewise.
2878         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
2879         (526_alu_shift_op): Likewise.
2880         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
2881         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
2882         (626te_alu_shift_op): Likewise.
2883         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
2884         (726te_alu_shift_op): Likewise.
2885         (726te_alu_shift_reg_op): Likewise.
2886         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
2887         (mp626_alu_shift_op): Likewise.
2888         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
2889         (pj4_alu_conds): Likewise.
2890         (pj4_shift): Likewise.
2891         (pj4_shift_conds): Likewise.
2892         (pj4_alu_shift): Likewise.
2893         (pj4_alu_shift_conds): Likewise.
2894         * config/aarch64/aarch64.md: Update for attribute change
2895         all occurrences of arlo_* and shift* types.
2897 2013-09-05  Mike Stump  <mikestump@comcast.net>
2899         * tree.h: Move documentation for tree_function_decl to tree-core.h
2900         with the declaration.
2902 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
2904         PR target/58139
2905         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
2906         looking for widest mode.
2908 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
2910         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
2912 2013-09-05  Richard Biener  <rguenther@suse.de>
2914         PR tree-optimization/58137
2915         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
2916         Do not create vectors of pointers.
2917         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
2918         types for the components of the vector initializer.
2919         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
2920         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
2922 2013-09-05  Martin Jambor  <mjambor@suse.cz>
2924         * ipa-prop.c (remove_described_reference): Accept missing references,
2925         return false if that hppens, otherwise return true.
2926         (cgraph_node_for_jfunc): New function.
2927         (try_decrement_rdesc_refcount): Likewise.
2928         (try_make_edge_direct_simple_call): Use them.
2929         (ipa_edge_removal_hook): Remove references from rdescs.
2930         (ipa_edge_duplication_hook): Clone rdescs and their references
2931         when the new edge has the same caller as the old one.
2932         * cgraph.c (cgraph_resolve_speculation): Remove speculative
2933         reference before removing any edges.
2935 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
2937         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
2938         initial store.
2939         * thumb2.md (thumb2_storewb_parisi): New pattern.
2941 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
2943         * config/aarch64/aarch64-option-extensions.def: Add
2944         AARCH64_OPT_EXTENSION of 'crc'.
2945         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
2946         (AARCH64_ISA_CRC): Ditto.
2947         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
2948         description of the CRC extension.
2950 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2952         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
2953         * config/rs6000/linux.h: Ditto.
2954         * alpha/linux.h: Ditto.
2955         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
2956         no_c99_libc_has_function.
2957         * config/c6x/uclinux-elf.h: Ditto.
2958         * config/lm32/uclinux-elf.h: Ditto.
2959         * config/m68k/uclinux.h: Ditto.
2960         * config/moxie/uclinux.h: Ditto.
2961         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
2962         (crisv32-*-linux*, cris-*-linux*): Ditto.
2963         * config/bfin/bfin.c: Include "tm_p.h".
2965 2013-09-05  Richard Biener  <rguenther@suse.de>
2967         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
2968         check for a definition without a basic-block.
2970 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2971             Sofiane Naci <sofiane.naci@arm.com>
2973         * config/aarch64/aarch64.md
2974         (*movti_aarch64): Rename r_2_f and f_2_r.
2975         (*movsf_aarch64): Likewise.
2976         (*movdf_aarch64): Likewise.
2977         (*movtf_aarch64): Likewise.
2978         (aarch64_movdi_<mode>low): Likewise.
2979         (aarch64_movdi_<mode>high): Likewise.
2980         (aarch64_mov<mode>high_di): Likewise.
2981         (aarch64_mov<mode>low_di): Likewise.
2982         (aarch64_movtilow_tilow): Likewise.
2983         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
2984         values to config/arm/types.md
2985         (attribute "conds"): Update for attribute change.
2986         (anddi3_insn): Likewise.
2987         (iordi3_insn): Likewise.
2988         (xordi3_insn): Likewise.
2989         (one_cmpldi2): Likewise.
2990         * config/arm/types.md (type): Add Neon types.
2991         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
2992         use "type" attribute.
2993         (movmisalign<mode>_neon_store): Likewise.
2994         (movmisalign<mode>_neon_load): Likewise.
2995         (vec_set<mode>_internal): Likewise.
2996         (vec_setv2di_internal): Likewise.
2997         (vec_extract<mode>): Likewise.
2998         (vec_extractv2di): Likewise.
2999         (add<mode>3_neon): Likewise.
3000         (adddi3_neon): Likewise.
3001         (sub<mode>3_neon): Likewise.
3002         (subdi3_neon): Likewise.
3003         (mul<mode>3_neon): Likewise.
3004         (mul<mode>3add<mode>_neon): Likewise.
3005         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
3006         (fma<VCVTF:mode>4)): Likewise.
3007         (fma<VCVTF:mode>4_intrinsic): Likewise.
3008         (fmsub<VCVTF:mode>4)): Likewise.
3009         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
3010         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
3011         (ior<mode>3): Likewise.
3012         (and<mode>3): Likewise.
3013         (anddi3_neon): Likewise.
3014         (orn<mode>3_neon): Likewise.
3015         (orndi3_neon): Likewise.
3016         (bic<mode>3_neon): Likewise.
3017         (bicdi3_neon): Likewise.
3018         (xor<mode>3): Likewise.
3019         (one_cmpl<mode>2): Likewise.
3020         (abs<mode>2): Likewise.
3021         (neg<mode>2): Likewise.
3022         (umin<mode>3_neon): Likewise.
3023         (umax<mode>3_neon): Likewise.
3024         (smin<mode>3_neon): Likewise.
3025         (smax<mode>3_neon): Likewise.
3026         (vashl<mode>3): Likewise.
3027         (vashr<mode>3_imm): Likewise.
3028         (vlshr<mode>3_imm): Likewise.
3029         (ashl<mode>3_signed): Likewise.
3030         (ashl<mode>3_unsigned): Likewise.
3031         (neon_load_count): Likewise.
3032         (ashldi3_neon_noclobber): Likewise.
3033         (signed_shift_di3_neon): Likewise.
3034         (unsigned_shift_di3_neon): Likewise.
3035         (ashrdi3_neon_imm_noclobber): Likewise.
3036         (lshrdi3_neon_imm_noclobber): Likewise.
3037         (widen_ssum<mode>3): Likewise.
3038         (widen_usum<mode>3): Likewise.
3039         (quad_halves_<code>v4si): Likewise.
3040         (quad_halves_<code>v4sf): Likewise.
3041         (quad_halves_<code>v8hi): Likewise.
3042         (quad_halves_<code>v16qi): Likewise.
3043         (reduc_splus_v2di): Likewise.
3044         (neon_vpadd_internal<mode>): Likewise.
3045         (neon_vpsmin<mode>): Likewise.
3046         (neon_vpsmax<mode>): Likewise.
3047         (neon_vpumin<mode>): Likewise.
3048         (neon_vpumax<mode>): Likewise.
3049         (ss_add<mode>_neon): Likewise.
3050         (us_add<mode>_neon): Likewise.
3051         (ss_sub<mode>_neon): Likewise.
3052         (us_sub<mode>_neon): Likewise.
3053         (neon_vadd<mode>_unspec): Likewise.
3054         (neon_vaddl<mode>): Likewise.
3055         (neon_vaddw<mode>): Likewise.
3056         (neon_vhadd<mode>): Likewise.
3057         (neon_vqadd<mode>): Likewise.
3058         (neon_vaddhn<mode>): Likewise.
3059         (neon_vmul<mode>): Likewise.
3060         (neon_vmla<mode>): Likewise.
3061         (neon_vmlal<mode>): Likewise.
3062         (neon_vmls<mode>): Likewise.
3063         (neon_vmlsl<mode>): Likewise.
3064         (neon_vqdmulh<mode>): Likewise.
3065         (neon_vqdmlal<mode>): Likewise.
3066         (neon_vqdmlsl<mode>): Likewise.
3067         (neon_vmull<mode>): Likewise.
3068         (neon_vqdmull<mode>): Likewise.
3069         (neon_vsub<mode>_unspec): Likewise.
3070         (neon_vsubl<mode>): Likewise.
3071         (neon_vsubw<mode>): Likewise.
3072         (neon_vqsub<mode>): Likewise.
3073         (neon_vhsub<mode>): Likewise.
3074         (neon_vsubhn<mode>): Likewise.
3075         (neon_vceq<mode>): Likewise.
3076         (neon_vcge<mode>): Likewise.
3077         (neon_vcgeu<mode>): Likewise.
3078         (neon_vcgt<mode>): Likewise.
3079         (neon_vcgtu<mode>): Likewise.
3080         (neon_vcle<mode>): Likewise.
3081         (neon_vclt<mode>): Likewise.
3082         (neon_vcage<mode>): Likewise.
3083         (neon_vcagt<mode>): Likewise.
3084         (neon_vtst<mode>): Likewise.
3085         (neon_vabd<mode>): Likewise.
3086         (neon_vabdl<mode>): Likewise.
3087         (neon_vaba<mode>): Likewise.
3088         (neon_vabal<mode>): Likewise.
3089         (neon_vmax<mode>): Likewise.
3090         (neon_vmin<mode>): Likewise.
3091         (neon_vpaddl<mode>): Likewise.
3092         (neon_vpadal<mode>): Likewise.
3093         (neon_vpmax<mode>): Likewise.
3094         (neon_vpmin<mode>): Likewise.
3095         (neon_vrecps<mode>): Likewise.
3096         (neon_vrsqrts<mode>): Likewise.
3097         (neon_vqabs<mode>): Likewise.
3098         (neon_vqneg<mode>): Likewise.
3099         (neon_vcls<mode>): Likewise.
3100         (clz<mode>2): Likewise.
3101         (popcount<mode>2): Likewise.
3102         (neon_vrecpe): Likewise.
3103         (neon_vrsqrte): Likewise.
3104         (neon_vget_lane<mode>_sext_internal): Likewise.
3105         (neon_vget_lane<mode>_zext_internal): Likewise.
3106         (neon_vdup_n<mode>): Likewise.
3107         (neon_vdup_nv2di): Likewise.
3108         (neon_vdpu_lane<mode>_internal): Likewise.
3109         (neon_vswp<mode>): Likewise.
3110         (float<mode><V_cvtto>2): Likewise.
3111         (floatuns<mode><V_cvtto>2): Likewise.
3112         (fix_trunc<mode><V_cvtto>)2): Likewise
3113         (fixuns_trunc<mode><V_cvtto)2): Likewise.
3114         (neon_vcvt<mode>): Likewise.
3115         (neon_vcvtv4sfv4hf): Likewise.
3116         (neon_vcvtv4hfv4sf): Likewise.
3117         (neon_vcvt_n<mode>): Likewise.
3118         (neon_vmovn<mode>): Likewise.
3119         (neon_vqmovn<mode>): Likewise.
3120         (neon_vqmovun<mode>): Likewise.
3121         (neon_vmovl<mode>): Likewise.
3122         (neon_vmul_lane<mode>): Likewise.
3123         (neon_vmull_lane<mode>): Likewise.
3124         (neon_vqdmull_lane<mode>): Likewise.
3125         (neon_vqdmulh_lane<mode>): Likewise.
3126         (neon_vmla_lane<mode>): Likewise.
3127         (neon_vmlal_lane<mode>): Likewise.
3128         (neon_vqdmlal_lane<mode>): Likewise.
3129         (neon_vmls_lane<mode>): Likewise.
3130         (neon_vmlsl_lane<mode>): Likewise.
3131         (neon_vqdmlsl_lane<mode>): Likewise.
3132         (neon_vext<mode>): Likewise.
3133         (neon_vrev64<mode>): Likewise.
3134         (neon_vrev32<mode>): Likewise.
3135         (neon_vrev16<mode>): Likewise.
3136         (neon_vbsl<mode>_internal): Likewise.
3137         (neon_vshl<mode>): Likewise.
3138         (neon_vqshl<mode>): Likewise.
3139         (neon_vshr_n<mode>): Likewise.
3140         (neon_vshrn_n<mode>): Likewise.
3141         (neon_vqshrn_n<mode>): Likewise.
3142         (neon_vqshrun_n<mode>): Likewise.
3143         (neon_vshl_n<mode>): Likewise.
3144         (neon_vqshl_n<mode>): Likewise.
3145         (neon_vqshlu_n<mode>): Likewise.
3146         (neon_vshll_n<mode>): Likewise.
3147         (neon_vsra_n<mode>): Likewise.
3148         (neon_vsri_n<mode>): Likewise.
3149         (neon_vsli_n<mode>): Likewise.
3150         (neon_vtbl1v8qi): Likewise.
3151         (neon_vtbl2v8qi): Likewise.
3152         (neon_vtbl3v8qi): Likewise.
3153         (neon_vtbl4v8qi): Likewise.
3154         (neon_vtbx1v8qi): Likewise.
3155         (neon_vtbx2v8qi): Likewise.
3156         (neon_vtbx3v8qi): Likewise.
3157         (neon_vtbx4v8qi): Likewise.
3158         (neon_vtrn<mode>_internal): Likewise.
3159         (neon_vzip<mode>_internal): Likewise.
3160         (neon_vuzp<mode>_internal): Likewise.
3161         (neon_vld1<mode>): Likewise.
3162         (neon_vld1_lane<mode>): Likewise.
3163         (neon_vld1_dup<mode>): Likewise.
3164         (neon_vld1_dupv2di): Likewise.
3165         (neon_vst1<mode>): Likewise.
3166         (neon_vst1_lane<mode>): Likewise.
3167         (neon_vld2<mode>): Likewise.
3168         (neon_vld2_lane<mode>): Likewise.
3169         (neon_vld2_dup<mode>): Likewise.
3170         (neon_vst2<mode>): Likewise.
3171         (neon_vst2_lane<mode>): Likewise.
3172         (neon_vld3<mode>): Likewise.
3173         (neon_vld3qa<mode>): Likewise.
3174         (neon_vld3qb<mode>): Likewise.
3175         (neon_vld3_lane<mode>): Likewise.
3176         (neon_vld3_dup<mode>): Likewise.
3177         (neon_vst3<mode>): Likewise.
3178         (neon_vst3qa<mode>): Likewise.
3179         (neon_vst3qb<mode>): Likewise.
3180         (neon_vst3_lane<mode>): Likewise.
3181         (neon_vld4<mode>): Likewise.
3182         (neon_vld4qa<mode>): Likewise.
3183         (neon_vld4qb<mode>): Likewise.
3184         (neon_vld4_lane<mode>): Likewise.
3185         (neon_vld4_dup<mode>): Likewise.
3186         (neon_vst4<mode>): Likewise.
3187         (neon_vst4qa<mode>): Likewise.
3188         (neon_vst4qb<mode>): Likewise.
3189         (neon_vst4_lane<mode>): Likewise.
3190         (neon_vec_unpack<US>_lo_<mode>): Likewise.
3191         (neon_vec_unpack<US>_hi_<mode>): Likewise.
3192         (neon_vec_<US>mult_lo_<mode>): Likewise.
3193         (neon_vec_<US>mult_hi_<mode>): Likewise.
3194         (neon_vec_<US>shiftl_<mode>): Likewise.
3195         (neon_unpack<US>_<mode>): Likewise.
3196         (neon_vec_<US>mult_<mode>): Likewise.
3197         (vec_pack_trunc_<mode>): Likewise.
3198         (neon_vec_pack_trunk_<mode>): Likewise.
3199         (neon_vabd<mode>_2): Likewise.
3200         (neon_vabd<mode>_3): Likewise.
3201         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3202         (thumb2_movsi_vfp): Likewise.
3203         (movdi_vfp): Likewise.
3204         (movdi_vfp_cortexa8): Likewise.
3205         (movhf_vfp_neon): Likewise.
3206         (movhf_vfp): Likewiwse.
3207         (movsf_vfp): Likewiwse.
3208         (thumb2_movsf_vfp): Likewiwse.
3209         (movdf_vfp): Likewise.
3210         (thumb2_movdf_vfp): Likewise.
3211         (movsfcc_vfp): Likewise.
3212         (thumb2_movsfcc_vfp): Likewise.
3213         (movdfcc_vfp): Likewise.
3214         (thumb2_movdfcc_vfp): Likewise.
3215         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
3216         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
3217         (v10_v2c): Likewise.
3218         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
3219         attribute change.
3220         (cortex_a15_neon_int_2): Likewise.
3221         (cortex_a15_neon_int_3): Likewise.
3222         (cortex_a15_neon_int_4): Likewise.
3223         (cortex_a15_neon_int_5): Likewise.
3224         (cortex_a15_neon_vqneg_vqabs): Likewise.
3225         (cortex_a15_neon_vmov): Likewise.
3226         (cortex_a15_neon_vaba): Likewise.
3227         (cortex_a15_neon_vaba_qqq): Likewise.
3228         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3229         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3230         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
3231         scalar_64_32_long_scalar): Likewise.
3232         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3233         (cortex_a15_neon_mla_qqq_8_16): Likewise.
3234         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3235         lotype_qdd_64_32_long): Likewise.
3236         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3237         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3238         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
3239         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3240         (cortex_a15_neon_shift_1): Likewise.
3241         (cortex_a15_neon_shift_2): Likewise.
3242         (cortex_a15_neon_shift_3): Likewise.
3243         (cortex_a15_neon_vshl_ddd): Likewise.
3244         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3245         (cortex_a15_neon_vsra_vrsra): Likewise.
3246         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
3247         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
3248         (cortex_a15_neon_fp_vmul_ddd): Likewise.
3249         (cortex_a15_neon_fp_vmul_qqd): Likewise.
3250         (cortex_a15_neon_fp_vmla_ddd): Likewise.
3251         (cortex_a15_neon_fp_vmla_qqq): Likewise.
3252         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3253         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3254         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3255         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3256         (cortex_a15_neon_bp_simple): Likewise.
3257         (cortex_a15_neon_bp_2cycle): Likewise.
3258         (cortex_a15_neon_bp_3cycle): Likewise.
3259         (cortex_a15_neon_vld1_1_2_regs): Likewise.
3260         (cortex_a15_neon_vld1_3_4_regs): Likewise.
3261         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3262         (cortex_a15_neon_vld2_4_regs): Likewise.
3263         (cortex_a15_neon_vld3_vld4): Likewise.
3264         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3265         (cortex_a15_neon_vst1_3_4_regs): Likewise.
3266         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
3267         (cortex_a15_neon_vst3_vst4): Likewise.
3268         (cortex_a15_neon_vld1_vld2_lane): Likewise.
3269         (cortex_a15_neon_vld3_vld4_lane" 10
3270         (cortex_a15_neon_vst1_vst2_lane): Likewise.
3271         (cortex_a15_neon_vst3_vst4_lane): Likewise.
3272         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3273         (cortex_a15_neon_ldm_2): Likewise.0
3274         (cortex_a15_neon_stm_2): Likewise.
3275         (cortex_a15_neon_mcr): Likewise.
3276         (cortex_a15_neon_mcr_2_mcrr): Likewise.
3277         (cortex_a15_neon_mrc): Likewise.
3278         (cortex_a15_neon_mrrc): Likewise.
3279         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3280         change.
3281         (cortex_a15_alu_shift): Likewise.
3282         (cortex_a15_alu_shift_reg): Likewise.
3283         (cortex_a15_mult32): Likewise.
3284         (cortex_a15_mult64): Likewise.
3285         (cortex_a15_block): Likewise.
3286         (cortex_a15_branch): Likewise.
3287         (cortex_a15_load1): Likewise.
3288         (cortex_a15_load3): Likewise.
3289         (cortex_a15_store1): Likewise.
3290         (cortex_a15_store3): Likewise.
3291         (cortex_a15_call): Likewise.
3292         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
3293         (cortex_a5_f2r): Likewise.
3294         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
3295         change.
3296         (cortex_a53_f2r): Likewise.
3297         * config/arm/cortex-a7.md
3298         (cortex_a7_branch): Update for attribute change.
3299         (cortex_a7_call): Likewise.
3300         (cortex_a7_alu_imm): Likewise.
3301         (cortex_a7_alu_reg): Likewise.
3302         (cortex_a7_alu_shift): Likewise.
3303         (cortex_a7_mul): Likewise.
3304         (cortex_a7_load1): Likewise.
3305         (cortex_a7_store1): Likewise.
3306         (cortex_a7_load2): Likewise.
3307         (cortex_a7_store2): Likewise.
3308         (cortex_a7_load3): Likewise.
3309         (cortex_a7_store3): Likewise.
3310         (cortex_a7_load4): Likewise.
3311         (cortex_a7_store4): Likewise.
3312         (cortex_a7_fpalu): Likewise.
3313         (cortex_a7_fconst): Likewise.
3314         (cortex_a7_fpmuls): Likewise.
3315         (cortex_a7_neon_mul): Likewise.
3316         (cortex_a7_fpmacs): Likewise.
3317         (cortex_a7_neon_mla: Likewise.
3318         (cortex_a7_fpmuld: Likewise.
3319         (cortex_a7_fpmacd: Likewise.
3320         (cortex_a7_fpfmad: Likewise.
3321         (cortex_a7_fdivs: Likewise.
3322         (cortex_a7_fdivd: Likewise.
3323         (cortex_a7_r2f: Likewise.
3324         (cortex_a7_f2r: Likewise.
3325         (cortex_a7_f_flags: Likewise.
3326         (cortex_a7_f_loads: Likewise.
3327         (cortex_a7_f_loadd: Likewise.
3328         (cortex_a7_f_stores: Likewise.
3329         (cortex_a7_f_stored: Likewise.
3330         (cortex_a7_neon): Likewise.
3331         * config/arm/cortex-a8-neon.md
3332         (cortex_a8_neon_mrc): Update for attribute change.
3333         (cortex_a8_neon_mrrc): Likewise.
3334         (cortex_a8_neon_int_1): Likewise.
3335         (cortex_a8_neon_int_2): Likewise.
3336         (cortex_a8_neon_int_3): Likewise.
3337         (cortex_a8_neon_int_4): Likewise.
3338         (cortex_a8_neon_int_5): Likewise.
3339         (cortex_a8_neon_vqneg_vqabs): Likewise.
3340         (cortex_a8_neon_vmov): Likewise.
3341         (cortex_a8_neon_vaba): Likewise.
3342         (cortex_a8_neon_vaba_qqq): Likewise.
3343         (cortex_a8_neon_vsma): Likewise.
3344         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3345         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3346         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
3347         long_scalar): Likewise.
3348         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3349         (cortex_a8_neon_mla_qqq_8_16): Likewise.
3350         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3351         long_scalar_qdd_64_32_long): Likewise.
3352         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3353         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3354         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
3355         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3356         (cortex_a8_neon_shift_1): Likewise.
3357         (cortex_a8_neon_shift_2): Likewise.
3358         (cortex_a8_neon_shift_3): Likewise.
3359         (cortex_a8_neon_vshl_ddd): Likewise.
3360         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3361         (cortex_a8_neon_vsra_vrsra): Likewise.
3362         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
3363         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
3364         (cortex_a8_neon_fp_vsum): Likewise.
3365         (cortex_a8_neon_fp_vmul_ddd): Likewise.
3366         (cortex_a8_neon_fp_vmul_qqd): Likewise.
3367         (cortex_a8_neon_fp_vmla_ddd): Likewise.
3368         (cortex_a8_neon_fp_vmla_qqq): Likewise.
3369         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
3370         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
3371         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3372         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3373         (cortex_a8_neon_bp_simple): Likewise.
3374         (cortex_a8_neon_bp_2cycle): Likewise.
3375         (cortex_a8_neon_bp_3cycle): Likewise.
3376         (cortex_a8_neon_ldr): Likewise.
3377         (cortex_a8_neon_str): Likewise.
3378         (cortex_a8_neon_vld1_1_2_regs): Likewise.
3379         (cortex_a8_neon_vld1_3_4_regs): Likewise.
3380         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3381         (cortex_a8_neon_vld2_4_regs): Likewise.
3382         (cortex_a8_neon_vld3_vld4): Likewise.
3383         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3384         (cortex_a8_neon_vst1_3_4_regs): Likewise.
3385         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
3386         (cortex_a8_neon_vst3_vst4): Likewise.
3387         (cortex_a8_neon_vld1_vld2_lane): Likewise.
3388         (cortex_a8_neon_vld3_vld4_lane): Likewise.
3389         (cortex_a8_neon_vst1_vst2_lane): Likewise.
3390         (cortex_a8_neon_vst3_vst4_lane): Likewise.
3391         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
3392         (cortex_a8_neon_mcr): Likewise.
3393         (cortex_a8_neon_mcr_2_mcrr): Likewise.
3394         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3395         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
3396         change.
3397         (ca9_neon_mrrc): Likewise.
3398         (cortex_a9_neon_int_1): Likewise.
3399         (cortex_a9_neon_int_2): Likewise.
3400         (cortex_a9_neon_int_3): Likewise.
3401         (cortex_a9_neon_int_4): Likewise.
3402         (cortex_a9_neon_int_5): Likewise.
3403         (cortex_a9_neon_vqneg_vqabs): Likewise.
3404         (cortex_a9_neon_vmov): Likewise.
3405         (cortex_a9_neon_vaba): Likewise.
3406         (cortex_a9_neon_vaba_qqq): Likewise.
3407         (cortex_a9_neon_vsma): Likewise.
3408         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3409         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3410         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
3411         long_scalar): Likewise.
3412         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3413         (cortex_a9_neon_mla_qqq_8_16): Likewise.
3414         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
3415         long_scalar_qdd_64_32_long): Likewise.
3416         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3417         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3418         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3419         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3420         (cortex_a9_neon_shift_1): Likewise.
3421         (cortex_a9_neon_shift_2): Likewise.
3422         (cortex_a9_neon_shift_3): Likewise.
3423         (cortex_a9_neon_vshl_ddd): Likewise.
3424         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3425         (cortex_a9_neon_vsra_vrsra): Likewise.
3426         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3427         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3428         (cortex_a9_neon_fp_vsum): Likewise.
3429         (cortex_a9_neon_fp_vmul_ddd): Likewise.
3430         (cortex_a9_neon_fp_vmul_qqd): Likewise.
3431         (cortex_a9_neon_fp_vmla_ddd): Likewise.
3432         (cortex_a9_neon_fp_vmla_qqq): Likewise.
3433         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3434         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3435         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3436         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3437         (cortex_a9_neon_bp_simple): Likewise.
3438         (cortex_a9_neon_bp_2cycle): Likewise.
3439         (cortex_a9_neon_bp_3cycle): Likewise.
3440         (cortex_a9_neon_ldr): Likewise.
3441         (cortex_a9_neon_str): Likewise.
3442         (cortex_a9_neon_vld1_1_2_regs): Likewise.
3443         (cortex_a9_neon_vld1_3_4_regs): Likewise.
3444         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3445         (cortex_a9_neon_vld2_4_regs): Likewise.
3446         (cortex_a9_neon_vld3_vld4): Likewise.
3447         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3448         (cortex_a9_neon_vst1_3_4_regs): Likewise.
3449         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3450         (cortex_a9_neon_vst3_vst4): Likewise.
3451         (cortex_a9_neon_vld1_vld2_lane): Likewise.
3452         (cortex_a9_neon_vld3_vld4_lane): Likewise.
3453         (cortex_a9_neon_vst1_vst2_lane): Likewise.
3454         (cortex_a9_neon_vst3_vst4_lane): Likewise.
3455         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3456         (cortex_a9_neon_mcr): Likewise.
3457         (cortex_a9_neon_mcr_2_mcrr): Likewise.
3458         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3459         (cortex_a9_fps): Likewise.
3460         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
3461         change.
3462         (cortex_m4_fmuls): Likewise.
3463         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
3464         change.
3465         (cortex_r4_mrc): Likewise.
3466         * config/arm/iterators.md: Update comment referring to neon_type.
3467         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
3468         (iwmmxt_movsi_insn): Likewise.
3469         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
3470         attribute change.
3471         (pj4_core_to_vfp): Likewise.
3472         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
3473         attribute change.
3474         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
3475         (vfp_fstore): Likewise.
3476         * doc/md.texi: Change references to neon_type to refer to type.
3478 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
3480         * tree.h (DECL_BUILT_IN): Fix typo in comment.
3482 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
3484         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
3485         lglobl if not weak.
3487 2013-09-04  Easwaran Raman  <eraman@google.com>
3489         PR middle-end/57370
3490         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
3491         (build_and_add_sum): Use it.
3492         (appears_later_in_bb): Simplify code.
3494 2013-09-04  Teresa Johnson  <tejohnson@google.com>
3496         * dumpfile.c (dump_finish): Don't close stderr/stdout.
3498 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
3500         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
3502 2013-09-04  Jan Hubicka  <jh@suse.cz>
3504         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
3505         * ipa-devirt.c: Include diganostic.h
3506         (odr_type_d): Add types and types_set.
3507         (hash_type_name): Work for types with vtables during LTO.
3508         (odr_hasher::remove): Fix comment; destroy types_set.
3509         (add_type_duplicate): New function,
3510         (get_odr_type): Use it.
3511         (dump_type_inheritance_graph): Dump type duplicates.
3512         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
3513         graph.
3514         * tree.c (types_same_for_odr): Give exact answers on types with
3515         virtual tables.
3517 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
3519         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
3520         explaining their differences.
3522 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
3524         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
3526 2013-09-03  Jeff Law  <law@redhat.com>
3528         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
3529         when not threading through a joiner block.  Pass joiner/no joiner
3530         state to register_jump_thread.
3531         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
3532         state from argument rather than implying on path length.
3533         Dump the entire jump thread path into debugging dump.
3534         * tree-flow.h (register_jump_thread): Update prototype.
3536 2013-08-29  Xinliang David Li  <davidxl@google.com>
3538         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
3539         Remove a trivial gcc_assert.
3541 2013-08-29  Xinliang David Li  <davidxl@google.com>
3543         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
3544         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
3545         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
3546         Delay base decl alignment adjustment.
3547         * tree-vectorizer.c (vect_destroy_datarefs): New function.
3548         * tree-vectorizer.h: New data structure.
3549         (set_dr_misalignment): New function.
3550         (dr_misalignment): Ditto.
3551         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
3552         (vectorizable_load): Ditto.
3553         (ensure_base_align): New function.
3554         (vectorize_loops): Add dbg_cnt support.
3555         (execute_vect_slp): Ditto.
3556         * dbgcnt.def: New debug counter.
3557         * Makefile: New dependency.
3559 2013-09-03  Meador Inge  <meadori@codesourcery.com>
3561         Revert:
3563         2013-08-30  Meador Inge  <meadori@codesourcery.com>
3565         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3567 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
3569         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
3570         function descriptor.
3572 2013-09-03  Richard Biener  <rguenther@suse.de>
3574         * tree-affine.c (add_elt_to_tree): Fix association issue,
3575         avoid useless converts and make sure to always return a
3576         properly typed result.
3578 2013-09-03  Richard Biener  <rguenther@suse.de>
3580         PR middle-end/57656
3581         * fold-const.c (negate_expr_p): Fix division case.
3582         (negate_expr): Likewise.
3584 2013-09-03  Richard Biener  <rguenther@suse.de>
3586         PR lto/58285
3587         * tree-streamer-out.c: Include tm.h.
3588         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
3590 2013-09-03  Jan Hubicka  <jh@suse.cz>
3592         * tree-profile.c (tree_profiling): Cleanup CFG when done.
3594 2013-09-03  Alan Modra  <amodra@gmail.com>
3596         * config.gcc (powerpc*-*-linux*): Add support for little-endian
3597         multilibs to big-endian target and vice versa.
3598         * config/rs6000/t-linux64: Use := assignment on all vars.
3599         (MULTILIB_EXTRA_OPTS): Remove fPIC.
3600         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
3601         * config/rs6000/t-linux64le: New file.
3602         * config/rs6000/t-linux64bele: New file.
3603         * config/rs6000/t-linux64lebe: New file.
3605 2013-09-02  Jan Hubicka  <jh@suse.cz>
3607         * ipa-inline-transform.c (inline_transform): Do not
3608         optimize_inline_calls when not optimizing.
3610 2013-09-02  Jan Hubicka  <jh@suse.cz>
3612         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
3613         duplicated nodes for assembler names.
3614         * symtab.c (symtab_unregister_node): Do not attempt to unlink
3615         hard registers from assembler name hash.
3617 2013-09-02  Jan Hubicka  <jh@suse.cz>
3619         * ipa-split.c (execute_split_functions): Split externally visible
3620         functions called once.
3622 2013-09-02  Martin Jambor  <mjambor@suse.cz>
3624         PR ipa/58106
3625         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
3626         linked list.  When finding the correct duplicate, also consider also
3627         the caller in additon to its inlined_to node.
3629 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
3631         * config/aarch64/aarch64-simd-builtins.def
3632         (dup_lane_scalar): Remove.
3633         * config/aarch64/aarch64-simd.md
3634         (aarch64_simd_dup): Add 'w->w' alternative.
3635         (aarch64_dup_lane<mode>): Allow for VALL.
3636         (aarch64_dup_lane_scalar<mode>): Remove.
3637         (aarch64_dup_lane_<vswap_width_name><mode>): New.
3638         (aarch64_get_lane_signed<mode>): Add w->w altenative.
3639         (aarch64_get_lane_unsigned<mode>): Likewise.
3640         (aarch64_get_lane<mode>): Likewise.
3641         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
3642         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
3643         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
3644         (VCON): Change container of V2SF.
3645         (vswap_width_name): Likewise.
3646         * config/aarch64/arm_neon.h
3647         (__aarch64_vdup_lane_any): New.
3648         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
3649         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
3650         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
3652 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
3654         PR middle-end/56382
3655         * expr.c (emit_move_complex): Do not move complex FP values as parts if
3656         the source or the destination is a single hard register.
3658 2013-09-02  Richard Biener  <rguenther@suse.de>
3660         PR middle-end/57511
3661         * tree-scalar-evolution.c (instantiate_scev_name): Allow
3662         non-linear SCEVs.
3664 2013-09-02  Richard Biener  <rguenther@suse.de>
3666         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
3667         arithmetic to sizetype.
3669 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
3671         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
3672         Find auto-increment use both before and after candidate.
3674 2013-09-02  Marek Polacek  <polacek@redhat.com>
3676         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
3678 2013-09-01  Jan Hubicka  <jh@suse.cz>
3680         * Makefile.in: Add ipa-profile.o
3681         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
3682         * cgraph.c (struct cgraph_propagate_frequency_data,
3683         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
3684         ipa-profile.c; replace cgraph_ by ipa_ prefix.
3685         * cgraph.h (cgraph_propagate_frequency): Remove.
3686         * ipa-inline-analysis.c: Include ipa-utils.h;
3687         drop duplicated cfgloop.h.
3688         (inline_update_callee_summaries): Update.
3689         * ipa-profile.c: New file.
3690         * ipa-utils.h (ipa_propagate_frequency): Declare.
3691         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
3692         data-streamer.h, value-prof.h.
3693         (symtab_remove_unreachable_nodes): Update profile.
3694         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
3695         account_time_size, cmp_counts, dump_histogram,
3696         ipa_profile_generate_summary, ipa_profile_write_summary,
3697         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
3698         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
3699         Move to ipa-profile.c.
3701 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
3703         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
3705 2013-09-01  Jan Hubicka  <jh@suse.cz>
3707         * common.opt (fdevirtualize-speculatively): New function.
3708         * invoke.texi (fdevirtualize-speculatively): Document.
3709         * ipa-devirt.c: Include ipa-inline.h
3710         (likely_target_p): New function.
3711         (ipa_devirt): New function.
3712         (gate_ipa_devirt): New function.
3713         (pass_data_ipa_devirt): New static var.
3714         (pass_ipa_devirt): Likewise.
3715         (make_pass_ipa_devirt): New function.
3716         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
3717         (common_handle_option): Disable devirtualization when
3718         value range profiling is available.
3719         * passes.def (pass_ipa_devirt): Add.
3720         * timever.def (TV_IPA_DEVIRT): New timevar.
3721         * tree-pass.h (make_pass_ipa_devirt):
3723 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
3725         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
3726         include sanitize(undefined).
3728 2013-08-31  Diego Novillo  <dnovillo@google.com>
3730         * Makefile.in (TREE_CORE_H): Define.
3731         (TREE_H): Use.
3732         (GTFILES): Add tree-core.h.
3733         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
3734         size the array.
3735         * tree-core.h: New file.
3736         Move all data structures, enum, typedefs, global
3737         declarations and constants from ...
3738         * tree.h: ... here.
3740 2013-08-31  Jan Hubicka  <jh@suse.cz>
3742         * bulitins.c (expand_builtin): Do not early exit for gcov
3743         instrumented functions.
3745 2013-08-31  Marek Polacek  <polacek@redhat.com>
3747         * ubsan.c: Include tm_p.h.
3749 2013-08-31  Jan Hubicka  <jh@suse.cz>
3751         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
3752         warning.
3754         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
3755         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
3756         * tree-cfg.c (verify_gimple_label): ... here.
3757         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
3758         (ipa_merge_profiles): New function.
3759         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
3760         (lto_input_function_body): Likewise.
3761         * ipa-utils.h (ipa_merge_profiles): Declare.
3762         * lto-streamer.h (lto_input_function_body): Update prototype.
3763         (emit_label_in_global_context_p): Remove.
3764         * lto-symtab.c: Include ipa-utils.h
3765         (lto_cgraph_replace_node): Use ipa_merge_profiles.
3767 2013-08-31  Jan Hubicka  <jh@suse.cz>
3769         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
3771 2013-08-31  Jan Hubicka  <jh@suse.cz>
3773         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
3775 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
3777         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
3778         "cmp" RTX before signed_comparison_operator check to account
3779         for "code" changes.
3781 2013-08-30  Jan Hubicka  <jh@suse.cz>
3783         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
3784         (detect_type_change_1): Rename to ...
3785         (detect_type_change): ... this one; early return on non-polymorphic
3786         types.
3787         (detect_type_change_ssa): Add comp_type parameter; update
3788         use of detect_type_change.
3789         (compute_complex_assign_jump_func): Add param_type parameter;
3790         update use of detect_type_change_ssa.
3791         (compute_complex_ancestor_jump_func): Likewise.
3792         (ipa_get_callee_param_type): New function.
3793         (ipa_compute_jump_functions_for_edge): Compute parameter type;
3794         update calls to the jump function computation functions.
3796 2013-08-30  Teresa Johnson  <tejohnson@google.com>
3797             Steven Bosscher  <steven@gcc.gnu.org>
3799         * cfgrtl.c (fixup_new_cold_bb): New routine.
3800         (commit_edge_insertions): Invoke fixup_partitions.
3801         (find_partition_fixes): New routine.
3802         (fixup_partitions): Ditto.
3803         (verify_hot_cold_block_grouping): Update comments.
3804         (rtl_verify_edges): Invoke find_partition_fixes.
3805         (rtl_verify_bb_pointers): Update comments.
3806         (rtl_verify_bb_layout): Ditto.
3807         * basic-block.h (probably_never_executed_edge_p): Declare.
3808         (fixup_partitions): Ditto.
3809         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
3810         * bb-reorder.c (sanitize_hot_paths): New function.
3811         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
3812         sanitize_hot_paths.
3813         * predict.c (probably_never_executed_edge_p): New routine.
3814         * cfg.c (check_bb_profile): Add partition insanity warnings.
3816 2013-08-30  Meador Inge  <meadori@codesourcery.com>
3818         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3820 2013-08-30  Marek Polacek  <polacek@redhat.com>
3822         * Makefile.in (ubsan.o): Add.
3823         (c-family/c-ubsan.o): Add.
3824         (builtins.o): Add ubsan.h dependency.
3825         * ubsan.h: New file.
3826         * ubsan.c: New file.
3827         * common.opt: Add -fsanitize=undefined option.
3828         (flag_sanitize): Add variable.
3829         (fsanitize=): Add option.  Add Driver.
3830         (fsanitize=thread): Remove option.
3831         (fsanitize=address): Likewise.
3832         (static-libubsan): New option.
3833         * doc/invoke.texi: Document the new flag and -static-libubsan.
3834         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
3835         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
3836         * builtin-attrs.def (ATTR_COLD): Define.
3837         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3838         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
3839         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
3840         * flag-types.h (sanitize_code): New enum.
3841         * opts.c (common_handle_option): Parse command line arguments
3842         of -fsanitize=.  Add -fsanitize=unreachable option.
3843         * varasm.c (get_variable_section): Adjust.
3844         (assemble_noswitch_variable): Likewise.
3845         (assemble_variable): Likewise.
3846         (output_constant_def_contents): Likewise.
3847         (categorize_decl_for_section): Likewise.
3848         (place_block_symbol): Likewise.
3849         (output_object_block): Likewise.
3850         * builtins.def: Likewise.
3851         * toplev.c (compile_file): Likewise.
3852         (process_options): Likewise.
3853         * cppbuiltin.c: Likewise.
3854         * tsan.c (tsan_pass): Likewise.
3855         (tsan_gate): Likewise.
3856         (tsan_gate_O0): Likewise.
3857         * cfgexpand.c (partition_stack_vars): Likewise.
3858         (expand_stack_vars): Likewise.
3859         (defer_stack_allocation): Likewise.
3860         (expand_used_vars): Likewise.
3861         * cfgcleanup.c (old_insns_match_p): Likewise.
3862         * asan.c (asan_finish_file): Likewise.
3863         (asan_instrument): Likewise.
3864         (gate_asan): Likewise.
3865         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
3866         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3867         (asan_global_struct): Use pointer_sized_int_node instead
3868         calling build_nonstandard_integer_type.
3869         (initialize_sanitizer_builtins): Likewise.
3870         (asan_finish_file): Likewise.
3871         * gcc.c: Document %{%:function(args):X}.
3872         (static_spec_functions): Add sanitize.
3873         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
3874         store funcval != NULL there.
3875         (do_spec_1): Adjust handle_spec_function caller.
3876         (handle_braces): Allow %:function(args) as condition.
3877         (sanitize_spec_function): New function.
3878         (ADD_STATIC_LIBUBSAN_LIBS): Define.
3879         (LIBUBSAN_SPEC): Likewise.
3880         (LIBUBSAN_EARLY_SPEC): Likewise.
3881         (SANITIZER_SPEC): Handle libubsan.
3882         (SANITIZER_EARLY_SPEC): Likewise.
3883         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
3884         instead of fsanitize=address.
3885         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
3886         instead of fsanitize=address*.
3887         * builtins.c: Include ubsan.h.
3888         (fold_builtin_0): Instrument __builtin_unreachable.
3889         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
3890         instead of flag_asan.
3891         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
3892         (pointer_sized_int_node): Define.
3893         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
3895 2013-08-30  Mike Stump  <mikestump@comcast.net>
3897         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
3898         with RE patterns.
3900 2013-08-29  Jan Hubicka  <jh@suse.cz>
3902         * cgraph.c (cgraph_function_body_availability): Handle weakref
3903         correctly.
3904         * passes.def: Remove pass_fixup_cfg.
3905         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
3906         track when we need to remove functions.
3907         (gate_ipa_inline): Execute inlining always; add comment why.
3908         (pass_data_ipa_inline): Remove TODO_remove_functions.
3909         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
3910         do not produce summaries.
3911         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
3912         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
3913         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
3914         constant pool and vtable.
3916 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
3918         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
3919         New arm_neon.h's internal macros to specify 64-bit constants.
3920         Avoid using stdint.h's macros.
3922 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
3924         * recog.c (verify_changes): Verify that changes[i].old is non-zero
3925         before applying REG_P.
3927 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
3929         PR tree-optimization/58277
3930         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
3931         after seeing too many stmts with vdef in between dombb and current
3932         bb, invalidate everything.
3934 2013-08-30  Richard Biener  <rguenther@suse.de>
3936         * fold-const.c (fold_single_bit_test): Fix overflow test.
3938 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
3940         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
3941         and which can live in a register, always retrieve the value on entry.
3942         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
3943         passed by invisible reference specially.
3944         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
3945         (vt_add_function_parameter): Correctly deal with a parameter passed by
3946         invisible reference.
3948 2013-08-30  Jan Hubicka  <jh@suse.cz>
3950         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
3952 2013-08-30  Richard Biener  <rguenther@suse.de>
3954         PR tree-optimization/58228
3955         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
3956         allow invariant loads in nested loop vectorization.
3958 2013-08-30  Richard Biener  <rguenther@suse.de>
3960         PR tree-optimization/58223
3961         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
3962         (has_anti_or_output_dependence): ... this and adjust to also
3963         look for output dependences.
3964         (mark_nodes_having_upstream_mem_writes): Adjust.
3965         (rdg_flag_uses): Likewise.
3967 2013-08-30  Richard Biener  <rguenther@suse.de>
3969         PR tree-optimization/58010
3970         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
3971         assert that we have a loop-closed PHI.
3973 2013-08-29  Jan Hubicka  <jh@suse.cz>
3975         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
3976         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
3977         * lto-section-in.c (lto_free_function_in_decl_state): New function.
3978         (lto_free_function_in_decl_state_for_node): New function.
3980 2013-08-29  Xinliang David Li  <davidxl@google.com>
3982         * loop-unroll.c (report_unroll_peel): Minor message change.
3983         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
3984         Emit alignment peeling message with default -fopt-info.
3985         (vect_loop_versioning): Emit loop version info message.
3986         * tree-vectorizer.c (vectorize_loops): Minor message change.
3987         (execute_vect_slp): Ditto.
3989 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
3991         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
3992         of the clone from the DECL_NAME of the original function.
3994 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
3996         * passes.c (register_pass): Add overload.
3997         * tree-pass.h (register_pass): Forward declare it.  Add comment.
3999 2013-08-29  Jan Hubicka  <jh@suse.cz>
4001         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
4002         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
4003         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
4004         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
4005         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
4006         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
4007         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
4008         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
4010 2013-08-29  Teresa Johnson  <tejohnson@google.com>
4012         * dumpfile.c (dump_loc): Output column number.
4013         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
4014         * doc/invoke.texi: Document optall -fopt-info flag.
4015         * profile.c (read_profile_edge_counts): Use new dump framework.
4016         (compute_branch_probabilities): Ditto.
4017         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
4018         when pass not in any opt group.
4019         * pass_manager.h (pass_manager::get_pass_profile): New method.
4020         * value-prof.c (check_counter): Use new dump framework.
4021         (check_ic_target): Ditto.
4022         * coverage.c (get_coverage_counts): Ditto.
4023         (coverage_init): Setup new dump framework.
4025 2013-08-29  Richard Biener  <rguenther@suse.de>
4027         PR tree-optimization/58246
4028         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
4029         handle the dominance check inside a basic-block.
4031 2013-08-29  Richard Biener  <rguenther@suse.de>
4033         PR middle-end/57287
4034         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
4035         of default defs that appear in abnormal PHI nodes.
4037 2013-08-29  Richard Biener  <rguenther@suse.de>
4039         PR tree-optimization/57685
4040         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
4041         single-use operands to avoid exponential complexity.
4043 2013-08-28  Dehao Chen  <dehao@google.com>
4045         * ipa-inline.c (edge_badness): Fix integer underflow.
4047 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
4049         * gtm-builtins.def (_ITM_free): Declare leaf.
4051 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
4053         PR target/58067
4054         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
4055         (*tls_local_dynamic_base_64_largepic): Likewise.
4056         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
4057         Remove predicate from call operand.
4058         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
4059         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
4061 2013-08-28  Jeff Law  <law@redhat.com>
4063         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
4064         checks for the number of predecessors and successors allowed.
4065         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
4066         which require copying a joiner block if there is a request which
4067         is a subpath that requires no joiner block copying.
4069 2013-08-28  Jan Hubicka  <jh@suse.cz>
4071         * lto-streamer-out.c (DFS_write_tree_body): Drop
4072         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
4073         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
4074         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
4075         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
4076         Do not read DECL_ERROR_ISSUED.
4077         (unpack_ts_decl_with_vis_value_fields): Do not read
4078         DECL_DEFER_OUTPUT.
4079         (lto_input_ts_binfo_tree_pointers): Do not read
4080         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
4081         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
4082         write DECL_ERROR_ISSUED..
4083         (pack_ts_decl_with_vis_value_fields): Do not write
4084         DECL_DEFER_OUTPUT.
4085         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
4086         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
4087         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
4088         * tree.h (tree_decl_common): Update comment.
4089         (DECL_ERROR_ISSUED): Remove.
4091 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
4093         PR middle-end/58257
4094         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
4096 2013-08-28  Jan Hubicka  <jh@suse.cz>
4098         * builtins.def (free): Declare leaf.
4100 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
4102         * gdbhooks.py: New.
4103         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
4104         * configure: Regenerate.
4106 2013-08-27  Martin Jambor  <mjambor@suse.cz>
4108         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
4109         (ipa_ancestor_jf_data): Likewise.
4110         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
4111         (ipa_get_jf_pass_through_type_preserved): New function.
4112         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
4113         (ipa_get_jf_ancestor_type_preserved): New function.
4114         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
4115         (ipa_get_jf_ancestor_result): Likewise.
4116         (propagate_vals_accross_pass_through): Use
4117         ipa_get_jf_pass_through_result to do all the value mappings.
4118         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
4119         type_preserved flag.
4120         (ipa_set_jf_cst_copy): New function.
4121         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
4122         (ipa_set_jf_arith_pass_through): Likewise.
4123         (ipa_set_ancestor_jf): Likewise.
4124         (compute_complex_assign_jump_func): Set type_preserved instead of
4125         punting.
4126         (ipa_compute_jump_functions_for_edge): Likewise.
4127         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
4128         (update_jump_functions_after_inlining): Update type_preserved.
4129         Explicitely create jump functions when combining one with pass_through.
4130         (ipa_write_jump_function): Stream the type_preserved flags.
4131         (ipa_read_jump_function): Likewise.
4133 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
4134             Aldy Hernandez  <aldyh@redhat.com>
4136         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
4137         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
4138         * function.h (struct function): Add has_force_vect_loops and
4139         has_simduid_loops.
4140         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
4141         * gimple.c (gimple_build_omp_critical): Add KIND argument and
4142         handle it.
4143         * gimple.def: Update CLAUSES comments.
4144         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
4145         (gimple_build_omp_for): Add argument to prototype.
4146         (gimple_omp_for_kind): New.
4147         (gimple_omp_for_set_kind): New.
4148         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
4149         GOVD_DATA_SHARE_CLASS.
4150         (enum omp_region_type): Add ORT_SIMD.
4151         (gimple_add_tmp_var): Handle ORT_SIMD.
4152         (gimplify_var_or_parm_decl): Same.
4153         (is_gimple_stmt): Same.
4154         (omp_firstprivatize_variable): Same.
4155         (omp_add_variable): Only use splay_tree_insert if lookup failed.
4156         (omp_notice_variable): Handle ORT_SIMD.
4157         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
4158         (omp_check_private): Handle ORT_SIMD.
4159         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
4160         OMP_CLAUSE_SAFELEN.
4161         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
4162         Handle OMP_CLAUSE_LASTPRIVATE.
4163         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
4164         OMP_CLAUSE_SAFELEN.
4165         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
4166         (gimplify_expr): Handle OMP_SIMD.
4167         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
4168         (expand_GOMP_SIMD_VF): New.
4169         (expand_GOMP_SIMD_LAST_LANE): New.
4170         * internal-fn.def (GOMP_SIMD_LANE): New.
4171         (GOMP_SIMD_VF): New.
4172         (GOMP_SIMD_LAST_LANE): New.
4173         * omp-low.c: Include target.h.
4174         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
4175         OMP_CLAUSE_SAFELEN.
4176         (check_omp_nesting_restrictions): Same.
4177         (omp_max_vf): New.
4178         (lower_rec_simd_input_clauses): New.
4179         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
4180         OMP_CLAUSE_LINEAR.
4181         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
4182         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
4183         (expand_omp_build_assign): New.
4184         (expand_omp_for_init_counts): New.
4185         (expand_omp_for_init_vars): New.
4186         (extract_omp_for_update_vars): New.
4187         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
4188         and rewrite accordingly.
4189         (expand_omp_simd): New.
4190         (expand_omp_for): Use expand_omp_simd.
4191         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
4192         (lower_omp_for): Do not lower the body.
4193         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
4194         in their own loops.
4195         * tree-flow.h (find_omp_clause): Remove prototype.
4196         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
4197         forcing vectorization of the loop, or if flag_tree_vectorize.
4198         (gate_tree_if_conversion): Similarly.
4199         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
4200         gimple_build_omp_for.
4201         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
4202         * tree-parloops (create_parallel_loop): Pass kind argument to
4203         gimple_build_omp_for.
4204         * tree-pretty-print.c (dump_omp_clause): Add cases for
4205         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
4206         OMP_CLAUSE__SIMDUID_.
4207         (dump_generic_node): Handle OMP_SIMD.
4208         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
4209         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
4210         unroll OMP_SIMD loops here.
4211         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
4212         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
4213         loop->safelen.
4214         (vect_analyze_data_refs): Handle simd loops.
4215         * tree-vect-loop.c (vectorizable_live_operation): Handle
4216         IFN_GOMP_SIMD*.
4217         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
4218         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
4219         (vectorizable_load): Same.
4220         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
4221         (struct simduid_to_vf): New.
4222         (simduid_to_vf::hash): New.
4223         (simduid_to-vf::equal): New.
4224         (struct simd_array_to_simduid): New.
4225         (simd_array_to_simduid::hash): New.
4226         (simd_array_to_simduid::equal): New.
4227         (adjust_simduid_builtins): New.
4228         (struct note_simd_array_uses_struct): New.
4229         (note_simd_array_uses_cb): New.
4230         (note_simd_array_uses): New.
4231         (vectorize_loops): Handle simd hints and adjust simd builtins
4232         accordingly.
4233         * tree-vectorizer.h (struct _stmt_vec_info): Add
4234         simd_lane_access_p field.
4235         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
4236         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
4237         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
4238         (omp_clause_code_name): Same.
4239         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
4240         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
4241         * tree.def (OMP_SIMD): New entry.
4242         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
4243         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
4244         (OMP_CLAUSE_DECL): Adjust range for new clauses.
4245         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
4246         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
4247         (OMP_CLAUSE_LINEAR_STEP): New.
4248         (OMP_CLAUSE_SAFELEN_EXPR): New.
4249         (OMP_CLAUSE__SIMDUID__DECL): New.
4250         (find_omp_clause): New prototype.
4252 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
4254         * config/i386/driver-i386.c (host_detect_local_cpu): Update
4255         Haswell processor detection.
4257 2013-08-27  Christian Widmer  <shadow@umbrox.de>
4259         PR target/57927
4260         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
4261         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
4262         AVX2 capable processors.
4264 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
4266         * config/aarch64/arm_neon.h: Replace all inline asm implementations
4267         of vget_low_* with implementations in terms of other intrinsics.
4269 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
4271         PR middle-end/57219
4272         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
4273         values to -1, 0 and 1.
4275 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
4277         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
4278         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
4279         (<optab><mode>3_insn): Remove.
4280         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
4281         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
4282         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
4283         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
4284         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
4285         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
4286         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
4287         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
4288         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
4289         (ror<mode>3_insn): Likewise.
4290         * config/aarch64/predicates.md (aarch64_simd_register): New.
4292 2013-08-27  Richard Biener  <rguenther@suse.de>
4294         PR tree-optimization/57521
4295         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
4296         one edge is non-critical.
4297         (find_phi_replacement_condition): Make sure to use a non-critical
4298         edge.  Cleanup and remove old bug workarounds.
4299         (bb_postdominates_preds): Remove.
4300         (if_convertible_loop_p_1): Do not compute post-dominators.
4301         (combine_blocks): Do not free post-dominators.
4302         (main_tree_if_conversion): Likewise.
4303         (pass_data_if_conversion): Add TODO_verify_ssa.
4305 2013-08-27  DJ Delorie  <dj@redhat.com>
4307         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
4309 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4311         * function.c (assign_parm_find_data_types): Set passed_mode and
4312         nominal_mode to the TYPE_MODE of nominal_type for the built
4313         pointer type in case of the struct-pass-by-reference.
4315 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
4317         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
4318         (UINT16_TYPE): Change default to "unsigned int".
4320         * config/avr/avr.opt (mfract-convert-truncate): New option.
4321         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
4322         is set, round negative fractional integers according to n1169
4323         when converting to integer types.
4325 2013-08-26  Jan Hubicka  <jh@suse.cz>
4327         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
4328         methods can not be called indirectly when their address is not taken.
4330 2013-08-26  Jan Hubicka  <jh@suse.cz>
4332         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
4333         ctor_for_folding.
4335 2013-08-26  Jan Hubicka  <jh@suse.cz>
4337         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
4338         can be unshared.
4340 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
4342         * reload.c (find_valid_class): Allow classes that do not include
4343         FIRST_PSEUDO_REGISTER - 1.
4345 2013-08-26  Jan Hubicka  <jh@suse.cz>
4347         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
4348         fix edge count/frequency when speculation failed; fix type check
4349         for the direct call.
4351 2013-08-26  Jan Hubicka  <jh@suse.cz>
4353         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
4355 2013-08-26  Jan Hubicka  <jh@suse.cz>
4357         * ipa-inline-transform.c (inline_transform): Be ready for basic block
4358         to be changed by edge redirection.
4360 2013-08-26  Jan Hubicka  <jh@suse.cz>
4362         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
4363         formating; add sanity check.
4364         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
4365         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
4366         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
4367         (resolve_noninline_speculation): Update callee keys, too.
4369 2013-08-26  Jan Hubicka  <jh@suse.cz>
4371         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
4372         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
4374 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
4376         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
4377         into proper place.
4379 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
4381         * config/i386/i386.c (ix86_debug_options): Remove prototype.
4382         (x86_64_elf_select_section): Ditto.
4383         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
4384         arguments.
4385         (ix86_pass_by_reference): Ditto.
4386         (output_set_got): Ditto.
4387         (ix86_unary_operator_ok): Ditto.
4388         (ix86_expand_builtin): Ditto.
4390 2013-08-23  Jan Hubicka  <jh@suse.cz>
4392         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
4394 2013-08-23  Jan Hubicka  <jh@suse.cz>
4396         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
4397         (tree_decl_with_vis): Add FINAL field.
4399 2013-08-23  Jeff Law  <law@redhat.com>
4401         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
4402         the debugging dump when the expression is fully redundant.
4404 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4406         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
4407         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
4408         * pretty-print.c (pp_formatted_text_data): Likewise.
4409         (pp_write_text_to_stream): Likewise.
4410         (pp_write_text_as_dot_label_to_stream): Likewise.
4411         (pp_append_r): Likewise.
4412         (pp_format): Likewise.
4413         (pp_flush): Likewise.
4414         (pp_clear_output_area): Likewise.
4415         (pp_append_text): Likewise.
4416         (pp_formatted_text): Likewise.
4417         (pp_remaining_character_count_for_line): Likewise.
4418         (pp_newline): Likewise.
4419         (pp_character): Likewise.
4420         (output_buffer::~output_buffer): Define.
4421         (pretty_printer::~pretty_printer): Destruct output buffer.
4422         * pretty-print.h (output_buffer::~output_buffer): Declare.
4423         (pretty_printer::~pretty_printer): Declare virtual.
4425 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
4427         PR other/57324
4428         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
4429         HOST_WIDE_INT_M1U): New macros.
4430         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
4431         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
4432         unsigned -1 for lshift.
4433         * cse.c (cse_insn): Likewise.
4434         * double-int.c (rshift_double, lshift_double): Likewise.
4435         * builtins.c (fold_builtin_bitop): Likewise.
4436         * combine.c (force_to_mode): Likewise.
4437         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
4438         * simplify-rtx.c (simplify_const_unary_operation,
4439         simplify_const_binary_operation): Likewise.
4440         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
4441         check_va_list_escapes): Likewise.
4442         * rtlanal.c (nonzero_bits1): Likewise.
4443         * expmed.c (expand_smod_pow2): Likewise.
4444         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
4446 2013-08-23  Jan Hubicka  <jh@suse.cz>
4448         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
4449         as having address taken.
4451 2013-08-23  Jan Hubicka  <jh@suse.cz>
4453         * ipa-utils.h (method_class_type): Declare.
4454         * ipa-devirt.c (method_class_type): Export.
4456         * cgraphunit.c (analyze_functions): Do basic devirtualization;
4457         do not walk base classes of anonymous types.
4459 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
4461         PR rtl-optimization/58220
4462         PR regression/58221
4463         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
4464         handle SEQUENCE insns properly.
4466 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4468         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
4469         definition.
4470         (pp_newline_and_indent): Likewise.
4471         (pp_separate_with): Likewise.
4472         * pretty-print.c (pp_newline_and_flush): Define.
4473         (pp_newline_and_indent): Likewise.
4474         (pp_separate_with): Likewise.
4476 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
4478         PR target/58218
4479         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
4480         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
4482 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
4484         * config/i386/predicates.md (ext_sse_reg_operand): New.
4485         * config/i386/i386.md (*movti_internal): Use
4486         predicate to determine if EVEX is needed.
4487         (*movsi_internal): Ditto.
4488         (*movdf_internal): Ditto.
4489         (*movsf_internal): Ditto.
4490         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
4492 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
4494         PR tree-optimization/58209
4495         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
4496         (find_tail_calls): Give up for pointer result types if m is non-NULL.
4497         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
4498         emit POINTER_PLUS_EXPR.
4499         (create_tailcall_accumulator): For pointer result type accumulate in
4500         sizetype type.
4502 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
4504         * configure.ac: Add backslashes missing from the last change.
4505         * configure: Regenerate.
4507 2013-08-22  Jan Hubicka  <jh@susue.cz>
4509         * ipa.c (function_and_variable_visibility): First remember function
4510         was global and then make it local.
4512 2013-08-22  Julian Brown  <julian@codesourcery.com>
4514         * configure.ac: Add aarch64 to list of arches which use "nop" in
4515         debug_line test.
4516         * configure: Regenerate.
4518 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4520         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
4521         gnu_libc_has_function.
4522         * config/s390/tpf.h: Likewise.
4524 2013-08-22  Jan Hubicka  <jh@susue.cz>
4526         * timevar.c (validate_phases): Add cast.
4528 2013-08-22  Jan Hubicka  <jh@susue.cz>
4530         * timevar.c (validate_phases): Use size_t for memory.
4531         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
4533 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4535         * pretty-print.h (output_buffer::output_buffer): Declare.
4536         (pretty_printer::pretty_printer): Likewise.
4537         (pp_construct): Remove.
4538         * pretty-print.c (output_buffer::output_buffer): Define.
4539         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
4540         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
4541         (print_gimple_expr): Likewise.
4542         (print_gimple_seq): Likewise.
4543         (gimple_dump_bb): Likewise.
4544         * sched-vis.c (dump_value_slim): Likewise.
4545         (dump_insn_slim): Likewise.
4546         (dump_rtl_slim): Likewise.
4547         (str_pattern_slim): Likewise.
4548         * tree-mudflap.c (mf_varname_tree): Likewise.
4549         * graph.c (print_graph_cfg): Likewise.
4550         (start_graph_dump): Likewise.
4551         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
4552         placement-new.
4553         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
4554         pretty printer initialization.
4555         * coretypes.h (diagnostic_context): Remove superflous type alias
4556         declaration.
4557         (pretty_printer): Likewise.  Declare directly as a class.
4558         (pretty_print_info): Remove declaration as class.
4559         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
4560         and pp_clear_output_area.
4561         (asan_add_global): Likewise.
4563 2013-08-22  Jan Hubicka  <jh@suse.cz>
4565         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
4566         * ipa-utils.h (update_type_inheritance_graph): Declare.
4567         (possible_polymorphic_call_target_p): Declare.
4568         (possible_polymorphic_call_target_p): New.
4569         * ipa-devirt.c: Update toplevel comments.
4570         (cached_polymorphic_call_targets): Move up.
4571         (odr_type_d): Move ID down.
4572         (polymorphic_type_binfo_p): Update comment.
4573         (odr_hasher::remove): Likewise;
4574         (get_odr_type): Set anonymous_namespace.
4575         (dump_odr_type): Dump it.
4576         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
4577         (maybe_record_node): Record node in cached_polymorphic_call_targets.
4578         (record_binfo): Add comment.
4579         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
4580         built.
4581         (devirt_node_removal_hook): Do not iCE when cache is freed.
4582         (possible_polymorphic_call_target_p): New predicate.
4583         (update_type_inheritance_graph): New function.
4585 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4586             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4587             Sergey Lega  <sergey.s.lega@intel.com>
4588             Anna Tikhonova  <anna.tikhonova@intel.com>
4589             Ilya Tocar  <ilya.tocar@intel.com>
4590             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4591             Ilya Verbin  <ilya.verbin@intel.com>
4592             Kirill Yukhin  <kirill.yukhin@intel.com>
4593             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4595         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
4596         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
4597         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
4598         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
4599         (OPTION_MASK_ISA_AVX2_UNSET): Update.
4600         (OPTION_MASK_ISA_AVX512F_UNSET): New.
4601         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
4602         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
4603         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
4604         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
4605         OPT_mavx512pf, OPT_mavx512er cases.
4606         * config/i386/constraints.md (v): New constraint.
4607         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
4608         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
4609         (bit_AVX512CD): New.
4610         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4611         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
4612         * config/i386/i386-c.c (ix86_target_macros_internal):
4613         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
4614         __AVX512PF__.
4615         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
4616         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
4617         * config/i386/i386.c (regclass_map, dbx_register_map)
4618         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
4619         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
4620         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
4621         -mavx512pf options.
4622         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
4623         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
4624         -mavx512pf options.  Fix formatting.
4625         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
4626         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
4627         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
4628         -mavx512cd, -mavx512pf options.
4629         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
4630         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
4631         (ix86_preferred_output_reload_class): Replace SSE_REGS with
4632         ALL_SSE_REGS.
4633         (ix86_hard_regno_mode_ok): Support 512-bit registers.
4634         (ix86_set_reg_reg_cost): Ditto.
4635         (x86_order_regs_for_local_alloc): Ditto.
4636         (MAX_VECT_LEN): Extend to 64-byte.
4637         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
4638         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
4639         (TARGET_AVX512ER, TARGET_AVX512CD): New.
4640         (BIGGEST_ALIGNMENT): Extend to 512-bits.
4641         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
4642         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
4643         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
4644         (SSE_REG_MODE_P): Support new modes.
4645         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
4646         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
4647         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
4648         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
4649         (REG_CLASS_CONTENTS): Add new registers.
4650         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
4651         (EXT_REX_SSE_REGNO_P): New.
4652         (HI_REGISTER_NAMES): Add new registers.
4653         * config/i386/i386.md: Define constants for new registers.
4654         (mode): Add new 512-bit modes.
4655         (prefix): Support evex prefix.
4656         (isa): Support avx512f, noavx512f, fma_avx512f.
4657         (ssemodesuffix): Add new 512-bit modes.
4658         (movxi): New.
4659         (*movxi_internal_avx512f): Ditto.
4660         (*movdi_internal): Replace constraint "x" with the new constraint "v".
4661         Support MODE_XI.
4662         (*movsi_internal): Likewise.
4663         (*movdf_internal): Likewise.
4664         (*movsf_internal): Likewise.
4665         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
4666         (<code><mode>3): Likewise.
4667         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
4668         New.
4669         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
4670         with the new constraint "v".
4671         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
4672         modes.
4673         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
4674         with the new constraint "v".
4675         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
4676         (<sse2>_storedqu<avxsizesuffix>): Likewise.
4677         (*<plusminus_insn><mode>3): Likewise.
4678         (<sse>_vm<plusminus_insn><mode>3): Likewise.
4679         (*mul<mode>3): Likewise.
4680         (<sse>_vmmul<mode>3): Likewise.
4681         (<sse>_div<mode>3): Likewise.
4682         (<sse>_vmdiv<mode>3): Likewise.
4683         (<sse>_sqrt<mode>2): Likewise.
4684         (<sse>_vmsqrt<mode>2): Likewise.
4685         (*<code><mode>3_finite): Likewise.
4686         (*<code><mode>3) <smaxmin>: Likewise.
4687         (<sse>_vm<code><mode>3): Likewise.
4688         (*<code><mode>3) <any_logic>: Likewise.
4689         (*fma_fmadd_<mode>): Likewise.
4690         (*fma_fmsub_<mode>): Likewise.
4691         (*fma_fnmadd_<mode>): Likewise.
4692         (*fma_fnmsub_<mode>): Likewise.
4693         (*fma_fmaddsub_<mode>): Likewise.
4694         (*fma_fmsubadd_<mode>): Likewise.
4695         (*fmai_fmadd_<mode>): Likewise.
4696         (*fmai_fmsub_<mode>): Likewise.
4697         (*fmai_fnmadd_<mode>): Likewise.
4698         (*fmai_fnmsub_<mode>): Likewise.
4699         (sse_cvtsi2ss): Likewise.
4700         (sse_cvtsi2ssq): Likewise.
4701         (sse_cvtss2si): Likewise.
4702         (sse_cvtss2si_2): Likewise.
4703         (sse_cvtss2siq): Likewise.
4704         (sse_cvtss2siq_2): Likewise.
4705         (sse_cvttss2si): Likewise.
4706         (sse_cvtss2siq_2): Likewise.
4707         (float<sseintvecmodelower><mode>2): Likewise.
4708         (sse2_cvtsd2si_2): Likewise.
4709         (sse2_cvtsd2siq_2): Likewise.
4710         (*<plusminus_insn><mode>3): Likewise.
4711         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
4712         (*<sse4_1_avx2>_mul<mode>3): Likewise.
4713         (ashr<mode>3): Likewise.
4714         (<shift_insn><mode>3): Likewise.
4715         (avx2_<code><mode>3): Likewise.
4716         (*avx2_<code><mode>3): Likewise.
4717         (*andnot<mode>3): Likewise.
4718         (*<code><mode>3) <any_logic>: Likewise.
4719         (abs<mode>2): Likewise.
4720         (avx2_permvar<mode>): Likewise.
4721         (avx2_perm<mode>_1): Likewise.
4722         (*avx_vpermilp<mode>): Likewise.
4723         (avx_vpermilvar<mode>3): Likewise.
4724         (avx2_ashrv<mode>): Likewise.
4725         (avx2_<shift_insn>v<mode>): Likewise.
4726         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
4727         -mavx512cd.
4728         * doc/rtl.texi: Document XImode.
4730 2013-08-21  Jeff Law  <law@redhat.com>
4732         * tree-flow.h (register_jump_thread): Pass vector of edges
4733         instead of each important edge.
4734         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
4735         thread path into a vector and pass that to register_jump_thread.
4736         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
4737         passed in edge vector to the current 3-edge form.
4739         Revert:
4740         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4742         * dce.c (fini_dce): Call df_analyze again just in case
4743         delete_unmarked_insns removed anything.
4745 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
4747         * reload.h (struct reg_equivs): Rename to ..
4748         (struct reg_equivs_s): .. this.
4750 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
4752         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
4754 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4756         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
4758 2013-08-21  Jeff Law  <law@redhat.com>
4760         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
4761         simplify assignments too.  If the RHS collapses to a singleton
4762         range, then return the value for the range.
4764 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
4766         * config/i386/sse.md (V16): Rename to...
4767         (VMOVE): this.
4768         (mov<mode>): Update iterator name.
4769         (*mov<mode>_internal): Ditto.
4770         (push<mode>1): Ditto.
4771         (movmisalign<mode>): Ditto.
4773 2013-08-20  Jan Hubicka  <jh@suse.cz>
4775         PR bootstrap/58186
4776         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
4777         entry for direct edges.
4778         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
4780 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4782         Revert my last two changes, r201865 and r201864:
4784         Revert r201865:
4785         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4787         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4788         instances can own GC refs.
4790         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4791         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4792         (gcc::context::gt_pch_nx): Likewise.
4793         (gcc::context::gt_pch_nx):  Likewise.
4794         * ggc.h (gt_ggc_mx <T>): New.
4795         (gt_pch_nx_with_op <T>): New.
4796         (gt_pch_nx <T>): New.
4797         * passes.c (opt_pass::gt_ggc_mx): New.
4798         (opt_pass::gt_pch_nx): New.
4799         (opt_pass::gt_pch_nx_with_op): New.
4800         (pass_manager::gt_ggc_mx): New.
4801         (pass_manager::gt_pch_nx): New.
4802         (pass_manager::gt_pch_nx_with_op): New.
4803         (pass_manager::operator new): Use
4804         ggc_internal_cleared_alloc_stat rather than xcalloc.
4805         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4806         (pass_manager::gt_ggc_mx): New.
4807         (pass_manager::gt_pch_nx): New.
4808         (pass_manager::gt_pch_nx_with_op): New.
4809         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4810         (opt_pass::operator new): New.
4811         (opt_pass::gt_ggc_mx): New.
4812         (opt_pass::gt_pch_nx): New.
4813         (opt_pass::gt_pch_nx_with_op): New.
4815         Revert r201864:
4816         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4818         * Makefile.in (GTFILES): Add context.h.
4819         * context.c (gcc::context::operator new): New.
4820         (gcc::context::gt_ggc_mx): New.
4821         (gcc::context::gt_pch_nx): New.
4822         (gcc::context::gt_pch_nx): New.
4823         * context.h (gcc::context): Add GTY((user)) marking.
4824         (gcc::context::operator new): New.
4825         (gcc::context::gt_ggc_mx): New.
4826         (gcc::context::gt_pch_nx): New.
4827         (gcc::context::gt_pch_nx): New.
4828         (g): Add GTY marking.
4829         (gt_ggc_mx (gcc::context *)): New.
4830         (gt_pch_nx (gcc::context *)): New.
4831         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4832         void *cookie)): New.
4833         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4835 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4837         * dce.c (fini_dce): Call df_analyze again just in case
4838         delete_unmarked_insns removed anything.
4840 2013-08-20  Teresa Johnson  <tejohnson@google.com>
4842         PR rtl-optimizations/57451
4843         * final.c (reemit_insn_block_notes): Prevent lexical blocks
4844         from crossing split section boundaries.
4846 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
4848         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
4849         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
4850         with MULTLIB_DEFAULTS.
4852 2013-08-20  Nick Clifton  <nickc@redhat.com>
4854         * target.def (narrow_volatile_bitfield): Note that the default
4855         value is false, not !TARGET_STRICT_ALIGN.
4856         * doc/tm.texi: Regenerate.
4858 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
4860         Fix LIB_SPEC for systems without libpthread.
4862         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
4863         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
4864         for Android.
4865         * config/i386/linux-common.h: Likewise.
4866         * config/mips/linux-common.h: Likewise.
4868 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
4870         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
4871         checks.
4873 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4875         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4876         instances can own GC refs.
4878         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4879         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4880         (gcc::context::gt_pch_nx): Likewise.
4881         (gcc::context::gt_pch_nx):  Likewise.
4882         * ggc.h (gt_ggc_mx <T>): New.
4883         (gt_pch_nx_with_op <T>): New.
4884         (gt_pch_nx <T>): New.
4885         * passes.c (opt_pass::gt_ggc_mx): New.
4886         (opt_pass::gt_pch_nx): New.
4887         (opt_pass::gt_pch_nx_with_op): New.
4888         (pass_manager::gt_ggc_mx): New.
4889         (pass_manager::gt_pch_nx): New.
4890         (pass_manager::gt_pch_nx_with_op): New.
4891         (pass_manager::operator new): Use
4892         ggc_internal_cleared_alloc_stat rather than xcalloc.
4893         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4894         (pass_manager::gt_ggc_mx): New.
4895         (pass_manager::gt_pch_nx): New.
4896         (pass_manager::gt_pch_nx_with_op): New.
4897         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4898         (opt_pass::operator new): New.
4899         (opt_pass::gt_ggc_mx): New.
4900         (opt_pass::gt_pch_nx): New.
4901         (opt_pass::gt_pch_nx_with_op): New.
4903 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4905         * Makefile.in (GTFILES): Add context.h.
4906         * context.c (gcc::context::operator new): New.
4907         (gcc::context::gt_ggc_mx): New.
4908         (gcc::context::gt_pch_nx): New.
4909         (gcc::context::gt_pch_nx): New.
4910         * context.h (gcc::context): Add GTY((user)) marking.
4911         (gcc::context::operator new): New.
4912         (gcc::context::gt_ggc_mx): New.
4913         (gcc::context::gt_pch_nx): New.
4914         (gcc::context::gt_pch_nx): New.
4915         (g): Add GTY marking.
4916         (gt_ggc_mx (gcc::context *)): New.
4917         (gt_pch_nx (gcc::context *)): New.
4918         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4919         void *cookie)): New.
4920         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4922 2013-08-20  Alan Modra  <amodra@gmail.com>
4924         PR target/57865
4925         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
4926         (rs6000_emit_epilogue): Likewise.
4928 2013-08-19  Dehao Chen  <dehao@google.com>
4930         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
4932 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
4933             Jakub Jelinek  <jakub@redhat.com>
4935         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
4936         (BUILT_IN_FABSD64): Likewise.
4937         (BUILT_IN_FABSD128): Likewise.
4938         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
4939         (fold_builtin_1): Likewise.
4940         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
4941         destination and source operands.
4942         (*abstd2_fpr): Likewise.
4943         (*nabstd2_fpr): Likewise.
4945 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
4947         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
4948         JUMP_P and INSN_P.
4950 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
4952         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
4953         incomplete.
4955 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4957         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
4958         * builtins.c (default_libc_has_function): New.
4959         (gnu_libc_has_function): Ditto.
4960         (no_c99_libc_has_function): Ditto.
4961         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
4962         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
4963         (fold_builtin_sincos): Likewise.
4964         (fold_builtin_cexp): Likewise.
4965         * builtins.def (DEF_C94_BUILTIN): Likewise.
4966         (DEF_C99_BUILTIN): Likewise.
4967         (DEF_C99_C90RES_BUILTIN): Likewise.
4968         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
4969         definitions to using this define.
4970         * config/darwin-protos.h (darwin_libc_has_function): New.
4971         * config/darwin.c (darwin_libc_has_function): Ditto.
4972         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
4973         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
4974         * config/darwin.h: Ditto.
4975         * config/elfos.h: Ditto.
4976         * config/freebsd.h: Ditto.
4977         * config/i386/cygming.h: Ditto.
4978         * config/i386/djgpp.h: Ditto.
4979         * config/i386/i386-interix.h: Ditto.
4980         * config/microblaze/microblaze.h: Ditto.
4981         * config/mmix/mmix.h: Ditto.
4982         * config/gnu-user.h: Ditto.
4983         * config/ia64/hpux.h: Ditto.
4984         * config/pa/pa-hpux.h: Ditto.
4985         * config/pdp11/pdp11.h: Ditto.
4986         * config/picochip/picochip.h: Ditto.
4987         * config/linux.h: Ditto.
4988         * config/netbsd.h: Ditto.
4989         * config/openbsd.h: Ditto.
4990         * config/rs6000/aix43.h: Ditto.
4991         * config/rs6000/aix51.h: Ditto.
4992         * config/rs6000/aix52.h: Ditto.
4993         * config/rs6000/aix53.h: Ditto.
4994         * config/rs6000/aix61.h: Ditto.
4995         * config/rs6000/darwin.h: Ditto.
4996         * config/rs6000/linux.h: Ditto.
4997         * config/rs6000/linux64.h: Ditto.
4998         * config/s390/tpf.h: Ditto.
4999         * config/sol2-10.h: Ditto.
5000         * config/sol2.h: Ditto.
5001         * config/vms/vms.h: Ditto.
5002         * config/vxworks.h: Ditto.
5003         * config/linux-android.c (linux_android_libc_has_function):
5004         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
5005         * config/linux-protos.h (linux_android_libc_has_function):
5006         New declaration.
5007         * config/i386/i386.c (ix86_libc_has_function): New.
5008         * config/i386/i386-protos.h
5009         (ix86_libc_has_function): New declaration.
5010         * config/i386/i386.md
5011         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
5012         ("isinf<mode>2): Likewise.
5013         * convert.c (convert_to_integer): Using new target hook
5014         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
5015         TARGET_C99_FUNCTIONS.
5016         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
5017         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
5018         * coretypes.h (function_class): New enum for different
5019         classes of functions.
5020         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
5021         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
5022         (TARGET_HAS_SINCOS): Likewise.
5023         (TARGET_LIBC_HAS_FUNCTION): New.
5024         * doc/tm.texi: Regenerated.
5025         * targhooks.h (default_libc_has_function): New declaration.
5026         (no_c99_libc_has_function): Ditto.
5027         (gnu_libc_has_function): Ditto.
5028         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
5029         and TARGET_HAS_SINCOS.
5031 2013-08-18  Jan Hubicka  <jh@suse.cz>
5033         * Makeifle-in (ipa-devirt.o): New.
5034         (GTFILES): Add ipa-utils.h and ipa-devirt.c
5035         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
5036         (analyze_functions): Look into possible targets of polymorphic call.
5037         * dumpfile.c (dump_files): Add type-inheritance dump.
5038         * dumpfile.h (TDI_inheritance): New.
5039         * ipa-devirt.c: New file.
5040         * ipa-utils.h (odr_type_d): Forward declare.
5041         (odr_type): New type.
5042         (build_type_inheritance_graph): Declare.
5043         (possible_polymorphic_call_targets): Declare and introduce inline
5044         variant when only edge is pased.
5045         (dump_possible_polymorphic_call_targets): Likewise.
5046         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
5047         * tree.c (type_in_anonymous_namespace_p): Break out from ...
5048         (types_same_for_odr): ... here.
5049         * tree.h (type_in_anonymous_namespace_p): Declare.
5051 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
5053         PR tree-optimization/58006
5054         * tree-parloops.c (take_address_of): Don't ICE if get_name
5055         returns NULL.
5056         (eliminate_local_variables_stmt): Remove clobber stmts.
5058 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
5060         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
5061         error message is issued for an alias to undefined symbol.
5063 2013-08-18  Jan Hubicka  <jh@suse.cz>
5065         * cgraph.c (cgraph_create_indirect_edge): Discover
5066         polymorphic calls and record basic info into indirect_info.
5067         * gimple-fold.c (gimple_fold_call): When doing BINFO based
5068         devirtualization, ignore objc function calls.
5069         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
5070         call with no parm index info.
5071         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
5072         * tree.c (virtual_method_call_p): New function.
5073         * tree.h (virtual_method_call_p): Declare.
5075 2013-08-16  Jan Hubicka  <jh@suse.cz>
5077         PR middle-end/58179
5078         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
5080 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
5082         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
5083         attribute.
5085 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
5087         * gengtype.c (type_for_name): Add special-case support for
5088         locating types within the "gcc::" namespace.
5089         (open_base_files): Emit a "using namespace gcc" directive.
5091 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
5093         PR target/58160
5094         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
5095         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
5097         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
5098         array instead of each individual operand as a separate argument.
5099         (emit_fusion_gpr_load): Likewise.
5100         (expand_fusion_gpr_load): Add new function declaration.
5102         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
5103         signature to have the operands passed as an array, instead of as
5104         separate arguments.  Allow ZERO_EXTEND to be in the memory
5105         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
5106         depend on the register live/dead flags when peepholes are run.
5107         (expand_fusion_gpr_load): New function to be called from the
5108         peephole2 pass, to change the register that addis sets to be the
5109         target register.
5110         (emit_fusion_gpr_load): Change the calling signature to have the
5111         operands passed as an array, instead of as separate arguments.
5112         Allow ZERO_EXTEND to be in the memory address, and also
5113         SIGN_EXTEND if -mpower8-fusion-sign.
5115         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
5116         unspec enumeration.
5117         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
5118         adjust the register addis loads up in the peephole2 pass.  Do not
5119         depend on the register live/dead state when the peephole pass is done.
5121 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
5123         * gengtype.c (create_user_defined_type): Ensure that the kind
5124         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
5125         declaration is seen before the GTY((user)) marking.
5127 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5129         PR target/58105
5130         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
5132 2013-08-16  Jan Hubicka  <jh@suse.cz>
5134         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
5135         arugment expected_type.
5136         (gimple_fold_call): Use it.
5137         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
5138         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
5139         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
5140         (try_make_edge_direct_virtual_call): Likewise.
5141         * tree.c (obj_type_ref_class): New.
5142         * tree.h (obj_type_ref_class): Use it.
5144 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5146         * sched-vis.c (rtl_slim_pp_initialized): Remove.
5147         (rtl_slim_pp): Likewise.
5148         (init_rtl_slim_pretty_print): Likewise.
5149         (dump_value_slim):  Don't call it.  Use local pretty printer.
5150         (dump_insn_slim): Likewise.
5151         (dump_rtl_slim): Likewise.
5152         (str_pattern_slim): Likewise.
5153         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
5154         Simplify.
5156 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
5158         PR tree-optimization/58164
5159         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
5160         walk gimple_goto_dest of GIMPLE_GOTO.
5162         PR tree-optimization/58165
5163         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
5164         bi_call must be the last stmt in a bb, don't split_block, instead
5165         use fallthru edge from it and give up if there is none.
5166         Release conds vector when returning early.
5168 2013-08-14  Xinliang David Li  <davidxl@google.com>
5170         * config/i386/i386.c (ix86_option_override_internal):
5171         Remove unused variable and field.
5173 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5175         PR target/57949
5176         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
5177         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
5178         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
5179         and Linux, correct BLKmode alignment when 128-bit alignment is
5180         required and compatibility flag is not set.
5181         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
5182         for zero-size arguments when compatibility flag is not set.
5184 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
5186         PR tree-optimization/58145
5187         * tree-sra.c (build_ref_for_offset): If prev_base has
5188         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
5190 2013-08-14  Xinliang David Li  <davidxl@google.com>
5192         * config/i386/i386.c (ix86_option_override_internal):
5193         Fix uninitialized variable error.
5195 2013-08-14  Xinliang David Li  <davidxl@google.com>
5197         * config/i386/i386.opt: Define two new options.
5198         * config/i386/x86-tune.def: Add arch selector field in macros.
5199         * config/i386/i386.h: Adjust macro definition.
5200         * config/i386/i386.c (ix86_option_override_internal):
5201         Refactor the code.
5202         (parse_mtune_ctrl_str): New function.
5203         (set_ix86_tune_features): New function.
5204         (ix86_function_specific_restore): Call the new helper function.
5206 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
5208         PR rtl-optimization/57662
5209         * sel-sched.c (code_motion_process_successors): When the current insn
5210         is removed after the recursive traversal, break from the loop.
5211         Add comments and debug printouts.
5213 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
5214             Alexandre Oliva  <aoliva@redhat.com>
5216         PR target/58067
5217         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
5218         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
5219         there also UNSPEC_PLTOFF.
5221 2013-08-14  Marek Polacek  <polacek@redhat.com>
5223         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
5224         NUM_CONDITIONS bit positions.
5226 2013-08-13  Cary Coutant  <ccoutant@google.com>
5228         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
5229         (attr_checksum): Hash vector contents instead of pointer.
5230         (attr_checksum_ordered): Likewise.
5232 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
5234         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
5235         when Pmode != word_mode.  Add length_address attribute.
5236         (sse3_monitor_<mode>): Merge from sse3_monitor and
5237         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
5238         Pmode != word_mode.  Update insn length attribute.
5239         * config/i386/i386.c (ix86_option_override_internal): Update
5240         ix86_gen_monitor selection for merged sse3_monitor insn.
5242 2013-08-13  Julian Brown  <julian@codesourcery.com>
5244         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
5245         perform invalid legitimization on greater-than-word-size modes for
5246         TARGET_E500_DOUBLE.
5248 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
5250         * ira.c (setup_class_translate_array): Use aclass instead of cl
5251         for classes not fully covered by allocno classes.
5253 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
5255         PR tree-optimization/57661
5256         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
5257         * tree-inline.c (tree_function_versioning): Initialize it.
5258         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
5259         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
5260         that is not being copied.
5262         PR sanitizer/56417
5263         * asan.c (instrument_strlen_call): Fix typo in comment.
5264         Use char * type even for the lhs of POINTER_PLUS_EXPR.
5266 2013-08-13  Steve Ellcey  <sellcey@mips.com>
5268         * config/mips/mips.md (prefetch): Use lw instead of ld on
5269         loongson in 32bit mode.
5271 2013-08-13  Nick Clifton  <nickc@redhat.com>
5273         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
5275 2013-08-13  Jan Hubicka  <jh@suse.cz>
5277         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
5278         introduced edge; fix typo in sanity check.
5279         (cgraph_resolve_speculation): Export; improve diagnostic.
5280         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
5281         speculation at type mismatch.
5282         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
5283         (cgraph_resolve_speculation): Declare.
5284         (symtab_can_be_discarded): New function.
5285         * value-prof.c (gimple_ic_transform): Remove actual transform code.
5286         * ipa-inline-transform.c (speculation_removed): New global var.
5287         (clone_inlined_nodes): See if speculation can be removed.
5288         (inline_call): If speculations was removed, we growths may not match.
5289         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
5290         (speculation_useful_p): New function.
5291         (resolve_noninline_speculation): New function.
5292         (inline_small_functions): Resolve useless speculations.
5293         * ipa-inline.h (speculation_useful_p): Declare
5294         * ipa.c (can_replace_by_local_alias): Simplify.
5295         (ipa_profile): Produce speculative calls in non-lto, too;
5296         add simple cost model; produce local aliases.
5298 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
5300         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
5301         PASS_MANAGER_H.
5303 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
5305         * config/i386/i386.c (ix86_function_versions): Use error + inform.
5307 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
5309         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
5310         iterator instead of X87MODEF.
5312 2013-08-12  Perez Read  <netfirewall@gmail.com>
5314         PR target/58132
5315         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
5316         operand 0 for intel asm alternative.
5317         (*movabs<mode>_2): Ditto for operand 1.
5319 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
5321         * config/aarch64/arm_none.h
5322         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
5324 2013-08-12  Nick Clifton  <nickc@redhat.com>
5326         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
5328 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5330         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
5331         expand for QI/HImode operand to produce more effictive code for
5332         unsigned char(short) --> float(double) conversion.
5334 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
5336         * doc/invoke.texi: Mention that -ftls-model does not force the final
5337         model.
5339 2013-08-12  Marek Polacek  <polacek@redhat.com>
5340             Marc Glisse  <marc.glisse@inria.fr>
5342         PR tree-optimization/57980
5343         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
5344         when creating -1 constant.
5346 2013-08-10  Jan Hubicka  <jh@suse.cz>
5348         Workaround binutils PR14342.
5349         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
5350         (gimple_init_edge_profiler): Likewise.
5351         (gimple_gen_ic_func_profiler): Likewise.
5353 2013-08-09  Jan Hubicka  <jh@suse.cz>
5355         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
5357 2013-08-09  Xinliang David Li  <davidxl@google.com>
5359         * config/i386/stringop.def: New file.
5360         * config/i386/stringop.opt: New file.
5361         * config/i386/i386-opts.h: Include stringopt.def.
5362         * config/i386/i386.opt: Include stringopt.opt.
5363         * config/i386/i386.c (ix86_option_override_internal):
5364         Override default size based stringop inline strategies with options.
5365         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
5366         New function.
5368 2013-08-09  Jan Hubicka  <jh@suse.cz>
5370         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
5372 2013-08-09  Jan Hubicka  <jh@suse.cz>
5374         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
5375         CGRAPH_FREQ_MAX.
5376         (dump_cgraph_node): Dump profile-id.
5377         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
5378         and common_target_probability.
5379         * lto-cgraph.c (lto_output_edge): Stream common targets.
5380         (lto_output_node): Stream profile ids.
5381         (input_node): Stream profile ids.
5382         (input_edge): Stream common targets.
5383         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
5384         * ipa.c: Include value-prof.h
5385         (ipa_profile_generate_summary): Turn indirect call statement histograms
5386         into common targets.
5387         (ipa_profile): Turn common targets into speculative edges.
5389 2013-08-09  Jan Hubicka  <jh@suse.cz>
5391         * cgraph.h (cgraph_node): Add profile_id.
5392         * value-prof.c (cgraph_node_map): Turn into pointer_map.
5393         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
5394         (del_node_map): Update.
5395         (find_func_by_funcdef_no): Replace by ...
5396         (find_func_by_profile_id): ... this one.
5397         (gimple_ic_transform): Do not remove useful histograms when
5398         speculation is not done; dump info when indirect call removal
5399         can happen at LTO.
5400         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
5401         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
5402         (__gcov_indirect_call_profiler_v2): .. this one.
5403         * profile.h (init_node_map): Update.
5404         * coverage.c (coverage_compute_profile_id): New function.
5405         * coverage.h (coverage_compute_profile_id): Declare.
5406         * tree-profile.c (init_ic_make_global_vars): Make
5407         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
5408         (gimple_init_edge_profiler): Update prototype of
5409         __gcov_indirect_call_profiler.
5410         (gimple_gen_ic_func_profiler): Simplify.
5411         (tree_profiling): Use init_node_map
5413 2013-08-09  Jan Hubicka  <jh@suse.cz>
5415         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
5416         non-speculative refs.
5417         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
5418         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
5419         (cgraph_set_call_stmt): Likewise.
5420         (cgraph_create_edge_1): Fix release checking compilatoin;
5421         clear lto_stmt_uid.
5422         (cgraph_free_edge): Free indirect info.
5423         (cgraph_turn_edge_to_speculative): New function.
5424         (cgraph_speculative_call_info): New function.
5425         (cgraph_make_edge_direct): Return direct edge; handle speculation.
5426         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
5427         (dump_cgraph_node): Dump speculation.
5428         (verify_edge_count_and_frequency): Accept speculative edges.
5429         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
5430         (verify_cgraph_node): Handle speculation.
5431         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
5432         (cgraph_set_call_stmt): Update prototype.
5433         (cgraph_make_edge_direct): Update prototype.
5434         (cgraph_speculative_call_info): Declare.
5435         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
5436         to change; update call of ipa_find_references.
5437         * ipa-ref.c (ipa_record_reference): Fix return value; clear
5438         lto_stmt_uid and speculative flags.
5439         (ipa_dump_references): Dump speculation.
5440         (ipa_clone_references): Clone speculative flag.
5441         (ipa_clone_referring): Likewise.
5442         (ipa_clone_ref): New function.
5443         (ipa_find_reference): Look into lto_stmt_uids
5444         (ipa_clear_stmts_in_references): Do not clear speculative calls.
5445         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
5446         (ipa_find_reference): Update declaration.
5447         (ipa_clone_ref): Declare.
5448         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
5449         stream speculative flag.
5450         (lto_output_ref): Stream statements uids and speculation.
5451         (input_ref): Likewise.
5452         (input_edge): Stream speuclation.
5453         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
5454         (cgraph_set_call_stmt_including_clones): Handle speculation.
5455         * ipa-inline.c (heap_edge_removal_hook): New function.
5456         (inline_small_functions): Register it.
5457         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
5458         also initialize refs.
5459         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
5460         edge to change.
5461         (try_make_edge_direct_simple_call): Likewise.
5462         (try_make_edge_direct_simple_call): Likewise.
5463         (update_indirect_edges_after_inlining): Likewise.
5464         (remove_described_reference): Look proper lto_stmt_uid.
5465         (propagate_controlled_uses): Likewise.
5466         (propagate_controlled_uses): Liekwise.
5467         * tree-inline.c (copy_bb): Copy speculative edges.
5468         (redirect_all_calls): New function.
5469         (copy_cfg_body): Do redirection after loop info is updated.
5470         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
5472 2013-08-09  Jan Hubicka  <jh@suse.cz>
5474         * lto-streamer-out.c (output_function): Renumber PHIs.
5475         * lto-streamer-in.c (input_function): Likewise.
5477 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
5479         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
5480         (get_lane_unsigned): Likewise.
5481         (dup_lane_scalar): Likewise.
5482         (get_lane): enable for VALL.
5483         * config/aarch64/aarch64-simd.md
5484         (aarch64_dup_lane_scalar<mode>): Remove.
5485         (aarch64_get_lane_signed<mode>): Likewise.
5486         (aarch64_get_lane_unsigned<mode>): Likewise.
5487         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
5488         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
5489         (aarch64_get_lane<mode>): Enable for all vector modes.
5490         (aarch64_get_lanedi): Remove misleading constraints.
5491         * config/aarch64/arm_neon.h
5492         (__aarch64_vget_lane_any): Define.
5493         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
5494         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
5495         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
5496         * config/aarch64/iterators.md (VDQQH): New.
5497         (VDQQHS): Likewise.
5498         (vwcore): Likewise.
5500 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
5502         * configure.ac: Add GAS check for LEON instructions on SPARC.
5503         * configure: Regenerate.
5504         * config.in: Likewise.
5505         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
5506         sparc*-*-* block.
5507         * config/sparc/sparc.opt (LEON, LEON3): New masks.
5508         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
5509         for LEON or LEON3.
5510         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
5511         (AS_LEON_FLAG): New macro.
5512         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
5513         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
5514         Deal with LEON and LEON3 for the memory model.
5515         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
5516         (atomic_compare_and_swap<mode>_1): Likewise.
5517         (*atomic_compare_and_swap<mode>_1): Likewise.
5519 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5521         * config/arm/neon.md (vcond): Fix floating-point vector
5522         comparisons against 0.
5524 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
5526         * lra-constraints.c (emit_spill_move): Remove assert.
5527         (process_alt_operands): Add more debugging
5528         output.  Increase reject for spilling into memory.  Decrease
5529         reject for reloading scratch.
5530         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
5532 2013-08-08  Steve Ellcey  <sellcey@mips.com>
5534         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
5535         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
5536         micromips incompatible.  Add nan2008.
5537         (MULTILIB_DIRNAMES): Add nan2008.
5538         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
5539         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
5540         and micromips incompatible.  Add nan2008.
5541         (MULTILIB_DIRNAMES): Add nan2008.
5542         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
5544 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
5546         PR rtl-optimization/58079
5547         * combine.c (combine_simplify_rtx): Avoid using SUBST if
5548         simplify_comparison has widened a comparison with an integer.
5550 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5552         * config/arm/neon.md (movmisalign<mode>): Disable when we
5553         don't allow unaligned accesses.
5554         (*movmisalign<mode>_neon_store): Likewise.
5555         (*movmisalign<mode>_neon_load): Likewise.
5556         (*movmisalign<mode>_neon_store): Likewise.
5557         (*movmisalign<mode>_neon_load): Likewise.
5559 2013-08-08  Jan Hubicka  <jh@suse.cz>
5561         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
5562         (make_pass_rebuild_cgraph_edges): Also clear references.
5563         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
5564         * ipa-inline-transform.c (inline_transform): Remove all references
5565         after inlining.
5566         * cgraphunit.c (expand_function): Remove all references after
5567         expansion.
5568         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
5569         (ipa_find_reference): Rewrite to iterator.
5570         (remove_stmt_references): Likewise.
5571         (ipa_clear_stmts_in_references): New function.
5572         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
5573         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
5574         clear references.
5575         * ipa-split.c (split_function): Remove references in split function.
5577 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
5579         PR target/57431
5580         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
5581         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
5583 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
5585         PR target/56979
5586         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
5587         suggested mode for the assignment isn't compatible with the
5588         registers required.
5590 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5592         PR target/58065
5593         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
5595 2013-08-07  Xinliang David Li  <davidxl@google.com>
5597         * config/i386/i386.opt: New option -mtune-ctrl=.
5598         * config/i386/x86-tune.def: New file.
5599         * config/i386/i386.h: include x86-tune.def.
5600         * config/i386/i386.c (ix86_option_override_internal):
5601         Parsing -mtune-ctrl= option and set tune features.
5603 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
5605         PR other/12081
5606         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
5607         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
5608         to gen_2arg_fn_t.
5610 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
5612         * rtl.h (update_alignments): Declare.
5613         * final.c (grow_label_align): New function extracted from...
5614         (shorten_branches): ...here.  Call it.
5615         (update_alignments): New function.
5616         * reorg.c (sibling_labels): New variable.
5617         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
5618         the new label along with it onto the sibling_labels vector.
5619         (fill_simple_delay_slots): Adjust call to get_label_before.
5620         (fill_slots_from_thread): Likewise.
5621         (relax_delay_slots): Likewise.
5622         (make_return_insns): Likewise.
5623         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
5625 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
5627         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
5628         document its semantics.
5629         (diagnostic_report_diagnostic): Adjust accordingly.
5631 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
5633         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
5634         (sparc_option_override): ...and port to new C++ pass API.
5635         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
5637 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
5639         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
5641 2013-08-06  Caroline Tice  <cmtice@google.com>
5643         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
5644         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
5645         * tree-pass.h: Add pass_vtable_verify.
5646         * varasm.c (assemble_variable): Add code to properly set the comdat
5647         section and name for the .vtable_map_vars section.
5648         (assemble_vtyv_preinit_initializer): New function.
5649         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
5650         LINK_ONCE flag.
5651         * output.h: Add function decl for assemble_vtv_preinit_initializer.
5652         * vtable-verify.c: New file.
5653         * vtable-verify.h: New file.
5654         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
5655         initialiation levels.
5656         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
5657         * passes.def: Insert pass_vtable_verify.
5658         * aclocal.m4: Reorder includes.
5659         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
5660         -fvtv-counts options.
5661         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
5662         as appropriate, if -fvtable-verify=... is used.
5663         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
5664         -fvtable-verify=... is used.
5665         * Makefile.in (OBJS):  Add vtable-verify.o to list.
5666         (vtable-verify.o): Add new build rule.
5667         (GTFILES): Add vtable-verify.c to list.
5668         * common.opt (fvtable-verify=): New flag.
5669         (vtv_priority): Values for fvtable-verify= flag.
5670         (fvtv-counts): New flag.
5671         (fvtv-debug): New flag.
5672         * tree.h (save_vtable_map_decl): New extern function decl.
5674 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
5676         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
5677         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
5678         (pass_data_rl78_devirt): ...new pass_data instance and...
5679         (make_pass_rl78_devirt): ...new function.
5680         (rl78_asm_file_start): Port pass registration to new C++ API.
5682 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
5684         * coretypes.h (rtl_opt_pass): Add.
5685         (gcc::context): Add.
5686         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
5687         (epiphany_init): Port to new C++ pass API.
5688         (epiphany_optimize_mode_switching): Likewise.
5689         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
5690         (pass_manager::get_pass_mode_switching): New.
5691         (pass_manager::get_pass_peephole2): New.
5692         * mode-switching.c (pass_mode_switching): Add clone method.
5693         * recog.c (pass_peephole2): Add clone method.
5694         (pass_split_all_insns): Add clone method.
5696 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
5698         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
5699         (mips_option_override): ...here, porting to new C++ API for passes.
5701 2013-08-06  Jan Hubicka  <jh@suse.cz>
5703         * cgraph.c (cgraph_get_body): New function based on lto.c
5704         implementation.
5705         * cgraph.h (cgraph_get_body): Declare.
5706         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
5707         LTO paths.
5708         * cgraphunit.c (expand_function): Get body prior expanding.
5709         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
5710         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
5711         really need.
5712         * passes.c (do_per_function_toporder): Get body.
5713         * tree-inline.c (expand_call_inline): Get body prior inlining it.
5714         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
5716 2013-08-06  Martin Jambor  <mjambor@suse.cz>
5718         PR fortran/57987
5719         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
5720         is not re-finalized.  Rename second parameter to no_collect.
5722 2013-08-06  Martin Jambor  <mjambor@suse.cz>
5724         PR middle-end/58041
5725         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
5726         MEM_REF has proper alignment information.
5728 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
5730         PR other/12081
5731         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
5732         class insn_gen_fn.
5733         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
5734         rtx (*) (rtx, ...) with insn_gen_fn.
5735         * genoutput.c (output_insn_data): Cast gen_? function pointers to
5736         insn_gen_fn::stored_funcptr.  Add initializer braces.
5738 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5740         Rewrite how instances of passes are cloned to remove assumptions
5741         about their sizes (thus allowing pass subclasses to have
5742         additional data fields, albeit non-GC-managed ones at this point).
5744         * passes.c (make_pass_instance): Now that passes have clone
5745         methods, rewrite this function to eliminate XNEW and memcpy
5746         calls that used hardcoded sizes.  Since this function no longer
5747         creates pass instances, rename it to...
5748         (add_pass_instance): ...this.  Document the old way that passes were
5749         numbered and flagged, and rework this function to continue using it.
5750         (next_pass_1): Add an initial_pass argument for use by
5751         add_pass_instance.
5752         (position_pass): When adding multiple instances of a pass, use
5753         the pass's clone method, rather than relying on the XNEW/memcpy
5754         within the former make_pass_instance (now add_pass_instance).
5755         (pass_manager::pass_manager): When invoking next_pass_1, also supply
5756         the initial instance of the current pass within the pass manager.
5758 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5760         This is the automated part of the conversion of passes from C
5761         structs to C++ classes.
5763         Patch autogenerated by refactor_passes.py from
5764         https://github.com/davidmalcolm/gcc-refactoring-scripts
5765         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
5767         * asan.c (pass_asan): Convert from a global struct to a subclass of
5768         gimple_opt_pass along with...
5769         (pass_data_asan): ...new pass_data instance and...
5770         (make_pass_asan): ...new function.
5771         (pass_asan_O0): Convert from a global struct to a subclass of
5772         gimple_opt_pass along with...
5773         (pass_data_asan_O0): ...new pass_data instance and...
5774         (make_pass_asan_O0): ...new function.
5775         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
5776         subclass of rtl_opt_pass along with...
5777         (pass_data_inc_dec): ...new pass_data instance and...
5778         (make_pass_inc_dec): ...new function.
5779         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
5780         a subclass of rtl_opt_pass along with...
5781         (pass_data_reorder_blocks): ...new pass_data instance and...
5782         (make_pass_reorder_blocks): ...new function.
5783         (pass_duplicate_computed_gotos): Convert from a global struct to a
5784         subclass of rtl_opt_pass along with...
5785         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
5786         (make_pass_duplicate_computed_gotos): ...new function.
5787         (pass_partition_blocks): Convert from a global struct to a subclass of
5788         rtl_opt_pass along with...
5789         (pass_data_partition_blocks): ...new pass_data instance and...
5790         (make_pass_partition_blocks): ...new function.
5791         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
5792         struct to a subclass of rtl_opt_pass along with...
5793         (pass_data_branch_target_load_optimize1): ...new pass_data instance
5794         and...
5795         (make_pass_branch_target_load_optimize1): ...new function.
5796         (pass_branch_target_load_optimize2): Convert from a global struct to a
5797         subclass of rtl_opt_pass along with...
5798         (pass_data_branch_target_load_optimize2): ...new pass_data instance
5799         and...
5800         (make_pass_branch_target_load_optimize2): ...new function.
5801         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
5802         of rtl_opt_pass along with...
5803         (pass_data_jump): ...new pass_data instance and...
5804         (make_pass_jump): ...new function.
5805         (pass_jump2): Convert from a global struct to a subclass of
5806         rtl_opt_pass along with...
5807         (pass_data_jump2): ...new pass_data instance and...
5808         (make_pass_jump2): ...new function.
5809         * cfgexpand.c (pass_expand): Convert from a global struct to a
5810         subclass of rtl_opt_pass along with...
5811         (pass_data_expand): ...new pass_data instance and...
5812         (make_pass_expand): ...new function.
5813         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
5814         of rtl_opt_pass along with...
5815         (pass_data_free_cfg): ...new pass_data instance and...
5816         (make_pass_free_cfg): ...new function.
5817         (pass_into_cfg_layout_mode): Convert from a global struct to a
5818         subclass of rtl_opt_pass along with...
5819         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
5820         (make_pass_into_cfg_layout_mode): ...new function.
5821         (pass_outof_cfg_layout_mode): Convert from a global struct to a
5822         subclass of rtl_opt_pass along with...
5823         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
5824         (make_pass_outof_cfg_layout_mode): ...new function.
5825         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
5826         struct to a subclass of gimple_opt_pass along with...
5827         (pass_data_build_cgraph_edges): ...new pass_data instance and...
5828         (make_pass_build_cgraph_edges): ...new function.
5829         (pass_rebuild_cgraph_edges): Convert from a global struct to a
5830         subclass of gimple_opt_pass along with...
5831         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
5832         (make_pass_rebuild_cgraph_edges): ...new function.
5833         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
5834         subclass of gimple_opt_pass along with...
5835         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
5836         and...
5837         (make_pass_remove_cgraph_callee_edges): ...new function.
5838         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
5839         struct to a subclass of rtl_opt_pass along with...
5840         (pass_data_stack_adjustments): ...new pass_data instance and...
5841         (make_pass_stack_adjustments): ...new function.
5842         * combine.c (pass_combine): Convert from a global struct to a subclass
5843         of rtl_opt_pass along with...
5844         (pass_data_combine): ...new pass_data instance and...
5845         (make_pass_combine): ...new function.
5846         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
5847         global struct to a subclass of rtl_opt_pass along with...
5848         (pass_data_compare_elim_after_reload): ...new pass_data instance
5849         and...
5850         (make_pass_compare_elim_after_reload): ...new function.
5851         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
5852         of rtl_opt_pass along with...
5853         (pass_data_rtl_cprop): ...new pass_data instance and...
5854         (make_pass_rtl_cprop): ...new function.
5855         * cse.c (pass_cse): Convert from a global struct to a subclass of
5856         rtl_opt_pass along with...
5857         (pass_data_cse): ...new pass_data instance and...
5858         (make_pass_cse): ...new function.
5859         (pass_cse2): Convert from a global struct to a subclass of
5860         rtl_opt_pass along with...
5861         (pass_data_cse2): ...new pass_data instance and...
5862         (make_pass_cse2): ...new function.
5863         (pass_cse_after_global_opts): Convert from a global struct to a
5864         subclass of rtl_opt_pass along with...
5865         (pass_data_cse_after_global_opts): ...new pass_data instance and...
5866         (make_pass_cse_after_global_opts): ...new function.
5867         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
5868         of rtl_opt_pass along with...
5869         (pass_data_ud_rtl_dce): ...new pass_data instance and...
5870         (make_pass_ud_rtl_dce): ...new function.
5871         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
5872         rtl_opt_pass along with...
5873         (pass_data_fast_rtl_dce): ...new pass_data instance and...
5874         (make_pass_fast_rtl_dce): ...new function.
5875         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
5876         a subclass of rtl_opt_pass along with...
5877         (pass_data_df_initialize_opt): ...new pass_data instance and...
5878         (make_pass_df_initialize_opt): ...new function.
5879         (pass_df_initialize_no_opt): Convert from a global struct to a
5880         subclass of rtl_opt_pass along with...
5881         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
5882         (make_pass_df_initialize_no_opt): ...new function.
5883         (pass_df_finish): Convert from a global struct to a subclass of
5884         rtl_opt_pass along with...
5885         (pass_data_df_finish): ...new pass_data instance and...
5886         (make_pass_df_finish): ...new function.
5887         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
5888         rtl_opt_pass along with...
5889         (pass_data_rtl_dse1): ...new pass_data instance and...
5890         (make_pass_rtl_dse1): ...new function.
5891         (pass_rtl_dse2): Convert from a global struct to a subclass of
5892         rtl_opt_pass along with...
5893         (pass_data_rtl_dse2): ...new pass_data instance and...
5894         (make_pass_rtl_dse2): ...new function.
5895         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
5896         subclass of rtl_opt_pass along with...
5897         (pass_data_dwarf2_frame): ...new pass_data instance and...
5898         (make_pass_dwarf2_frame): ...new function.
5899         * except.c (pass_set_nothrow_function_flags): Convert from a global
5900         struct to a subclass of rtl_opt_pass along with...
5901         (pass_data_set_nothrow_function_flags): ...new pass_data instance
5902         and...
5903         (make_pass_set_nothrow_function_flags): ...new function.
5904         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
5905         subclass of rtl_opt_pass along with...
5906         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
5907         and...
5908         (make_pass_convert_to_eh_region_ranges): ...new function.
5909         * final.c (pass_compute_alignments): Convert from a global struct to a
5910         subclass of rtl_opt_pass along with...
5911         (pass_data_compute_alignments): ...new pass_data instance and...
5912         (make_pass_compute_alignments): ...new function.
5913         (pass_final): Convert from a global struct to a subclass of
5914         rtl_opt_pass along with...
5915         (pass_data_final): ...new pass_data instance and...
5916         (make_pass_final): ...new function.
5917         (pass_shorten_branches): Convert from a global struct to a subclass of
5918         rtl_opt_pass along with...
5919         (pass_data_shorten_branches): ...new pass_data instance and...
5920         (make_pass_shorten_branches): ...new function.
5921         (pass_clean_state): Convert from a global struct to a subclass of
5922         rtl_opt_pass along with...
5923         (pass_data_clean_state): ...new pass_data instance and...
5924         (make_pass_clean_state): ...new function.
5925         * function.c (pass_instantiate_virtual_regs): Convert from a global
5926         struct to a subclass of rtl_opt_pass along with...
5927         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
5928         (make_pass_instantiate_virtual_regs): ...new function.
5929         (pass_leaf_regs): Convert from a global struct to a subclass of
5930         rtl_opt_pass along with...
5931         (pass_data_leaf_regs): ...new pass_data instance and...
5932         (make_pass_leaf_regs): ...new function.
5933         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
5934         subclass of rtl_opt_pass along with...
5935         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
5936         and...
5937         (make_pass_thread_prologue_and_epilogue): ...new function.
5938         (pass_match_asm_constraints): Convert from a global struct to a
5939         subclass of rtl_opt_pass along with...
5940         (pass_data_match_asm_constraints): ...new pass_data instance and...
5941         (make_pass_match_asm_constraints): ...new function.
5942         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
5943         subclass of rtl_opt_pass along with...
5944         (pass_data_rtl_fwprop): ...new pass_data instance and...
5945         (make_pass_rtl_fwprop): ...new function.
5946         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
5947         rtl_opt_pass along with...
5948         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
5949         (make_pass_rtl_fwprop_addr): ...new function.
5950         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
5951         rtl_opt_pass along with...
5952         (pass_data_rtl_pre): ...new pass_data instance and...
5953         (make_pass_rtl_pre): ...new function.
5954         (pass_rtl_hoist): Convert from a global struct to a subclass of
5955         rtl_opt_pass along with...
5956         (pass_data_rtl_hoist): ...new pass_data instance and...
5957         (make_pass_rtl_hoist): ...new function.
5958         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
5959         subclass of gimple_opt_pass along with...
5960         (pass_data_lower_cf): ...new pass_data instance and...
5961         (make_pass_lower_cf): ...new function.
5962         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
5963         from a global struct to a subclass of gimple_opt_pass along with...
5964         (pass_data_strength_reduction): ...new pass_data instance and...
5965         (make_pass_strength_reduction): ...new function.
5966         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
5967         of rtl_opt_pass along with...
5968         (pass_data_rtl_ifcvt): ...new pass_data instance and...
5969         (make_pass_rtl_ifcvt): ...new function.
5970         (pass_if_after_combine): Convert from a global struct to a subclass of
5971         rtl_opt_pass along with...
5972         (pass_data_if_after_combine): ...new pass_data instance and...
5973         (make_pass_if_after_combine): ...new function.
5974         (pass_if_after_reload): Convert from a global struct to a subclass of
5975         rtl_opt_pass along with...
5976         (pass_data_if_after_reload): ...new pass_data instance and...
5977         (make_pass_if_after_reload): ...new function.
5978         * init-regs.c (pass_initialize_regs): Convert from a global struct to
5979         a subclass of rtl_opt_pass along with...
5980         (pass_data_initialize_regs): ...new pass_data instance and...
5981         (make_pass_initialize_regs): ...new function.
5982         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
5983         of ipa_opt_pass_d along with...
5984         (pass_data_ipa_cp): ...new pass_data instance and...
5985         (make_pass_ipa_cp): ...new function.
5986         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
5987         global struct to a subclass of gimple_opt_pass along with...
5988         (pass_data_inline_parameters): ...new pass_data instance and...
5989         (make_pass_inline_parameters): ...new function.
5990         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
5991         subclass of gimple_opt_pass along with...
5992         (pass_data_early_inline): ...new pass_data instance and...
5993         (make_pass_early_inline): ...new function.
5994         (pass_ipa_inline): Convert from a global struct to a subclass of
5995         ipa_opt_pass_d along with...
5996         (pass_data_ipa_inline): ...new pass_data instance and...
5997         (make_pass_ipa_inline): ...new function.
5998         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
5999         struct to a subclass of gimple_opt_pass along with...
6000         (pass_data_local_pure_const): ...new pass_data instance and...
6001         (make_pass_local_pure_const): ...new function.
6002         (pass_ipa_pure_const): Convert from a global struct to a subclass of
6003         ipa_opt_pass_d along with...
6004         (pass_data_ipa_pure_const): ...new pass_data instance and...
6005         (make_pass_ipa_pure_const): ...new function.
6006         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
6007         to a subclass of ipa_opt_pass_d along with...
6008         (pass_data_ipa_reference): ...new pass_data instance and...
6009         (make_pass_ipa_reference): ...new function.
6010         * ipa-split.c (pass_split_functions): Convert from a global struct to
6011         a subclass of gimple_opt_pass along with...
6012         (pass_data_split_functions): ...new pass_data instance and...
6013         (make_pass_split_functions): ...new function.
6014         (pass_feedback_split_functions): Convert from a global struct to a
6015         subclass of gimple_opt_pass along with...
6016         (pass_data_feedback_split_functions): ...new pass_data instance and...
6017         (make_pass_feedback_split_functions): ...new function.
6018         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
6019         global struct to a subclass of simple_ipa_opt_pass along with...
6020         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
6021         instance and...
6022         (make_pass_ipa_function_and_variable_visibility): ...new function.
6023         (pass_ipa_free_inline_summary): Convert from a global struct to a
6024         subclass of simple_ipa_opt_pass along with...
6025         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
6026         (make_pass_ipa_free_inline_summary): ...new function.
6027         (pass_ipa_whole_program_visibility): Convert from a global struct to a
6028         subclass of ipa_opt_pass_d along with...
6029         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
6030         and...
6031         (make_pass_ipa_whole_program_visibility): ...new function.
6032         (pass_ipa_profile): Convert from a global struct to a subclass of
6033         ipa_opt_pass_d along with...
6034         (pass_data_ipa_profile): ...new pass_data instance and...
6035         (make_pass_ipa_profile): ...new function.
6036         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
6037         ipa_opt_pass_d along with...
6038         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
6039         (make_pass_ipa_cdtor_merge): ...new function.
6040         * ira.c (pass_ira): Convert from a global struct to a subclass of
6041         rtl_opt_pass along with...
6042         (pass_data_ira): ...new pass_data instance and...
6043         (make_pass_ira): ...new function.
6044         (pass_reload): Convert from a global struct to a subclass of
6045         rtl_opt_pass along with...
6046         (pass_data_reload): ...new pass_data instance and...
6047         (make_pass_reload): ...new function.
6048         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
6049         subclass of rtl_opt_pass along with...
6050         (pass_data_cleanup_barriers): ...new pass_data instance and...
6051         (make_pass_cleanup_barriers): ...new function.
6052         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
6053         of rtl_opt_pass along with...
6054         (pass_data_loop2): ...new pass_data instance and...
6055         (make_pass_loop2): ...new function.
6056         (pass_rtl_loop_init): Convert from a global struct to a subclass of
6057         rtl_opt_pass along with...
6058         (pass_data_rtl_loop_init): ...new pass_data instance and...
6059         (make_pass_rtl_loop_init): ...new function.
6060         (pass_rtl_loop_done): Convert from a global struct to a subclass of
6061         rtl_opt_pass along with...
6062         (pass_data_rtl_loop_done): ...new pass_data instance and...
6063         (make_pass_rtl_loop_done): ...new function.
6064         (pass_rtl_move_loop_invariants): Convert from a global struct to a
6065         subclass of rtl_opt_pass along with...
6066         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
6067         (make_pass_rtl_move_loop_invariants): ...new function.
6068         (pass_rtl_unswitch): Convert from a global struct to a subclass of
6069         rtl_opt_pass along with...
6070         (pass_data_rtl_unswitch): ...new pass_data instance and...
6071         (make_pass_rtl_unswitch): ...new function.
6072         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
6073         subclass of rtl_opt_pass along with...
6074         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
6075         and...
6076         (make_pass_rtl_unroll_and_peel_loops): ...new function.
6077         (pass_rtl_doloop): Convert from a global struct to a subclass of
6078         rtl_opt_pass along with...
6079         (pass_data_rtl_doloop): ...new pass_data instance and...
6080         (make_pass_rtl_doloop): ...new function.
6081         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
6082         a subclass of rtl_opt_pass along with...
6083         (pass_data_lower_subreg): ...new pass_data instance and...
6084         (make_pass_lower_subreg): ...new function.
6085         (pass_lower_subreg2): Convert from a global struct to a subclass of
6086         rtl_opt_pass along with...
6087         (pass_data_lower_subreg2): ...new pass_data instance and...
6088         (make_pass_lower_subreg2): ...new function.
6089         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
6090         struct to a subclass of ipa_opt_pass_d along with...
6091         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
6092         (make_pass_ipa_lto_gimple_out): ...new function.
6093         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
6094         of ipa_opt_pass_d along with...
6095         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
6096         (make_pass_ipa_lto_finish_out): ...new function.
6097         * mode-switching.c (pass_mode_switching): Convert from a global struct
6098         to a subclass of rtl_opt_pass along with...
6099         (pass_data_mode_switching): ...new pass_data instance and...
6100         (make_pass_mode_switching): ...new function.
6101         * modulo-sched.c (pass_sms): Convert from a global struct to a
6102         subclass of rtl_opt_pass along with...
6103         (pass_data_sms): ...new pass_data instance and...
6104         (make_pass_sms): ...new function.
6105         * omp-low.c (pass_expand_omp): Convert from a global struct to a
6106         subclass of gimple_opt_pass along with...
6107         (pass_data_expand_omp): ...new pass_data instance and...
6108         (make_pass_expand_omp): ...new function.
6109         (pass_lower_omp): Convert from a global struct to a subclass of
6110         gimple_opt_pass along with...
6111         (pass_data_lower_omp): ...new pass_data instance and...
6112         (make_pass_lower_omp): ...new function.
6113         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
6114         of gimple_opt_pass along with...
6115         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
6116         (make_pass_diagnose_omp_blocks): ...new function.
6117         * passes.c (pass_early_local_passes): Convert from a global struct to
6118         a subclass of simple_ipa_opt_pass along with...
6119         (pass_data_early_local_passes): ...new pass_data instance and...
6120         (make_pass_early_local_passes): ...new function.
6121         (pass_all_early_optimizations): Convert from a global struct to a
6122         subclass of gimple_opt_pass along with...
6123         (pass_data_all_early_optimizations): ...new pass_data instance and...
6124         (make_pass_all_early_optimizations): ...new function.
6125         (pass_all_optimizations): Convert from a global struct to a subclass
6126         of gimple_opt_pass along with...
6127         (pass_data_all_optimizations): ...new pass_data instance and...
6128         (make_pass_all_optimizations): ...new function.
6129         (pass_all_optimizations_g): Convert from a global struct to a subclass
6130         of gimple_opt_pass along with...
6131         (pass_data_all_optimizations_g): ...new pass_data instance and...
6132         (make_pass_all_optimizations_g): ...new function.
6133         (pass_rest_of_compilation): Convert from a global struct to a subclass
6134         of rtl_opt_pass along with...
6135         (pass_data_rest_of_compilation): ...new pass_data instance and...
6136         (make_pass_rest_of_compilation): ...new function.
6137         (pass_postreload): Convert from a global struct to a subclass of
6138         rtl_opt_pass along with...
6139         (pass_data_postreload): ...new pass_data instance and...
6140         (make_pass_postreload): ...new function.
6141         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
6142         subclass of rtl_opt_pass along with...
6143         (pass_data_gcse2): ...new pass_data instance and...
6144         (make_pass_gcse2): ...new function.
6145         * postreload.c (pass_postreload_cse): Convert from a global struct to
6146         a subclass of rtl_opt_pass along with...
6147         (pass_data_postreload_cse): ...new pass_data instance and...
6148         (make_pass_postreload_cse): ...new function.
6149         * predict.c (pass_profile): Convert from a global struct to a subclass
6150         of gimple_opt_pass along with...
6151         (pass_data_profile): ...new pass_data instance and...
6152         (make_pass_profile): ...new function.
6153         (pass_strip_predict_hints): Convert from a global struct to a subclass
6154         of gimple_opt_pass along with...
6155         (pass_data_strip_predict_hints): ...new pass_data instance and...
6156         (make_pass_strip_predict_hints): ...new function.
6157         * recog.c (pass_peephole2): Convert from a global struct to a subclass
6158         of rtl_opt_pass along with...
6159         (pass_data_peephole2): ...new pass_data instance and...
6160         (make_pass_peephole2): ...new function.
6161         (pass_split_all_insns): Convert from a global struct to a subclass of
6162         rtl_opt_pass along with...
6163         (pass_data_split_all_insns): ...new pass_data instance and...
6164         (make_pass_split_all_insns): ...new function.
6165         (pass_split_after_reload): Convert from a global struct to a subclass
6166         of rtl_opt_pass along with...
6167         (pass_data_split_after_reload): ...new pass_data instance and...
6168         (make_pass_split_after_reload): ...new function.
6169         (pass_split_before_regstack): Convert from a global struct to a
6170         subclass of rtl_opt_pass along with...
6171         (pass_data_split_before_regstack): ...new pass_data instance and...
6172         (make_pass_split_before_regstack): ...new function.
6173         (pass_split_before_sched2): Convert from a global struct to a subclass
6174         of rtl_opt_pass along with...
6175         (pass_data_split_before_sched2): ...new pass_data instance and...
6176         (make_pass_split_before_sched2): ...new function.
6177         (pass_split_for_shorten_branches): Convert from a global struct to a
6178         subclass of rtl_opt_pass along with...
6179         (pass_data_split_for_shorten_branches): ...new pass_data instance
6180         and...
6181         (make_pass_split_for_shorten_branches): ...new function.
6182         * ree.c (pass_ree): Convert from a global struct to a subclass of
6183         rtl_opt_pass along with...
6184         (pass_data_ree): ...new pass_data instance and...
6185         (make_pass_ree): ...new function.
6186         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
6187         subclass of rtl_opt_pass along with...
6188         (pass_data_stack_regs): ...new pass_data instance and...
6189         (make_pass_stack_regs): ...new function.
6190         (pass_stack_regs_run): Convert from a global struct to a subclass of
6191         rtl_opt_pass along with...
6192         (pass_data_stack_regs_run): ...new pass_data instance and...
6193         (make_pass_stack_regs_run): ...new function.
6194         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
6195         subclass of rtl_opt_pass along with...
6196         (pass_data_cprop_hardreg): ...new pass_data instance and...
6197         (make_pass_cprop_hardreg): ...new function.
6198         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
6199         subclass of rtl_opt_pass along with...
6200         (pass_data_reginfo_init): ...new pass_data instance and...
6201         (make_pass_reginfo_init): ...new function.
6202         * regmove.c (pass_regmove): Convert from a global struct to a subclass
6203         of rtl_opt_pass along with...
6204         (pass_data_regmove): ...new pass_data instance and...
6205         (make_pass_regmove): ...new function.
6206         * regrename.c (pass_regrename): Convert from a global struct to a
6207         subclass of rtl_opt_pass along with...
6208         (pass_data_regrename): ...new pass_data instance and...
6209         (make_pass_regrename): ...new function.
6210         * reorg.c (pass_delay_slots): Convert from a global struct to a
6211         subclass of rtl_opt_pass along with...
6212         (pass_data_delay_slots): ...new pass_data instance and...
6213         (make_pass_delay_slots): ...new function.
6214         (pass_machine_reorg): Convert from a global struct to a subclass of
6215         rtl_opt_pass along with...
6216         (pass_data_machine_reorg): ...new pass_data instance and...
6217         (make_pass_machine_reorg): ...new function.
6218         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
6219         of rtl_opt_pass along with...
6220         (pass_data_sched): ...new pass_data instance and...
6221         (make_pass_sched): ...new function.
6222         (pass_sched2): Convert from a global struct to a subclass of
6223         rtl_opt_pass along with...
6224         (pass_data_sched2): ...new pass_data instance and...
6225         (make_pass_sched2): ...new function.
6226         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
6227         to a subclass of rtl_opt_pass along with...
6228         (pass_data_stack_ptr_mod): ...new pass_data instance and...
6229         (make_pass_stack_ptr_mod): ...new function.
6230         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
6231         to a subclass of rtl_opt_pass along with...
6232         (pass_data_rtl_store_motion): ...new pass_data instance and...
6233         (make_pass_rtl_store_motion): ...new function.
6234         * tracer.c (pass_tracer): Convert from a global struct to a subclass
6235         of gimple_opt_pass along with...
6236         (pass_data_tracer): ...new pass_data instance and...
6237         (make_pass_tracer): ...new function.
6238         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
6239         to a subclass of gimple_opt_pass along with...
6240         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
6241         (make_pass_diagnose_tm_blocks): ...new function.
6242         (pass_lower_tm): Convert from a global struct to a subclass of
6243         gimple_opt_pass along with...
6244         (pass_data_lower_tm): ...new pass_data instance and...
6245         (make_pass_lower_tm): ...new function.
6246         (pass_tm_init): Convert from a global struct to a subclass of
6247         gimple_opt_pass along with...
6248         (pass_data_tm_init): ...new pass_data instance and...
6249         (make_pass_tm_init): ...new function.
6250         (pass_tm_mark): Convert from a global struct to a subclass of
6251         gimple_opt_pass along with...
6252         (pass_data_tm_mark): ...new pass_data instance and...
6253         (make_pass_tm_mark): ...new function.
6254         (pass_tm_edges): Convert from a global struct to a subclass of
6255         gimple_opt_pass along with...
6256         (pass_data_tm_edges): ...new pass_data instance and...
6257         (make_pass_tm_edges): ...new function.
6258         (pass_tm_memopt): Convert from a global struct to a subclass of
6259         gimple_opt_pass along with...
6260         (pass_data_tm_memopt): ...new pass_data instance and...
6261         (make_pass_tm_memopt): ...new function.
6262         (pass_ipa_tm): Convert from a global struct to a subclass of
6263         simple_ipa_opt_pass along with...
6264         (pass_data_ipa_tm): ...new pass_data instance and...
6265         (make_pass_ipa_tm): ...new function.
6266         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
6267         subclass of gimple_opt_pass along with...
6268         (pass_data_call_cdce): ...new pass_data instance and...
6269         (make_pass_call_cdce): ...new function.
6270         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
6271         subclass of gimple_opt_pass along with...
6272         (pass_data_build_cfg): ...new pass_data instance and...
6273         (make_pass_build_cfg): ...new function.
6274         (pass_split_crit_edges): Convert from a global struct to a subclass of
6275         gimple_opt_pass along with...
6276         (pass_data_split_crit_edges): ...new pass_data instance and...
6277         (make_pass_split_crit_edges): ...new function.
6278         (pass_warn_function_return): Convert from a global struct to a
6279         subclass of gimple_opt_pass along with...
6280         (pass_data_warn_function_return): ...new pass_data instance and...
6281         (make_pass_warn_function_return): ...new function.
6282         (pass_warn_function_noreturn): Convert from a global struct to a
6283         subclass of gimple_opt_pass along with...
6284         (pass_data_warn_function_noreturn): ...new pass_data instance and...
6285         (make_pass_warn_function_noreturn): ...new function.
6286         (pass_warn_unused_result): Convert from a global struct to a subclass
6287         of gimple_opt_pass along with...
6288         (pass_data_warn_unused_result): ...new pass_data instance and...
6289         (make_pass_warn_unused_result): ...new function.
6290         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
6291         a subclass of gimple_opt_pass along with...
6292         (pass_data_merge_phi): ...new pass_data instance and...
6293         (make_pass_merge_phi): ...new function.
6294         * tree-complex.c (pass_lower_complex): Convert from a global struct to
6295         a subclass of gimple_opt_pass along with...
6296         (pass_data_lower_complex): ...new pass_data instance and...
6297         (make_pass_lower_complex): ...new function.
6298         (pass_lower_complex_O0): Convert from a global struct to a subclass of
6299         gimple_opt_pass along with...
6300         (pass_data_lower_complex_O0): ...new pass_data instance and...
6301         (make_pass_lower_complex_O0): ...new function.
6302         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
6303         subclass of gimple_opt_pass along with...
6304         (pass_data_lower_eh): ...new pass_data instance and...
6305         (make_pass_lower_eh): ...new function.
6306         (pass_refactor_eh): Convert from a global struct to a subclass of
6307         gimple_opt_pass along with...
6308         (pass_data_refactor_eh): ...new pass_data instance and...
6309         (make_pass_refactor_eh): ...new function.
6310         (pass_lower_resx): Convert from a global struct to a subclass of
6311         gimple_opt_pass along with...
6312         (pass_data_lower_resx): ...new pass_data instance and...
6313         (make_pass_lower_resx): ...new function.
6314         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
6315         of gimple_opt_pass along with...
6316         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
6317         (make_pass_lower_eh_dispatch): ...new function.
6318         (pass_cleanup_eh): Convert from a global struct to a subclass of
6319         gimple_opt_pass along with...
6320         (pass_data_cleanup_eh): ...new pass_data instance and...
6321         (make_pass_cleanup_eh): ...new function.
6322         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
6323         to a subclass of simple_ipa_opt_pass along with...
6324         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
6325         (make_pass_ipa_lower_emutls): ...new function.
6326         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
6327         a subclass of gimple_opt_pass along with...
6328         (pass_data_if_conversion): ...new pass_data instance and...
6329         (make_pass_if_conversion): ...new function.
6330         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
6331         subclass of gimple_opt_pass along with...
6332         (pass_data_build_ssa): ...new pass_data instance and...
6333         (make_pass_build_ssa): ...new function.
6334         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
6335         global struct to a subclass of gimple_opt_pass along with...
6336         (pass_data_loop_distribution): ...new pass_data instance and...
6337         (make_pass_loop_distribution): ...new function.
6338         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
6339         subclass of gimple_opt_pass along with...
6340         (pass_data_mudflap_1): ...new pass_data instance and...
6341         (make_pass_mudflap_1): ...new function.
6342         (pass_mudflap_2): Convert from a global struct to a subclass of
6343         gimple_opt_pass along with...
6344         (pass_data_mudflap_2): ...new pass_data instance and...
6345         (make_pass_mudflap_2): ...new function.
6346         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
6347         subclass of gimple_opt_pass along with...
6348         (pass_data_mudflap_1): ...new pass_data instance and...
6349         (make_pass_mudflap_1): ...new function.
6350         (pass_mudflap_2): Convert from a global struct to a subclass of
6351         gimple_opt_pass along with...
6352         (pass_data_mudflap_2): ...new pass_data instance and...
6353         (make_pass_mudflap_2): ...new function.
6354         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
6355         gimple_opt_pass along with...
6356         (pass_data_nrv): ...new pass_data instance and...
6357         (make_pass_nrv): ...new function.
6358         (pass_return_slot): Convert from a global struct to a subclass of
6359         gimple_opt_pass along with...
6360         (pass_data_return_slot): ...new pass_data instance and...
6361         (make_pass_return_slot): ...new function.
6362         * tree-object-size.c (pass_object_sizes): Convert from a global struct
6363         to a subclass of gimple_opt_pass along with...
6364         (pass_data_object_sizes): ...new pass_data instance and...
6365         (make_pass_object_sizes): ...new function.
6366         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
6367         global struct to a subclass of gimple_opt_pass along with...
6368         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
6369         and...
6370         (make_pass_cleanup_cfg_post_optimizing): ...new function.
6371         (pass_fixup_cfg): Convert from a global struct to a subclass of
6372         gimple_opt_pass along with...
6373         (pass_data_fixup_cfg): ...new pass_data instance and...
6374         (make_pass_fixup_cfg): ...new function.
6375         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
6376         (make_pass_mudflap_1): ...new function.
6377         (pass_mudflap_2): Replace declaration with that of...
6378         (make_pass_mudflap_2): ...new function.
6379         (pass_asan): Replace declaration with that of...
6380         (make_pass_asan): ...new function.
6381         (pass_asan_O0): Replace declaration with that of...
6382         (make_pass_asan_O0): ...new function.
6383         (pass_tsan): Replace declaration with that of...
6384         (make_pass_tsan): ...new function.
6385         (pass_tsan_O0): Replace declaration with that of...
6386         (make_pass_tsan_O0): ...new function.
6387         (pass_lower_cf): Replace declaration with that of...
6388         (make_pass_lower_cf): ...new function.
6389         (pass_refactor_eh): Replace declaration with that of...
6390         (make_pass_refactor_eh): ...new function.
6391         (pass_lower_eh): Replace declaration with that of...
6392         (make_pass_lower_eh): ...new function.
6393         (pass_lower_eh_dispatch): Replace declaration with that of...
6394         (make_pass_lower_eh_dispatch): ...new function.
6395         (pass_lower_resx): Replace declaration with that of...
6396         (make_pass_lower_resx): ...new function.
6397         (pass_build_cfg): Replace declaration with that of...
6398         (make_pass_build_cfg): ...new function.
6399         (pass_early_tree_profile): Replace declaration with that of...
6400         (make_pass_early_tree_profile): ...new function.
6401         (pass_cleanup_eh): Replace declaration with that of...
6402         (make_pass_cleanup_eh): ...new function.
6403         (pass_sra): Replace declaration with that of...
6404         (make_pass_sra): ...new function.
6405         (pass_sra_early): Replace declaration with that of...
6406         (make_pass_sra_early): ...new function.
6407         (pass_early_ipa_sra): Replace declaration with that of...
6408         (make_pass_early_ipa_sra): ...new function.
6409         (pass_tail_recursion): Replace declaration with that of...
6410         (make_pass_tail_recursion): ...new function.
6411         (pass_tail_calls): Replace declaration with that of...
6412         (make_pass_tail_calls): ...new function.
6413         (pass_tree_loop): Replace declaration with that of...
6414         (make_pass_tree_loop): ...new function.
6415         (pass_tree_loop_init): Replace declaration with that of...
6416         (make_pass_tree_loop_init): ...new function.
6417         (pass_lim): Replace declaration with that of...
6418         (make_pass_lim): ...new function.
6419         (pass_tree_unswitch): Replace declaration with that of...
6420         (make_pass_tree_unswitch): ...new function.
6421         (pass_predcom): Replace declaration with that of...
6422         (make_pass_predcom): ...new function.
6423         (pass_iv_canon): Replace declaration with that of...
6424         (make_pass_iv_canon): ...new function.
6425         (pass_scev_cprop): Replace declaration with that of...
6426         (make_pass_scev_cprop): ...new function.
6427         (pass_empty_loop): Replace declaration with that of...
6428         (make_pass_empty_loop): ...new function.
6429         (pass_record_bounds): Replace declaration with that of...
6430         (make_pass_record_bounds): ...new function.
6431         (pass_graphite): Replace declaration with that of...
6432         (make_pass_graphite): ...new function.
6433         (pass_graphite_transforms): Replace declaration with that of...
6434         (make_pass_graphite_transforms): ...new function.
6435         (pass_if_conversion): Replace declaration with that of...
6436         (make_pass_if_conversion): ...new function.
6437         (pass_loop_distribution): Replace declaration with that of...
6438         (make_pass_loop_distribution): ...new function.
6439         (pass_vectorize): Replace declaration with that of...
6440         (make_pass_vectorize): ...new function.
6441         (pass_slp_vectorize): Replace declaration with that of...
6442         (make_pass_slp_vectorize): ...new function.
6443         (pass_complete_unroll): Replace declaration with that of...
6444         (make_pass_complete_unroll): ...new function.
6445         (pass_complete_unrolli): Replace declaration with that of...
6446         (make_pass_complete_unrolli): ...new function.
6447         (pass_parallelize_loops): Replace declaration with that of...
6448         (make_pass_parallelize_loops): ...new function.
6449         (pass_loop_prefetch): Replace declaration with that of...
6450         (make_pass_loop_prefetch): ...new function.
6451         (pass_iv_optimize): Replace declaration with that of...
6452         (make_pass_iv_optimize): ...new function.
6453         (pass_tree_loop_done): Replace declaration with that of...
6454         (make_pass_tree_loop_done): ...new function.
6455         (pass_ch): Replace declaration with that of...
6456         (make_pass_ch): ...new function.
6457         (pass_ccp): Replace declaration with that of...
6458         (make_pass_ccp): ...new function.
6459         (pass_phi_only_cprop): Replace declaration with that of...
6460         (make_pass_phi_only_cprop): ...new function.
6461         (pass_build_ssa): Replace declaration with that of...
6462         (make_pass_build_ssa): ...new function.
6463         (pass_build_alias): Replace declaration with that of...
6464         (make_pass_build_alias): ...new function.
6465         (pass_build_ealias): Replace declaration with that of...
6466         (make_pass_build_ealias): ...new function.
6467         (pass_dominator): Replace declaration with that of...
6468         (make_pass_dominator): ...new function.
6469         (pass_dce): Replace declaration with that of...
6470         (make_pass_dce): ...new function.
6471         (pass_dce_loop): Replace declaration with that of...
6472         (make_pass_dce_loop): ...new function.
6473         (pass_cd_dce): Replace declaration with that of...
6474         (make_pass_cd_dce): ...new function.
6475         (pass_call_cdce): Replace declaration with that of...
6476         (make_pass_call_cdce): ...new function.
6477         (pass_merge_phi): Replace declaration with that of...
6478         (make_pass_merge_phi): ...new function.
6479         (pass_split_crit_edges): Replace declaration with that of...
6480         (make_pass_split_crit_edges): ...new function.
6481         (pass_pre): Replace declaration with that of...
6482         (make_pass_pre): ...new function.
6483         (pass_profile): Replace declaration with that of...
6484         (make_pass_profile): ...new function.
6485         (pass_strip_predict_hints): Replace declaration with that of...
6486         (make_pass_strip_predict_hints): ...new function.
6487         (pass_lower_complex_O0): Replace declaration with that of...
6488         (make_pass_lower_complex_O0): ...new function.
6489         (pass_lower_complex): Replace declaration with that of...
6490         (make_pass_lower_complex): ...new function.
6491         (pass_lower_vector): Replace declaration with that of...
6492         (make_pass_lower_vector): ...new function.
6493         (pass_lower_vector_ssa): Replace declaration with that of...
6494         (make_pass_lower_vector_ssa): ...new function.
6495         (pass_lower_omp): Replace declaration with that of...
6496         (make_pass_lower_omp): ...new function.
6497         (pass_diagnose_omp_blocks): Replace declaration with that of...
6498         (make_pass_diagnose_omp_blocks): ...new function.
6499         (pass_expand_omp): Replace declaration with that of...
6500         (make_pass_expand_omp): ...new function.
6501         (pass_expand_omp_ssa): Replace declaration with that of...
6502         (make_pass_expand_omp_ssa): ...new function.
6503         (pass_object_sizes): Replace declaration with that of...
6504         (make_pass_object_sizes): ...new function.
6505         (pass_strlen): Replace declaration with that of...
6506         (make_pass_strlen): ...new function.
6507         (pass_fold_builtins): Replace declaration with that of...
6508         (make_pass_fold_builtins): ...new function.
6509         (pass_stdarg): Replace declaration with that of...
6510         (make_pass_stdarg): ...new function.
6511         (pass_early_warn_uninitialized): Replace declaration with that of...
6512         (make_pass_early_warn_uninitialized): ...new function.
6513         (pass_late_warn_uninitialized): Replace declaration with that of...
6514         (make_pass_late_warn_uninitialized): ...new function.
6515         (pass_cse_reciprocals): Replace declaration with that of...
6516         (make_pass_cse_reciprocals): ...new function.
6517         (pass_cse_sincos): Replace declaration with that of...
6518         (make_pass_cse_sincos): ...new function.
6519         (pass_optimize_bswap): Replace declaration with that of...
6520         (make_pass_optimize_bswap): ...new function.
6521         (pass_optimize_widening_mul): Replace declaration with that of...
6522         (make_pass_optimize_widening_mul): ...new function.
6523         (pass_warn_function_return): Replace declaration with that of...
6524         (make_pass_warn_function_return): ...new function.
6525         (pass_warn_function_noreturn): Replace declaration with that of...
6526         (make_pass_warn_function_noreturn): ...new function.
6527         (pass_cselim): Replace declaration with that of...
6528         (make_pass_cselim): ...new function.
6529         (pass_phiopt): Replace declaration with that of...
6530         (make_pass_phiopt): ...new function.
6531         (pass_forwprop): Replace declaration with that of...
6532         (make_pass_forwprop): ...new function.
6533         (pass_phiprop): Replace declaration with that of...
6534         (make_pass_phiprop): ...new function.
6535         (pass_tree_ifcombine): Replace declaration with that of...
6536         (make_pass_tree_ifcombine): ...new function.
6537         (pass_dse): Replace declaration with that of...
6538         (make_pass_dse): ...new function.
6539         (pass_nrv): Replace declaration with that of...
6540         (make_pass_nrv): ...new function.
6541         (pass_rename_ssa_copies): Replace declaration with that of...
6542         (make_pass_rename_ssa_copies): ...new function.
6543         (pass_sink_code): Replace declaration with that of...
6544         (make_pass_sink_code): ...new function.
6545         (pass_fre): Replace declaration with that of...
6546         (make_pass_fre): ...new function.
6547         (pass_check_data_deps): Replace declaration with that of...
6548         (make_pass_check_data_deps): ...new function.
6549         (pass_copy_prop): Replace declaration with that of...
6550         (make_pass_copy_prop): ...new function.
6551         (pass_vrp): Replace declaration with that of...
6552         (make_pass_vrp): ...new function.
6553         (pass_uncprop): Replace declaration with that of...
6554         (make_pass_uncprop): ...new function.
6555         (pass_return_slot): Replace declaration with that of...
6556         (make_pass_return_slot): ...new function.
6557         (pass_reassoc): Replace declaration with that of...
6558         (make_pass_reassoc): ...new function.
6559         (pass_rebuild_cgraph_edges): Replace declaration with that of...
6560         (make_pass_rebuild_cgraph_edges): ...new function.
6561         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
6562         (make_pass_remove_cgraph_callee_edges): ...new function.
6563         (pass_build_cgraph_edges): Replace declaration with that of...
6564         (make_pass_build_cgraph_edges): ...new function.
6565         (pass_local_pure_const): Replace declaration with that of...
6566         (make_pass_local_pure_const): ...new function.
6567         (pass_tracer): Replace declaration with that of...
6568         (make_pass_tracer): ...new function.
6569         (pass_warn_unused_result): Replace declaration with that of...
6570         (make_pass_warn_unused_result): ...new function.
6571         (pass_diagnose_tm_blocks): Replace declaration with that of...
6572         (make_pass_diagnose_tm_blocks): ...new function.
6573         (pass_lower_tm): Replace declaration with that of...
6574         (make_pass_lower_tm): ...new function.
6575         (pass_tm_init): Replace declaration with that of...
6576         (make_pass_tm_init): ...new function.
6577         (pass_tm_mark): Replace declaration with that of...
6578         (make_pass_tm_mark): ...new function.
6579         (pass_tm_memopt): Replace declaration with that of...
6580         (make_pass_tm_memopt): ...new function.
6581         (pass_tm_edges): Replace declaration with that of...
6582         (make_pass_tm_edges): ...new function.
6583         (pass_split_functions): Replace declaration with that of...
6584         (make_pass_split_functions): ...new function.
6585         (pass_feedback_split_functions): Replace declaration with that of...
6586         (make_pass_feedback_split_functions): ...new function.
6587         (pass_strength_reduction): Replace declaration with that of...
6588         (make_pass_strength_reduction): ...new function.
6589         (pass_ipa_lower_emutls): Replace declaration with that of...
6590         (make_pass_ipa_lower_emutls): ...new function.
6591         (pass_ipa_function_and_variable_visibility): Replace declaration with
6592         that of...
6593         (make_pass_ipa_function_and_variable_visibility): ...new function.
6594         (pass_ipa_tree_profile): Replace declaration with that of...
6595         (make_pass_ipa_tree_profile): ...new function.
6596         (pass_early_local_passes): Replace declaration with that of...
6597         (make_pass_early_local_passes): ...new function.
6598         (pass_ipa_whole_program_visibility): Replace declaration with that
6599         of...
6600         (make_pass_ipa_whole_program_visibility): ...new function.
6601         (pass_ipa_lto_gimple_out): Replace declaration with that of...
6602         (make_pass_ipa_lto_gimple_out): ...new function.
6603         (pass_ipa_increase_alignment): Replace declaration with that of...
6604         (make_pass_ipa_increase_alignment): ...new function.
6605         (pass_ipa_inline): Replace declaration with that of...
6606         (make_pass_ipa_inline): ...new function.
6607         (pass_ipa_free_lang_data): Replace declaration with that of...
6608         (make_pass_ipa_free_lang_data): ...new function.
6609         (pass_ipa_free_inline_summary): Replace declaration with that of...
6610         (make_pass_ipa_free_inline_summary): ...new function.
6611         (pass_ipa_cp): Replace declaration with that of...
6612         (make_pass_ipa_cp): ...new function.
6613         (pass_ipa_reference): Replace declaration with that of...
6614         (make_pass_ipa_reference): ...new function.
6615         (pass_ipa_pure_const): Replace declaration with that of...
6616         (make_pass_ipa_pure_const): ...new function.
6617         (pass_ipa_pta): Replace declaration with that of...
6618         (make_pass_ipa_pta): ...new function.
6619         (pass_ipa_lto_finish_out): Replace declaration with that of...
6620         (make_pass_ipa_lto_finish_out): ...new function.
6621         (pass_ipa_tm): Replace declaration with that of...
6622         (make_pass_ipa_tm): ...new function.
6623         (pass_ipa_profile): Replace declaration with that of...
6624         (make_pass_ipa_profile): ...new function.
6625         (pass_ipa_cdtor_merge): Replace declaration with that of...
6626         (make_pass_ipa_cdtor_merge): ...new function.
6627         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
6628         of...
6629         (make_pass_cleanup_cfg_post_optimizing): ...new function.
6630         (pass_init_datastructures): Replace declaration with that of...
6631         (make_pass_init_datastructures): ...new function.
6632         (pass_fixup_cfg): Replace declaration with that of...
6633         (make_pass_fixup_cfg): ...new function.
6634         (pass_expand): Replace declaration with that of...
6635         (make_pass_expand): ...new function.
6636         (pass_instantiate_virtual_regs): Replace declaration with that of...
6637         (make_pass_instantiate_virtual_regs): ...new function.
6638         (pass_rtl_fwprop): Replace declaration with that of...
6639         (make_pass_rtl_fwprop): ...new function.
6640         (pass_rtl_fwprop_addr): Replace declaration with that of...
6641         (make_pass_rtl_fwprop_addr): ...new function.
6642         (pass_jump): Replace declaration with that of...
6643         (make_pass_jump): ...new function.
6644         (pass_jump2): Replace declaration with that of...
6645         (make_pass_jump2): ...new function.
6646         (pass_lower_subreg): Replace declaration with that of...
6647         (make_pass_lower_subreg): ...new function.
6648         (pass_cse): Replace declaration with that of...
6649         (make_pass_cse): ...new function.
6650         (pass_fast_rtl_dce): Replace declaration with that of...
6651         (make_pass_fast_rtl_dce): ...new function.
6652         (pass_ud_rtl_dce): Replace declaration with that of...
6653         (make_pass_ud_rtl_dce): ...new function.
6654         (pass_rtl_dce): Replace declaration with that of...
6655         (make_pass_rtl_dce): ...new function.
6656         (pass_rtl_dse1): Replace declaration with that of...
6657         (make_pass_rtl_dse1): ...new function.
6658         (pass_rtl_dse2): Replace declaration with that of...
6659         (make_pass_rtl_dse2): ...new function.
6660         (pass_rtl_dse3): Replace declaration with that of...
6661         (make_pass_rtl_dse3): ...new function.
6662         (pass_rtl_cprop): Replace declaration with that of...
6663         (make_pass_rtl_cprop): ...new function.
6664         (pass_rtl_pre): Replace declaration with that of...
6665         (make_pass_rtl_pre): ...new function.
6666         (pass_rtl_hoist): Replace declaration with that of...
6667         (make_pass_rtl_hoist): ...new function.
6668         (pass_rtl_store_motion): Replace declaration with that of...
6669         (make_pass_rtl_store_motion): ...new function.
6670         (pass_cse_after_global_opts): Replace declaration with that of...
6671         (make_pass_cse_after_global_opts): ...new function.
6672         (pass_rtl_ifcvt): Replace declaration with that of...
6673         (make_pass_rtl_ifcvt): ...new function.
6674         (pass_into_cfg_layout_mode): Replace declaration with that of...
6675         (make_pass_into_cfg_layout_mode): ...new function.
6676         (pass_outof_cfg_layout_mode): Replace declaration with that of...
6677         (make_pass_outof_cfg_layout_mode): ...new function.
6678         (pass_loop2): Replace declaration with that of...
6679         (make_pass_loop2): ...new function.
6680         (pass_rtl_loop_init): Replace declaration with that of...
6681         (make_pass_rtl_loop_init): ...new function.
6682         (pass_rtl_move_loop_invariants): Replace declaration with that of...
6683         (make_pass_rtl_move_loop_invariants): ...new function.
6684         (pass_rtl_unswitch): Replace declaration with that of...
6685         (make_pass_rtl_unswitch): ...new function.
6686         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
6687         (make_pass_rtl_unroll_and_peel_loops): ...new function.
6688         (pass_rtl_doloop): Replace declaration with that of...
6689         (make_pass_rtl_doloop): ...new function.
6690         (pass_rtl_loop_done): Replace declaration with that of...
6691         (make_pass_rtl_loop_done): ...new function.
6692         (pass_web): Replace declaration with that of...
6693         (make_pass_web): ...new function.
6694         (pass_cse2): Replace declaration with that of...
6695         (make_pass_cse2): ...new function.
6696         (pass_df_initialize_opt): Replace declaration with that of...
6697         (make_pass_df_initialize_opt): ...new function.
6698         (pass_df_initialize_no_opt): Replace declaration with that of...
6699         (make_pass_df_initialize_no_opt): ...new function.
6700         (pass_reginfo_init): Replace declaration with that of...
6701         (make_pass_reginfo_init): ...new function.
6702         (pass_inc_dec): Replace declaration with that of...
6703         (make_pass_inc_dec): ...new function.
6704         (pass_stack_ptr_mod): Replace declaration with that of...
6705         (make_pass_stack_ptr_mod): ...new function.
6706         (pass_initialize_regs): Replace declaration with that of...
6707         (make_pass_initialize_regs): ...new function.
6708         (pass_combine): Replace declaration with that of...
6709         (make_pass_combine): ...new function.
6710         (pass_if_after_combine): Replace declaration with that of...
6711         (make_pass_if_after_combine): ...new function.
6712         (pass_ree): Replace declaration with that of...
6713         (make_pass_ree): ...new function.
6714         (pass_partition_blocks): Replace declaration with that of...
6715         (make_pass_partition_blocks): ...new function.
6716         (pass_match_asm_constraints): Replace declaration with that of...
6717         (make_pass_match_asm_constraints): ...new function.
6718         (pass_regmove): Replace declaration with that of...
6719         (make_pass_regmove): ...new function.
6720         (pass_split_all_insns): Replace declaration with that of...
6721         (make_pass_split_all_insns): ...new function.
6722         (pass_fast_rtl_byte_dce): Replace declaration with that of...
6723         (make_pass_fast_rtl_byte_dce): ...new function.
6724         (pass_lower_subreg2): Replace declaration with that of...
6725         (make_pass_lower_subreg2): ...new function.
6726         (pass_mode_switching): Replace declaration with that of...
6727         (make_pass_mode_switching): ...new function.
6728         (pass_sms): Replace declaration with that of...
6729         (make_pass_sms): ...new function.
6730         (pass_sched): Replace declaration with that of...
6731         (make_pass_sched): ...new function.
6732         (pass_ira): Replace declaration with that of...
6733         (make_pass_ira): ...new function.
6734         (pass_reload): Replace declaration with that of...
6735         (make_pass_reload): ...new function.
6736         (pass_clean_state): Replace declaration with that of...
6737         (make_pass_clean_state): ...new function.
6738         (pass_branch_prob): Replace declaration with that of...
6739         (make_pass_branch_prob): ...new function.
6740         (pass_value_profile_transformations): Replace declaration with that
6741         of...
6742         (make_pass_value_profile_transformations): ...new function.
6743         (pass_postreload_cse): Replace declaration with that of...
6744         (make_pass_postreload_cse): ...new function.
6745         (pass_gcse2): Replace declaration with that of...
6746         (make_pass_gcse2): ...new function.
6747         (pass_split_after_reload): Replace declaration with that of...
6748         (make_pass_split_after_reload): ...new function.
6749         (pass_branch_target_load_optimize1): Replace declaration with that
6750         of...
6751         (make_pass_branch_target_load_optimize1): ...new function.
6752         (pass_thread_prologue_and_epilogue): Replace declaration with that
6753         of...
6754         (make_pass_thread_prologue_and_epilogue): ...new function.
6755         (pass_stack_adjustments): Replace declaration with that of...
6756         (make_pass_stack_adjustments): ...new function.
6757         (pass_peephole2): Replace declaration with that of...
6758         (make_pass_peephole2): ...new function.
6759         (pass_if_after_reload): Replace declaration with that of...
6760         (make_pass_if_after_reload): ...new function.
6761         (pass_regrename): Replace declaration with that of...
6762         (make_pass_regrename): ...new function.
6763         (pass_cprop_hardreg): Replace declaration with that of...
6764         (make_pass_cprop_hardreg): ...new function.
6765         (pass_reorder_blocks): Replace declaration with that of...
6766         (make_pass_reorder_blocks): ...new function.
6767         (pass_branch_target_load_optimize2): Replace declaration with that
6768         of...
6769         (make_pass_branch_target_load_optimize2): ...new function.
6770         (pass_leaf_regs): Replace declaration with that of...
6771         (make_pass_leaf_regs): ...new function.
6772         (pass_split_before_sched2): Replace declaration with that of...
6773         (make_pass_split_before_sched2): ...new function.
6774         (pass_compare_elim_after_reload): Replace declaration with that of...
6775         (make_pass_compare_elim_after_reload): ...new function.
6776         (pass_sched2): Replace declaration with that of...
6777         (make_pass_sched2): ...new function.
6778         (pass_stack_regs): Replace declaration with that of...
6779         (make_pass_stack_regs): ...new function.
6780         (pass_stack_regs_run): Replace declaration with that of...
6781         (make_pass_stack_regs_run): ...new function.
6782         (pass_df_finish): Replace declaration with that of...
6783         (make_pass_df_finish): ...new function.
6784         (pass_compute_alignments): Replace declaration with that of...
6785         (make_pass_compute_alignments): ...new function.
6786         (pass_duplicate_computed_gotos): Replace declaration with that of...
6787         (make_pass_duplicate_computed_gotos): ...new function.
6788         (pass_variable_tracking): Replace declaration with that of...
6789         (make_pass_variable_tracking): ...new function.
6790         (pass_free_cfg): Replace declaration with that of...
6791         (make_pass_free_cfg): ...new function.
6792         (pass_machine_reorg): Replace declaration with that of...
6793         (make_pass_machine_reorg): ...new function.
6794         (pass_cleanup_barriers): Replace declaration with that of...
6795         (make_pass_cleanup_barriers): ...new function.
6796         (pass_delay_slots): Replace declaration with that of...
6797         (make_pass_delay_slots): ...new function.
6798         (pass_split_for_shorten_branches): Replace declaration with that of...
6799         (make_pass_split_for_shorten_branches): ...new function.
6800         (pass_split_before_regstack): Replace declaration with that of...
6801         (make_pass_split_before_regstack): ...new function.
6802         (pass_convert_to_eh_region_ranges): Replace declaration with that
6803         of...
6804         (make_pass_convert_to_eh_region_ranges): ...new function.
6805         (pass_shorten_branches): Replace declaration with that of...
6806         (make_pass_shorten_branches): ...new function.
6807         (pass_set_nothrow_function_flags): Replace declaration with that of...
6808         (make_pass_set_nothrow_function_flags): ...new function.
6809         (pass_dwarf2_frame): Replace declaration with that of...
6810         (make_pass_dwarf2_frame): ...new function.
6811         (pass_final): Replace declaration with that of...
6812         (make_pass_final): ...new function.
6813         (pass_rtl_seqabstr): Replace declaration with that of...
6814         (make_pass_rtl_seqabstr): ...new function.
6815         (pass_release_ssa_names): Replace declaration with that of...
6816         (make_pass_release_ssa_names): ...new function.
6817         (pass_early_inline): Replace declaration with that of...
6818         (make_pass_early_inline): ...new function.
6819         (pass_inline_parameters): Replace declaration with that of...
6820         (make_pass_inline_parameters): ...new function.
6821         (pass_update_address_taken): Replace declaration with that of...
6822         (make_pass_update_address_taken): ...new function.
6823         (pass_convert_switch): Replace declaration with that of...
6824         (make_pass_convert_switch): ...new function.
6825         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
6826         to a subclass of simple_ipa_opt_pass along with...
6827         (pass_data_ipa_tree_profile): ...new pass_data instance and...
6828         (make_pass_ipa_tree_profile): ...new function.
6829         * tree-sra.c (pass_sra_early): Convert from a global struct to a
6830         subclass of gimple_opt_pass along with...
6831         (pass_data_sra_early): ...new pass_data instance and...
6832         (make_pass_sra_early): ...new function.
6833         (pass_sra): Convert from a global struct to a subclass of
6834         gimple_opt_pass along with...
6835         (pass_data_sra): ...new pass_data instance and...
6836         (make_pass_sra): ...new function.
6837         (pass_early_ipa_sra): Convert from a global struct to a subclass of
6838         gimple_opt_pass along with...
6839         (pass_data_early_ipa_sra): ...new pass_data instance and...
6840         (make_pass_early_ipa_sra): ...new function.
6841         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
6842         subclass of gimple_opt_pass along with...
6843         (pass_data_ccp): ...new pass_data instance and...
6844         (make_pass_ccp): ...new function.
6845         (pass_fold_builtins): Convert from a global struct to a subclass of
6846         gimple_opt_pass along with...
6847         (pass_data_fold_builtins): ...new pass_data instance and...
6848         (make_pass_fold_builtins): ...new function.
6849         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
6850         subclass of gimple_opt_pass along with...
6851         (pass_data_copy_prop): ...new pass_data instance and...
6852         (make_pass_copy_prop): ...new function.
6853         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
6854         global struct to a subclass of gimple_opt_pass along with...
6855         (pass_data_rename_ssa_copies): ...new pass_data instance and...
6856         (make_pass_rename_ssa_copies): ...new function.
6857         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
6858         subclass of gimple_opt_pass along with...
6859         (pass_data_dce): ...new pass_data instance and...
6860         (make_pass_dce): ...new function.
6861         (pass_dce_loop): Convert from a global struct to a subclass of
6862         gimple_opt_pass along with...
6863         (pass_data_dce_loop): ...new pass_data instance and...
6864         (make_pass_dce_loop): ...new function.
6865         (pass_cd_dce): Convert from a global struct to a subclass of
6866         gimple_opt_pass along with...
6867         (pass_data_cd_dce): ...new pass_data instance and...
6868         (make_pass_cd_dce): ...new function.
6869         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
6870         subclass of gimple_opt_pass along with...
6871         (pass_data_dominator): ...new pass_data instance and...
6872         (make_pass_dominator): ...new function.
6873         (pass_phi_only_cprop): Convert from a global struct to a subclass of
6874         gimple_opt_pass along with...
6875         (pass_data_phi_only_cprop): ...new pass_data instance and...
6876         (make_pass_phi_only_cprop): ...new function.
6877         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
6878         subclass of gimple_opt_pass along with...
6879         (pass_data_dse): ...new pass_data instance and...
6880         (make_pass_dse): ...new function.
6881         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
6882         a subclass of gimple_opt_pass along with...
6883         (pass_data_forwprop): ...new pass_data instance and...
6884         (make_pass_forwprop): ...new function.
6885         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
6886         struct to a subclass of gimple_opt_pass along with...
6887         (pass_data_tree_ifcombine): ...new pass_data instance and...
6888         (make_pass_tree_ifcombine): ...new function.
6889         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
6890         subclass of gimple_opt_pass along with...
6891         (pass_data_ch): ...new pass_data instance and...
6892         (make_pass_ch): ...new function.
6893         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
6894         subclass of gimple_opt_pass along with...
6895         (pass_data_tree_loop): ...new pass_data instance and...
6896         (make_pass_tree_loop): ...new function.
6897         (pass_tree_loop_init): Convert from a global struct to a subclass of
6898         gimple_opt_pass along with...
6899         (pass_data_tree_loop_init): ...new pass_data instance and...
6900         (make_pass_tree_loop_init): ...new function.
6901         (pass_lim): Convert from a global struct to a subclass of
6902         gimple_opt_pass along with...
6903         (pass_data_lim): ...new pass_data instance and...
6904         (make_pass_lim): ...new function.
6905         (pass_tree_unswitch): Convert from a global struct to a subclass of
6906         gimple_opt_pass along with...
6907         (pass_data_tree_unswitch): ...new pass_data instance and...
6908         (make_pass_tree_unswitch): ...new function.
6909         (pass_predcom): Convert from a global struct to a subclass of
6910         gimple_opt_pass along with...
6911         (pass_data_predcom): ...new pass_data instance and...
6912         (make_pass_predcom): ...new function.
6913         (pass_vectorize): Convert from a global struct to a subclass of
6914         gimple_opt_pass along with...
6915         (pass_data_vectorize): ...new pass_data instance and...
6916         (make_pass_vectorize): ...new function.
6917         (pass_graphite): Convert from a global struct to a subclass of
6918         gimple_opt_pass along with...
6919         (pass_data_graphite): ...new pass_data instance and...
6920         (make_pass_graphite): ...new function.
6921         (pass_graphite_transforms): Convert from a global struct to a subclass
6922         of gimple_opt_pass along with...
6923         (pass_data_graphite_transforms): ...new pass_data instance and...
6924         (make_pass_graphite_transforms): ...new function.
6925         (pass_check_data_deps): Convert from a global struct to a subclass of
6926         gimple_opt_pass along with...
6927         (pass_data_check_data_deps): ...new pass_data instance and...
6928         (make_pass_check_data_deps): ...new function.
6929         (pass_iv_canon): Convert from a global struct to a subclass of
6930         gimple_opt_pass along with...
6931         (pass_data_iv_canon): ...new pass_data instance and...
6932         (make_pass_iv_canon): ...new function.
6933         (pass_scev_cprop): Convert from a global struct to a subclass of
6934         gimple_opt_pass along with...
6935         (pass_data_scev_cprop): ...new pass_data instance and...
6936         (make_pass_scev_cprop): ...new function.
6937         (pass_record_bounds): Convert from a global struct to a subclass of
6938         gimple_opt_pass along with...
6939         (pass_data_record_bounds): ...new pass_data instance and...
6940         (make_pass_record_bounds): ...new function.
6941         (pass_complete_unroll): Convert from a global struct to a subclass of
6942         gimple_opt_pass along with...
6943         (pass_data_complete_unroll): ...new pass_data instance and...
6944         (make_pass_complete_unroll): ...new function.
6945         (pass_complete_unrolli): Convert from a global struct to a subclass of
6946         gimple_opt_pass along with...
6947         (pass_data_complete_unrolli): ...new pass_data instance and...
6948         (make_pass_complete_unrolli): ...new function.
6949         (pass_parallelize_loops): Convert from a global struct to a subclass
6950         of gimple_opt_pass along with...
6951         (pass_data_parallelize_loops): ...new pass_data instance and...
6952         (make_pass_parallelize_loops): ...new function.
6953         (pass_loop_prefetch): Convert from a global struct to a subclass of
6954         gimple_opt_pass along with...
6955         (pass_data_loop_prefetch): ...new pass_data instance and...
6956         (make_pass_loop_prefetch): ...new function.
6957         (pass_iv_optimize): Convert from a global struct to a subclass of
6958         gimple_opt_pass along with...
6959         (pass_data_iv_optimize): ...new pass_data instance and...
6960         (make_pass_iv_optimize): ...new function.
6961         (pass_tree_loop_done): Convert from a global struct to a subclass of
6962         gimple_opt_pass along with...
6963         (pass_data_tree_loop_done): ...new pass_data instance and...
6964         (make_pass_tree_loop_done): ...new function.
6965         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
6966         struct to a subclass of gimple_opt_pass along with...
6967         (pass_data_cse_reciprocals): ...new pass_data instance and...
6968         (make_pass_cse_reciprocals): ...new function.
6969         (pass_cse_sincos): Convert from a global struct to a subclass of
6970         gimple_opt_pass along with...
6971         (pass_data_cse_sincos): ...new pass_data instance and...
6972         (make_pass_cse_sincos): ...new function.
6973         (pass_optimize_bswap): Convert from a global struct to a subclass of
6974         gimple_opt_pass along with...
6975         (pass_data_optimize_bswap): ...new pass_data instance and...
6976         (make_pass_optimize_bswap): ...new function.
6977         (pass_optimize_widening_mul): Convert from a global struct to a
6978         subclass of gimple_opt_pass along with...
6979         (pass_data_optimize_widening_mul): ...new pass_data instance and...
6980         (make_pass_optimize_widening_mul): ...new function.
6981         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
6982         subclass of gimple_opt_pass along with...
6983         (pass_data_phiopt): ...new pass_data instance and...
6984         (make_pass_phiopt): ...new function.
6985         (pass_cselim): Convert from a global struct to a subclass of
6986         gimple_opt_pass along with...
6987         (pass_data_cselim): ...new pass_data instance and...
6988         (make_pass_cselim): ...new function.
6989         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
6990         subclass of gimple_opt_pass along with...
6991         (pass_data_phiprop): ...new pass_data instance and...
6992         (make_pass_phiprop): ...new function.
6993         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
6994         subclass of gimple_opt_pass along with...
6995         (pass_data_pre): ...new pass_data instance and...
6996         (make_pass_pre): ...new function.
6997         (pass_fre): Convert from a global struct to a subclass of
6998         gimple_opt_pass along with...
6999         (pass_data_fre): ...new pass_data instance and...
7000         (make_pass_fre): ...new function.
7001         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
7002         subclass of gimple_opt_pass along with...
7003         (pass_data_reassoc): ...new pass_data instance and...
7004         (make_pass_reassoc): ...new function.
7005         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
7006         subclass of gimple_opt_pass along with...
7007         (pass_data_sink_code): ...new pass_data instance and...
7008         (make_pass_sink_code): ...new function.
7009         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
7010         subclass of gimple_opt_pass along with...
7011         (pass_data_strlen): ...new pass_data instance and...
7012         (make_pass_strlen): ...new function.
7013         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
7014         struct to a subclass of gimple_opt_pass along with...
7015         (pass_data_build_alias): ...new pass_data instance and...
7016         (make_pass_build_alias): ...new function.
7017         (pass_build_ealias): Convert from a global struct to a subclass of
7018         gimple_opt_pass along with...
7019         (pass_data_build_ealias): ...new pass_data instance and...
7020         (make_pass_build_ealias): ...new function.
7021         (pass_ipa_pta): Convert from a global struct to a subclass of
7022         simple_ipa_opt_pass along with...
7023         (pass_data_ipa_pta): ...new pass_data instance and...
7024         (make_pass_ipa_pta): ...new function.
7025         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
7026         subclass of gimple_opt_pass along with...
7027         (pass_data_uncprop): ...new pass_data instance and...
7028         (make_pass_uncprop): ...new function.
7029         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
7030         global struct to a subclass of gimple_opt_pass along with...
7031         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
7032         (make_pass_late_warn_uninitialized): ...new function.
7033         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
7034         to a subclass of gimple_opt_pass along with...
7035         (pass_data_init_datastructures): ...new pass_data instance and...
7036         (make_pass_init_datastructures): ...new function.
7037         (pass_early_warn_uninitialized): Convert from a global struct to a
7038         subclass of gimple_opt_pass along with...
7039         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
7040         (make_pass_early_warn_uninitialized): ...new function.
7041         (pass_update_address_taken): Convert from a global struct to a
7042         subclass of gimple_opt_pass along with...
7043         (pass_data_update_address_taken): ...new pass_data instance and...
7044         (make_pass_update_address_taken): ...new function.
7045         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
7046         struct to a subclass of gimple_opt_pass along with...
7047         (pass_data_release_ssa_names): ...new pass_data instance and...
7048         (make_pass_release_ssa_names): ...new function.
7049         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
7050         subclass of gimple_opt_pass along with...
7051         (pass_data_stdarg): ...new pass_data instance and...
7052         (make_pass_stdarg): ...new function.
7053         * tree-switch-conversion.c (pass_convert_switch): Convert from a
7054         global struct to a subclass of gimple_opt_pass along with...
7055         (pass_data_convert_switch): ...new pass_data instance and...
7056         (make_pass_convert_switch): ...new function.
7057         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
7058         to a subclass of gimple_opt_pass along with...
7059         (pass_data_tail_recursion): ...new pass_data instance and...
7060         (make_pass_tail_recursion): ...new function.
7061         (pass_tail_calls): Convert from a global struct to a subclass of
7062         gimple_opt_pass along with...
7063         (pass_data_tail_calls): ...new pass_data instance and...
7064         (make_pass_tail_calls): ...new function.
7065         * tree-vect-generic.c (pass_lower_vector): Convert from a global
7066         struct to a subclass of gimple_opt_pass along with...
7067         (pass_data_lower_vector): ...new pass_data instance and...
7068         (make_pass_lower_vector): ...new function.
7069         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
7070         gimple_opt_pass along with...
7071         (pass_data_lower_vector_ssa): ...new pass_data instance and...
7072         (make_pass_lower_vector_ssa): ...new function.
7073         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
7074         to a subclass of gimple_opt_pass along with...
7075         (pass_data_slp_vectorize): ...new pass_data instance and...
7076         (make_pass_slp_vectorize): ...new function.
7077         (pass_ipa_increase_alignment): Convert from a global struct to a
7078         subclass of simple_ipa_opt_pass along with...
7079         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
7080         (make_pass_ipa_increase_alignment): ...new function.
7081         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
7082         gimple_opt_pass along with...
7083         (pass_data_vrp): ...new pass_data instance and...
7084         (make_pass_vrp): ...new function.
7085         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
7086         subclass of simple_ipa_opt_pass along with...
7087         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
7088         (make_pass_ipa_free_lang_data): ...new function.
7089         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
7090         gimple_opt_pass along with...
7091         (pass_data_tsan): ...new pass_data instance and...
7092         (make_pass_tsan): ...new function.
7093         (pass_tsan_O0): Convert from a global struct to a subclass of
7094         gimple_opt_pass along with...
7095         (pass_data_tsan_O0): ...new pass_data instance and...
7096         (make_pass_tsan_O0): ...new function.
7097         * var-tracking.c (pass_variable_tracking): Convert from a global
7098         struct to a subclass of rtl_opt_pass along with...
7099         (pass_data_variable_tracking): ...new pass_data instance and...
7100         (make_pass_variable_tracking): ...new function.
7101         * web.c (pass_web): Convert from a global struct to a subclass of
7102         rtl_opt_pass along with...
7103         (pass_data_web): ...new pass_data instance and...
7104         (make_pass_web): ...new function.
7105         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
7106         declaration with that of...
7107         (make_pass_mode_switch_use): ...new function.
7108         (pass_resolve_sw_modes): Replace declaration with that of...
7109         (make_pass_resolve_sw_modes): ...new function.
7110         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
7111         from a global struct to a subclass of rtl_opt_pass along with...
7112         (pass_data_mode_switch_use): ...new pass_data instance and...
7113         (make_pass_mode_switch_use): ...new function.
7114         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
7115         from a global struct to a subclass of rtl_opt_pass along with...
7116         (pass_data_resolve_sw_modes): ...new pass_data instance and...
7117         (make_pass_resolve_sw_modes): ...new function.
7118         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
7119         struct to a subclass of rtl_opt_pass along with...
7120         (pass_data_insert_vzeroupper): ...new pass_data instance and...
7121         (make_pass_insert_vzeroupper): ...new function.
7122         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
7123         global struct to a subclass of rtl_opt_pass along with...
7124         (pass_data_work_around_errata): ...new pass_data instance and...
7125         (make_pass_work_around_errata): ...new function.
7126         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
7127         struct to a subclass of rtl_opt_pass along with...
7128         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
7129         (make_pass_mips_machine_reorg2): ...new function.
7131 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
7133         * passes.c (pass_manager::operator new): New.
7135 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
7137         Handwritten part of conversion of passes to C++ classes.
7139         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
7140         (toplev.o): Add dep on PASS_MANAGER_H.
7141         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
7142         of early local pases to reflect this moving from a global to a
7143         member of gcc::pass_manager.
7144         (cgraph_add_new_function): Likewise.
7145         * lto-cgraph.c (lto_output_node): Update for conversion of
7146         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
7147         * passes.c (opt_pass::clone): New.
7148         (opt_pass::gate): New.
7149         (opt_pass::execute): New.
7150         (opt_pass::opt_pass): New.
7151         (pass_manager::execute_early_local_passes): New.
7152         (pass_manager::execute_pass_mode_switching): new.
7153         (finish_optimization_passes): Convert to...
7154         (pass_manager::finish_optimization_passes): ...this.
7155         (finish_optimization_passes): Update for conversion of passes to
7156         C++ classes.
7157         (register_dump_files_1): Use has_gate since we cannot portably
7158         check a vtable entry against NULL.
7159         (dump_one_pass): Likewise.
7160         (ipa_write_summaries_2): Likewise.
7161         (ipa_write_optimization_summaries_1): Likewise.
7162         (ipa_read_summaries_1): Likewise.
7163         (ipa_read_optimization_summaries_1): Likewise.
7164         (execute_ipa_stmt_fixups): Likewise.
7165         (pass_manager::pass_manager): Rewrite pass-creation, invoking
7166         pass-creation functions rather than wiring up globals, and
7167         storing the results in fields of pass_manager generated using
7168         pass-instances.def.
7169         (pass_manager::dump_profile_report): Update for conversion of
7170         passes to C++ classes.
7171         (pass_manager::execute_ipa_summary_passes): Likewise.
7172         (execute_one_ipa_transform_pass): Likewise.
7173         (execute_one_pass): Use has_gate and has_execute since we cannot
7174         portably check a vtable entry against NULL.
7175         * pass_manager.h (pass_manager::finish_optimization_passes): New.
7176         (pass_manager): Use pass-instances.def to add fields for the
7177         various pass instances.
7178         * toplev.c (finalize): Update for move of
7179         finish_optimization_passes to a method of gcc::pass_manager.
7180         * toplev.h (finish_optimization_passes): Move to method of class
7181         pass_manager.
7182         * tree-pass.h (struct pass_data): New.
7183         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
7184         (opt_pass::gate): Convert to virtual function.
7185         (opt_pass::~opt_pass): New.
7186         (opt_pass::clone): New.
7187         (opt_pass::execute): Convert to virtual function.
7188         (opt_pass::opt_pass): New.
7189         (opt_pass::ctxt_): new.
7190         (gimple_opt_pass): Convert to subclass of opt_pass.
7191         (gimple_opt_pass::gimple_opt_pass): New.
7192         (rtl_opt_pass): Convert to subclass of opt_pass.
7193         (rtl_opt_pass::rtl_opt_pass): New.
7194         (ipa_opt_pass_d): Convert to subclass of opt_pass.
7195         (ipa_opt_pass_d::ipa_opt_pass_d): New.
7196         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
7197         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
7198         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
7199         invocation of pass_mode_switching to reflect this moving from a
7200         global to a member of gcc::pass_manager.
7201         (ix86_option_override): Rework how pass_insert_vzeroupper is
7202         added to the pass_manager to reflect autogenerated changes.
7203         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
7205 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
7207         PR rtl-optimization/57708
7208         * recog.c (peep2_find_free_register): Validate all regs in a
7209         multi-reg mode.
7211 2013-08-05  Jan Hubicka  <jh@suse.cz>
7213         PR lto/57602
7214         * cgraph.c (verify_cgraph_node): Accept local flags from other
7215         partitions.
7216         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
7217         (function_and_variable_visibility): Likewise.
7218         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
7220 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7222         * graph.c (init_graph_slim_pretty_print): Remove.
7223         (print_graph_cfg): Do not call it.  Use local pretty printer.
7224         (start_graph_dump): Likewise.
7226 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7228         * gimple-pretty-print.c (buffer): Remove.
7229         (initialized): Likewise.
7230         (maybe_init_pretty_print): Likewise.
7231         (print_gimple_stmt): Do not call it.  Use non-static local
7232         pretty_printer variable.
7233         (print_gimple_expr): Likewise.
7234         (print_gimple_seq): Likewise.
7235         (gimple_dump_bb): Likewise.
7237 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7239         * asan.c (asan_pp): Remove.
7240         (asan_pp_initialized): Likewise.
7241         (asan_pp_initialize): Likewise.
7242         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
7243         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
7244         (asan_add_global): Likewise.
7246 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7248         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
7249         * diagnostic.h (diagnostic_flush_buffer): Adjust.
7250         * pretty-print.c (pp_formatted_text_data): Likewise.
7251         (pp_indent): Rename from pp_base_indent.
7252         (pp_format): Rename from pp_base_format.
7253         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
7254         (pp_format_verbatim): Rename from pp_base_format_verbatim.
7255         (pp_flush): Rename from pp_base_flush.
7256         (pp_set_line_maximum_length): Rename from
7257         pp_base_set_line_maximum_length.
7258         (pp_clear_output_area): Rename from pp_base_clear_output_area.
7259         (pp_set_prefix): Rename from pp_base_set_prefix.
7260         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
7261         (pp_emit_prefix): Rename from pp_base_emit_prefix.
7262         (pp_append_text): Rename from pp_base_append_text.
7263         (pp_formatted_text): Rename from pp_base_formatted_text.
7264         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
7265         (pp_remaining_character_count_for_line): Rename from
7266         pp_base_remaining_character_count_for_line.
7267         (pp_newline): Rename from pp_base_newline.
7268         (pp_character): Rename from pp_base_character.
7269         (pp_string): Rename from pp_base_string.
7270         (pp_maybe_space): Rename from pp_base_maybe_space.
7271         * asan.c (asan_pp_string): Adjust.
7272         (asan_emit_stack_protection): Likewise.
7273         (asan_add_global): Likewise.
7274         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
7275         * tree-mudflap.c (mf_varname_tree): Likewise.
7276         * tree-pretty-print.c (pp_tree_identifier): Rename from
7277         pp_base_tree_identifier.
7278         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
7279         Declare as function.
7281 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7283         * pretty-print.h (pp_bar_bar): New.
7284         (pp_ampersand_ampersand): Likewise.
7285         (pp_less_equal): Likewise.
7286         (pp_greater_equal): Likewise.
7287         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
7288         printer functions instead of pp_string or operators and punctuators.
7289         (dump_gimple_call): Likewise.
7290         (dump_gimple_omp_for): Likewise.
7291         (dump_gimple_transaction): Likewise.
7292         (dump_gimple_phi): Likewise.
7293         (pp_gimple_stmt_1): Likewise.
7294         * sched-vis.c (print_insn): Likewise.
7295         * tree-mudflap.c (mf_varname_tree): Likewise.
7296         * tree-pretty-print.c (dump_block_node): Likewise.
7297         (dump_generic_node): Likewise.
7299 2013-08-02  Jan Hubicka  <jh@suse.cz>
7301         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
7302         boundaries.
7303         * lto-streamer-out.c (tree_is_indexable): Results decls and
7304         parm decls are not indexable.
7305         (DFS_write_tree_body): Do not follow args and results.
7306         (hash_tree): Likewise.
7307         (output_functions): Rearrange so struct function is needed
7308         only when real body is output; be able to also ouptut abstract
7309         functions; output DECL_ARGUMENTS and DECL_RESULT.
7310         (lto_output): When not in WPA, ale store abstract functions.
7311         (write_symbol): Do not care about RESULT_DECL.
7312         (output_symbol_p): Handle correctly sbtract decls.
7313         * lto-streamer-in.c (input_function): Rearrange so struct
7314         function can be NULL at entry; allow streaming of
7315         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
7316         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
7317         sanity check during LTO.
7318         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
7319         RESULT_DECl and DECL_ARGUMENTS.
7320         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
7321         Likewise.
7323 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
7325         * pretty-print.h (pp_underscore): New.
7326         (pp_comma): Tidy.
7327         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
7328         printer functions instead of pp_character.
7329         (dump_binary_rhs): Likewise.
7330         (dump_ternary_rhs): Likewise.
7331         (dump_gimple_call_args): Likewise.
7332         (pp_points_to_solution): Likewise.
7333         (dump_gimple_call): Likewise.
7334         (dump_gimple_switch): Likewise.
7335         (dump_gimple_cond): Likewise.
7336         (dump_gimple_bind): Likewise.
7337         (dump_gimple_try): Likewise.
7338         (dump_gimple_omp_for): Likewise.
7339         (dump_gimple_omp_continue): Likewise.
7340         (dump_gimple_omp_single): Likewise.
7341         (dump_gimple_omp_sections): Likewise.
7342         (dump_gimple_omp_block): Likewise.
7343         (dump_gimple_omp_critical): Likewise.
7344         (dump_gimple_transaction): Likewise.
7345         (dump_gimple_asm): Likewise.
7346         (dump_gimple_phi): Likewise.
7347         (dump_gimple_omp_parallel): Likewise.
7348         (dump_gimple_omp_task): Likewise.
7349         (dump_gimple_omp_atomic_load): Likewise.
7350         (dump_gimple_omp_atomic_store): Likewise.
7351         (dump_gimple_mem_ops): Likewise.
7352         (pp_gimple_stmt_1): Likewise.
7353         (pp_cfg_jump): Likewise.
7354         (dump_implicit_edges): Likewise.
7355         (gimple_dump_bb_for_graph): Likewise.
7356         * graph.c (draw_cfg_node): Likewise.
7357         * langhooks.c (lhd_print_error_function): Likewise.
7358         * sched-vis.c (print_exp): Likewise.
7359         (print_value): Likewise.
7360         (print_pattern): Likewise.
7361         (print_insn): Likewise.
7362         (rtl_dump_bb_for_graph): Likewise.
7363         * tree-pretty-print.c (dump_function_declaration): Likewise.
7364         (dump_array_domain): Likewise.
7365         (dump_omp_clause): Likewise.
7366         (dump_location): Likewise.
7367         (dump_generic_node): Likewise.
7368         (print_struct_decl): Likewise.
7369         * diagnostic.c (diagnostic_show_locus): Use pp_space.
7371 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7373         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
7374         candidate table when replacing a candidate statement.
7375         (replace_rhs_if_not_dup): Likewise.
7376         (replace_one_candidate): Likewise.
7378 2013-08-02  Jan Hubicka  <jh@suse.cz>
7379             Martin Liska  <marxin.liska@gmail.com>
7381         * cgraphunit.c (add_new_function): Fix logic when adding from
7382         late IPA pass.
7383         (assemble_thunk): Rename to ...
7384         (expand_thunk); .. this one; export; get it working with
7385         general functions; make produced gimple valid.
7386         * cgraph.h (expand_thunk): Declare.
7388 2013-08-02  Jan Hubicka  <jh@suse.cz>
7390         * ipa-cp.c (gather_context_independent_values): Use
7391         ipa_get_param_move_cost.
7392         (get_replacement_map): Remove PARAM; move parameter folding
7393         into tree-inline.c
7394         (create_specialized_node): Update.
7395         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
7396         assert that we have gimple body; update move_cost.
7397         (count_formal_params): Assert that we have gimple body.
7398         (ipa_dump_param): New function.
7399         (ipa_alloc_node_params): Break out from ...
7400         (ipa_initialize_node_params): ... here.
7401         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
7402         (ipa_write_node_info): Stream move costs.
7403         (ipa_read_node_info): Read move costs.
7404         (ipa_update_after_lto_read): Do not recompute node params.
7405         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
7406         (ipa_get_param): Check we are not in WPA.
7407         (ipa_get_param_move_cost): New.
7408         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
7409         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
7410         parm numbers to be present.
7412 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
7414         PR rtl-optimization/58048
7415         * lra-constraints.c (process_alt_operands): Don't check asm
7416         operand on register.
7418 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
7420         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
7421         the implied StoreLoad barrier for atomic operations if before.
7423 2013-08-02  Jan Hubicka  <jh@suse.cz>
7424             Martin Liska  <marxin.liska@gmail.com>
7426         * cgraph.c (cgraph_function_body_availability): Do not check
7427         cgraph flags.
7428         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
7429         symtab_node_availability): Declare.
7430         * ipa.c (can_replace_by_local_alias): New.
7431         (function_and_variable_visibility): Use it.
7432         * symtab.c (symtab_for_node_and_aliases,
7433         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
7435 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
7437         PR rtl-optimization/57963
7438         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
7439         (lra_constraints): Use them.
7441 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
7443         * config/arm/types.md (define_attr "type"): Add "load_acq"
7444         and "store_rel".
7445         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
7446         changes.
7447         (cortex_a53_store1): Likewise.
7449 2013-08-01  Jan Hubicka  <jh@suse.cz>
7451         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
7452         partitions are not needed.
7454 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
7456         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
7457         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
7458         MAYBE_NON_Q_CLASS_P where appropriate.
7460 2013-08-01  Jan Hubicka  <jh@suse.cz>
7462         * cgraph.h (release_function_body): Declare.
7463         * tree.c (free_lang_data_in_decl): Free, parameters and return values
7464         of unused delcarations.
7466 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7468         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
7469         RTL form when subtracting a constant.
7471 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7473         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
7474         Generate canonical plus rtx with negated immediate instead of minus
7475         where appropriate.
7476         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
7478 2013-08-01  Jan Hubicka  <jh@suse.cz>
7480         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
7481         (cgraph_release_function_body): Likewise.
7482         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
7483         * cgraph.h (cgrpah_node): Rename abstract_and_needed
7484         to used_as_abstract_origin.
7485         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
7486         symbols used as abstract origins.
7487         * cgraphunit.c (analyze_functions): Update.
7488         * ipa.c (symtab_remove_unreachable_nodes): Recompute
7489         used_as_abstract_origin.
7490         * tree-inline.c (tree_function_versioning): Update
7491         used_as_abstract_origin; be ready for DECL_RESULT and
7492         DECL_ARGUMENTS to be NULL.
7494         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
7495         for abstract functions.
7496         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
7497         real symbols.
7499 2013-08-01  Jan Hubicka  <jh@suse.cz>
7501         * profile.c (compute_value_histograms): Fix thinko.
7503 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
7505         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
7506         aarch-common-protos.h to extra_headers.
7507         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
7508         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
7509         * config/aarch64/t-aarch64 (aarch-common.o): Define.
7511 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
7513         * config/aarch64/aarch64.md (define_attr "type"): Delete.
7514         Include "../arm/types.md".  Define "type" attribute for all patterns.
7515         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
7516         attribute changes.
7518 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
7520         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
7521         to support power8 load fusion.
7522         (fusion_gpr_mem_load): Likewise.
7524         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
7526         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
7527         declarations for power8 load fusion.
7528         (emit_fusion_gpr_load): Likewise.
7530         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
7531         tuning for power8, turn on fusion mode by default.  Turn on sign
7532         extending fusion mode if normal fusion mode is on, and we are at
7533         -O2 or -O3.
7534         (fusion_gpr_load_p): New function, return true if we can fuse an
7535         addis instruction with a dependent load to a GPR.
7536         (emit_fusion_gpr_load): Emit the instructions for power8 load
7537         fusion to GPRs.
7539         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
7540         (VSX load fusion peepholes): New peepholes to fuse together an
7541         addi instruction with a VSX load instruction.
7543         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
7544         peepholes to fuse an addis instruction with a load to a GPR base
7545         register.  If we are supporting sign extending fusions, convert
7546         sign extending loads to zero extending loads and add an explicit
7547         sign extension.
7549 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
7551         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
7552         aarch-common-protos.h to extra_headers.
7553         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
7554         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
7555         (arm_early_store_addr_dep): Likewise.
7556         (arm_no_early_alu_shift_dep): Likewise.
7557         (arm_no_early_alu_shift_value_dep): Likewise.
7558         (arm_no_early_mul_dep): Likewise.
7559         (arm_no_early_store_addr_dep): Likewise.
7560         (arm_mac_accumulator_is_mul_result): Likewise.
7561         (arm_mac_accumulator_is_result): Likewise.
7562         * config/arm/aarch-common.c: ... here.  New file.
7563         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
7564         here to ...
7565         (arm_early_store_addr_dep): Likewise.
7566         (arm_no_early_alu_shift_dep): Likewise.
7567         (arm_no_early_alu_shift_value_dep): Likewise.
7568         (arm_no_early_mul_dep): Likewise.
7569         (arm_no_early_store_addr_dep): Likewise.
7570         (arm_mac_accumulator_is_mul_result): Likewise.
7571         (arm_mac_accumulator_is_result): Likewise.
7572         * config/arm/aarch-common-protos.h: ... here.  New file.
7573         * config/arm/t-arm (aarch-common.o): Define.
7575 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
7577         * config/arm/arm.md: Include new file "types.md".
7578         (define_attr "type"): Move from here to ...
7579         (define_attr "mul32"): Likewise.
7580         (define_attr "mul64"): Likewise.
7581         * config/arm/types.md: ... here.  New file.
7583 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7585         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
7586         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
7588 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7590         * gen-pass-instances.awk: Fix offset of substr().
7592 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
7594         * Makefile.in (pass-instances.def): New.
7595         (passes.o): Replace dependency on passes.def with one on
7596         pass-instances.def
7598         * gen-pass-instances.awk: New.
7600         * passes.c (pass_manager::pass_manager): Use pass-instances.def
7601         rather than passes.def, updating local definition of NEXT_PASS
7602         macro to add an extra NUM parameter (currently unused).
7604 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
7606         * Makefile.in (PASS_MANAGER_H): New.
7607         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
7608         (passes.o): Likewise.
7609         (statistics.o): Likewise.
7610         (cgraphunit.o): Likewise.
7611         (context.o): Depend on PASS_MANAGER_H.
7613         * pass_manager.h: New.
7615         * cgraphunit.c (cgraph_add_new_function): Update for moves
7616         of globals to fields of pass_manager.
7617         (analyze_function): Likewise.
7618         (expand_function): Likewise.
7619         (ipa_passes): Likewise.
7620         (compile): Likewise.
7622         * context.c (context::context): New.
7623         * context.h  (context::context): New.
7624         (context::get_passes): New.
7625         (context::passes_): New.
7627         * lto-cgraph.c (input_node): Update for moves of globals to
7628         fields of pass_manager.
7630         * passes.c (all_passes): Remove, in favor of a field of the
7631         same name within the new class pass_manager.
7632         (all_small_ipa_passes): Likewise.
7633         (all_lowering_passes): Likewise.
7634         (all_regular_ipa_passes): Likewise.
7635         (all_late_ipa_passes): Likewise.
7636         (all_lto_gen_passes): Likewise.
7637         (passes_by_id): Likewise.
7638         (passes_by_id_size): Likewise.
7639         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7640         the new class pass_manager.
7641         (set_pass_for_id): Convert to...
7642         (pass_manager::set_pass_for_id): ...method.
7643         (get_pass_for_id): Convert to...
7644         (pass_manager::get_pass_for_id): ...method.
7645         (register_one_dump_file): Move body of implementation into...
7646         (pass_manager::register_one_dump_file): ...here.
7647         (register_dump_files_1): Convert to...
7648         (pass_manager::register_dump_files_1): ...method.
7649         (register_dump_files): Convert to...
7650         (pass_manager::register_dump_files): ...method.
7651         (create_pass_tab): Update for moves of globals to fields of
7652         pass_manager.
7653         (dump_passes): Move body of implementation into...
7654         (pass_manager::dump_passes): ...here.
7655         (register_pass): Move body of implementation into...
7656         (pass_manager::register_pass): ...here.
7657         (init_optimization_passes): Convert into...
7658         (pass_manager::pass_manager): ...constructor for new
7659         pass_manager class, and initialize the pass_lists array.
7660         (check_profile_consistency): Update for moves of globals to
7661         fields of pass_manager.
7662         (dump_profile_report): Move body of implementation into...
7663         (pass_manager::dump_profile_report): ...here.
7664         (ipa_write_summaries_1): Update for moves of pass lists from
7665         being globals to fields of pass_manager.
7666         (ipa_write_optimization_summaries): Likewise.
7667         (ipa_read_summaries):  Likewise.
7668         (ipa_read_optimization_summaries): Likewise.
7669         (execute_all_ipa_stmt_fixups): Likewise.
7671         * statistics.c (statistics_fini): Update for moves of globals to
7672         fields of pass_manager.
7674         * toplev.c (general_init): Replace call to
7675         init_optimization_passes with construction of the pass_manager
7676         instance.
7678         * tree-pass.h (all_passes): Remove, in favor of a field of the
7679         same name within the new class pass_manager.
7680         (all_small_ipa_passes): Likewise.
7681         (all_lowering_passes): Likewise.
7682         (all_regular_ipa_passes): Likewise.
7683         (all_lto_gen_passes): Likewise.
7684         (all_late_ipa_passes): Likewise.
7685         (passes_by_id): Likewise.
7686         (passes_by_id_size): Likewise.
7687         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7688         the new class pass_manager.
7689         (get_pass_for_id): Remove.
7691 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
7693         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
7694         configs.
7696 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
7698         * arm.md (mulhi3): New expand pattern.
7700 2013-07-30  Jan Hubicka  <jh@suse.cz>
7701             Martin Liska  <marxin.liska@gmail.com>
7703         * profile.c (compute_value_histograms): Do not ICE when
7704         there is mismatch only on some counters.
7706 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7708         PR rtl-optimization/57637
7709         * function.c (move_insn_for_shrink_wrap): Also check the
7710         GEN set of the LIVE problem for the liveness analysis
7711         if it exists, otherwise give up.
7713 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7715         PR tree-optimization/57993
7716         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
7717         replaced statement in the candidate table.
7718         (phi_add_costs): Return infinite cost when the hidden basis does
7719         not dominate all phis on which the candidate is dependent.
7720         (replace_one_candidate): Record replaced statement in the
7721         candidate table.
7723 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
7725         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
7726         (ashlv2si3): New expander.
7727         (*ashlv2si3_i): New define_insn_and_split.
7728         * predicates.md (float_operation): Allow patterns with three
7729         basic sub-patterns.
7731         PR rtl-optimization/58021
7732         * mode-switching.c (create_pre_exit): Always split off preceding
7733         insns if we are not at the basic block head.
7735 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
7737         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
7738         (UCLIBC_DYNAMIC_LINKER): New macro.
7739         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
7740         `-mnan=2008'.
7741         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
7742         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
7743         `-mnan=2008'.
7744         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
7745         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
7746         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
7747         for SF and DF modes.  Use ieee_quad_format for TF mode.
7748         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
7749         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
7750         (mips_option_override): Handle `-mnan=legacy'.
7751         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
7752         `-mabs=2008' and `-mnan=2008'.
7753         (OPTION_DEFAULT_SPECS): Add "nan" default.
7754         (ASM_SPEC): Handle `-mnan='.
7755         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
7756         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
7757         comment accordingly.
7758         (neg<mode>2): Likewise.
7759         * config/mips/mips.opt (mabs, mnan): New options.
7760         * doc/install.texi (Configuration): Document `--with-nan=' option.
7761         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
7762         `-mnan=' options.
7763         (MIPS Options): Document them.
7764         * config.gcc <mips*-*-*>: Handle `--with-nan='.
7765         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
7766         * configure: Regenerate.
7767         * config.in: Regenerate.
7769 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
7771         * config/i386/i386.md (float post-reload splitters): Do not check
7772         for subregs of SSE registers.
7774 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
7775             H.J. Lu  <hongjiu.lu@intel.com>
7777         PR target/57954
7778         PR target/57988
7779         * config/i386/i386.md (post-reload splitter
7780         to avoid partial SSE reg dependency stalls): New pattern.
7782 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
7784         * config/s390/s390.md ("movcc"): Swap load and store instructions.
7786 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
7788         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
7789         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
7791 2013-07-26  Cary Coutant  <ccoutant@google.com>
7793         * dwarf2out.c (die_checksum_ordered): Don't include template
7794         instantiations in signature.
7795         (is_template_parameter): New function.
7796         (is_template_instantiation): New function.
7797         (generate_skeleton_bottom_up): Don't include template instantiations
7798         in type unit DIE.
7799         (generate_skeleton): Likewise.
7800         (break_out_comdat_types): Move recursive call to break out nested
7801         types earlier.
7802         (prune_unused_types_mark_generic_parms_dies): Call
7803         is_template_parameter.
7805 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
7807         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
7808         uses vector registers.
7809         * config/aarch64/iterators.md: Add attributes rtn and vas.
7811 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7812             Richard Earnshaw  <richard.earnshaw@arm.com>
7814         * combine.c (simplify_comparison): Re-canonicalize operands
7815         where appropriate.
7816         * config/arm/arm.md (movcond_addsi): New splitter.
7818 2013-07-25  Sterling Augustine  <saugustine@google.com>
7820         * dwarf2out.c (size_of_pubnames): Move code to...
7821         (include_pubname_in_output): ...here.  New.
7822         (want_pubnames): Rearrange.
7823         (output_pubnames): Call include_pubname_in_output.  Move assertion.
7825 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7827         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
7829 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7831         PR target/38836
7832         * doc/extend.texi: Remove obsolete builtins. Fix
7833         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
7835 2013-07-25  Jan Hubicka  <jh@suse.cz>
7837         * cgraph.c (release_function_body): Break out from ...
7838         (cgraph_release_function_body): ... this one; also release DECL_RESULT
7839         and DECL_ARGUMENTS.
7840         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
7841         old_tree in the map.
7842         (create_specialized_node): Update.
7843         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
7844         into index.
7845         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
7846         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
7847         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
7848         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
7849         DECL_RESULT.
7851 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7853         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
7854         addsi3_carryin_alt2_<optab>): Correct output template.
7856 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7858         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
7859         Adjust for arm_restrict_it.
7860         Remove trailing whitespace.
7862 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
7864         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
7865         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
7867         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
7869 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
7871         PR rtl-optimization/57960
7872         * lra-constraints.c (process_alt_operands): Use the right mode
7873         when checking strict_low.
7875 2013-07-25  Jan Hubicka  <jh@suse.cz>
7877         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
7878         * cgraph.c (cgraph_remove_node): Do not release function body
7879         when in cgraph streaming.
7880         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
7881         in other partitions are not considered reachable; fix handling of
7882         clones.
7884 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7886         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
7888 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7890         PR target/19599
7891         PR target/57731
7892         PR target/57837
7893         * config/arm/arm.md ("*sibcall_insn): Replace use of
7894         Ss with US. Adjust output for v5 and v4t.
7895         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
7897         * config/arm/constraints.md ("Ss"): Rename to US.
7899 2013-07-25  Terry Guo  <terry.guo@arm.com>
7901         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
7902         shift_add/shift_sub0/shift_sub1 RTXs.
7904 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
7905             Anton Blanchard  <anton@au1.ibm.com>
7907         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
7908         (altivec_vpks<VI_char>ss): Likewise.
7909         (altivec_vpks<VI_char>us): Likewise.
7910         (altivec_vpku<VI_char>us): Likewise.
7911         (altivec_vpku<VI_char>um): Likewise.
7913 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
7915         Introduce context class.
7917         * Makefile.in (CONTEXT_H): New.
7918         (OBJS): Add context.o.
7919         (toplev.o): Add CONTEXT_H to dependencies.
7920         (context.o): New.
7922         * toplev.c (general_init): Create the singleton gcc::context instance.
7924         * context.c: New.
7926         * context.h: New.
7928 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
7930         PR rtl-optimization/57968
7931         * mode-switching.c (create_pre_exit): Allow instructions that
7932         don't set a return register to need a non-exit mode.
7934 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7935             Anton Blanchard  <anton@au1.ibm.com>
7937         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
7938         operands to vperm for little endian.
7939         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
7940         of lvsl to create the control mask for a vperm for little endian.
7942 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7943             Anton Blanchard  <anton@au1.ibm.com>
7945         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
7946         two operands for little-endian.
7948 2013-07-23  Steve Ellcey  <sellcey@mips.com>
7950         * config/mips/mips.c (mips_case_values_threshold): New.
7951         (TARGET_CASE_VALUES_THRESHOLD): Define.
7953 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7954             Anton Blanchard  <anton@au1.ibm.com>
7956         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
7957         selection of field for vector splat in little endian mode.
7959 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7961         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
7962         expanders to rs6000.md.
7963         (ior<mode>3): Likewise.
7964         (and<mode>3): Likewise.
7965         (one_cmpl<mode>2): Likewise.
7966         (nor<mode>3): Likewise.
7967         (andc<mode>3): Likewise.
7968         (eqv<mode>3): Likewise.
7969         (nand<mode>3): Likewise.
7970         (orc<mode>3): Likewise.
7972         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
7973         declaration.
7975         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
7976         to split multi-word logical operations.
7977         (rs6000_split_logical_di): Likewise.
7978         (rs6000_split_logical): Likewise.
7980         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
7981         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
7982         and allow TImode operations in 32-bit.
7983         (vsx_and<mode>3_64bit): Likewise.
7984         (vsx_ior<mode>3_32bit): Likewise.
7985         (vsx_ior<mode>3_64bit): Likewise.
7986         (vsx_xor<mode>3_32bit): Likewise.
7987         (vsx_xor<mode>3_64bit): Likewise.
7988         (vsx_one_cmpl<mode>2_32bit): Likewise.
7989         (vsx_one_cmpl<mode>2_64bit): Likewise.
7990         (vsx_nor<mode>3_32bit): Likewise.
7991         (vsx_nor<mode>3_64bit): Likewise.
7992         (vsx_andc<mode>3_32bit): Likewise.
7993         (vsx_andc<mode>3_64bit): Likewise.
7994         (vsx_eqv<mode>3_32bit): Likewise.
7995         (vsx_eqv<mode>3_64bit): Likewise.
7996         (vsx_nand<mode>3_32bit): Likewise.
7997         (vsx_nand<mode>3_64bit): Likewise.
7998         (vsx_orc<mode>3_32bit): Likewise.
7999         (vsx_orc<mode>3_64bit): Likewise.
8001         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
8002         logical types in GPRs.
8004         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
8005         logical insns to rs6000.md, and allow TImode operations in
8006         32-bit.
8007         (altivec_ior<mode>3): Likewise.
8008         (altivec_xor<mode>3): Likewise.
8009         (altivec_one_cmpl<mode>2): Likewise.
8010         (altivec_nor<mode>3): Likewise.
8011         (altivec_andc<mode>3): Likewise.
8013         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
8014         attributes for moving the 128-bit logical operations into
8015         rs6000.md.
8016         (BOOL_REGS_OUTPUT): Likewise.
8017         (BOOL_REGS_OP1): Likewise.
8018         (BOOL_REGS_OP2): Likewise.
8019         (BOOL_REGS_UNARY): Likewise.
8020         (BOOL_REGS_AND_CR0): Likewise.
8021         (one_cmpl<mode>2): Add support for DI logical operations on
8022         32-bit, splitting the operations to 32-bit.
8023         (anddi3): Likewise.
8024         (iordi3): Likewise.
8025         (xordi3): Likewise.
8026         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
8027         changes to combine the 32/64-bit code, allow logical operations on
8028         TI mode in 32-bit, and to use similar match_operator patterns like
8029         scalar mode uses.  Combine the Altivec and VSX code for logical
8030         operations, and move it here.
8031         (ior<mode>3, 128-bit types): Likewise.
8032         (xor<mode>3, 128-bit types): Likewise.
8033         (one_cmpl<mode>3, 128-bit types): Likewise.
8034         (nor<mode>3, 128-bit types): Likewise.
8035         (andc<mode>3, 128-bit types): Likewise.
8036         (eqv<mode>3, 128-bit types): Likewise.
8037         (nand<mode>3, 128-bit types): Likewise.
8038         (orc<mode>3, 128-bit types): Likewise.
8039         (and<mode>3_internal): Likewise.
8040         (bool<mode>3_internal): Likewise.
8041         (boolc<mode>3_internal1): Likewise.
8042         (boolc<mode>3_internal2): Likewise.
8043         (boolcc<mode>3_internal1): Likewise.
8044         (boolcc<mode>3_internal2): Likewise.
8045         (eqv<mode>3_internal1): Likewise.
8046         (eqv<mode>3_internal2): Likewise.
8047         (one_cmpl1<mode>3_internal): Likewise.
8049 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
8051         * config/microblaze/microblaze.c (microblaze_expand_prologue):
8052         Rename flag_stack_usage to flag_stack_usage_info.
8054 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
8056         * config/microblaze/sync.md: New file.
8057         * config/microblaze/microblaze.md: Include sync.md
8058         * config/microblaze/microblaze.c: Add print_operand 'y'.
8059         * config/microblaze/constraints.md: Add memory_contraint
8060         'Q' which is a single register.
8062 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
8064         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
8066 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
8068         * reload.c (find_reloads): Exit loop once we find this operand
8069         cannot be reloaded somehow for this alternative.
8071         * reload.c (find_reloads): Exit loop once we find a hard register.
8073         * rtlanal.c (computed_jump_p): Exit loop once we find label
8074         reference is used.
8076         * i386.c (ix86_pad_returns): Exit loop after setting replace.
8078         * cfgloopmanip.c (remove_path): Exit loop after setting
8079         irred_invalidated.
8081         * gensupport.c (subst_dup): Avoid loop if code is not
8082         MATCH_DUP nor MATCH_OP_DUP.
8084 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
8086         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
8088 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8090         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
8091         true for SP_REGNUM if mode == ptr_mode.
8092         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
8093         with value R0_REGNUM + 31.
8095 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8097         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
8098         pad pointer-typed argument downward.
8100 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8102         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
8103         and __ILP32__ when the ILP32 model is in use.
8105 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8107         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
8108         (aarch64_load_symref_appropriately): In the case of
8109         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
8110         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
8111         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
8112         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
8113         if 'mode' doesn't equal to 'ptr_mode'.
8114         (aarch64_output_mi_thunk): Add an assertion on the alignment of
8115         'vcall_offset'; change to call aarch64_emit_move differently depending
8116         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
8117         to calculate the upper bound of 'vcall_offset'.
8118         (aarch64_cannot_force_const_mem): Change to also return true if
8119         mode != ptr_mode.
8120         (aarch64_legitimize_reload_address): In the case of large
8121         displacements, add new local variable 'xmode' and an assertion
8122         based on it; change to use 'xmode' to generate the new rtx and
8123         reload.
8124         (aarch64_asm_trampoline_template): Change to generate the template
8125         differently depending on TARGET_ILP32 or not; change to use
8126         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
8127         (aarch64_trampoline_size): Removed.
8128         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
8129         and replace immediate literals with it.  Change to use 'ptr_mode'
8130         instead of 'DImode' and call convert_memory_address if the mode
8131         of 'fnaddr' doesn't equal to 'ptr_mode'.
8132         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
8133         to output symbol.
8134         (aarch64_elf_asm_destructor): Likewise.
8135         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
8136         on TARGET_ILP32 instead of aarch64_trampoline_size.
8137         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
8138         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
8139         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
8140         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
8141         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
8142         (storewb_pair<GPI:mode>_<P:mode>): ... this.
8143         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
8144         depending on the value of 'mode'.
8145         (add_losym_<mode>): New.
8146         (ldr_got_small_<mode>): New, based on ldr_got_small.
8147         (ldr_got_small): Remove.
8148         (ldr_got_small_sidi): New.
8149         * config/aarch64/iterators.md (P): New.
8150         (PTR): Change to 'ptr_mode' in the condition.
8152 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
8154         * config.gcc (aarch64*-*-*): Support --with-abi.
8155         (aarch64*-*-elf): Support --with-multilib-list.
8156         (aarch64*-*-linux*): Likewise.
8157         (supported_defaults): Add abi to aarch64*-*-*.
8158         * configure.ac: Mention AArch64 for --with-multilib-list.
8159         * configure: Re-generated.
8160         * config/aarch64/biarchilp32.h: New file.
8161         * config/aarch64/biarchlp64.h: New file.
8162         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
8163         (ABI_SPEC): Ditto.
8164         (MULTILIB_DEFAULTS): Ditto.
8165         (DRIVER_SELF_SPECS): Ditto.
8166         (ASM_SPEC): Update to also substitute -mabi.
8167         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
8168         file whose name depends on -mabi= and -mbig-endian.
8169         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
8170         TARGET_ILP32.
8171         (POINTER_SIZE): New define.
8172         (POINTERS_EXTEND_UNSIGNED): Ditto.
8173         (enum aarch64_abi_type): New enumeration tag.
8174         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
8175         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
8176         (TARGET_ILP32): New define.
8177         * config/aarch64/aarch64.opt (mabi): New.
8178         (aarch64_abi): New.
8179         (ilp32, lp64): New values for -mabi.
8180         * config/aarch64/t-aarch64 (comma): New define.
8181         (MULTILIB_OPTIONS): Ditto.
8182         (MULTILIB_DIRNAMES): Ditto.
8183         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
8184         * doc/invoke.texi: Document -mabi for AArch64.
8186 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
8188         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
8190 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
8191             Anton Blanchard  <anton@au1.ibm.com>
8193         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
8194         endianness when selecting field to splat.
8196 2013-07-22  Eric Christopher  <echristo@gmail.com>
8198         * dwarf2out.c (die_odr_checksum): New function to use
8199         CHECKSUM_ macros and ULEB128 for DIE tag.
8200         (generate_type_signature): Use.
8202 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
8204         * config.gcc (sparc*-*-*): Accept leon3 processor.
8205         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
8206         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
8207         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
8208         * config/sparc/sparc.opt (enum processor_type): Add leon3.
8209         (mfix-ut699): Adjust comment.
8210         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
8211         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
8212         (CPP_CPU_SPEC): Likewise.
8213         (ASM_CPU_SPEC): Likewise.
8214         * config/sparc/sparc.c (leon3_cost): New constant.
8215         (sparc_option_override): Add leon3 support.
8216         (mem_ref): New function.
8217         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
8218         (sparc_do_work_around_errata): Look into the instruction in the delay
8219         slot and adjust accordingly.  Add fix for the data cache nullify issues
8220         of the UT699.  Change insertion position for the NOP.
8221         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
8222         (leon3_load): New reservation.
8223         (leon_store): Bump latency to 2.
8224         (grfpu): New automaton.
8225         (grfpu_alu): New unit.
8226         (grfpu_ds): Likewise.
8227         (leon_fp_alu): Adjust.
8228         (leon_fp_mult): Delete.
8229         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
8230         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
8231         * config/sparc/sparc.md (cpu): Add leon3.
8232         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
8233         (swapsi): Likewise.
8234         (atomic_test_and_set): Likewise.
8235         (ldstub): Likewise.
8237 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
8239         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
8240         default for R5900 targets.
8241         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
8242         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
8243         * config/mips/mips.c (mips_option_override): Report an error for
8244         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
8245         for -march=r5900 -mhard-float.
8247 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
8249         * df-problems.c (can_move_insns_across): Exit loop once we
8250         find a non-fixed, non-global register.
8252         * ipa-pure-const.c (propagate_nothrow): Exit loop after
8253         setting can_throw.
8255         * omega.c (omega_eliminate_red): Break after setting red_found.
8256         (omega_problem_has_red_equations): Similarly after setting found.
8257         (omega_query_variable): Similarly after setting coupled.
8259 2013-07-22  Marek Polacek  <polacek@redhat.com>
8261         * gimplify.c: Don't include gimple.h twice.
8263 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8265         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
8266         instead of TARGET_THUMB1.
8267         (Pz): New constraint.
8268         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
8269         encodings.
8270         (compare_negsi_si): Likewise.
8271         (compare_addsi2_op0): Likewise.
8272         (compare_addsi2_op1): Likewise.
8273         (addsi3_carryin_<optab>): Likewise.
8274         (addsi3_carryin_alt2_<optab>): Likewise.
8275         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
8276         for arm_restrict_it.
8277         (subsi3_carryin): Likewise.
8278         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
8279         (minmax_arithsi): Disable for arm_restrict_it.
8280         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
8281         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
8282         (satsi_<SAT:code>_shift): Likewise.
8283         (arm_shiftsi3): Add alternative for 16-bit encoding.
8284         (arm32_movhf): Disable for arm_restrict_it.
8285         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
8286         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
8288 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
8290         * config/arm/arm.md (attribute "insn"): Delete.
8291         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
8292         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
8293         (not_shiftsi): Update for attribute change.
8294         (not_shiftsi_compare0): Likewise.
8295         (not_shiftsi_compare0_scratch): Likewise.
8296         (arm_one_cmplsi2): Likewise.
8297         (thumb1_one_cmplsi2): Likewise.
8298         (notsi_compare0): Likewise.
8299         (notsi_compare0_scratch): Likewise.
8300         (thumb1_movdi_insn): Likewise.
8301         (arm_movsi_insn): Likewise.
8302         (movhi_insn_arch4): Likewise.
8303         (movhi_bytes): Likewise.
8304         (arm_movqi_insn): Likewise.
8305         (thumb1_movqi_insn): Likewise.
8306         (arm32_movhf): Likewise.
8307         (thumb1_movhf): Likewise.
8308         (arm_movsf_soft_insn): Likewise.
8309         (thumb1_movsf_insn): Likewise.
8310         (thumb_movdf_insn): Likewise.
8311         (movsicc_insn): Likewise.
8312         (movsfcc_soft_insn): Likewise.
8313         (and_scc): Likewise.
8314         (cond_move): Likewise.
8315         (if_move_not): Likewise.
8316         (if_not_move): Likewise.
8317         (if_shift_move): Likewise.
8318         (if_move_shift): Likewise.
8319         (if_shift_shift): Likewise.
8320         (if_not_arith): Likewise.
8321         (if_arith_not): Likewise.
8322         (cond_move_not): Likewise.
8323         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
8324         (neon_mov<mode>): Likewise.
8325         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
8326         (thumb2_movsi_vfp): Likewise.
8327         (movsf_vfp): Likewise.
8328         (thumb2_movsf_vfp): Likewise.
8329         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8330         change.
8331         (cortexa7_older_only): Likewise.
8332         (cortexa7_younger): Likewise.
8333         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
8334         (1020alu_shift_op): Likewise.
8335         (1020alu_shift_reg_op): Likewise.
8336         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
8337         (alu_shift_op): Likewise.
8338         (alu_shift_reg_op): Likewise.
8339         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
8340         (11_alu_shift_op): Likewise.
8341         (11_alu_shift_reg_op): Likewise.
8342         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
8343         (9_alu_shift_reg_op): Likewise.
8344         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
8345         change.
8346         (cortex_a15_alu_shift): Likewise.
8347         (cortex_a15_alu_shift_reg): Likewise.
8348         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
8349         (cortex_a5_alu_shift): Likewise.
8350         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
8351         change.
8352         (cortex_a53_alu_shift): Likewise.
8353         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
8354         change.
8355         (cortex_a7_alu_reg): Likewise.
8356         (cortex_a7_alu_shift): Likewise.
8357         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
8358         (cortex_a8_alu_shift): Likewise.
8359         (cortex_a8_alu_shift_reg): Likewise.
8360         (cortex_a8_mov): Likewise.
8361         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
8362         (cortex_a9_dp_shift): Likewise.
8363         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
8364         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
8365         (cortex_r4_mov): Likewise.
8366         (cortex_r4_alu_shift): Likewise.
8367         (cortex_r4_alu_shift_reg): Likewise.
8368         * config/arm/fa526.md (526_alu_op): Update for attribute change.
8369         (526_alu_shift_op): Likewise.
8370         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
8371         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
8372         (626te_alu_shift_op): Likewise.
8373         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
8374         (726te_alu_op): Likewise.
8375         (726te_alu_shift_op): Likewise.
8376         (726te_alu_shift_reg_op): Likewise.
8377         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
8378         (mp626_alu_shift_op): Likewise.
8379         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
8380         (pj4_alu_e1_conds): Likewise.
8381         (pj4_alu): Likewise.
8382         (pj4_alu_conds): Likewise.
8383         (pj4_shift): Likewise.
8384         (pj4_shift_conds): Likewise.
8385         (pj4_alu_shift): Likewise.
8386         (pj4_alu_shift_conds): Likewise.
8388 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8390         * config/arm/predicates.md (shiftable_operator_strict_it):
8391         New predicate.
8392         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
8393         Disable cond_exec version for arm_restrict_it.
8394         (thumb2_smaxsi3): Convert to generate cond_exec.
8395         (thumb2_sminsi3): Likewise.
8396         (thumb32_umaxsi3): Likewise.
8397         (thumb2_uminsi3): Likewise.
8398         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
8399         (thumb2_neg_abssi2): Likewise.
8400         (thumb2_mov_scc): Add alternative for 16-bit encoding.
8401         (thumb2_movsicc_insn): Adjust alternatives.
8402         (thumb2_mov_negscc): Disable for arm_restrict_it.
8403         (thumb2_mov_negscc_strict_it): New pattern.
8404         (thumb2_mov_notscc_strict_it): New pattern.
8405         (thumb2_mov_notscc): Disable for arm_restrict_it.
8406         (thumb2_ior_scc): Likewise.
8407         (thumb2_ior_scc_strict_it): New pattern.
8408         (thumb2_cond_move): Adjust for arm_restrict_it.
8409         (thumb2_cond_arith): Disable for arm_restrict_it.
8410         (thumb2_cond_arith_strict_it): New pattern.
8411         (thumb2_cond_sub): Adjust for arm_restrict_it.
8412         (thumb2_movcond): Likewise.
8413         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
8414         (thumb2_zero_extendhisi2_v6): Likewise.
8415         (thumb2_zero_extendqisi2_v6): Likewise.
8416         (orsi_notsi_si): Likewise.
8417         (orsi_not_shiftsi_si): Likewise.
8419 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
8421         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
8422         instruction sequence is 1 byte shorter.
8424 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
8426         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
8427         it is not needed after split.
8429 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
8431         PR target/51784
8432         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
8433         second label for nonlocal goto receivers. Don't output pic base labels
8434         unless we're producing PIC; mark that action unreachable().
8435         (ix86_save_reg): If the function contains a nonlocal label, save the
8436         PIC base reg.
8437         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
8438         * config/darwin.c (emitted_pic_label_num): New GTY.
8439         (update_pic_label_number_if_needed): New.
8440         (machopic_output_function_base_name): Adjust for nonlocal receiver
8441         case.
8442         (machopic_should_output_picbase_label): New.
8443         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
8444         (nonlocal_goto_receiver): New insn and split.
8446 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
8448         * config/aarch64/aarch64-builtins.c
8449         (aarch64_fold_builtin): Fold abs in all modes.
8450         * config/aarch64/aarch64-simd-builtins.def
8451         (abs): Enable for all modes.
8452         * config/aarch64/arm_neon.h
8453         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
8454         (vabs_f64): Add missing intrinsic.
8456 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
8458         * config/aarch64/arm_neon.h (vabs_s64): New function
8460 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
8462         PR target/57516
8463         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
8464         * config/avr/avr.md (adjust_len): Add `round'.
8465         * config/avr/avr-protos.h (avr_out_round): New prototype.
8466         (avr_out_plus): Add `out_label' argument.
8467         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
8468         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
8469         Handle the case where `insn' is just a pattern.
8470         (avr_out_bitop): Handle the case where `insn' is just a pattern.
8471         (avr_out_round): New function.
8472         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
8474 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
8476         * config/microblaze/microblaze.c (microblaze_expand_prologue):
8477         Add check for flag_stack_usage to handle -fstack-usage support
8479 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
8481         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
8482         interaction for new Power8 flags and VSX.
8484 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
8486         PR middle-end/57698
8487         * tree-inline.c (expand_call_inline): Emit errors during
8488         early_inlining only if optimization is not turned on.
8490 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8492         * passes.def: New.
8494         * passes.c (init_optimization_passes): Move the construction of
8495         the pass hierarchy into a new passes.def file.
8497         * Makefile.in (passes.o): Add dependency on passes.def.
8499 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8501         * passes.c (init_optimization_passes): Introduce macros for
8502         constructing the tree of passes (INSERT_PASSES_AFTER,
8503         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
8504         TERMINATE_PASS_LIST).
8506 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
8507             Wei Mi  <wmi@google.com>
8509         PR rtl-optimization/57878
8510         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
8511         top.
8512         (reload_pseudo_compare_func): Check nregs first for reload
8513         pseudos.
8515 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8517         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
8519 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
8521         * read-rtl.c (validate_const_int): Once an invalid character is
8522         seen, quit the loop.
8524         * gengtype.c (write_roots): Similarly once we find the "deletable"
8525         or "if_marked" option.
8527 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
8529         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
8530         "xtab" and "sat".  Move value "clz" from here to ...
8531         (attriubte "type"): ... here.
8532         (satsi_<SAT:code>): Delete "insn" attribute.
8533         (satsi_<SAT:code>_shift): Likewise.
8534         (arm_zero_extendqisi2addsi): Likewise.
8535         (arm_extendqisi2addsi): Likewise.
8536         (clzsi2): Update for attribute changes.
8537         (rbitsi2): Likewise.
8538         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
8539         attribute.
8540         (arm_usatsihi): Likewise.
8541         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
8543 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
8545         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
8546         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
8547         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
8548         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
8549         in alphabetical order.
8550         (attribute "core_cycles"): Update for attribute changes.
8551         (arm_addsi3): Likewise.
8552         (addsi3_compare0): Likewise.
8553         (addsi3_compare0_scratch): Likewise.
8554         (addsi3_compare_op1): Likewise.
8555         (addsi3_compare_op2): Likewise.
8556         (compare_addsi2_op0): Likewise.
8557         (compare_addsi2_op1): Likewise.
8558         (addsi3_carryin_shift_<optab>): Likewise.
8559         (subsi3_carryin_shift): Likewise.
8560         (rsbsi3_carryin_shift): Likewise.
8561         (arm_subsi3_insn): Likewise.
8562         (subsi3_compare0): Likewise.
8563         (subsi3_compare): Likewise.
8564         (arm_andsi3_insn): Likewise.
8565         (thumb1_andsi3_insn): Likewise.
8566         (andsi3_compare0): Likewise.
8567         (andsi3_compare0_scratch): Likewise.
8568         (zeroextractsi_compare0_scratch
8569         (andsi_not_shiftsi_si): Likewise.
8570         (iorsi3_insn): Likewise.
8571         (iorsi3_compare0): Likewise.
8572         (iorsi3_compare0_scratch): Likewise.
8573         (arm_xorsi3): Likewise.
8574         (thumb1_xorsi3_insn): Likewise.
8575         (xorsi3_compare0): Likewise.
8576         (xorsi3_compare0_scratch): Likewise.
8577         (satsi_<SAT:code>_shift): Likewise.
8578         (rrx): Likewise.
8579         (arm_shiftsi3): Likewise.
8580         (shiftsi3_compare0): Likewise.
8581         (not_shiftsi): Likewise.
8582         (not_shiftsi_compare0): Likewise.
8583         (not_shiftsi_compare0_scratch): Likewise.
8584         (arm_one_cmplsi2): Likewise.
8585         (thumb_one_complsi2): Likewise.
8586         (notsi_compare0): Likewise.
8587         (notsi_compare0_scratch): Likewise.
8588         (thumb1_zero_extendhisi2): Likewise.
8589         (arm_zero_extendhisi2): Likewise.
8590         (arm_zero_extendhisi2_v6): Likewise.
8591         (arm_zero_extendhisi2addsi): Likewise.
8592         (thumb1_zero_extendqisi2): Likewise.
8593         (thumb1_zero_extendqisi2_v6): Likewise.
8594         (arm_zero_extendqisi2): Likewise.
8595         (arm_zero_extendqisi2_v6): Likewise.
8596         (arm_zero_extendqisi2addsi): Likewise.
8597         (thumb1_extendhisi2): Likewise.
8598         (arm_extendhisi2): Likewise.
8599         (arm_extendhisi2_v6): Likewise.
8600         (arm_extendqisi): Likewise.
8601         (arm_extendqisi_v6): Likewise.
8602         (arm_extendqisi2addsi): Likewise.
8603         (thumb1_extendqisi2): Likewise.
8604         (thumb1_movdi_insn): Likewise.
8605         (arm_movsi_insn): Likewise.
8606         (movsi_compare0): Likewise.
8607         (movhi_insn_arch4): Likewise.
8608         (movhi_bytes): Likewise.
8609         (arm_movqi_insn): Likewise.
8610         (thumb1_movqi_insn): Likewise.
8611         (arm32_movhf): Likewise.
8612         (thumb1_movhf): Likewise.
8613         (arm_movsf_soft_insn): Likewise.
8614         (thumb1_movsf_insn): Likewise.
8615         (movdf_soft_insn): Likewise.
8616         (thumb_movdf_insn): Likewise.
8617         (arm_cmpsi_insn): Likewise.
8618         (cmpsi_shiftsi): Likewise.
8619         (cmpsi_shiftsi_swp): Likewise.
8620         (arm_cmpsi_negshiftsi_si): Likewise.
8621         (movsicc_insn): Likewise.
8622         (movsfcc_soft_insn): Likewise.
8623         (arith_shiftsi): Likewise.
8624         (arith_shiftsi_compare0
8625         (arith_shiftsi_compare0_scratch
8626         (sub_shiftsi): Likewise.
8627         (sub_shiftsi_compare0
8628         (sub_shiftsi_compare0_scratch
8629         (and_scc): Likewise.
8630         (cond_move): Likewise.
8631         (if_plus_move): Likewise.
8632         (if_move_plus): Likewise.
8633         (if_move_not): Likewise.
8634         (if_not_move): Likewise.
8635         (if_shift_move): Likewise.
8636         (if_move_shift): Likewise.
8637         (if_shift_shift): Likewise.
8638         (if_not_arith): Likewise.
8639         (if_arith_not): Likewise.
8640         (cond_move_not): Likewise.
8641         (thumb1_ashlsi3): Set type attribute.
8642         (thumb1_ashrsi3): Likewise.
8643         (thumb1_lshrsi3): Likewise.
8644         (thumb1_rotrsi3): Likewise.
8645         (shiftsi3_compare0_scratch): Likewise.
8646         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
8647         (neon_mov<mode>): Likewise.
8648         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
8649         attribute changes.
8650         (thumb2_movsi_insn): Likewise.
8651         (thumb2_cmpsi_neg_shiftsi): Likewise.
8652         (thumb2_extendqisi_v6): Likewise.
8653         (thumb2_zero_extendhisi2_v6): Likewise.
8654         (thumb2_zero_extendqisi2_v6): Likewise.
8655         (thumb2_shiftsi3_short): Likewise.
8656         (thumb2_addsi3_compare0_scratch): Likewise.
8657         (orsi_not_shiftsi_si): Likewise.
8658         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
8659         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
8660         changes.
8661         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
8662         (1020alu_shift_op): Likewise.
8663         (1020alu_shift_reg_op): Likewise.
8664         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
8665         (alu_shift_op): Likewise.
8666         (alu_shift_reg_op): Likewise.
8667         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
8668         (11_alu_shift_op): Likewise.
8669         (11_alu_shift_reg_op): Likewise.
8670         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
8671         (9_alu_shift_reg_op): Likewise.
8672         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
8673         changes.
8674         (cortex_a15_alu_shift): Likewise.
8675         (cortex_a15_alu_shift_reg): Likewise.
8676         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
8677         changes.
8678         (cortex_a5_alu_shift): Likewise.
8679         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
8680         changes.
8681         (cortex_a53_alu_shift): Likewise.
8682         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
8683         changes.
8684         (cortex_a7_alu_reg): Likewise.
8685         (cortex_a7_alu_shift): Likewise.
8686         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
8687         changes.
8688         (cortex_a8_alu_shift): Likewise.
8689         (cortex_a8_alu_shift_reg): Likewise.
8690         (cortex_a8_mov): Likewise.
8691         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
8692         (cortex_a9_dp_shift): Likewise.
8693         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
8694         changes.
8695         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
8696         changes.
8697         (cortex_r4_mov): Likewise.
8698         (cortex_r4_alu_shift): Likewise.
8699         (cortex_r4_alu_shift_reg): Likewise.
8700         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
8701         (526_alu_shift_op): Likewise.
8702         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
8703         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
8704         (626te_alu_shift_op): Likewise.
8705         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
8706         (726te_alu_op): Likewise.
8707         (726te_alu_shift_op): Likewise.
8708         (726te_alu_shift_reg_op): Likewise.
8709         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
8710         (mp626_alu_shift_op): Likewise.
8711         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
8712         (pj4_alu_e1_conds): Likewise.
8713         (pj4_alu): Likewise.
8714         (pj4_alu_conds): Likewise.
8715         (pj4_shift): Likewise.
8716         (pj4_shift_conds): Likewise.
8717         (pj4_alu_shift): Likewise.
8718         (pj4_alu_shift_conds): Likewise.
8719         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8720         changes.
8721         (cortexa7_older_only): Likewise.
8722         (cortexa7_younger): Likewise.
8724 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8726         * ipa-pure-const.c (generate_summary): Rename to...
8727         (pure_const_generate_summary): ... this.
8729 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
8731         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
8733 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
8735         PR target/57909
8736         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
8737         usage in HI mode.
8739 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8741         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
8742         enabled without -march=zEC12.
8743         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
8744         flags to be set.
8746 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8748         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
8749         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
8750         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
8751         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
8752         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
8753         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
8754         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
8755         ISA_HAS_FP4.
8756         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
8757         and ISA_HAS_NMADD3_NMSUB3.
8758         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
8759         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
8760         (nmsub4<mode>, nmsub3<mode>): Likewise.
8761         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
8763 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8765         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
8766         TARGET_MIPS5400 checking.
8768 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
8769             Peter Bergner  <bergner@vnet.ibm.com>
8771         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
8772         registers in the comment.
8773         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
8774         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
8775         rather than FIRST_PSEUDO_REGISTERS.
8777 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
8779         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
8780         enable extra ISA flags with TARGET_HTM.
8782 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8784         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
8785         Fix comment typos.
8787 2013-07-15  Cong Hou  <congh@google.com>
8789         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
8790         in compare function for sorting.
8792 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
8794         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
8795         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
8796         * config/rs6000/rs6000.opt: Add -mhtm option.
8797         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
8798         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
8799         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8800         __HTM__ if the HTM instructions are available.
8801         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
8802         htm_spr_reg_operand): New define_predicates.
8803         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
8804         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
8805         Include htm.md.
8806         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
8807         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
8808         HTM builtin functions.
8809         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
8810         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
8811         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
8812         (rs6000_builtin_mask_calculate): Likewise.
8813         (rs6000_option_override_internal): Likewise.
8814         (bdesc_htm): Add new HTM builtin support.
8815         (htm_spr_num): New function.
8816         (htm_spr_regno): Likewise.
8817         (rs6000_htm_spr_icode): Likewise.
8818         (htm_expand_builtin): Likewise.
8819         (htm_init_builtins): Likewise.
8820         (rs6000_expand_builtin): Add support for HTM builtin functions.
8821         (rs6000_init_builtins): Likewise.
8822         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
8823         option.
8824         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
8825         (TARGET_HTM, MASK_HTM): Define macros.
8826         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
8827         (FIXED_REGISTERS): Likewise.
8828         (CALL_USED_REGISTERS): Likewise.
8829         (CALL_REALLY_USED_REGISTERS): Likewise.
8830         (REG_ALLOC_ORDER): Likewise.
8831         (enum reg_class): Likewise.
8832         (REG_CLASS_NAMES): Likewise.
8833         (REG_CLASS_CONTENTS): Likewise.
8834         (REGISTER_NAMES): Likewise.
8835         (ADDITIONAL_REGISTER_NAMES): Likewise.
8836         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
8837         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
8838         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
8839         * config/rs6000/htm.md: New file.
8840         * config/rs6000/htmintrin.h: New file.
8841         * config/rs6000/htmxlintrin.h: New file.
8843 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8845         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8846         Define SYMBOL_TINY_GOT, update comment.
8847         * config/aarch64/aarch64.c
8848         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
8849         (aarch64_expand_mov_immediate): Likewise.
8850         (aarch64_print_operand): Likewise.
8851         (aarch64_classify_symbol): Likewise.
8852         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
8853         (ldr_got_tiny): Define.
8855 2013-07-13  Tobias Grosser  <tobias@grosser.es>
8857         PR tree-optimization/54094
8858         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
8859           scheduling dimension for the parallelism check from the polyhedral
8860           information in the AST.
8861         * graphite-dependences.c (carries_deps): Do not assume the schedule is
8862           in 2D + 1 form.
8864 2013-07-13  Jason Merrill  <jason@redhat.com>
8866         * print-tree.c (debug_vec_tree): Use debug_raw.
8867         (debug_raw (vec<tree, va_gc> &)): New.
8868         (debug_raw (vec<tree, va_gc> *)): New.
8869         * tree.h: Declare them.
8871 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
8873         * ifcvt.c (ifcvt_after_combine): New static variable.
8874         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
8875         for size.
8876         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
8877         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
8878         rest_of_handle_if_after_reload): Pass new argument for if_convert.
8880 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
8882         * config/mips/mips.c (mips_expand_call): Remove empty statement.
8884 2013-07-12  Michael Matz  <matz@suse.de>
8886         PR middle-end/55771
8887         * convert.c (convert_to_real): Reject non-float inner types.
8889 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
8891         * config/aarch64/aarch64-protos.h
8892         (aarch64_simd_immediate_valid_for_move): Remove.
8893         * config/aarch64/aarch64.c (simd_immediate_info): New member.
8894         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
8895         cases.
8896         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
8898 2013-07-11  Steve Ellcey  <sellcey@mips.com>
8900         * config/mips/mips.c (mips_conditional_register_usage): Do not
8901         use t[0-7] registers in MIPS16 mode when optimizing for size.
8903 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
8905         * config/i386/i386.c (dispatch_function_versions): Fix array
8906         indexing of function_version_info to match actual_versions.
8908 2013-07-11  Teresa Johnson  <tejohnson@google.com>
8910         * vec.h (struct va_gc): Move release out-of-line.
8911         (va_gc::release): Call ggc_free on released vec.
8913 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
8915         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
8916         Require GOT register as additional operand in UNSPEC.
8917         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
8918         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8919         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8920         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
8921         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
8922         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
8923         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
8925 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
8927         PR target/57631
8928         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
8929         name seen by assembler/linker rather if available.
8931 2013-07-11  Andreas Schwab  <schwab@suse.de>
8933         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
8935 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
8937         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
8939 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
8941         * doc/tm.texi.in: Move hook documentation to ....
8942         * target.def: ... here.
8944         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
8945         text on @hook line.
8946         * doc/tm.texi: Regenerate.
8948 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
8950         PR c++/57869
8951         * doc/invoke.texi: Document -Wconditionally-supported.
8953 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8955         PR target/57844
8956         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
8957         of my_fp.
8959 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8961         PR target/57506
8962         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
8963         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
8964         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
8965         Remove duplicate devices.
8966         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
8967         * config/avr/t-multilib: Regenerate.
8968         * config/avr/avr-tables.opt: Regenerate.
8969         * doc/avr-mmcu.texi: Regenerate.
8971 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8973         PR target/56987
8974         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
8976 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
8978         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
8979         the cost of MULT when optimizing for size.
8981 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8983         * config/cr16/cr16-protos.h: Don't include target.h.
8985 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
8987         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
8988         adjust register size for TDmode and TFmode for VSX registers.
8990 2013-07-08  Kai Tietz  <ktietz@redhat.com>
8992         PR target/56892
8993         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
8994         hook_bool_const_tree_true.
8996 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8998         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
8999         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
9000         * config/s390/s390.md: Define FPR*_REGNUM constants.
9001         Fix FPR2_REGNUM constant (18 -> 17).
9002         ("*trunc<BFP:mode><DFP_ALL:mode>2")
9003         ("*trunc<DFP_ALL:mode><BFP:mode>2")
9004         ("trunc<BFP:mode><DFP_ALL:mode>2")
9005         ("trunc<DFP_ALL:mode><BFP:mode>2")
9006         ("*extend<BFP:mode><DFP_ALL:mode>2")
9007         ("*extend<DFP_ALL:mode><BFP:mode>2")
9008         ("extend<BFP:mode><DFP_ALL:mode>2")
9009         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
9010         FPR4_REGNUM.
9012 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
9014         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
9016 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9018         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
9019         and cfun_fpr_bit_p to cfun_fpr_save_p.
9020         (s390_frame_area, s390_register_info, s390_frame_info)
9021         (s390_emit_prologue, s390_emit_epilogue)
9022         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
9023         register numbers.
9024         * config/s390/s390.h: Define *_REGNUM macros for floating point
9025         register numbers.
9027 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
9029         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
9031 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
9033         PR rtl-optimization/57786
9034         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
9035         and break out of the loop when it is set to false.
9037 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
9039         PR target/57819
9040         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
9041         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
9042         (const_int 63)) 0)).
9043         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
9044         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
9045         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
9047         PR rtl-optimization/57829
9048         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
9049         mask bits outside of mode are just sign-extension from mode to HWI.
9051 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
9053         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
9054         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
9055         adjust_address instead of change_address to keep info about alignment.
9056         (emit_strmov): Remove.
9057         (emit_memmov): New function.
9058         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
9059         (expand_movmem_epilogue): Likewise and return updated rtx for
9060         destination.
9061         (expand_constant_movmem_prologue): Likewise and return updated rtx for
9062         destination and source.
9063         (decide_alignment): Refactor, handle vector_loop.
9064         (ix86_expand_movmem): Likewise.
9065         (ix86_expand_setmem): Likewise.
9066         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
9068 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
9070         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
9071         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
9073 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
9075         * config/i386/sse.md (sse_movlhps): Change alternative 3
9076         of operand 2 to "m".
9078 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
9080         PR target/57807
9081         * config/i386/sse.md (iptr): New mode attribute.
9082         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
9083         (<sse>_vm<plusminus_insn><mode>3): Ditto.
9084         (<sse>_vmmul<mode>3): Ditto.
9085         (<sse>_vmdiv<mode>3): Ditto.
9086         (sse_vmrcpv4sf2): Ditto.
9087         (<sse>_vmsqrt<mode>2): Ditto.
9088         (sse_vmrsqrtv4sf2): Ditto.
9089         (<sse>_vm<code><mode>3): Ditto.
9090         (avx_vmcmp<mode>3): Ditto.
9091         (<sse>_vmmaskcmp<mode>3): Ditto.
9092         (<sse>_comi): Ditto.
9093         (<sse>_ucomi): Ditto.
9094         (*xop_vmfrcz_<mode>): Ditto.
9095         (*fmai_fmadd_<mode>): Ditto.
9096         (*fmai_fmsub_<mode>): Ditto.
9097         (*fmai_fnmadd_<mode>): Ditto.
9098         (*fmai_fnmsub_<mode>): Ditto.
9099         (*fma4i_vmfmadd_<mode>): Ditto.
9100         (*fma4i_vmfmsub_<mode>): Ditto.
9101         (*fma4i_vmfnmadd_<mode>): Ditto.
9102         (*fma4i_vmfnmsub_<mode>): Ditto.
9103         (*xop_vmfrcz_<mode>): Ditto.
9104         (sse_cvtps2pi): Ditto.
9105         (sse_cvttps2pi): Ditto.
9106         (sse_cvtss2si): Ditto.
9107         (sse_cvtss2si_2): Ditto.
9108         (sse_cvtss2siq_2): Ditto.
9109         (sse_cvttss2si): Ditto.
9110         (sse_cvttss2siq): Ditto.
9111         (sse_cvtsd2si): Ditto.
9112         (sse_cvtsd2si_2): Ditto.
9113         (sse_cvtsd2siq_2): Ditto.
9114         (sse_cvttsd2si): Ditto.
9115         (sse_cvttsd2siq): Ditto.
9116         (sse_cvtsd2ss): Ditto.
9117         (sse_cvtss2sd): Ditto.
9118         (avx2_pbroadcast<mode>): Ditto.
9119         (avx2_pbroadcast<mode>_1): Ditto.
9120         (*avx_vperm_broadcast_v4sf): Ditto.
9122         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
9123         (sse_movlhps): Ditto.
9124         (sse_storehps): Ditto.
9125         (sse_loadhps): Ditto.
9126         (sse_storelps): Ditto.
9127         (sse_loadlps): Ditto.
9128         (*vec_concatv4sf): Ditto.
9129         (*vec_interleave_highv2df): Ditto.
9130         (*vec_interleave_lowv2df): Ditto.
9131         (*vec_extractv2df_1_sse): Ditto.
9132         (*vec_extractv2df_0_sse): Ditto.
9133         (sse2_storelpd): Ditto.
9134         (sse2_loadlpd): Ditto.
9135         (sse2_movsd): Ditto.
9136         (*vec_concatv4si): Ditto.
9137         (vec_concatv2di): Ditto.
9139         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
9140         for Intel asm dialect.
9141         (mmx_punpcklwd): Ditto.
9142         (mmx_punpckldq): Ditto.
9144         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
9145         for intel assembler dialect.
9147 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
9149         PR target/29776
9150         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
9151         for BUILT_IN_C{LZ,LRSB}*.
9152         * tree.h (CASE_INT_FN): Add FN##IMAX case.
9153         * tree-vrp.c (extract_range_basic): Handle
9154         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
9155         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
9156         fall thru to code calling set_value*.
9157         * builtins.c (expand_builtin): Remove *IMAX cases.
9158         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
9159         if width is bigger than 2*HWI.
9161 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
9163         PR rtl-optimization/55342
9164         * lra-int.h (lra_subreg_reload_pseudos): New.
9165         * lra.c: Add undoing optional reloads to the block diagram.
9166         (lra_subreg_reload_pseudos): New.
9167         (lra_optional_reload_pseudos): Change comments.
9168         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
9169         lra_optional_reload_pseudos after undo transformations.
9170         * lra-assigns.c (pseudo_prefix_title): New.
9171         (lra_setup_reg_renumber): Use it.
9172         (spill_for): Ditto.  Check subreg reload pseudos too.
9173         (assign_by_spills): Consider subreg reload pseudos too.
9174         * lra-constraints.c (simplify_operand_subreg): Use
9175         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
9176         (curr_insn_transform): Recognize and do optional reloads.
9177         (undo_optional_reloads): New.
9178         (lra_undo_inheritance): Call undo_optional_reloads.
9180 2013-07-05  Thomas Quinot  <quinot@adacore.com>
9182         * tree-complex.c (expand_complex_operations_1): Fix typo.
9184 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
9186         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
9187         (tune_params): New member 'const vec_costs'.
9188         * config/aarch64/aarch64.c (generic_vector_cost): New.
9189         (generic_tunings): New member 'generic_vector_cost'.
9190         (aarch64_builtin_vectorization_cost): New.
9191         (aarch64_add_stmt_cost): New.
9192         (TARGET_VECTORIZE_ADD_STMT_COST): New.
9193         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
9195 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
9197         PR target/57777
9198         * config/i386/predicates.md (vsib_address_operand): Disallow
9199         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
9201 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
9203         PR middle-end/55030
9204         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
9205         expand_builtin_setjmp_receiver.
9206         (expand_label): Adjust, call expand_builtin_setjmp_receiver
9207         with NULL for the label parameter.
9208         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
9209         the frame-pointer.  Adjust comments.
9210         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
9211         only if LABEL is non-NULL.
9213 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
9215         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
9216         (ARM_ABI_AAPCS64): Ditto.
9217         (arm_abi): Ditto.
9218         (ARM_DEFAULT_ABI): Ditto.
9220 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
9222         * config/aarch64/aarch64-builtins.c
9223         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
9224         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
9225         (st1): Likewise.
9226         * config/aarch64/aarch64-simd.md
9227         (aarch64_ld1<VALL:mode>): New.
9228         (aarch64_st1<VALL:mode>): Likewise.
9229         * config/aarch64/arm_neon.h
9230         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
9232 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
9234         * config/i386/i386.c (gate_insert_vzeroupper): Check if
9235         target ISA is AVX.
9236         (ix86_option_override_internal):Turn on all -mavx target flags by
9237         default as they are dependent on AVX anyway.
9239 2013-07-02  Cary Coutant  <ccoutant@google.com>
9241         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
9242         deterministic hash.
9243         (loc_checksum_ordered): Likewise.
9244         (hash_loc_operands): Remove inline keyword.
9246 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
9248         PR tree-optimization/57741
9249         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
9250         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
9251         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
9252         Allow REAL_CST step_exprs if flag_associative_math.
9253         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
9255 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
9257         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
9259 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
9261         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
9263 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9265         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
9266         encoding.
9267         (iorsi3_insn): Likewise.
9268         (arm_xorsi3): Likewise.
9270 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
9272         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
9273         to ...
9274         (attribute "type"): ... here, and prefix with "wmmx_".
9275         (attribute "core_cycles"): Update for attribute changes.
9276         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
9277         (tbcstv4hi): Likewise.
9278         (tbcstv2si): Likewise.
9279         (iwmmxt_iordi3): Likewise.
9280         (iwmmxt_xordi3): Likewise.
9281         (iwmmxt_anddi3): Likewise.
9282         (iwmmxt_nanddi3): Likewise.
9283         (iwmmxt_arm_movdi): Likewise.
9284         (iwmmxt_movsi_insn): Likewise.
9285         (mov<mode>_internal): Likewise.
9286         (and<mode>3_iwmmxt): Likewise.
9287         (ior<mode>3_iwmmxt): Likewise.
9288         (xor<mode>3_iwmmxt): Likewise.
9289         (add<mode>3_iwmmxt): Likewise.
9290         (ssaddv8qi3): Likewise.
9291         (ssaddv4hi3): Likewise.
9292         (ssaddv2si3): Likewise.
9293         (usaddv8qi3): Likewise.
9294         (usaddv4hi3): Likewise.
9295         (usaddv2si3): Likewise.
9296         (sub<mode>3_iwmmxt): Likewise.
9297         (sssubv8qi3): Likewise.
9298         (sssubv4hi3): Likewise.
9299         (sssubv2si3): Likewise.
9300         (ussubv8qi3): Likewise.
9301         (ussubv4hi3): Likewise.
9302         (ussubv2si3): Likewise.
9303         (mulv4hi3_iwmmxt): Likewise.
9304         (smulv4hi3_highpart): Likewise.
9305         (umulv4hi3_highpart): Likewise.
9306         (iwmmxt_wmacs): Likewise.
9307         (iwmmxt_wmacsz): Likewise.
9308         (iwmmxt_wmacu): Likewise.
9309         (iwmmxt_wmacuz): Likewise.
9310         (iwmmxt_clrdi): Likewise.
9311         (iwmmxt_clrv8qi): Likewise.
9312         (iwmmxt_clr4hi): Likewise.
9313         (iwmmxt_clr2si): Likewise.
9314         (iwmmxt_uavgrndv8qi3): Likewise.
9315         (iwmmxt_uavgrndv4hi3): Likewise.
9316         (iwmmxt_uavgv8qi3): Likewise.
9317         (iwmmxt_uavgv4hi3): Likewise.
9318         (iwmmxt_tinsrb): Likewise.
9319         (iwmmxt_tinsrh): Likewise.
9320         (iwmmxt_tinsrw): Likewise.
9321         (iwmmxt_textrmub): Likewise.
9322         (iwmmxt_textrmsb): Likewise.
9323         (iwmmxt_textrmuh): Likewise.
9324         (iwmmxt_textrmsh): Likewise.
9325         (iwmmxt_textrmw): Likewise.
9326         (iwmxxt_wshufh): Likewise.
9327         (eqv8qi3): Likewise.
9328         (eqv4hi3): Likewise.
9329         (eqv2si3): Likewise.
9330         (gtuv8qi3): Likewise.
9331         (gtuv4hi3): Likewise.
9332         (gtuv2si3): Likewise.
9333         (gtv8qi3): Likewise.
9334         (gtv4hi3): Likewise.
9335         (gtv2si3): Likewise.
9336         (smax<mode>3_iwmmxt): Likewise.
9337         (umax<mode>3_iwmmxt): Likewise.
9338         (smin<mode>3_iwmmxt): Likewise.
9339         (umin<mode>3_iwmmxt): Likewise.
9340         (iwmmxt_wpackhss): Likewise.
9341         (iwmmxt_wpackwss): Likewise.
9342         (iwmmxt_wpackdss): Likewise.
9343         (iwmmxt_wpackhus): Likewise.
9344         (iwmmxt_wpackwus): Likewise.
9345         (iwmmxt_wpackdus): Likewise.
9346         (iwmmxt_wunpckihb): Likewise.
9347         (iwmmxt_wunpckihh): Likewise.
9348         (iwmmxt_wunpckihw): Likewise.
9349         (iwmmxt_wunpckilb): Likewise.
9350         (iwmmxt_wunpckilh): Likewise.
9351         (iwmmxt_wunpckilw): Likewise.
9352         (iwmmxt_wunpckehub): Likewise.
9353         (iwmmxt_wunpckehuh): Likewise.
9354         (iwmmxt_wunpckehuw): Likewise.
9355         (iwmmxt_wunpckehsb): Likewise.
9356         (iwmmxt_wunpckehsh): Likewise.
9357         (iwmmxt_wunpckehsw): Likewise.
9358         (iwmmxt_wunpckelub): Likewise.
9359         (iwmmxt_wunpckeluh): Likewise.
9360         (iwmmxt_wunpckeluw): Likewise.
9361         (iwmmxt_wunpckelsb): Likewise.
9362         (iwmmxt_wunpckelsh): Likewise.
9363         (iwmmxt_wunpckelsw): Likewise.
9364         (ror<mode>3): Likewise.
9365         (ashr<mode>3_iwmmxt): Likewise.
9366         (lshr<mode>3_iwmmxt): Likewise.
9367         (ashl<mode>3_iwmmxt): Likewise.
9368         (ror<mode>3_di): Likewise.
9369         (ashr<mode>3_di): Likewise.
9370         (lshr<mode>3_di): Likewise.
9371         (ashl<mode>3_di): Likewise.
9372         (iwmmxt_wmadds): Likewise.
9373         (iwmmxt_wmaddu): Likewise.
9374         (iwmmxt_tmia): Likewise.
9375         (iwmmxt_tmiaph): Likewise.
9376         (iwmmxt_tmiabb): Likewise.
9377         (iwmmxt_tmiatb): Likewise.
9378         (iwmmxt_tmiabt): Likewise.
9379         (iwmmxt_tmiatt): Likewise.
9380         (iwmmxt_tmovmskb): Likewise.
9381         (iwmmxt_tmovmskh): Likewise.
9382         (iwmmxt_tmovmskw): Likewise.
9383         (iwmmxt_waccb): Likewise.
9384         (iwmmxt_wacch): Likewise.
9385         (iwmmxt_waccw): Likewise.
9386         (iwmmxt_waligni): Likewise.
9387         (iwmmxt_walignr): Likewise.
9388         (iwmmxt_walignr0): Likewise.
9389         (iwmmxt_walignr1): Likewise.
9390         (iwmmxt_walignr2): Likewise.
9391         (iwmmxt_walignr3): Likewise.
9392         (iwmmxt_wsadb): Likewise.
9393         (iwmmxt_wsadh): Likewise.
9394         (iwmmxt_wsadbz): Likewise.
9395         (iwmmxt_wsadhz): Likewise.
9396         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
9397         (iwmmxt_wabsdiffb): Likewise.
9398         (iwmmxt_wabsdiffh): Likewise.
9399         (iwmmxt_wabsdiffw): Likewise.
9400         (iwmmxt_waddsubhx): Likewise
9401         (iwmmxt_wsubaddhx): Likewise.
9402         (addc<mode>3): Likewise.
9403         (iwmmxt_avg4): Likewise.
9404         (iwmmxt_avg4r): Likewise.
9405         (iwmmxt_wmaddsx): Likewise.
9406         (iwmmxt_wmaddux): Likewise.
9407         (iwmmxt_wmaddsn): Likewise.
9408         (iwmmxt_wmaddun): Likewise.
9409         (iwmmxt_wmulwsm): Likewise.
9410         (iwmmxt_wmulwum): Likewise.
9411         (iwmmxt_wmulsmr): Likewise.
9412         (iwmmxt_wmulumr): Likewise.
9413         (iwmmxt_wmulwsmr): Likewise.
9414         (iwmmxt_wmulwumr): Likewise.
9415         (iwmmxt_wmulwl): Likewise.
9416         (iwmmxt_wqmulm): Likewise.
9417         (iwmmxt_wqmulwm): Likewise.
9418         (iwmmxt_wqmulmr): Likewise.
9419         (iwmmxt_wqmulwmr): Likewise.
9420         (iwmmxt_waddbhusm): Likewise.
9421         (iwmmxt_waddbhusl): Likewise.
9422         (iwmmxt_wqmiabb): Likewise.
9423         (iwmmxt_wqmiabt): Likewise.
9424         (iwmmxt_wqmiatb): Likewise.
9425         (iwmmxt_wqmiatt): Likewise.
9426         (iwmmxt_wqmiabbn): Likewise.
9427         (iwmmxt_wqmiabtn): Likewise.
9428         (iwmmxt_wqmiatbn): Likewise.
9429         (iwmmxt_wqmiattn): Likewise.
9430         (iwmmxt_wmiabb): Likewise.
9431         (iwmmxt_wmiabt): Likewise.
9432         (iwmmxt_wmiatb): Likewise.
9433         (iwmmxt_wmiatt): Likewise.
9434         (iwmmxt_wmiabbn): Likewise.
9435         (iwmmxt_wmiabtn): Likewise.
9436         (iwmmxt_wmiatbn): Likewise.
9437         (iwmmxt_wmiattn): Likewise.
9438         (iwmmxt_wmiawbb): Likewise.
9439         (iwmmxt_wmiawbt): Likewise.
9440         (iwmmxt_wmiawtb): Likewise.
9441         (iwmmxt_wmiawtt): Likewise.
9442         (iwmmxt_wmiawbbn): Likewise.
9443         (iwmmxt_wmiawbtn): Likewise.
9444         (iwmmxt_wmiawtbn): Likewise.
9445         (iwmmxt_wmiawttn): Likewise.
9446         (iwmmxt_wmerge): Likewise.
9447         (iwmmxt_tandc<mode>3): Likewise.
9448         (iwmmxt_torc<mode>3): Likewise.
9449         (iwmmxt_torvsc<mode>3): Likewise.
9450         (iwmmxt_textrc<mode>3): Likewise.
9451         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
9452         (wmmxt_pack): Likewise.
9453         (wmmxt_mult_c1): Likewise.
9454         (wmmxt_mult_c2): Likewise.
9455         (wmmxt_alu_c1): Likewise.
9456         (wmmxt_alu_c2): Likewise.
9457         (wmmxt_alu_c3): Likewise.
9458         (wmmxt_transfer_c1): Likewise.
9459         (wmmxt_transfer_c2): Likewise.
9460         (wmmxt_transfer_c3): Likewise.
9461         (marvell_f_iwmmxt_wstr): Likewise.
9462         (marvell_f_iwmmxt_wldr): Likewise.
9464 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
9466         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
9468 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
9470         Revert:
9471         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
9472         * lra-constraints.c (need_for_split_p): Check call used hard regs
9473         living through calls.
9475         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
9476         call used regs for call insn.
9478 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
9480         PR target/57736
9481         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
9482         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
9484 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
9486         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
9488 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
9490         * lra-constraints.c (need_for_split_p): Check call used hard regs
9491         living through calls.
9493 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
9495         PR target/57744
9496         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
9497         to tie with any other modes.  Eliminate Altivec vector mode tests,
9498         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
9499         code, to return 0 if testing MODE2 for a condition, if we've
9500         already tested MODE1 for the same condition.
9502 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9504         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
9505         layout.
9507 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9509         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
9510         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
9512 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9514         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
9515         Define.
9516         (aarch64_symbolic_constant_p): Remove.
9517         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
9518         static.  Fix line length and white space.
9519         (aarch64_symbolic_constant_p): Remove.
9520         * config/aarch64/predicates.md (aarch64_valid_symref):
9521         Use aarch64_classify_symbol_expression.
9523 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9525         * config/arm/constraints.md (Ts): New constraint.
9526         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
9527         16-bit encodings.
9528         (compare_scc): Use "Ts" constraint for operand 0.
9529         (ior_scc_scc): Likewise.
9530         (and_scc_scc): Likewise.
9531         (and_scc_scc_nodom): Likewise.
9532         (ior_scc_scc_cmp): Likewise for operand 7.
9533         (and_scc_scc_cmp): Likewise.
9534         * config/arm/thumb2.md (thumb2_movsi_insn):
9535         Add alternatives for 16-bit encodings.
9536         (thumb2_movhi_insn): Likewise.
9537         (thumb2_movsicc_insn): Likewise.
9538         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
9539         (thumb2_negscc): Use "Ts" constraint.
9540         Move mvn instruction outside cond_exec block.
9541         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
9542         for 16-bit encodings.
9544 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9546         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
9547         encoding.
9548         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
9549         (mulsi3subsi): Likewise.
9550         (mulsidi3adddi): Likewise.
9551         (mulsidi3_v6): Likewise.
9552         (umulsidi3_v6): Likewise.
9553         (umulsidi3adddi_v6): Likewise.
9554         (smulsi3_highpart_v6): Likewise.
9555         (umulsi3_highpart_v6): Likewise.
9556         (mulhisi3tb): Likewise.
9557         (mulhisi3bt): Likewise.
9558         (mulhisi3tt): Likewise.
9559         (maddhisi4): Likewise.
9560         (maddhisi4tb): Likewise.
9561         (maddhisi4tt): Likewise.
9562         (maddhidi4): Likewise.
9563         (maddhidi4tb): Likewise.
9564         (maddhidi4tt): Likewise.
9565         (zeroextractsi_compare0_scratch): Likewise.
9566         (insv_zero): Likewise.
9567         (insv_t2): Likewise.
9568         (anddi_notzesidi_di): Likewise.
9569         (anddi_notsesidi_di): Likewise.
9570         (andsi_notsi_si): Likewise.
9571         (iordi_zesidi_di): Likewise.
9572         (xordi_zesidi_di): Likewise.
9573         (andsi_iorsi3_notsi): Likewise.
9574         (smax_0): Likewise.
9575         (smax_m1): Likewise.
9576         (smin_0): Likewise.
9577         (not_shiftsi): Likewise.
9578         (unaligned_loadsi): Likewise.
9579         (unaligned_loadhis): Likewise.
9580         (unaligned_loadhiu): Likewise.
9581         (unaligned_storesi): Likewise.
9582         (unaligned_storehi): Likewise.
9583         (extv_reg): Likewise.
9584         (extzv_t2): Likewise.
9585         (divsi3): Likewise.
9586         (udivsi3): Likewise.
9587         (arm_zero_extendhisi2addsi): Likewise.
9588         (arm_zero_extendqisi2addsi): Likewise.
9589         (compareqi_eq0): Likewise.
9590         (arm_extendhisi2_v6): Likewise.
9591         (arm_extendqisi2addsi): Likewise.
9592         (arm_movt): Likewise.
9593         (thumb2_ldrd): Likewise.
9594         (thumb2_ldrd_base): Likewise.
9595         (thumb2_ldrd_base_neg): Likewise.
9596         (thumb2_strd): Likewise.
9597         (thumb2_strd_base): Likewise.
9598         (thumb2_strd_base_neg): Likewise.
9599         (arm_negsi2): Add alternative for 16-bit encoding.
9600         (arm_one_cmplsi2): Likewise.
9602 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9604         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
9605         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
9606         (movdfcc): Likewise.
9607         * config/arm/vfp.md (*thumb2_movsf_vfp):
9608         Disable predication for arm_restrict_it.
9609         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
9610         (*thumb2_movdfcc_vfp): Likewise.
9611         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
9612         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
9613         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
9614         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
9615         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
9616         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
9617         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
9618         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
9619         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
9620         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
9621         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
9622         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
9623         Disable predication for arm_restrict_it.
9625 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
9627         * config/i386/bmiintrin.h (_bextr_u32): New.
9628         (_bextr_u64): Ditto.
9630 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
9632         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
9633         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
9634         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
9635         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
9636         * config/mips/n32-elf.h: ...this new file.
9638 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
9640         PR target/57224
9641         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
9642         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
9644 2013-06-27  Catherine Moore  <clm@codesourcery.com>
9646         * config/mips/mips-tables.opt: Regenerate.
9647         * config/mips/mips-cpus.def: Add m14ke and m14kec.
9648         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
9649         * doc/invoke.texi: Add -m14kc.
9651 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
9653         PR target/57623
9654         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
9655         constraints of operand 1 and 2.
9657         PR target/57623
9658         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
9659         to match RTL canonicalization.  Swap predicates and
9660         constraints of operand 1 and 2.
9662 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
9664         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
9665         Process OP_INOUT regs for splitting too.
9667 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
9669         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
9670         decl before the loop, initialize to NULL.
9671         (vectorizable_load): Initialize ptr_incr to NULL.
9673 2013-06-27  Martin Jambor  <mjambor@suse.cz>
9675         PR lto/57208
9676         * ipa-ref.h (ipa_maybe_record_reference): Declare.
9677         * ipa-ref.c (ipa_maybe_record_reference): New function.
9678         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
9679         * ipa-cp.c (create_specialized_node): Record potential references from
9680         aggvals.
9681         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
9683 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
9685         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
9686         parameter 'mode' of type 'enum machine_mode mode'; change to pass
9687         'mode' to force_reg.
9688         (aarch64_add_offset): Update calls to aarch64_force_temporary.
9689         (aarch64_expand_mov_immediate): Likewise.
9691 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
9693         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
9694         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
9696 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9698         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
9699         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9700         (struct machine_function): Add tbegin_p.
9701         (s390_canonicalize_comparison): Fold CC mode compares to
9702         conditional jump if possible.
9703         (s390_emit_jump): Return the emitted jump.
9704         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
9705         Handle CCRAWmode compares.
9706         (s390_option_override): Default to -mhtm if available.
9707         (s390_reg_clobbered_rtx): Handle floating point regs as well.
9708         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
9709         FPRs instead of df_regs_ever_live_p.
9710         (s390_optimize_nonescaping_tx): New function.
9711         (s390_init_frame_layout): Extend clobbered_regs array to cover
9712         FPRs as well.
9713         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
9714         (s390_expand_tbegin): New function.
9715         (enum s390_builtin): New enum definition.
9716         (code_for_builtin): New array definition.
9717         (s390_init_builtins): New function.
9718         (s390_expand_builtin): New function.
9719         (TARGET_INIT_BUILTINS): Define.
9720         (TARGET_EXPAND_BUILTIN): Define.
9721         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
9722         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
9723         (s390_alc_comparison): Likewise.
9724         * config/s390/s390-modes.def: Add CCRAWmode.
9725         * config/s390/s390.h (processor_flags): Add PF_TX.
9726         (TARGET_CPU_HTM): Define macro.
9727         (TARGET_HTM): Define macro.
9728         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
9729         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
9730         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9731         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
9732         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
9733         (TBEGIN_MASK, TBEGINC_MASK): New constants.
9734         ("*cc_to_int"): Move up.
9735         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
9736         constants other than 0.
9737         ("*ccraw_to_int"): New insn and splitter definition.
9738         ("tbegin", "tbegin_nofloat", "tbegin_retry")
9739         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
9740         ("tx_assist"): New expander.
9741         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
9742         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
9743         * config/s390/s390.opt: Add -mhtm option.
9744         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
9745         * config/s390/htmxlintrin.h: New file.
9746         * config/s390/htmintrin.h: New file.
9747         * config/s390/s390intrin.h: New file.
9748         * doc/extend.texi: Document htm builtins.
9749         * config.gcc: Add the new header files to extra_headers.
9751 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
9753         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
9754         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
9756 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
9757             Pat Haugen  <pthaugen@us.ibm.com>
9758             Peter Bergner  <bergner@vnet.ibm.com>
9760         * config/rs6000/power8.md: New.
9761         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
9762         setting for power8 entry.
9763         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
9764         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
9765         test for Power4/Power5 only.
9766         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
9767         support.
9768         (force_new_group): Adjust comment.
9769         * config/rs6000/rs6000.md: Include power8.md.
9771 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
9773         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
9774         * config/arm/arm-protos.h (arm_max_conditional_execute): New
9775         declaration.
9776         (tune_params): Update comment.
9777         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
9778         (arm_max_conditional_execute): New function.
9779         (thumb2_final_prescan_insn): Use max_insn_skipped and
9780         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
9782 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
9784         PR tree-optimization/57705
9785         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
9786         SSA_NAME step, provided that it is not defined inside the loop.
9787         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
9788         (get_initial_def_for_induction): Handle SSA_NAME IV step.
9790 2013-06-25  Martin Jambor  <mjambor@suse.cz>
9792         PR middle-end/57670
9793         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
9794         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
9795         calls in the dump.
9796         (ipa_note_param_call): Initialize member_ptr flag.
9797         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
9798         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
9799         (ipa_write_indirect_edge_info): Stream member_ptr flag.
9800         (ipa_read_indirect_edge_info): Likewise.
9802 2013-06-25  Richard Biener  <rguenther@suse.de>
9804         PR middle-end/56977
9805         * passes.c (init_optimization_passes): Move pass_fold_builtins
9806         and pass_dce earlier with -Og.
9808 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
9810         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
9811         <BIT_FIELD_REF>: Remove trailing TAB.
9812         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
9813         remove blank line.
9815 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9817         PR tree-optimization/57358
9818         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
9819         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
9820         (ipa_analyze_params_uses): Generate pessimistic info when true.
9822 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9824         PR tree-optimization/57539
9825         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
9826         global.inlined_to of the new node to it.  All callers changed.
9827         * ipa-inline-transform.c (clone_inlined_nodes): New variable
9828         inlining_into, pass it to cgraph_clone_node.
9829         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
9830         ipa_free_edge_args_substructures.
9831         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
9832         rdesc linked list.  Do not assert rdesc edges have inlined caller.
9833         Assert we have found an rdesc in the rdesc list.
9835 2013-06-24  Richard Biener  <rguenther@suse.de>
9837         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
9838         (pointer_set_lookup): Declare.
9839         (class pointer_map): New template class implementing a
9840         generic pointer to T map.
9841         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
9842         pointer_map<T>::contains, pointer_map<T>::insert,
9843         pointer_map<T>::traverse): New functions.
9844         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
9845         (pointer_set_lookup): New function.
9846         (pointer_set_contains): Use pointer_set_lookup.
9847         (pointer_set_insert): Likewise.
9848         (insert_aux): Remove.
9849         (struct pointer_map_t): Embed a pointer_set_t.
9850         (pointer_map_create): Adjust.
9851         (pointer_map_destroy): Likewise.
9852         (pointer_map_contains): Likewise.
9853         (pointer_map_insert): Likewise.
9854         (pointer_map_traverse): Likewise.
9855         * tree-streamer.h (struct streamer_tree_cache_d): Use a
9856         pointer_map<unsigned> instead of a pointer_map_t.
9857         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
9858         (streamer_tree_cache_lookup): Likewise.
9859         (streamer_tree_cache_create): Likewise.
9860         (streamer_tree_cache_delete): Likewise.
9861         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
9862         pointer_map<unsigned> instead of a pointer_map_t.
9863         (lto_init_tree_ref_encoder): Adjust.
9864         (lto_destroy_tree_ref_encoder): Likewise.
9865         * lto-section-out.c (lto_output_decl_index): Likewise.
9866         (lto_record_function_out_decl_state): Likewise.
9867         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
9869 2013-06-24  Richard Biener  <rguenther@suse.de>
9871         PR tree-optimization/57488
9872         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
9874 2013-06-24  Alan Modra  <amodra@gmail.com>
9876         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
9877         (gen_easy_altivec_constant): Likewise.
9878         * config/rs6000/predicates.md (easy_vector_constant_add_self,
9879         easy_vector_constant_msb): Likewise.
9881 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
9883         PR target/57688
9884         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
9885         add missing return true.
9887 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
9889         PR target/52483
9890         * config/sh/predicates.md (general_extend_operand): Invoke
9891         general_movsrc_operand for memory operands.
9892         (general_movsrc_operand): Allow reg+reg addressing, do not use
9893         general_operand for memory operands.
9895 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
9897         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
9898         when current target options does not apply.
9899         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
9900         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
9901         * config/i386/bmiintrin.h: Pass appropriate target
9902         attributes to header.
9903         * config/i386/mmintrin.h: Ditto.
9904         * config/i386/nmmintrin.h: Ditto.
9905         * config/i386/avx2intrin.h: Ditto.
9906         * config/i386/fxsrintrin.h: Ditto.
9907         * config/i386/tbmintrin.h: Ditto.
9908         * config/i386/xsaveintrin.h: Ditto.
9909         * config/i386/f16cintrin.h: Ditto.
9910         * config/i386/xtestintrin.h: Ditto.
9911         * config/i386/xsaveoptintrin.h: Ditto.
9912         * config/i386/bmi2intrin.h: Ditto.
9913         * config/i386/lzcntintrin.h: Ditto.
9914         * config/i386/smmintrin.h: Ditto.
9915         * config/i386/wmmintrin.h: Ditto.
9916         * config/i386/x86intrin.h: Remove all header include guards.
9917         * config/i386/prfchwintrin.h: Ditto.
9918         * config/i386/pmmintrin.h: Ditto.
9919         * config/i386/tmmintrin.h: Ditto.
9920         * config/i386/xmmintrin.h: Ditto.
9921         * config/i386/popcntintrin.h: Ditto.
9922         * config/i386/rdseedintrin.h: Ditto.
9923         * config/i386/ammintrin.h: Ditto.
9924         * config/i386/emmintrin.h: Ditto.
9925         * config/i386/immintrin.h: Remove all header include guards.
9926         * config/i386/fma4intrin.h: Ditto.
9927         * config/i386/lwpintrin.h: Ditto.
9928         * config/i386/xopintrin.h: Ditto.
9929         * config/i386/ia32intrin.h: Ditto.
9930         * config/i386/avxintrin.h: Ditto.
9931         * config/i386/rtmintrin.h: Ditto.
9932         * config/i386/fmaintrin.h: Ditto.
9933         * config/i386/mm3dnow.h: Ditto.
9935 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
9937         * common/config/i386/i386-common.c: Handle LZCNT.
9939 2013-06-22  Andi Kleen  <ak@linux.intel.com>
9941         * doc/extend.texi: Use __atomic_store_n instead of
9942         __atomic_store in HLE example.
9944 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
9946         * config/sh/sh.c: Remove <cstdlib> workaround.
9948 2013-06-21  Andi Kleen  <ak@linux.intel.com>
9950         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
9952 2013-06-21  Andi Kleen  <ak@linux.intel.com>
9954         * doc/extend.texi: Document that __atomic_clear and
9955           __atomic_test_and_set should only be used with bool.
9957 2013-06-20  Jan Hubicka  <jh@suse.cz>
9959         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
9960         types_same_for_odr.
9961         * tree.c (decls_same_for_odr): New function.
9962         (same_for_edr): New function.
9963         (types_same_for_odr): New function.
9964         (get_binfo_at_offset): Use it.
9965         * tree.h (types_same_for_odr): Declare.
9967 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
9968             Jason Merrill  <jason@redhat.com>
9970         * system.h: Include <cstdlib> as well as <stdlib.h>.
9972 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
9974         PR target/57655
9975         * config/i386/i386.c (construct_container): Report error if
9976         long double is used with disabled x87 float returns.
9978 2013-06-20  Jan Hubicka  <jh@suse.cz>
9980         * lto-cgraph.c (input_symtab): Do not set cgraph state.
9982 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
9984         PR rtl-optimization/57425
9985         PR rtl-optimization/57569
9986         * alias.c (write_dependence_p): Remove parameters mem_mode and
9987         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
9988         Changed all callers.
9989         (canon_anti_dependence): Get comments and semantics in sync.
9990         Add parameter mem_canonicalized.  Changed all callers.
9991         * rtl.h (canon_anti_dependence): Update prototype.
9993 2013-06-20  Richard Biener  <rguenther@suse.de>
9995         * data-streamer-in.c (streamer_read_uhwi): Optimize single
9996         byte case, inline streamer_read_uchar and defer section
9997         overrun check.
9999 2013-06-20  Richard Biener  <rguenther@suse.de>
10001         PR tree-optimization/57584
10002         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
10003         SSA names into the expanded expression that take part in
10004         abnormal coalescing.
10006 2013-06-19  Sharad Singhai  <singhai@google.com>
10008         * gcov.c (print_usage): Handle new option.
10009         (process_args): Ditto.
10010         (get_gcov_intermediate_filename): New function.
10011         (output_intermediate_file): New function.
10012         (output_gcov_file): New function
10013         (generate_results): Handle new option.
10014         (release_function): Relase demangled name.
10015         (read_graph_file): Handle demangled name.
10016         (output_lines): Ditto.
10017         * doc/gcov.texi: Document gcov intermediate format.
10019 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
10021         PR bootstrap/57604
10022         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
10023         (lra_emit_add): Use the functions.  Add comment about Y as an
10024         address segment.
10026 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
10028         PR driver/57652
10029         * collect2.c (collect_atexit): New.
10030         (collect_exit): Delete.
10031         (main): Register collect_atexit with atexit.
10032         (collect_wait): Change collect_exit to exit.
10033         (do_wait): Same.
10034         * collect2.h (collect_exit): Delete.
10035         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
10037 2013-06-19  Wei Mi  <wmi@google.com>
10039         PR rtl-optimization/57518
10040         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
10041         if regno is used in paradoxical subreg.
10042         (update_equiv_regs): Check pdx_subregs[regno] before
10043         set a reg to be equivalent with a mem.
10045 2013-06-19  Matthias Klose  <doko@ubuntu.com>
10047         PR driver/57651
10048         * file-find.h (find_a_file): Add a mode parameter.
10049         * file-find.c (find_a_file): Likewise.
10050         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
10051         with X_OK for the executables.
10052         * collect2.c (main): Call find_a_file with X_OK.
10054 2013-06-19  Steve Ellcey  <sellcey@mips.com>
10056         PR target/56942
10057         * config/mips/mips.md (casesi_internal_mips16_<mode>):
10058         Use NEXT_INSN instead of next_real_insn.
10060 2013-06-19  Jan Hubicka  <jh@suse.cz>
10062         * cgraph.h (const_value_known_p): Replace by ...
10063         (ctor_for_folding): .. this one.
10064         * cgraphunit.c (process_function_and_variable_attributes): Use it.
10065         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
10066         * expr.c (expand_expr_real_1): Likewise.
10067         (string_constant): Likewise.
10068         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
10069         * ipa.c (process_references): Likewise.
10070         (symtab_remove_unreachable_nodes): Likewise.
10071         * ipa-inline-analysis.c (param_change_prob): Likewise.
10072         * gimple-fold.c (canonicalize_constructor_val): Likewise.
10073         (get_base_constructor): Likwise.
10074         * varpool.c (varpool_remove_node): Likewise.
10075         (varpool_remove_initializer): LIkewise.
10076         (dump_varpool_node): LIkwise.
10077         (const_value_known_p): Rewrite to ...
10078         (ctor_for_folding): ... this one.
10080 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
10082         PR driver/57651
10083         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
10084         PERSONALITY in $PATH derived prefixes.
10086 2013-06-19  Jeff Law  <law@redhat.com>
10088         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
10089         in comment.
10091         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
10092         (simplify_bitwise_binary): Use it to simpify certain binary ops on
10093         booleans.
10095 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
10097         * config/arm/vfp.md: Move VFP instruction classification documentation
10098         to ...
10099         * config/arm/arm.md: ... here.  Update instruction classification
10100         documentation.
10102 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
10104         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
10105         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
10106         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
10108 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
10110         PR target/57609
10111         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10112         with NEXT_INSN.  Use tablejump_p to check for jump table data
10113         insns.
10115 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
10117         PR c++/56544
10118         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
10119         that now in C++ the value is correct per the C++ standards.
10121 2013-06-19  Richard Biener  <rguenther@suse.de>
10123         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
10124         for global context.
10126 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10128         Revert:
10129         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10131         PR target/57609
10132         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10133         with next_active_insn.
10135 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
10137         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
10138         functions are inlined during failures to flag an error.
10139         * tree-inline.c (expand_call_inline): Allow the error to be flagged
10140         in early inline pass.
10142 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
10144         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
10145         in comments.
10147 2013-06-18  Julian Brown  <julian@codesourcery.com>
10149         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
10150         Permit virtual register pre-reload if !strict.
10151         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
10152         change.
10153         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
10154         prototype.
10155         * config/arm/neon.md (movmisalign<mode>): Use
10156         neon_perm_struct_or_reg_operand instead of
10157         neon_struct_or_register_operand.
10158         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
10159         neon_permissive_struct_operand instead of neon_struct_operand.
10160         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
10161         neon_vector_mem_operand.
10162         * config/arm/predicates.md (neon_struct_operand): Adjust call to
10163         neon_vector_mem_operand.
10164         (neon_permissive_struct_operand): New.
10165         (neon_struct_or_register_operand): Rename to...
10166         (neon_perm_struct_or_reg_operand): This. Adjust call to
10167         neon_vector_mem_operand.
10169 2013-06-18  Richard Biener  <rguenther@suse.de>
10171         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
10172         * lto-streamer.h: Include pointer-set.h.
10173         (struct lto_decl_slot): Remove.
10174         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
10175         Remove next_index entry.
10176         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
10177         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
10178         (lto_init_tree_ref_encoder): Adjust.
10179         (lto_destroy_tree_ref_encoder): Likewise.
10180         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
10181         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
10182         (lto_output_decl_index): Adjust.
10183         (lto_new_out_decl_state): Likewise.
10184         (lto_record_function_out_decl_state): Likewise.
10185         * lto-streamer-out.c (copy_function): Likewise.
10187 2013-06-18  Richard Biener  <rguenther@suse.de>
10189         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
10190         * cgraphunit.c: Include cfgloop.h.
10191         (init_lowered_empty_function): Initialize the loop tree.
10192         (assemble_thunk): Insert new BBs into loops.
10194 2013-06-18  Richard Biener  <rguenther@suse.de>
10196         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
10197         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
10198         the map from cache entry to cache index optional.
10199         (streamer_tree_cache_replace_tree): Adjust accordingly.
10200         (streamer_tree_cache_append): Likewise.
10201         (streamer_tree_cache_delete): Likewise.
10202         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
10203         streamer cache map from cache entry to cache index.
10204         * lto-streamer-out.c (create_output_block): Adjust.
10206 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
10208         * config/arm/arm.md (attribute "insn"): Move multiplication and
10209         division attributes to...
10210         (attribute "type"): ... here.  Remove mult.
10211         (attribute "mul32"): New attribute.
10212         (attribute "mul64"): Add umaal.
10213         (*arm_mulsi3): Update attributes.
10214         (*arm_mulsi3_v6): Likewise.
10215         (*thumb_mulsi3): Likewise.
10216         (*thumb_mulsi3_v6): Likewise.
10217         (*mulsi3_compare0): Likewise.
10218         (*mulsi3_compare0_v6): Likewise.
10219         (*mulsi_compare0_scratch): Likewise.
10220         (*mulsi_compare0_scratch_v6): Likewise.
10221         (*mulsi3addsi): Likewise.
10222         (*mulsi3addsi_v6): Likewise.
10223         (*mulsi3addsi_compare0): Likewise.
10224         (*mulsi3addsi_compare0_v6): Likewise.
10225         (*mulsi3addsi_compare0_scratch): Likewise.
10226         (*mulsi3addsi_compare0_scratch_v6): Likewise.
10227         (*mulsi3subsi): Likewise.
10228         (*mulsidi3adddi): Likewise.
10229         (*mulsi3addsi_v6): Likewise.
10230         (*mulsidi3adddi_v6): Likewise.
10231         (*mulsidi3_nov6): Likewise.
10232         (*mulsidi3_v6): Likewise.
10233         (*umulsidi3_nov6): Likewise.
10234         (*umulsidi3_v6): Likewise.
10235         (*umulsidi3adddi): Likewise.
10236         (*umulsidi3adddi_v6): Likewise.
10237         (*smulsi3_highpart_nov6): Likewise.
10238         (*smulsi3_highpart_v6): Likewise.
10239         (*umulsi3_highpart_nov6): Likewise.
10240         (*umulsi3_highpart_v6): Likewise.
10241         (mulhisi3): Likewise.
10242         (*mulhisi3tb): Likewise.
10243         (*mulhisi3bt): Likewise.
10244         (*mulhisi3tt): Likewise.
10245         (maddhisi4): Likewise.
10246         (*maddhisi4tb): Likewise.
10247         (*maddhisi4tt): Likewise.
10248         (maddhidi4): Likewise.
10249         (*maddhidi4tb): Likewise.
10250         (*maddhidi4tt): Likewise.
10251         (divsi3): Likewise.
10252         (udivsi3): Likewise.
10253         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
10254         (thumb2_mulsi_short_compare0): Likewise.
10255         (thumb2_mulsi_short_compare0_scratch): Likewise.
10256         * config/arm/arm1020e.md (1020mult1): Update attribute change.
10257         (1020mult2): Likewise.
10258         (1020mult3): Likewise.
10259         (1020mult4): Likewise.
10260         (1020mult5): Likewise.
10261         (1020mult6): Likewise.
10262         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
10263         change.
10264         (cortex_a15_mult64): Likewise.
10265         (cortex_a15_sdiv): Likewise.
10266         (cortex_a15_udiv): Likewise.
10267         * config/arm/arm1026ejs.md (mult1): Update attribute change.
10268         (mult2): Likewise.
10269         (mult3): Likewise.
10270         (mult4): Likewise.
10271         (mult5): Likewise.
10272         (mult6): Likewise.
10273         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
10274         (pj4_ir_div): Likewise.
10275         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
10276         (11_mult2): Likewise.
10277         (11_mult3): Likewise.
10278         (11_mult4): Likewise.
10279         (11_mult5): Likewise.
10280         (11_mult6): Likewise.
10281         (11_mult7): Likewise.
10282         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
10283         (cortex_a8_mla): Likewise.
10284         (cortex_a8_mull): Likewise.
10285         (cortex_a8_smulwy): Likewise.
10286         (cortex_a8_smlald): Likewise.
10287         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
10288         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
10289         (cortex_r4_mul_3): Likewise.
10290         (cortex_r4_mla_4): Likewise.
10291         (cortex_r4_mla_3): Likewise.
10292         (cortex_r4_smlald): Likewise.
10293         (cortex_r4_mull): Likewise.
10294         (cortex_r4_sdiv): Likewise.
10295         (cortex_r4_udiv): Likewise.
10296         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
10297         (cortex_a7_idiv): Likewise.
10298         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
10299         (9_mult2): Likewise.
10300         (9_mult3): Likewise.
10301         (9_mult4): Likewise.
10302         (9_mult5): Likewise.
10303         (9_mult6): Likewise.
10304         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
10305         (cortex_a53_sdiv): Likewise.
10306         (cortex_a53_udiv): Likewise.
10307         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
10308         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
10309         (mp626_mult2): Likewise.
10310         (mp626_mult3): Likewise.
10311         (mp626_mult4): Likewise.
10312         * config/arm/fa526.md (526_mult1): Update attribute change.
10313         (526_mult2): Likewise.
10314         * config/arm/arm-generic.md (mult): Update attribute change.
10315         (mult_ldsched_strongarm): Likewise.
10316         (mult_ldsched): Likewise.
10317         (multi_cycle): Likewise.
10318         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
10319         * config/arm/fa606te.md (606te_mult1): Update attribute change.
10320         (606te_mult2): Likewise.
10321         (606te_mult3): Likewise.
10322         (606te_mult4): Likewise.
10323         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
10324         (cortex_a9_mac16): Likewise.
10325         (cortex_a9_multiply): Likewise.
10326         (cortex_a9_mac): Likewise.
10327         (cortex_a9_multiply_long): Likewise.
10328         * config/arm/fa626te.md (626te_mult1): Update attribute change.
10329         (626te_mult2): Likewise.
10330         (626te_mult3): Likewise.
10331         (626te_mult4): Likewise.
10333 2013-06-18  Richard Biener  <rguenther@suse.de>
10335         PR lto/57334
10336         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
10338 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10340         PR target/57609
10341         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
10342         with next_active_insn.
10344 2013-06-18  Alan Modra  <amodra@gmail.com>
10346         * config/rs6000/rs6000.h (enum data_align): New.
10347         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
10348         (DATA_ABI_ALIGNMENT): Define.
10349         (CONSTANT_ALIGNMENT): Correct comment.
10350         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
10351         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
10353 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
10355         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
10356         ATTRIBUTE_UNUSED marking.
10358 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
10360         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
10361         alternative and update.
10362         (aarch64_dup_lanedi): Delete.
10363         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
10364         * config/aarch64/aarch64-simd-builtins.def: Update.
10366 2013-06-17  Richard Biener  <rguenther@suse.de>
10368         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
10369         (lto_input_scc): Declare.
10370         (lto_input_tree_1): Likewise.
10371         (struct lto_stats_d): Add num_tree_bodies_output and
10372         num_pickle_refs_output.
10373         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
10374         (lto_read_tree_1): Split out from ...
10375         (lto_read_tree): ... this.
10376         (lto_input_scc): New function.
10377         (lto_input_tree_1): Split out from ...
10378         (lto_input_tree): ... this.  Handle LTO_tree_scc.
10379         (lto_data_in_create): Create the streamer cache without hashes.
10380         * lto-streamer-out.c (create_output_block): Create the streamer
10381         cache with hashes when not doing WPA.
10382         (lto_write_tree_1): Split out from ...
10383         (lto_write_tree): ... this.
10384         (get_symbol_initial_value): New function.
10385         (lto_output_tree_1): Split out from ...
10386         (lto_output_tree): ... this.  Write trees as series of SCCs
10387         using a DFS walk via DFS_write_tree.
10388         (struct sccs, struct scc_entry): New types.
10389         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
10390         (DFS_write_tree_body): New function.
10391         (DFS_write_tree): Likewise.
10392         (hash_tree): Likewise.
10393         (scc_entry_compare): Likewise.
10394         (hash_scc): Likewise.
10395         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
10396         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
10397         TREE_CHAIN as regular reference.
10398         (streamer_read_integer_cst): Remove.
10399         (streamer_get_pickled_tree): Adjust.
10400         * tree-streamer-out.c (streamer_write_chain): Disable streaming
10401         of DECL_EXTERNALs in BLOCK_VARS for now.
10402         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
10403         reference.
10404         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
10405         Add hash value argument and record that if hashes are recorded
10406         in the cache.
10407         (streamer_tree_cache_insert_1): Adjust.
10408         (streamer_tree_cache_insert): Likewise.
10409         (streamer_tree_cache_insert_at): Rename to ...
10410         (streamer_tree_cache_replace_tree): ... this and adjust.
10411         (streamer_tree_cache_append): Adjust.
10412         (record_common_node): Likewise.
10413         (streamer_tree_cache_create): Add argument whether to
10414         record hash values together with trees.
10415         (streamer_tree_cache_delete): Adjust.
10416         * tree-streamer.h (struct streamer_tree_cache_d): Add
10417         vector of hashes.
10418         (streamer_read_integer_cst): Remove.
10419         (streamer_tree_cache_insert): Adjust.
10420         (streamer_tree_cache_append): Likewise.
10421         (streamer_tree_cache_insert_at): Rename to ...
10422         (streamer_tree_cache_replace_tree): ... this and adjust.
10423         (streamer_tree_cache_create): Add argument whether to record hashes.
10424         (streamer_tree_cache_get): Rename to ...
10425         (streamer_tree_cache_get_tree): ... this.
10426         (streamer_tree_cache_get_hash): New function.
10427         * tree.c (cache_integer_cst): New function.
10428         * tree.h (cache_integer_cst): Declare.
10429         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
10430         * lto-symtab.c (lto_varpool_replace_node): Only release
10431         DECL_INITIAL of non-prevailing decls.
10432         * varpool.c (varpool_remove_initializer): Do not release
10433         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
10435 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
10437         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
10438         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
10439         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
10440         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
10441         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
10442         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
10443         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
10444         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
10445         instead of TARGET_64BIT.
10446         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
10447         Require ISA_HAS_<D>DIV.
10449 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
10451         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
10452         (mips*-*-linux*): Move default with_llsc setting to where other
10453         defaults are set.
10454         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
10455         with_arch block.
10456         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
10457         Likewise.  Remove default with_tune setting.  Move default float
10458         setting to its own block.  Handle with_llsc in the same block as above.
10460 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
10462         PR rtl-optimization/57425
10463         PR rtl-optimization/57569
10464         * alias.c (write_dependence_p): Add new parameters mem_mode,
10465         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
10466         Changed all callers.
10467         (canon_anti_dependence): New function.
10468         * cse.c (check_dependence): Use canon_anti_dependence.
10469         * cselib.c (cselib_invalidate_mem): Likewise.
10470         * rtl.h (canon_anti_dependence): Declare.
10472 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
10474         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
10475         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
10476         ".set mips3" for 64-bit targets.
10478 2013-06-15  Dehao Chen  <dehao@google.com>
10480         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
10481         * gimple-low.c (gimple_check_call_matching_types): Likewise.
10482         (gimple_check_call_args): Likewise.
10483         * value-prof.c (check_ic_target): Likewise.
10484         * ipa-inline.c (early_inliner): Likewise.
10485         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
10486         * cgraph.c (cgraph_create_edge_1): Likewise.
10487         (cgraph_make_edge_direct): Likewise.
10489 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
10491         PR target/57615
10492         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
10493         rs6000_output_move_128bit to handle emitting quad memory
10494         operations.  Set attribute length to 8 bytes.
10496 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
10498         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
10499         New pattern.
10500         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
10501         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
10502         (aarch64_<su>mlsl<mode>): Likewise.
10504 2013-06-14  Mike Stump  <mikestump@comcast.net>
10506         * Makefile.in (TARGET_H): Add insn-codes.h.
10508 2013-06-14  Alan Modra  <amodra@gmail.com>
10510         PR middle-end/57134
10511         PR middle-end/57586
10512         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
10513         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
10514         bitfield expansion when EXPAND_MEMORY.
10515         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
10517 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
10519         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
10520         test for clearing quad memory on 32-bit later.
10522 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
10524         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
10525         (fold_negate_expr): Likewise.
10526         (fold_real_zero_addition_p): Handle vectors.
10527         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
10529 2013-06-14  Alan Modra  <amodra@gmail.com>
10531         * varasm.c (force_const_mem): Revert 2013-06-07 change.
10533 2013-06-13  Jan Hubicka  <jh@suse.cz>
10535         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
10536         Local comdats are not externally visible.
10537         * symtab.c (dump_symtab_base): Dump externally visible.
10538         (verify_symtab_base): Verify back links in the symtab hash.
10540 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
10542         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
10543         CONVERT_EXPR as equal nodes.
10545 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
10547         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
10549 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
10551         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
10552         Generalize to complex and vector.
10553         * tree.c (build_all_ones_cst): New function.
10554         * tree.h (build_all_ones_cst): Declare it.
10556 2013-06-13  Alan Modra  <amodra@gmail.com>
10558         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
10559         * config/rs6000/rs6000.md (signbittf2): New insn.
10560         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
10561         (abstf2_internal, cmptf_internal2): Likewise.
10562         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
10564 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
10565             Pat Haugen  <pthaugen@us.ibm.com>
10566             Peter Bergner  <bergner@vnet.ibm.com>
10568         * config/rs6000/rs6000.c (emit_load_locked): Add support for
10569         power8 byte, half-word, and quad-word atomic instructions.
10570         (emit_store_conditional): Likewise.
10571         (rs6000_expand_atomic_compare_and_swap): Likewise.
10572         (rs6000_expand_atomic_op): Likewise.
10574         * config/rs6000/sync.md (larx): Add new modes for power8.
10575         (stcx): Likewise.
10576         (AINT): New mode iterator to include TImode as well as normal
10577         integer modes on power8.
10578         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
10579         that VSX registers are not considered.  Use AINT mode iterator
10580         instead of INT1 to allow inclusion of quad word atomic operations
10581         on power8.
10582         (load_locked<mode>): Likewise.
10583         (store_conditional<mode>): Likewise.
10584         (atomic_compare_and_swap<mode>): Likewise.
10585         (atomic_exchange<mode>): Likewise.
10586         (atomic_nand<mode>): Likewise.
10587         (atomic_fetch_<fetchop_name><mode>): Likewise.
10588         (atomic_nand_fetch<mode>): Likewise.
10589         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
10590         each type.
10591         (ATOMIC): On power8, add QImode, HImode modes.
10592         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
10593         modes that promote to SImode.
10594         (load_lockedti): Convert TImode arguments to PTImode, so that we
10595         get a guaranteed even/odd register pair.
10596         (load_lockedpti): Likewise.
10597         (store_conditionalti): Likewise.
10598         (store_conditionalpti): Likewise.
10600         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
10601         atomic load/store instructions.
10602         (HSI): Likewise.
10604 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
10606         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
10607         loads.
10608         (insn_count): New attribute, with most cases extracted from...
10609         (length): ...here.  Redefine most cases in terms of insn_count.
10610         (single_insn): Delete.
10611         (can_delay): Use insn_count to check for single instructions.
10612         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
10613         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
10614         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
10615         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
10616         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
10617         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
10618         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
10619         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
10620         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
10621         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
10622         rather than "length".
10623         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
10624         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
10625         Use "insn_count" rather than "length".
10626         * config/mips/mips-dsp.md
10627         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
10628         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
10629         length attributes.
10631 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
10633         PR tree-optimization/57361
10634         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
10636 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
10638         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
10639         to split.
10640         (aarch64_simd_combine<mode>): New instruction expansion.
10641         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
10642         function prototype.
10643         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
10644         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
10646 2013-06-12  Jan Hubicka  <jh@suse.cz>
10648         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
10649         decl has when in streaming stage.
10650         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
10651         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
10653 2013-06-12  Roland Stigge  <stigge@antcom.de>
10655         PR target/57578
10656         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
10658 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
10660         PR tree-optimization/57537
10661         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
10662         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
10664 2013-06-12  Richard Biener  <rguenther@suse.de>
10666         * data-streamer.h (streamer_write_char_stream): CSE
10667         obs->current_pointer.
10668         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
10669         streamer_write_char_stream manually and optimize the resulting loop.
10670         (streamer_write_hwi_stream): Likewise.
10672 2013-06-12  Jan Hubicka  <jh@suse.cz>
10674         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
10675         * cgraph.h (varpool_create_empty_node): Declare.
10676         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
10677         duplicated nodes.
10678         * symtab.c (symtab_unregister_node): Be lax about missin entries
10679         in node hash.
10680         (symtab_get_node): Update comment.
10681         * varpool.c (varpool_create_empty_node): Break out from ...
10682         (varpool_node_for_decl): ... here.
10683         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
10685 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
10687         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
10688         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
10689         part.  Use straight-line flow at the end.
10690         <COMPONENT_REF>: Remove superfluous else.
10691         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
10693 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
10695         PR target/56564
10696         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
10697         target hook even for !TREE_PUBLIC decls.  If no resolution info
10698         is available, return false for common and external decls.
10700 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10702         * config/rl78/constraints.md (U): New constraint.
10703         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
10704         valloc attribute.
10706 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
10708         PR target/57589
10709         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
10710         to allow returning address to AT_PLATFORM name.
10712 2013-06-11  Jan Hubicka  <jh@suse.cz>
10714         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
10715         * cgraph.h (symtab_node_base): Add weakref flag.
10716         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
10717         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
10718         (output_weakrefs): Use weakref flag.
10719         * fold-const.c (simple_operand_p): Handle WEAK.
10720         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
10721         * ipa.c (varpool_externally_visible_p): Drop weakref.
10722         (function_and_variable_visibility): Update comment; fix weakref
10723         sanity checks; do not clear DECL_WEAK on them.
10724         * lto-cgraph.c (lto_output_node): update.
10725         (lto_output_varpool_node): Update.
10726         (input_overwrite_node): Update.
10727         (input_node): Update.
10728         (input_varpool_node): Update.
10729         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
10730         (lto_symtab_merge_symbols): Add sanity check.
10731         (lto_symtab_prevailing_decl): Do not special case weakrefs.
10732         * passes.c (rest_of_decl_compilation): Set static flag, too.
10733         * symtab.c (dump_symtab_base): Dump weakref.
10734         (verify_symtab_base): Sanity check weakrefs.
10735         (symtab_make_decl_local): Remove duplicated code.
10736         (symtab_alias_ultimate_target): Simplify.
10737         * varpool.c (varpool_create_variable_alias): Set weakref flag.
10739 2013-06-11  Tom de Vries  <tom@codesourcery.com>
10741         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
10742         sequence_vect == NULL.
10744 2013-06-11  DJ Delorie  <dj@redhat.com>
10746         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
10747         (rl78_unwind_word_mode): New.
10749 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
10751         * final.c (debug_prefix_maps): Make static.
10753 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
10755         * function.c (initial_trampoline): Remove stray copy.
10757 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
10759         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
10761 2013-06-11  Martin Jambor  <mjambor@suse.cz>
10763         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
10764         within bounds at the beginning of the function.
10766 2013-06-11  Alan Modra  <amodra@gmail.com>
10768         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
10769         reporting.
10770         (get_named_section): Don't NULL !DECL_P decl.
10772 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
10774         * doc/invoke.texi (core-avx2): Document.
10775         (slm): Likewise.
10776         (atom): Updated with MOVBE.
10778 2013-06-11  Richard Biener  <rguenther@suse.de>
10780         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
10782 2013-06-11  Anton Blanchard  <anton@samba.org>
10784         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
10785         correct shift value in little-endian mode.
10787 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
10789         PR target/56564
10790         * varasm.c (get_variable_align): Move #endif to the right place.
10792 2013-06-10  Cary Coutant  <ccoutant@google.com>
10794         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
10795         for hash so that hash table traversal order is deterministic.
10797 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
10798             Pat Haugen  <pthaugen@us.ibm.com>
10799             Peter Bergner  <bergner@vnet.ibm.com>
10801         * config/rs6000/vector.md (GPR move splitter): Do not split moves
10802         of vectors in GPRS if they are direct moves or quad word load or
10803         store moves.
10805         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
10806         declaration.
10807         (direct_move_p): Likewise.
10808         (quad_load_store_p): Likewise.
10810         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
10811         classes into bins based on the physical register type.
10812         (reg_class_to_reg_type): Likewise.
10813         (IS_STD_REG_TYPE): Likewise.
10814         (IS_FP_VECT_REG_TYPE): Likewise.
10815         (reload_fpr_gpr): Arrays to determine what insn to use if we can
10816         use direct move instructions.
10817         (reload_gpr_vsx): Likewise.
10818         (reload_vsx_gpr): Likewise.
10819         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
10820         information that is a simplification of register classes.  Also
10821         precalculate direct move reload helpers.
10822         (direct_move_p): New function to return true if the operation can
10823         be done as a direct move instruciton.
10824         (quad_load_store_p): New function to return true if the operation
10825         is a quad memory operation.
10826         (rs6000_legitimize_address): If quad memory, only allow register
10827         indirect for TImode addresses.
10828         (rs6000_legitimate_address_p): Likewise.
10829         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
10830         (rs6000_reload_register_type): Likewise.
10831         (register_to_reg_type): Return register type.
10832         (rs6000_secondary_reload_simple_move): New helper function for
10833         secondary reload and secondary memory needed to identify anything
10834         that is a simple move, and does not need reloading.
10835         (rs6000_secondary_reload_direct_move): New helper function for
10836         secondary reload to identify cases that can be done with several
10837         instructions via the direct move instructions.
10838         (rs6000_secondary_reload_move): New helper function for secondary
10839         reload to identify moves between register types that can be done.
10840         (rs6000_secondary_reload): Add support for quad memory operations
10841         and for direct move.
10842         (rs6000_secondary_memory_needed): Likewise.
10843         (rs6000_debug_secondary_memory_needed): Change argument names.
10844         (rs6000_output_move_128bit): New function to return the move to
10845         use for 128-bit moves, including knowing about the various
10846         limitations of quad memory operations.
10848         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
10849         memory operations.  call rs6000_output_move_128bit for the actual
10850         instruciton(s) to generate.
10851         (vsx_movti_64bit): Likewise.
10853         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
10854         (UNSPEC_P8V_MTVSRWZ): Likewise.
10855         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
10856         (UNSPEC_P8V_MTVSRD): Likewise.
10857         (UNSPEC_P8V_XXPERMDI): Likewise.
10858         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
10859         (UNSPEC_FUSION_GPR): Likewise.
10860         (FMOVE128_GPR): New iterator for direct move.
10861         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
10862         (f32_sv): Likewise.
10863         (f32_dm): Likewise.
10864         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
10865         loads and direct move instructions.
10866         (zero_extendsidi2_lfiwzx): Likewise.
10867         (extendsidi2_lfiwax): Likewise.
10868         (extendsidi2_nocell): Likewise.
10869         (floatsi<mode>2_lfiwax): Likewise.
10870         (lfiwax): Likewise.
10871         (floatunssi<mode>2_lfiwzx): Likewise.
10872         (lfiwzx): Likewise.
10873         (fix_trunc<mode>_stfiwx): Likewise.
10874         (fixuns_trunc<mode>_stfiwx): Likewise.
10875         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
10876         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
10877         (parity<mode>2_cmpb): Set length/type attr.
10878         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
10879         for 'mr.' to fast_compare.
10880         (bpermd_<mode>): Change type attr to popcnt.
10881         (p8_fmrgow_<mode>): New insns for power8 direct move support.
10882         (p8_mtvsrwz_1): Likewise.
10883         (p8_mtvsrwz_2): Likewise.
10884         (reload_fpr_from_gpr<mode>): Likewise.
10885         (p8_mtvsrd_1): Likewise.
10886         (p8_mtvsrd_2): Likewise.
10887         (p8_xxpermdi_<mode>): Likewise.
10888         (reload_vsx_from_gpr<mode>): Likewise.
10889         (reload_vsx_from_gprsf): Likewise.
10890         (p8_mfvsrd_3_<mode>): LIkewise.
10891         (reload_gpr_from_vsx<mode>): Likewise.
10892         (reload_gpr_from_vsxsf): Likewise.
10893         (p8_mfvsrd_4_disf): Likewise.
10894         (multi-word GPR splits): Do not split direct moves or quad memory
10895         operations.
10897 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
10899         * tree-into-ssa.c (interesting_blocks): Make static.
10901 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
10903         PR target/56564
10904         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
10905         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
10906         Use DATA_ABI_ALIGNMENT for that case instead if defined.
10907         (get_variable_align): New function.
10908         (get_variable_section, emit_bss, emit_common,
10909         assemble_variable_contents, place_block_symbol): Use
10910         get_variable_align instead of DECL_ALIGN.
10911         (assemble_noswitch_variable): Add align argument, use it
10912         instead of DECL_ALIGN.
10913         (assemble_variable): Adjust caller.  Use get_variable_align
10914         instead of DECL_ALIGN.
10915         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
10916         caller.
10917         (DATA_ABI_ALIGNMENT): Define.
10918         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
10919         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
10920         opt is false, only return the psABI mandated alignment increase.
10921         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
10922         (DATA_ABI_ALIGNMENT): ... this.
10923         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
10924         (DATA_ABI_ALIGNMENT): ... this.
10925         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
10926         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
10927         (DATA_ABI_ALIGNMENT): ... this.
10928         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
10929         * doc/tm.texi: Regenerated.
10931 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
10933         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
10934         cmp_code to construct REG_EQUAL note.
10936 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
10938         PR target/57568
10939         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
10940         that operands[2] doesn't overlap with operands[0].
10942 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
10943             Jan Hubicka  <jh@suse.cz>
10945         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
10946         hack to mark symbols as used.
10948 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
10950         PR rtl-optimization/57559
10951         * lra-constraints.c (process_alt_operands): Don't discourage
10952         memory with known offset for offsetable memory constraint.
10953         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
10955 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
10957         * varasm.c (struct oc_local_state): Reorder fields.
10958         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
10959         and adjust accordingly.
10960         (output_constructor): Reorder initialization code and adjust call to
10961         output_constructor_bitfield.
10963 2013-06-07  Jan Hubicka  <jh@suse.cz>
10965         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
10967 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
10969         * tree-object-size.c (unknown): Make const.
10971 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10973         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
10974         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
10975         for last alternative in the cpu_facility attribute.
10977 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10979         PR target/56315
10980         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
10981         (xordi3): Change operand 2 constraint to arm_xordi_operand.
10982         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
10983         * config/arm/constraints.md (Dg): New constraint.
10984         * config/arm/neon.md (xordi3_neon): Remove.
10985         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
10986         * config/arm/predicates.md (arm_xordi_operand): New predicate.
10988 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10990         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
10991         Clean up alternatives.
10993 2013-06-07  Alan Modra  <amodra@gmail.com>
10995         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
10996         va_list_gpr_size.
10998 2013-06-07  Alan Modra  <amodra@gmail.com>
11000         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
11002 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11004         * config/arm/constraints.md (Df): New constraint.
11005         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
11006         Correct length attribute for last two alternatives.
11008 2013-06-07  Alan Modra  <amodra@gmail.com>
11010         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
11011         override user -mfp-in-toc.
11012         (offsettable_ok_by_alignment): Consider just the current access
11013         rather than the whole object, unless BLKmode.  Handle
11014         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
11015         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
11016         for -mcmodel=medium.
11017         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
11018         override user -mfp-in-toc or -msum-in-toc.  Default to
11019         -mno-fp-in-toc for -mcmodel=medium.
11021 2013-06-06  DJ Delorie  <dj@redhat.com>
11023         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
11024         TARGET_VALID_POINTER_MODE.
11026 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
11027             Pat Haugen  <pthaugen@us.ibm.com>
11028             Peter Bergner  <bergner@vnet.ibm.com>
11030         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
11031         Document new power8 builtins.
11033         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
11034         condition code register, to allow 128-bit logical operations to be
11035         done in the VSX or GPR registers.
11036         (nor<mode>3): Use the canonical form for nor.
11037         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
11038         vclz*, and vpopcnt* vector instructions.
11039         (nand<mode>3): Likewise.
11040         (orc<mode>3): Likewise.
11041         (clz<mode>2): LIkewise.
11042         (popcount<mode>2): Likewise.
11044         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
11045         that only the GPRs are recognized.
11047         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11048         support for new power8 builtins.
11050         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
11051         builtin functions.
11052         (xscvdpspn): Likewise.
11053         (vclz): Likewise.
11054         (vclzb): Likewise.
11055         (vclzh): Likewise.
11056         (vclzw): Likewise.
11057         (vclzd): Likewise.
11058         (vpopcnt): Likewise.
11059         (vpopcntb): Likewise.
11060         (vpopcnth): Likewise.
11061         (vpopcntw): Likewise.
11062         (vpopcntd): Likewise.
11063         (vgbbd): Likewise.
11064         (vmrgew): Likewise.
11065         (vmrgow): Likewise.
11066         (eqv): Likewise.
11067         (eqv_v16qi3): Likewise.
11068         (eqv_v8hi3): Likewise.
11069         (eqv_v4si3): Likewise.
11070         (eqv_v2di3): Likewise.
11071         (eqv_v4sf3): Likewise.
11072         (eqv_v2df3): Likewise.
11073         (nand): Likewise.
11074         (nand_v16qi3): Likewise.
11075         (nand_v8hi3): Likewise.
11076         (nand_v4si3): Likewise.
11077         (nand_v2di3): Likewise.
11078         (nand_v4sf3): Likewise.
11079         (nand_v2df3): Likewise.
11080         (orc): Likewise.
11081         (orc_v16qi3): Likewise.
11082         (orc_v8hi3): Likewise.
11083         (orc_v4si3): Likewise.
11084         (orc_v2di3): Likewise.
11085         (orc_v4sf3): Likewise.
11086         (orc_v2df3): Likewise.
11088         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
11089         allow power8 quad mode in 64-bit.
11090         (rs6000_builtin_vectorized_function): Add support to vectorize
11091         ISA 2.07 count leading zeros, population count builtins.
11092         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
11093         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
11094         (builtin_function_type): Add vgbbd builtin function which takes an
11095         unsigned argument.
11096         (altivec_expand_vec_perm_const): Add support for new power8 merge
11097         instructions.
11099         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
11100         that does not include TImdoe for use with 32-bit.
11101         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
11102         instructions.
11103         (UNSPEC_VSX_CVDPSPN): Likewise.
11104         (vsx_xscvdpspn): Likewise.
11105         (vsx_xscvspdpn): Likewise.
11106         (vsx_xscvdpspn_scalar): Likewise.
11107         (vsx_xscvspdpn_directmove): Likewise.
11108         (vsx_and<mode>3): Split logical operations into 32-bit and
11109         64-bit. Add support to do logical operations on TImode as well as
11110         VSX vector types.  Allow logical operations to be done in either
11111         VSX registers or in general purpose registers in 64-bit mode.  Add
11112         splitters if GPRs were used. For AND, add clobber of CCmode to
11113         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
11114         encoding.
11115         (vsx_and<mode>3_32bit): Likewise.
11116         (vsx_and<mode>3_64bit): Likewise.
11117         (vsx_ior<mode>3): Likewise.
11118         (vsx_ior<mode>3_32bit): Likewise.
11119         (vsx_ior<mode>3_64bit): Likewise.
11120         (vsx_xor<mode>3): Likewise.
11121         (vsx_xor<mode>3_32bit): Likewise.
11122         (vsx_xor<mode>3_64bit): Likewise.
11123         (vsx_one_cmpl<mode>2): Likewise.
11124         (vsx_one_cmpl<mode>2_32bit): Likewise.
11125         (vsx_one_cmpl<mode>2_64bit): Likewise.
11126         (vsx_nor<mode>3): Likewise.
11127         (vsx_nor<mode>3_32bit): Likewise.
11128         (vsx_nor<mode>3_64bit): Likewise.
11129         (vsx_andc<mode>3): Likewise.
11130         (vsx_andc<mode>3_32bit): Likewise.
11131         (vsx_andc<mode>3_64bit): Likewise.
11132         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
11133         and xxlorc instructions.
11134         (vsx_eqv<mode>3_64bit): Likewise.
11135         (vsx_nand<mode>3_32bit): Likewise.
11136         (vsx_nand<mode>3_64bit): Likewise.
11137         (vsx_orc<mode>3_32bit): Likewise.
11138         (vsx_orc<mode>3_64bit): Likewise.
11140         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
11142         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
11143         instruction.
11144         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
11145         (p8_vmrgow): Likewise.
11146         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
11147         GPRs to be split under VSX.
11148         (p8v_clz<mode>2): Add power8 count leading zero support.
11149         (p8v_popcount<mode>2): Add power8 population count support.
11150         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
11151         support.
11153         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
11154         instruction.
11156         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
11157         builtin functions.
11158         (vec_nand): Likewise.
11159         (vec_vclz): Likewise.
11160         (vec_vclzb): Likewise.
11161         (vec_vclzd): Likewise.
11162         (vec_vclzh): Likewise.
11163         (vec_vclzw): Likewise.
11164         (vec_vgbbd): Likewise.
11165         (vec_vmrgew): Likewise.
11166         (vec_vmrgow): Likewise.
11167         (vec_vpopcnt): Likewise.
11168         (vec_vpopcntb): Likewise.
11169         (vec_vpopcntd): Likewise.
11170         (vec_vpopcnth): Likewise.
11171         (vec_vpopcntw): Likewise.
11173 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
11175         PR rtl-optimization/57468
11176         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
11177         spilled pseudos.
11179 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
11181         PR rtl-optimization/57459
11182         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
11183         type when setting live regs.
11185 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
11187         * config/s390/s390.opt (mlra): New option.
11188         * config/s390/s390.c (s390_decompose_address): Check displacement
11189         for all registers for LRA.
11190         (s390_secondary_reload): Don't used secondary reloads for LRA.
11191         (s390_lra_p): New function.
11192         (TARGET_LRA_P): Define.
11193         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
11194         of attribute cpu_facility to zarch for the last alternative.
11195         (*cmpmem_short): Ditto.
11197 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
11199         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
11200         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
11201         (arm_expand_prologue): Likewise.
11203 2013-06-06  Teresa Johnson  <tejohnson@google.com>
11205         PR c++/53743
11206         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
11207         as this is now done by redirect_edge_and_branch_force.
11208         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
11209         barriers, and fix interaction with splitting.
11210         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
11211         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
11212         reflect changes made in the routine.
11213         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
11214         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
11215         since this is called in cfglayout mode, and replace partition fixup
11216         with assert as that is now done by force_nonfallthru_and_redirect.
11217         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
11218         already be marked with region crossing note.
11219         (insert_section_boundary_note): Make non-static, gate on flag
11220         has_bb_partition, rewrite to also check for multiple partitions.
11221         (rest_of_handle_reorder_blocks): Remove call to
11222         insert_section_boundary_note, now done later during free_cfg.
11223         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
11224         * bb-reorder.h (insert_section_boundary_note): Declare.
11225         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
11226         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
11227         invoke insert_section_boundary_note.
11228         (try_redirect_by_replacing_jump): Remove unnecessary
11229         check for region crossing note.
11230         (fixup_partition_crossing): New function.
11231         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
11232         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
11233         in non-cfglayout mode.
11234         (force_nonfallthru_and_redirect): Fixup partition boundaries,
11235         remove old code that tried to do this. Emit barrier correctly
11236         when we are in cfglayout mode.
11237         (last_bb_in_partition): New function.
11238         (rtl_split_edge): Correctly fixup partition boundaries.
11239         (commit_one_edge_insertion): Remove old code that tried to
11240         fixup region crossing edge since this is now handled in
11241         split_block, and set up insertion point correctly since
11242         block may now end in a jump.
11243         (verify_hot_cold_block_grouping): Guard against checking when not in
11244         linearized RTL mode.
11245         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
11246         notes.
11247         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
11248         rtl_verify_flow_info, so not called in cfglayout mode.
11249         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
11250         (fixup_reorder_chain): Remove old code that attempted to fixup region
11251         crossing note as this is now handled in force_nonfallthru_and_redirect.
11252         (duplicate_insn_chain): Don't duplicate switch section notes.
11253         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
11254         note.
11255         * basic-block.h (emit_barrier_after_bb): Declare.
11257 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11259         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
11260         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
11261         arm_usatsihi): Adjust alternatives for arm_restrict_it.
11263 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11265         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
11266         where appropriate.
11267         * config/arm/ldmstm.md: Regenerate.
11269 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11271         * config/arm/sync.md (atomic_loaddi_1):
11272         Disable predication for arm_restrict_it.
11273         (arm_load_exclusive<mode>): Likewise.
11274         (arm_load_exclusivesi): Likewise.
11275         (arm_load_exclusivedi): Likewise.
11276         (arm_load_acquire_exclusive<mode>): Likewise.
11277         (arm_load_acquire_exclusivesi): Likewise.
11278         (arm_load_acquire_exclusivedi): Likewise.
11279         (arm_store_exclusive<mode>): Likewise.
11280         (arm_store_exclusive<mode>): Likewise.
11281         (arm_store_release_exclusivedi): Likewise.
11282         (arm_store_release_exclusive<mode>): Likewise.
11284 2013-06-06  Richard Biener  <rguenther@suse.de>
11286         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
11287         after LTO_null.
11288         (lto_tag_is_tree_code_p): Adjust.
11289         (lto_tag_is_gimple_code_p): Likewise.
11290         (lto_gimple_code_to_tag): Likewise.
11291         (lto_tag_to_gimple_code): Likewise.
11292         (lto_tree_code_to_tag): Likewise.
11293         (lto_tag_to_tree_code): Likewise.
11294         * data-streamer.h (streamer_write_hwi_in_range): Use
11295         uhwi streaming to stream the normalized range.
11296         (streamer_read_hwi_in_range): Likewise.
11298 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11300         * config/arm/arm.md (enabled_for_depr_it): New attribute.
11301         (predicable_short_it): Likewise.
11302         (predicated): Likewise.
11303         (enabled): Handle above.
11304         (define_cond_exec): Set predicated attribute to yes.
11306 2013-06-05  Mike Stump  <mikestump@comcast.net>
11308         * gdbinit.in (__FUNCTION__): Add.
11310 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
11312         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
11313         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
11315 2013-06-05  Jan Hubicka  <jh@suse.cz>
11317         * varasm.c (mark_decl_referenced): Revert the removal until targets
11318         are fixed.
11320 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
11322         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
11323         instead of mark_decl_referenced.
11325 2013-06-05  Jan Hubicka  <jh@suse.cz>
11327         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
11328         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
11329         and symtab_used_from_object_file_p.
11330         (cgraph_make_node_local_1): Clear forced_by_abi.
11331         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
11332         * cgraph.h (symtab_node_base): Add forced_by_abi.
11333         (decide_is_variable_needed): Remove.
11334         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
11335         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
11336         (decide_is_symbol_needed): ... this one; handle symbols in general;
11337         always analyze virtuals; honnor forced_by_abi.
11338         (cgraph_finalize_function): Update.
11339         (varpool_finalize_decl): Update.
11340         (symbol_defined_and_needed): Remove.
11341         (analyze_functions): Update.
11342         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
11343         output_refs, input_overwrite_node): Handle forced_by_abi.
11344         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
11345         (address_taken_from_non_vtable_p): ... this one.
11346         (comdat_can_be_unshared_p_1): New function.
11347         (cgraph_comdat_can_be_unshared_p): Rename to ...
11348         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
11349         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
11350         (function_and_variable_visibility): Clear forced_by_abi as needed.
11351         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
11352         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
11353         * symtab.c (dump_symtab_base): Dump forced_by_abi.
11354         * varpool.c (decide_is_variable_needed): Remove.
11356 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11358         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
11359         (arm_option_override): Override arm_restrict_it where appropriate.
11360         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
11361         * config/arm/arm.opt (mrestrict-it): New command-line option.
11362         * doc/invoke.texi: Document -mrestrict-it.
11364 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
11366         * tsan.c (tsan_atomic_table): Make const.
11368 2013-06-05  Richard Biener  <rguenther@suse.de>
11370         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
11371         index associated with the tree we are supposed to replace.
11372         * tree-streamer-out.c (pack_ts_base_value_fields): Output
11373         TREE_ASM_WRITTEN as zero for everything but SSA names.
11375 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
11377         * tree-ssa-structalias.c (call_stmt_vars): Make static.
11379 2013-06-04  Jan Hubicka  <jh@suse.cz>
11381         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
11382         (input_node, input_varpool_node): Handle correctly external same
11383         body aliases.
11384         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
11385         nodes at ltrans stage.
11387 2013-06-04  Jan Hubicka  <jh@suse.cz>
11389         * ipa-inline.c (update_caller_keys): Fix availability test.
11390         (update_callee_keys): Likewise.
11391         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
11392         to follow ELF standard.
11394 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
11396         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
11397         (mips64r5900el-*-elf*): New configurations.
11398         * config/mips/mips-cpus.def (r5900): New processor.
11399         * config/mips/mips-tables.opt: Regenerate.
11400         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
11401         (mips_issue_rate): Handle PROCESSOR_R5900.
11402         (mips_reorg_process_insns): Force reorder mode for the R5900.
11403         * config/mips/mips.h (TARGET_MIPS5900): Define.
11404         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
11405         TARGET_MIPS5900.
11406         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
11407         TARGET_MIPS5900.
11408         * config/mips/mips.md (processor): Add r5900.
11409         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
11411 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11413         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
11414         into function to generate MOVI instruction.
11415         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
11416         (aarch64_preferred_simd_mode): Turn into wrapper.
11417         (aarch64_output_scalar_simd_mov_immediate): New function.
11418         * config/aarch64/aarch64-protos.h: Add prototype for above.
11420 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11422         * config/aarch64/aarch64.c (simd_immediate_info): Remove
11423         element_char member.
11424         (sizetochar): Return signed char.
11425         (aarch64_simd_valid_immediate): Remove elchar and other
11426         unnecessary variables.
11427         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
11428         Calculate element_char as required.
11429         * config/aarch64/aarch64-protos.h: Update and move prototype
11430         for aarch64_output_simd_mov_immediate.
11431         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11432         Update arguments.
11434 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11436         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
11437         information completed by aarch64_simd_valid_immediate.
11438         (aarch64_legitimate_constant_p): Update arguments.
11439         (aarch64_simd_valid_immediate): Work with struct rather than many
11440         pointers.
11441         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
11442         (aarch64_simd_make_constant): Update arguments.
11443         (aarch64_output_simd_mov_immediate): Work with struct rather than
11444         many pointers.  Output immediate directly rather than as operand.
11445         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
11446         Update prototype.
11447         * config/aarch64/constraints.md (Dn): Update arguments.
11449 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11451         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
11452         longer static.
11453         (aarch64_simd_immediate_valid_for_move): Remove.
11454         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
11455         (aarch64_simd_make_constant): Update call.
11456         (aarch64_output_simd_mov_immediate): Update call.
11457         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
11458         Add prototype.
11459         * config/aarch64/constraints.md (Dn): Update call.
11461 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
11463         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
11464         return type to bool for prototype.
11465         (aarch64_legitimate_constant_p): Check for true instead of not -1.
11466         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
11467         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
11469 2013-06-04  Catherine Moore  <clm@codesourcery.com>
11471         * config/mips/mips.opt (meva): New.
11472         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
11473         (ASM_SPEC): Handle -meva.
11474         * doc/invoke.texi (meva):  Document.
11476 2013-06-04  Alan Modra  <amodra@gmail.com>
11478         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
11479         constant output.
11481 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11483         * rtl.def: Add extra fourth optional field to define_cond_exec.
11484         * gensupport.c (process_one_cond_exec): Process attributes from
11485         define_cond_exec.
11486         * doc/md.texi: Document fourth field in define_cond_exec.
11488 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
11490         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
11491         out the processing order as in store_bit_field_1.
11493 2013-06-04  Jan Hubicka  <jh@suse.cz>
11495         PR middle-end/57500
11496         * cgraphunit.c (cgraph_process_same_body_aliases): Create
11497         non-VAR_DECL node if it does not exist yet.
11499 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
11501         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
11502         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
11503         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
11504         target_cpu_default setting.
11506 2013-06-03  Teresa Johnson  <tejohnson@google.com>
11508         * dumpfile.c (opt_info_switch_p): Change -fopt-info
11509         default to -fopt-info=optimized instead of all.
11510         * doc/invoke.texi: Ditto.
11511         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
11512         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
11513         (execute_vect_slp): Emit BB vectorization success under
11514         MSG_OPTIMIZED_LOCATIONS.
11515         * tree-vect-slp.c (vect_slp_transform_bb): Change
11516         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
11517         * tree-vect-loop.c (vect_transform_loop): Ditto.
11519 2013-06-03  Jason Merrill  <jason@redhat.com>
11521         PR c++/57415
11522         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11523         Use TARGET_EXPR for C++.
11525 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
11527         PR rtl-optimization/57268
11528         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
11529         if DEBUG_INSN_P (insn).
11531         Reapply
11532         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
11534         PR rtl-optimization/57268
11535         * sched-deps.c (sched_analyze_2): Flush dependence lists if
11536         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11538 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11540         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
11541         (ix86_avoid_lea_for_addr): Likewise.
11542         (exact_dependency_1): Likewise.
11543         (ix86_adjust_cost): Likewise.
11544         (swap_top_of_ready_list): Fix formatting and !reload_completed check
11545         removed.
11546         (do_reorder_for_imul): Fix typo, formatting and
11547         !reload_completed check removed.
11548         (ix86_sched_reorder): Fix typo and formatting.
11549         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
11550         list.
11552 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
11554         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
11556 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
11558         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
11559         <STRING_CST>: Likewise.
11560         <VECTOR_CST>: Likewise.
11562 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
11563             Mikael Morin  <mikael@gcc.gnu.org>
11565         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
11566         * config.in: Regenerated.
11567         * configure: Regenerated.
11569 2013-06-01  Jan Hubicka  <jh@suse.cz>
11571         PR middle-end/57366
11572         * cgraphunit.c (compile): When weakref is not supported,
11573         set up transparent aliases before final output pass.
11574         * varasm.c (assemble_alias): Do not try to do it here.
11576 2013-06-01  Jan Hubicka  <jh@suse.cz>
11578         PR middle-end/57467
11579         * passes.c (for_per_function): Skip unanalyzed functions.
11581 2013-06-01  Jan Hubicka  <jh@suse.cz>
11583         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
11584         (lto_symtab_merge_symbols_1): ... this one.
11585         (lto_symtab_merge_cgraph_nodes): Rename to ...
11586         (lto_symtab_merge_symbols): ... this one; simplify.
11587         * cgraph.c (same_body_aliases_done): Rename to ...
11588         (cpp_implicit_aliases_done): ... this one.
11589         (cgraph_create_function_alias): Update.
11590         (cgraph_same_body_alias): Update.
11591         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
11592         (verify_edge_corresponds_to_fndecl): Simplify.
11593         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
11594         (cgraph_node): Remove same_body_alias.
11595         (varpool_node): Remove alias_of and extra_name_alias.
11596         (same_body_aliases_done): Rename to ..
11597         (cpp_implicit_aliases_done): ... this one.
11598         (symtab_alias_ultimate_target): Add default parameter.
11599         (symtab_resolve_alias): New function.
11600         (fixup_same_cpp_alias_visibility): Declare.
11601         (cgraph_function_node): Add default parameter.
11602         (cgraph_node_asm_name): Likewise.
11603         (cgraph_function_or_thunk_node): Add default parameter; do
11604         not ICE when it is NULL.
11605         (varpool_variable_node): Likewise.
11606         * tree-emutls.c (create_emultls_var): Update.
11607         (ipa_lower_emutls): Update.
11608         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
11609         (cgraph_reset_node): Reset alias info.
11610         (cgraph_finalize_function): Update.
11611         (fixup_same_cpp_alias_visibility): Move to symtab.c.
11612         (analyze_function): Simplify.
11613         (cgraph_process_same_body_aliases): Simplify.
11614         (analyze_functions): Fixup same body aliases.
11615         (handle_alias_pairs): Simplify.
11616         (assemble_thunk): Update.
11617         (assemble_thunks_and_aliases): Update.
11618         (output_weakrefs): Rewrite.
11619         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
11620         (lto_output_varpool_node): Likewise.
11621         (compute_ltrans_boundary): Remve assert.
11622         (get_alias_symbol): New functoin.
11623         (input_node): Rewrite alias handling.
11624         (input_varpool_node): Likewise.
11625         * ipa-pure-const.c (propagate_pure_const): Fix formating.
11626         * ipa.c (process_references): Handle weakrefs correctly.
11627         (symtab_remove_unreachable_nodes): Likewise.
11628         * trans-mem.c (get_cg_data): Update.
11629         (ipa_tm_create_version_alias): Update.
11630         (ipa_tm_execute): Update.
11631         * symtab.c (dump_symtab_base): Dump aliases.
11632         (verify_symtab_base): Verify aliases.
11633         (symtab_node_availability): New function.
11634         (symtab_alias_ultimate_target): Simplify.
11635         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
11636         handle all the fixup cases.
11637         (symtab_resolve_alias): New function.
11638         * passes.c (ipa_write_summaries): Handle weakrefs.
11639         * varpool.c (varpool_analyze_node): Simplify.
11640         (assemble_aliases): Update.
11641         (varpool_create_variable_alias): Simplify.
11642         (varpool_extra_name_alias): Simplify.
11643         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
11644         (lto_symtab_merge_symbols): ... this one.
11646 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
11648         Revert
11649         PR rtl-optimization/57268
11650         * sched-deps.c (sched_analyze_2): Flush dependence lists if
11651         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11653 2013-06-01  Tobias Burnus  <burnus@net-b.de>
11655         Partially reverted:
11656         2013-05-31  Tobias Burnus  <burnus@net-b.de>
11658         PR middle-end/57073
11659         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
11660         further up.
11662 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
11664         PR rtl-optimization/57268
11665         * sched-deps.c (sched_analyze_2): Flush dependence lists if
11666         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
11668 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
11670         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
11671         unordered comparison operators when -fno-trapping-math is in effect
11672         on the e500.
11673         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
11674         and implement unordered comparison operators properly on the e500.
11676 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
11678         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
11679         for constant scalar integers.
11680         (simplify_relational_operation_1): Likewise.
11682 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
11684         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
11685         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
11686         Fix comment.
11688 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11689             Igor Zamyatin  <igor.zamyatin@intel.com>
11691         Silvermont (SLM) architecture performance tuning.
11692         * config/i386/i386.h (enum ix86_tune_indices): Add
11693         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
11694         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
11696         * config/i386/i386.c (initial_ix86_tune_features)
11697         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
11698         (ix86_lea_outperforms): Handle Silvermont tuning.
11699         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
11700         call.
11701         (ix86_use_lea_for_mov): Likewise.
11702         (ix86_avoid_lea_for_addr): Likewise.
11703         (ix86_lea_for_add_ok): Likewise.
11704         (exact_dependency_1): New function.
11705         (exact_store_load_dependency): Likewise.
11706         (ix86_adjust_cost): Handle Silvermont tuning.
11707         (do_reoder_for_imul): Likewise.
11708         (swap_top_of_ready_list): New function.
11709         (ix86_sched_reorder): Changed to handle Silvermont tuning.
11711         * config/i386/i386.md (peepholes that split memory operand in fp
11712         converts): New.
11714 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11716         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11717         Remove un-necessary braces.
11719 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11721         * config/aarch64/aarch64.c (aarch64_classify_symbol):
11722         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
11724 2013-05-31  Tobias Burnus  <burnus@net-b.de>
11726         PR middle-end/57073
11727         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
11729 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11731         PR target/56315
11732         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
11733         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
11734         * config/arm/neon.md (iordi3_neon): Remove.
11735         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
11736         * config/arm/predicates.md (imm_for_neon_logic_operand):
11737         Move to earlier in the file.
11738         (neon_logic_op2): Likewise.
11739         (arm_iordi_operand_neon): New predicate.
11741 2013-05-31  Richard Biener  <rguenther@suse.de>
11743         PR tree-optimization/57478
11744         PR tree-optimization/57453
11745         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
11746         are life as well.
11748 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11750         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
11751         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
11753 2013-05-30  Tobias Burnus  <burnus@net-b.de>
11754             Thomas Koenig  <tkoenig@gcc.gnu.org>
11756         PR middle-end/57073
11757         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
11758         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
11760 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
11762         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
11764 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
11766         * target.def (register_usage_leveling_p): New hook.
11767         * targhooks.c (default_register_usage_leveling_p): New.
11768         * targhooks.h (default_register_usage_leveling_p): New prototype.
11769         * lra-assigns.c (register_usage_leveling_p): Use the hook.
11770         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
11771         * doc/tm.texi: Update.
11772         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
11774 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
11776         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
11777         (*insv_reg<mode>): New define_insn.
11779 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
11781         PR rtl-optimization/57439
11782         * postreload.c (move2add_valid_value_p): Check that we have
11783         a zero subreg_regno_offset when accessing the register in
11784         the requested mode.
11786 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11787             Igor Zamyatin  <igor.zamyatin@intel.com>
11789         Silvermont (SLM) architecture pipeline model, tuning and
11790         insn selection.
11791         * config.gcc: Add slm config options and target.
11793         * config/i386/slm.md: New.
11795         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
11797         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
11798         PROCESSOR_SLM.
11799         (ix86_target_macros_internal): Likewise.
11801         * config/i386/i386.c (slm_cost): New cost.
11802         (m_SLM): New macro flag.
11803         (initial_ix86_tune_features): Set m_SLM.
11804         (x86_accumulate_outgoing_args): Likewise.
11805         (x86_arch_always_fancy_math_387): Likewise.
11806         (processor_target_table): Add slm cost.
11807         (cpu_names): Add slm cpu name.
11808         (x86_option_override_internal): Set SLM ISA.
11809         (ix86_issue_rate): New case PROCESSOR_SLM.
11810         (ia32_multipass_dfa_lookahead): Likewise.
11811         (fold_builtin_cpu): Add slm.
11813         * config/i386/i386.h (TARGET_SLM): New target macro.
11814         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
11815         (processor_type): Add PROCESSOR_SLM.
11817         * config/i386/i386.md (cpu): Add new value "slm".
11818         (slm.md): Include slm.md.
11820 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
11821             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11823         * config/arm/arm-protos.h: Add and update function protos.
11824         * config/arm/arm.c (use_simple_return_p): New added.
11825         (thumb2_expand_return): Check simple_return flag.
11826         * config/arm/arm.md: Add simple_return and conditional simple_return.
11827         * config/arm/iterators.md: Add iterator for return and simple_return.
11829 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11831         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
11832         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
11833         (arm_emit_vfp_multi_reg_pop): Likewise.
11834         (thumb2_emit_ldrd_pop): Likewise.
11835         (arm_expand_epilogue): Add misc REG_CFA notes.
11836         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
11838 2013-05-29  Lawrence Crowl  <crowl@google.com>
11840         * config/arm/t-arm: Update for below.
11842         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
11843         Change type to hash_table.  Update dependent calls and types.
11845         * config/i386/t-cygming: Update for below.
11847         * config/i386/t-interix: Update for below.
11849         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
11850         Change type to hash_table.  Update dependent calls and types.
11851         (i386_find_on_wrapper_list::wrappers): Likewise.
11853         * config/ia64/t-ia64: Update for below.
11855         * config/ia64/ia64.c (bundle_state_table):
11856         Change type to hash_table.  Update dependent calls and types.
11858         * config/mips/mips.c (mips_reorg_process_insns::htab):
11859         Change type to hash_table.  Update dependent calls and types.
11861         * config/sol2.c (solaris_comdat_htab):
11862         Change type to hash_table.  Update dependent calls and types.
11864         * config/t-sol2: Update for above.
11866 2013-05-29  Teresa Johnson  <tejohnson@google.com>
11868         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
11869         functions are not yet marked as defined.
11871 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
11872             Pat Haugen  <pthaugen@us.ibm.com>
11873             Peter Bergner  <bergner@vnet.ibm.com>
11875         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
11876         instructions.
11877         (VEC_A): Likewise.
11878         (VEC_C): Likewise.
11879         (vrotl<mode>3): Likewise.
11880         (vashl<mode>3): Likewise.
11881         (vlshr<mode>3): Likewise.
11882         (vashr<mode>3): Likewise.
11884         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11885         support for power8 V2DI builtins.
11887         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
11888         power8 V2DI builtins.
11889         (vupkhsw): Likewise.
11890         (vupklsw): Likewise.
11891         (vaddudm): Likewise.
11892         (vminsd): Likewise.
11893         (vmaxsd): Likewise.
11894         (vminud): Likewise.
11895         (vmaxud): Likewise.
11896         (vpkudum): Likewise.
11897         (vpksdss): Likewise.
11898         (vpkudus): Likewise.
11899         (vpksdus): Likewise.
11900         (vrld): Likewise.
11901         (vsld): Likewise.
11902         (vsrd): Likewise.
11903         (vsrad): Likewise.
11904         (vsubudm): Likewise.
11905         (vcmpequd): Likewise.
11906         (vcmpgtsd): Likewise.
11907         (vcmpgtud): Likewise.
11908         (vcmpequd_p): Likewise.
11909         (vcmpgtsd_p): Likewise.
11910         (vcmpgtud_p): Likewise.
11911         (vupkhsw): Likewise.
11912         (vupklsw): Likewise.
11913         (vaddudm): Likewise.
11914         (vmaxsd): Likewise.
11915         (vmaxud): Likewise.
11916         (vminsd): Likewise.
11917         (vminud): Likewise.
11918         (vpksdss): Likewise.
11919         (vpksdus): Likewise.
11920         (vpkudum): Likewise.
11921         (vpkudus): Likewise.
11922         (vrld): Likewise.
11923         (vsld): Likewise.
11924         (vsrad): Likewise.
11925         (vsrd): Likewise.
11926         (vsubudm): Likewise.
11928         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
11929         support for power8 V2DI instructions.
11931         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
11932         power8 V2DI instructions.  Combine pack and unpack insns to use an
11933         iterator for each mode.  Check whether a particular mode supports
11934         Altivec instructions instead of just checking TARGET_ALTIVEC.
11935         (UNSPEC_VPKUWUM): Likewise.
11936         (UNSPEC_VPKSHSS): Likewise.
11937         (UNSPEC_VPKSWSS): Likewise.
11938         (UNSPEC_VPKUHUS): Likewise.
11939         (UNSPEC_VPKSHUS): Likewise.
11940         (UNSPEC_VPKUWUS): Likewise.
11941         (UNSPEC_VPKSWUS): Likewise.
11942         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
11943         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
11944         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
11945         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
11946         (UNSPEC_VUPKHSB): Likewise.
11947         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
11948         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
11949         (UNSPEC_VUPKHSH): Likewise.
11950         (UNSPEC_VUPKLSB): Likewise.
11951         (UNSPEC_VUPKLSH): Likewise.
11952         (VI2): Likewise.
11953         (VI_char): Likewise.
11954         (VI_scalar): Likewise.
11955         (VI_unit): Likewise.
11956         (VP): Likewise.
11957         (VP_small): Likewise.
11958         (VP_small_lc): Likewise.
11959         (VU_char): Likewise.
11960         (add<mode>3): Likewise.
11961         (altivec_vaddcuw): Likewise.
11962         (altivec_vaddu<VI_char>s): Likewise.
11963         (altivec_vadds<VI_char>s): Likewise.
11964         (sub<mode>3): Likewise.
11965         (altivec_vsubcuw): Likewise.
11966         (altivec_vsubu<VI_char>s): Likewise.
11967         (altivec_vsubs<VI_char>s): Likewise.
11968         (altivec_vavgs<VI_char>): Likewise.
11969         (altivec_vcmpbfp): Likewise.
11970         (altivec_eq<mode>): Likewise.
11971         (altivec_gt<mode>): Likewise.
11972         (altivec_gtu<mode>): Likewise.
11973         (umax<mode>3): Likewise.
11974         (smax<mode>3): Likewise.
11975         (umin<mode>3): Likewise.
11976         (smin<mode>3): Likewise.
11977         (altivec_vpkuhum): Likewise.
11978         (altivec_vpkuwum): Likewise.
11979         (altivec_vpkshss): Likewise.
11980         (altivec_vpkswss): Likewise.
11981         (altivec_vpkuhus): Likewise.
11982         (altivec_vpkshus): Likewise.
11983         (altivec_vpkuwus): Likewise.
11984         (altivec_vpkswus): Likewise.
11985         (altivec_vpks<VI_char>ss): Likewise.
11986         (altivec_vpks<VI_char>us): Likewise.
11987         (altivec_vpku<VI_char>us): Likewise.
11988         (altivec_vpku<VI_char>um): Likewise.
11989         (altivec_vrl<VI_char>): Likewise.
11990         (altivec_vsl<VI_char>): Likewise.
11991         (altivec_vsr<VI_char>): Likewise.
11992         (altivec_vsra<VI_char>): Likewise.
11993         (altivec_vsldoi_<mode>): Likewise.
11994         (altivec_vupkhsb): Likewise.
11995         (altivec_vupkhs<VU_char>): Likewise.
11996         (altivec_vupkls<VU_char>): Likewise.
11997         (altivec_vupkhsh): Likewise.
11998         (altivec_vupklsb): Likewise.
11999         (altivec_vupklsh): Likewise.
12000         (altivec_vcmpequ<VI_char>_p): Likewise.
12001         (altivec_vcmpgts<VI_char>_p): Likewise.
12002         (altivec_vcmpgtu<VI_char>_p): Likewise.
12003         (abs<mode>2): Likewise.
12004         (vec_unpacks_hi_v16qi): Likewise.
12005         (vec_unpacks_hi_v8hi): Likewise.
12006         (vec_unpacks_lo_v16qi): Likewise.
12007         (vec_unpacks_hi_<VP_small_lc>): Likewise.
12008         (vec_unpacks_lo_v8hi): Likewise.
12009         (vec_unpacks_lo_<VP_small_lc>): Likewise.
12010         (vec_pack_trunc_v8h): Likewise.
12011         (vec_pack_trunc_v4si): Likewise.
12012         (vec_pack_trunc_<mode>): Likewise.
12014         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
12015         V2DI builtins.
12016         (vec_vmaxsd): Likewise.
12017         (vec_vmaxud): Likewise.
12018         (vec_vminsd): Likewise.
12019         (vec_vminud): Likewise.
12020         (vec_vpksdss): Likewise.
12021         (vec_vpksdus): Likewise.
12022         (vec_vpkudum): Likewise.
12023         (vec_vpkudus): Likewise.
12024         (vec_vrld): Likewise.
12025         (vec_vsld): Likewise.
12026         (vec_vsrad): Likewise.
12027         (vec_vsrd): Likewise.
12028         (vec_vsubudm): Likewise.
12029         (vec_vupkhsw): Likewise.
12030         (vec_vupklsw): Likewise.
12032 2013-05-29  Jan Hubicka  <jh@suse.cz>
12034         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
12035         flags; reorder rest of fields in more consistent way.
12036         (varpool_node): Remove analyzed, finalized and alias.
12037         (cgraph_ndoe): Likewise.
12038         (symtab_alias_ultimate_target): New function.
12039         (cgraph_function_node): Move offline.
12040         (cgraph_reset_node): Declare.
12041         (cgraph_comdat_can_be_unshared_p): Remove.
12042         (varpool_remove_initializer): Declare.
12043         (varpool_first_defined_variable, varpool_next_defined_variable
12044         cgraph_first_defined_function, cgraph_next_defined_function): Update.
12045         (cgraph_function_with_gimple_body_p): Update.
12046         (varpool_all_refs_explicit_p): Update.
12047         (symtab_alias_target): New function.
12048         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
12049         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
12050         (cgraph_function_or_thunk_node): Simplify using
12051         symtab_alias_ultimate_target.
12052         (varpool_variable_node): Likewise.
12053         * cgraph.c (cgraph_create_function_alias): Update.
12054         (cgraph_add_thunk): Update.
12055         (cgraph_remove_node): Update.
12056         (dump_cgraph_node): Do not dump removed flags.
12057         (cgraph_function_body_availability): Update.
12058         (cgraph_propagate_frequency): Update.
12059         (verify_cgraph_node): Check sanity of local flag.
12060         (cgraph_function_node): Move here from cgraph.h; revamp for
12061         cgraph_function_or_thunk_node.
12062         * lto-symtab.c (lto_varpool_replace_node): Update.
12063         (lto_symtab_resolve_can_prevail_p): Update.
12064         (lto_symtab_merge_cgraph_nodes): Update.
12065         * ipa-cp.c (determine_versionability, initialize_node_lattices,
12066         propagate_constants_accross_call, devirtualization_time_bonus,
12067         ipcp_propagate_stage): Update.
12068         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
12069         * ipa-inline-transform.c (clone_inlined_nodes,
12070         preserve_function_body_p): Update.
12071         * ipa-reference.c (propagate): Update.
12072         (write_node_summary_p): Update.
12073         * toplev.c (wrapup_global_declaration_2): Update.
12074         * cgraphunit.c (cgraph_analyze_function): Rename to ...
12075         (analyze_function) ... this one.
12076         (cgraph_process_new_functions): Update.
12077         (cgraph_reset_node): Export.
12078         (cgraph_finalize_function): Update.
12079         (cgraph_add_new_function): Update.
12080         (process_function_and_variable_attributes): Update.
12081         (varpool_finalize_decl): Update.
12082         (symbol_finalized): Remove.
12083         (symbol_finalized_and_needed): Rename to ...
12084         (symbol_defined_and_needed): ... update.
12085         (cgraph_analyze_functions): Update.
12086         (handle_alias_pairs): Update.
12087         (mark_functions_to_output): Update.
12088         (assemble_thunk): Update.
12089         (output_in_order): Update.
12090         (output_weakrefs): Update.
12091         (finalize_compilation_unit): Update.
12092         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
12093         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
12094         input_node, input_varpool_node): Update.
12095         * dbxout.c (dbxout_expand_expr): Update.
12096         * cgraphclones.c (cgraph_clone_node): Update.
12097         (cgraph_copy_node_for_versioning): Update.
12098         (cgraph_materialize_clone): Update.
12099         (cgraph_materialize_all_clones): Update.
12100         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
12101         propagate_pure_const, propagate_nothrow): Update.
12102         * lto-streamer-out.c (lto_output, write_symbol): Update.
12103         * ipa-utils.c (ipa_reverse_postorder): Update.
12104         * ipa-inline.c (can_inline_edge_p): Update.
12105         (update_caller_keys, ipa_inline): Update.
12106         * dwarf2out.c (reference_to_unused,
12107         premark_types_used_by_global_vars_helper): Update.
12108         * tree-eh.c (tree_could_trap_p): Update.
12109         * ipa-split.c (consider_split, execute_split_functions): Update.
12110         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
12111         has_addr_references_p): Update; move ahead in file for better
12112         readability.
12113         (process_references): Simplify.
12114         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
12115         bodies are removed.
12116         (cgraph_comdat_can_be_unshared_p): Make static.
12117         (cgraph_externally_visible_p): Update.
12118         (varpool_externally_visible_p): Update.
12119         (function_and_variable_visibility): Update.
12120         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
12121         ipa_tm_mark_force_output_node): Update.
12122         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
12123         estimate_edge_devirt_benefit, inline_generate_summary,
12124         inline_write_summary): Update.
12125         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
12126         * ipa-prop.c (ipa_compute_jump_functions): Update.
12127         (ipa_print_node_params, ipa_prop_read_section,
12128         ipa_update_after_lto_read, read_replacements_section): Update.
12129         * varasm.c (mark_decl_referenced): Update.
12130         (assemble_alias, dump_tm_clone_pairs): Update.
12131         * tree-inline.c (copy_bb): Update.
12132         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
12133         Update.
12134         * symtab.c (dump_symtab_base): Print new flags.
12135         (verify_symtab_base): Verify new flags.
12136         (symtab_alias_ultimate_target): New function.
12137         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
12138         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
12139         Update.
12140         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
12141         Update.
12142         * i386.c (ix86_get_function_versions_dispatcher,
12143         ix86_generate_version_dispatcher_body): Update.
12144         (fold_builtin_cpu): Use varpool_add_new_variable.
12145         * varpool.c (varpool_remove_initializer): Break out from ...
12146         (varpool_remove_node): ... this one.
12147         (dump_varpool_node, varpool_node_for_asm,
12148         cgraph_variable_initializer_availability, varpool_analyze_node,
12149         varpool_assemble_decl, varpool_remove_unreferenced_decls,
12150         varpool_finalize_named_section_flags, varpool_create_variable_alias):
12151         Update.
12153 2013-05-29  Jan Hubicka  <jh@suse.cz>
12155         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
12157 2013-05-29  Easwaran Raman  <eraman@google.com>
12159         PR tree-optimization/57442
12160         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
12161         when control exits the main loop.
12163 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
12165         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
12166         and RX600.
12167         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
12168         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
12169         * rx/t-rx: Add rx100 under multi library matches option for nofpu
12170         option.
12172 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12174         PR tree-optimization/57441
12175         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
12176         Don't limit size of incr_vec to number of candidates.
12178 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
12180         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
12181         and mips16 directories.
12182         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
12183         (MULTILIB_DIRNAMES): Ditto.
12184         (MULTILIB_EXCEPTIONS): Add new exceptions.
12185         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
12186         (MULTILIB_DIRNAMES): Ditto.
12187         (MULTILIB_EXCEPTIONS): Add new exceptions.
12189 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12190             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12192         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
12193         SYMBOL_TINY_ABSOLUTE.
12194         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
12195         SYMBOL_TINY_ABSOLUTE.
12196         (aarch64_expand_mov_immediate): Likewise.
12197         (aarch64_classify_symbol): Likewise.
12198         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
12199         Permit SYMBOL_TINY_ABSOLUTE.
12200         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
12202 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12203             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12205         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
12206         Refactor if/switch.  Replace gcc_assert with if.
12208 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12210         * config/i386/i386.c (initial_ix86_tune_features): Enable
12211         FP Reassociation for AMD bdver1 and bdver2.
12213 2013-05-29  Martin Jambor  <mjambor@suse.cz>
12215         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
12216         and IMAGPART_EXPR do not occur within other handled_components.
12218 2013-05-29  Richard Biener  <rguenther@suse.de>
12220         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
12221         access on whether the use is in the BB we currently try to
12222         vectorize.
12223         (vect_bb_vectorization_profitable_p): Pass the BB we currently
12224         vectorize to vect_bb_slp_scalar_cost.
12226 2013-05-29  Richard Biener  <rguenther@suse.de>
12228         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
12229         computing scalar cost offsetted by stmts that are kept live
12230         by scalar uses.
12231         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
12232         for computation of scalar cost.
12234 2013-05-28  Steve Ellcey  <sellcey@mips.com>
12236         * config/mips/mips-cpus.def (mips32r2): Change processor type.
12238 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
12240         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
12241         array notation built-in reduction functions.
12242         * doc/passes.texi (Passes): Added documentation about changes done
12243         for Cilk Plus.
12244         * doc/invoke.texi (C Dialect Options): Added documentation about
12245         the -fcilkplus flag.
12246         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
12247         (BUILTINS_DEF): Depend on cilkplus.def.
12248         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
12249         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
12250         * cilkplus.def: New file.
12252 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
12254         PR rtl-optimization/57439
12255         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
12257 2013-05-28  Easwaran Raman  <eraman@google.com>
12259         PR tree-optimization/57337
12260         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
12261         (find_insert_point): Correctly identify the insertion point
12262         when two statements with the same UID is compared.
12264 2013-05-28  Richard Biener  <rguenther@suse.de>
12266         PR tree-optimization/56787
12267         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
12268         from the list of data references.
12269         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
12270         clobbers.
12271         (vect_analyze_loop_operations): Likewise.
12272         (vect_transform_loop): Remove clobbers.
12274 2013-05-28  Martin Jambor  <mjambor@suse.cz>
12276         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
12277         and REALPART_EXPRs have scalar type.
12279 2013-05-28  Richard Biener  <rguenther@suse.de>
12281         PR tree-optimization/57411
12282         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
12283         virtual operands.
12284         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
12285         virtual operand propagation.
12287 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
12289         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
12290         destination register for bmasksi_vis.
12291         (vector_init_bshuffle): Likewise.
12292         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
12294 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
12296         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
12297         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
12298         mode if the instruction isn't available in the original mode.
12299         * config/sparc/sparc.opt (mfix-ut699): New option.
12300         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
12301         (divdf3): Turn into expander.
12302         (divdf3_nofix): New insn.
12303         (divdf3_fix): Likewise.
12304         (divsf3): Disable if -mfix-ut699.
12305         (sqrtdf2): Turn into expander.
12306         (sqrtdf2_nofix): New insn.
12307         (sqrtdf2_fix): Likewise.
12308         (sqrtsf2): Disable if -mfix-ut699.
12310 2013-05-27  Richard Biener  <rguenther@suse.de>
12312         PR middle-end/57412
12313         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
12314         block for the new loop.
12316 2013-05-27  Richard Biener  <rguenther@suse.de>
12318         PR tree-optimization/57343
12319         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
12320         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
12321         (number_of_iterations_cond): Do not build the folded tree.
12323 2013-05-27  Richard Biener  <rguenther@suse.de>
12325         Revert
12326         PR middle-end/57381
12327         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
12328         OEP_CONSTANT_ADDRESS_OF retained.
12330         PR tree-optimization/57417
12331         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
12332         for unchanged base.
12333         (set_ssa_val_to): Compare addresses using
12334         get_addr_base_and_unit_offset.
12336 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
12338         PR rtl-optimization/56833
12339         * postreload.c (move2add_record_mode): New function.
12340         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
12341         (move2add_use_add2_insn): Use move2add_record_sym_value.
12342         (move2add_use_add3_insn): Likewise.
12343         (reload_cse_move2add): Use move2add_valid_value_p and
12344         move2add_record_mode.  Invalidate call-clobbered and REG_INC
12345         affected regs by setting reg_mode to VOIDmode.
12346         (move2add_note_store): Don't pretend the inside of a SUBREG is
12347         the actual destination.  Invalidate single/leading registers by
12348         setting reg_mode to VOIDmode.
12349         Use move2add_record_sym_value, move2add_valid_value_p and
12350         move2add_record_mode.
12352 2013-05-27  Richard Biener  <rguenther@suse.de>
12354         PR tree-optimization/57396
12355         * tree-affine.c (double_int_constant_multiple_p): Properly
12356         return false for val == 0 and div != 0.
12358 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
12360         * config/mips/mips.h: Use #elif in preprocessor conditions.
12362 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
12364         PR target/53916
12365         * config/mips/constraints.md (kl): New constraint.
12366         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
12367         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
12368         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
12369         from LO for MIPS16.
12370         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
12372 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
12374         PR target/55777
12375         * config/mips/mips.c (mips_can_inline_p): New function.
12376         (TARGET_CAN_INLINE_P): Define.
12378 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
12380         * sched-int.h (ds_t, dw_t): Make unsigned int.
12381         Fix documentation that describes how all the ds_t bits are used.
12382         Reserve the last bit for delayed-branch scheduling.
12383         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
12384         (BITS_PER_DEP_WEAK): Fix definition and documentation.
12385         (gen_dep_weak_1): Remove prototype.
12386         * sched-deps.c (get_dep_weak_1): Make static.
12387         * target.def (speculate_insn, needs_block_p, gen_spec_check,
12388         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
12389         * doc/tm.texi: Regenerate.
12390         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
12392 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
12394         PR debug/56950
12395         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
12397 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
12398             Sandra Loosemore  <sandra@codesourcery.com>
12400         * config.gcc (powerpc-*): Allow native for with-cpu.
12402 2013-05-24  Jeff Law  <law@redhat.com>
12404         PR tree-optimization/57124
12405         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
12406         conversion feeding a condition if the range has an overflow
12407         if -fstrict-overflow.  Add warnings for when we do make the
12408         transformation.
12410 2013-05-24  Dehao Chen  <dehao@google.com>
12412         * tree-cfg.c (locus_discrim_map): Fix the typo.
12413         (locus_discrim_hasher): Likewise.
12414         (locus_discrim_hasher::hash): Likewise.
12415         (locus_discrim_hasher::equal): Likewise.
12417 2013-05-24  Martin Jambor  <mjambor@suse.cz>
12419         PR tree-optimization/57294
12420         * cgraph.h (ipa_record_stmt_references): Declare.
12421         * cgraphbuild.c (ipa_record_stmt_references): New function.
12422         (build_cgraph_edges): Use ipa_record_stmt_references.
12423         (rebuild_cgraph_edges): Likewise.
12424         (cgraph_rebuild_references): Likewise.
12425         * ipa-prop.c (ipa_modify_call_arguments): Discard references
12426         associated with the old statement and build references from the
12427         newly built statements.
12428         * ipa-ref.c (ipa_remove_stmt_references): New function.
12429         * ipa-ref.h (ipa_remove_stmt_references): Declare.
12431 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
12433         * lra-constraints.c (emit_spill_move): Use smaller mode for
12434         mem-mem moves.
12435         (check_and_process_move): Consider mem-reg moves for secondary
12436         too.
12437         (curr_insn_transform): Don't lose insns emitted before for
12438         secondary memory moves.
12439         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
12440         reg set up in the current insn.
12442 2013-05-24  Dehao Chen  <dehao@google.com>
12444         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
12445         hash function.
12446         (locus_descrim_hasher::equal): Likewise.
12447         (build_gimple_cfg): New discriminator assignment algorithm.
12448         (make_edges): Likewise.
12449         (next_discriminator_for_locus): Likewise.
12450         (same_line_p): Likewise.
12451         (assign_discriminators): Likewise.
12452         (make_cond_expr_edges): Likewise.
12453         (make_gimple_switch_edges): Likewise.
12454         (make_goto_expr_edges): Likewise.
12455         (make_gimple_asm_edges): Likewise.
12457 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
12459         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
12460         X format specifier to only display bottom 16 bits.
12461         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
12462         immediate to match for operand 2, since it will be masked.
12464 2013-05-24  Richard Biener  <rguenther@suse.de>
12466         PR tree-optimization/57287
12467         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
12468         all SSA names that occur in abnormal PHIs.
12470 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12472         PR tree-ssa/57385
12473         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
12474         that index is not negative.
12476 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
12478         PR rtl-optimization/55177
12479         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
12480         (simplify_byte_swapping_operation): New.
12481         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
12482         (simplify_relational_operation_1): Deal with BSWAP.
12484 2013-05-23  Richard Henderson  <rth@redhat.com>
12486         PR target/56742
12487         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
12488         (ix86_reorg): Call it.
12490 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
12492         PR target/57379
12493         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
12494         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
12495         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
12497 2013-05-23  Christian Bruel  <christian.bruel@st.com>
12499         PR debug/57351
12500         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
12502 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12503             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12505         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
12506         * config/aarch64/constraints.md (Usa): Remove.
12507         * doc/md.texi (AArch64 Usa): Remove.
12509 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
12510             Marcus Shawcroft  <marcus.shawcroft@arm.com>
12512         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
12513         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
12514         * config/aarch64/predicates.md (aarch64_const_address): Remove.
12515         (aarch64_mov_operand): Use aarch64_mov_operand_p.
12517 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
12519         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
12520         instruction (AdvSIMD).
12521         * config/aarch64/aarch64-builtins.c
12522         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
12523         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
12525 2013-05-23  Martin Jambor  <mjambor@suse.cz>
12527         PR middle-end/57347
12528         * tree.h (contains_bitfld_component_ref_p): Declare.
12529         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
12530         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
12531         caller.
12532         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
12533         not access a bit-field.  Assert all final offsets are byte-aligned.
12535 2013-05-23  Richard Biener  <rguenther@suse.de>
12537         PR tree-optimization/57380
12538         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
12539         least one invariant or re-used load.
12540         * passes.c (init_optimization_passes): Move pass_phiprop before
12541         pass_forwprop.
12543 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
12545         * config/aarch64/aarch64-simd.md
12546         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
12548 2013-05-23  Richard Biener  <rguenther@suse.de>
12550         PR middle-end/57381
12551         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
12552         OEP_CONSTANT_ADDRESS_OF retained.
12554 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
12556         PR middle-end/57344
12557         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
12558         don't lower unit.  Handle unit not being always BITS_PER_WORD.
12560 2013-05-23  Richard Biener  <rguenther@suse.de>
12562         PR rtl-optimization/57341
12563         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
12564         instead of true_dependence.
12566 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
12568         * bb-reorder.c (branch_threshold): Make const.
12569         (exec_threshold): Ditto.
12571 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
12572             Pat Haugen  <pthaugen@us.ibm.com>
12573             Peter Bergner  <bergner@vnet.ibm.com>
12575         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
12576         documentation for the power8 crypto builtins.
12578         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
12580         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
12581         macros for defining power8 builtin functions.
12582         (BU_P8V_AV_2): Likewise.
12583         (BU_P8V_AV_P): Likewise.
12584         (BU_P8V_VSX_1): Likewise.
12585         (BU_P8V_OVERLOAD_1): Likewise.
12586         (BU_P8V_OVERLOAD_2): Likewise.
12587         (BU_CRYPTO_1): Likewise.
12588         (BU_CRYPTO_2): Likewise.
12589         (BU_CRYPTO_3): Likewise.
12590         (BU_CRYPTO_OVERLOAD_1): Likewise.
12591         (BU_CRYPTO_OVERLOAD_2): Likewise.
12592         (XSCVSPDP): Fix typo, point to the correct instruction.
12593         (VCIPHER): Add power8 crypto builtins.
12594         (VCIPHERLAST): Likewise.
12595         (VNCIPHER): Likewise.
12596         (VNCIPHERLAST): Likewise.
12597         (VPMSUMB): Likewise.
12598         (VPMSUMH): Likewise.
12599         (VPMSUMW): Likewise.
12600         (VPERMXOR_V2DI): Likewise.
12601         (VPERMXOR_V4SI: Likewise.
12602         (VPERMXOR_V8HI: Likewise.
12603         (VPERMXOR_V16QI: Likewise.
12604         (VSHASIGMAW): Likewise.
12605         (VSHASIGMAD): Likewise.
12606         (VPMSUM): Likewise.
12607         (VPERMXOR): Likewise.
12608         (VSHASIGMA): Likewise.
12610         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12611         __CRYPTO__ if the crypto instructions are available.
12612         (altivec_overloaded_builtins): Add support for overloaded power8
12613         builtins.
12615         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
12616         support for power8 crypto builtins.
12617         (builtin_function_type): Likewise.
12618         (altivec_init_builtins): Add support for builtins that take vector
12619         long long (V2DI) arguments.
12621         * config/rs6000/crypto.md: New file, define power8 crypto
12622         instructions.
12624 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
12625             Pat Haugen  <pthaugen@us.ibm.com>
12626             Peter Bergner  <bergner@vnet.ibm.com>
12628         * doc/invoke.texi (Option Summary): Add power8 options.
12629         (RS/6000 and PowerPC Options): Likewise.
12631         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
12632         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
12633         wm, wn, wr documentation.
12635         * config/rs6000/constraints.md (wm): New constraint for VSX
12636         registers if direct move instructions are enabled.
12637         (wn): New constraint for no registers.
12638         (wq): New constraint for quad word even GPR registers.
12639         (wr): New constraint if 64-bit instructions are enabled.
12640         (wv): New constraint if power8 vector instructions are enabled.
12641         (wQ): New constraint for quad word memory locations.
12643         * config/rs6000/predicates.md (const_0_to_15_operand): New
12644         constraint for 0..15 for crypto instructions.
12645         (gpc_reg_operand): If VSX allow registers in VSX registers as well
12646         as GPR and floating point registers.
12647         (int_reg_operand): New predicate to match only GPR registers.
12648         (base_reg_operand): New predicate to match base registers.
12649         (quad_int_reg_operand): New predicate to match even GPR registers
12650         for quad memory operations.
12651         (vsx_reg_or_cint_operand): New predicate to allow vector logical
12652         operations in both GPR and VSX registers.
12653         (quad_memory_operand): New predicate for quad memory operations.
12654         (reg_or_indexed_operand): New predicate for direct move support.
12656         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
12657         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
12658         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
12659         (POWERPC_MASKS): Add power8 options.
12660         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
12661         various options.
12663         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
12664         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
12666         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
12667         (-mpower8-fusion): New power8 options.
12668         (-mpower8-fusion-sign): Likewise.
12669         (-mpower8-vector): Likewise.
12670         (-mcrypto): Likewise.
12671         (-mdirect-move): Likewise.
12672         (-mquad-memory): Likewise.
12674         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
12675         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
12676         registers.
12677         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
12678         (rs6000_debug_vector_unit): Add p8_vector.
12679         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
12680         definitions.  Also print fusion state.
12681         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
12682         (rs6000_builtin_mask_calculate): Add power8 builtin support.
12683         (rs6000_option_override_internal): Add support for power8.
12684         (rs6000_common_init_builtins): Add debugging for skipped builtins
12685         if -mdebug=builtin.
12686         (rs6000_adjust_cost): Add power8 support.
12687         (rs6000_issue_rate): Likewise.
12688         (insn_must_be_first_in_group): Likewise.
12689         (insn_must_be_last_in_group): Likewise.
12690         (force_new_group): Likewise.
12691         (rs6000_register_move_cost): Likewise.
12692         (rs6000_opt_masks): Likewise.
12694         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
12695         power8 capable assembler, default to power7 options.
12696         (TARGET_DIRECT_MOVE): Likewise.
12697         (TARGET_CRYPTO): Likewise.
12698         (TARGET_P8_VECTOR): Likewise.
12699         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
12700         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
12701         (VECTOR_MEM_P8_VECTOR_P): Likewise.
12702         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
12703         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
12704         (TARGET_XSCVDPSPN): Likewise.
12705         (TARGET_XSCVSPDPN): Likewsie.
12706         (TARGET_SYNC_HI_QI): Likewise.
12707         (TARGET_SYNC_TI): Likewise.
12708         (MASK_CRYPTO): Likewise.
12709         (MASK_DIRECT_MOVE): Likewise.
12710         (MASK_P8_FUSION): Likewise.
12711         (MASK_P8_VECTOR): Likewise.
12712         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
12713         temporary used by some of the direct move instructions to get two FP
12714         temporary registers does not force creation of a stack frame.
12715         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
12716         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
12717         that any VSX registers are tieable, even if they are also an
12718         Altivec vector mode.
12719         (r6000_reg_class_enum): Add wm, wr, wv constraints.
12720         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
12721         (RS6000_BTM_CRYPTO): Likewise.
12722         (RS6000_BTM_COMMON): Likewise.
12724         * config/rs6000/rs6000.md (cpu attribute): Add power8.
12725         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
12726         (enum rs6000_vector): Add power8 vector support.
12728 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12730         PR target/19599
12731         PR target/57340
12732         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
12733         (any_sibcall_could_use_r3): this and handle indirect calls.
12734         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
12736 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12738         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
12740 2013-05-22  Richard Biener  <rguenther@suse.de>
12742         PR middle-end/57349
12743         * profile.c (branch_prob): Do not split blocks that are
12744         abnormally receiving from ECF_RETURNS_TWICE functions.
12746 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
12748         * recog.c (offsettable_address_addr_space_p): Fix calculation of
12749         address mode.  Move pointer mode initialization to the same place.
12751 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
12753         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
12754         while it has any effect.
12756 2013-05-21  Easwaran Raman  <eraman@google.com>
12758         PR tree-optimization/57322
12759         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
12760         UID of the statement added to the BB to be 1.
12762 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
12764         PR tree-optimization/57331
12765         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
12766         of conversion from pointer type to integral type with integer.
12768 2013-05-21  Martin Jambor  <mjambor@suse.cz>
12770         PR lto/57289
12771         * ipa-prop.c (ipa_read_node_info): Process param_used and
12772         controlled_uses in the same order as when writing.
12774 2013-05-21  Magnus Granberg  <baldrick@free.fr>
12776         PR plugins/56754
12777         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
12779 2013-05-21  Richard Biener  <rguenther@suse.de>
12781         PR tree-optimization/57318
12782         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
12783         estimate stmts with side-effects as likely eliminated.
12785 2013-05-21  Richard Biener  <rguenther@suse.de>
12787         PR tree-optimization/57330
12788         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
12789         preserve the call stmts fntype.
12791 2013-05-21  Richard Biener  <rguenther@suse.de>
12793         PR tree-optimization/57303
12794         * tree-ssa-sink.c (statement_sink_location): Improve killing
12795         stmt detection and properly handle self-assignments.
12797 2013-05-21  Christian Bruel  <christian.bruel@st.com>
12799         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
12800         spanning registers. LEAF_REG_REMAP is supported only for contiguous
12801         registers. Set register size out of the PARALLEL loop.
12803 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
12805         PR target/56547
12806         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
12807         (*fmasf4, *fmasf4_media): New insns.
12809 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12811         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
12812         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
12813         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
12814         (mips_idiv_insns): Update the comments to say that the returned
12815         instruction counts are in units of BASE_INSN_LENGTH.
12816         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
12817         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
12818         using 2 rather than 4 as the length of indirect MIPS16 and
12819         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
12820         length of a NOP.  Don't divide MIPS16 lengths by 2.
12821         (mips16_split_long_branches): Assume a branch is long if the
12822         length is greater than 4 rather than 8.
12823         * config/mips/mips.md (length): Give MIPS16 lengths directly,
12824         rather than multiplying them by 2.  Multiply instruction counts
12825         by BASE_INSN_LENGTH rather than 4.
12826         (*jump_mips16, tls_get_tp_mips16_<mode>)
12827         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
12829 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12831         * config/mips/mips.md (extended_mips16): Remove branch case.
12832         (length): Remove duplicated extended_mips16 test.
12834 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12836         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
12838 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
12840         * recog.h (Recog_data): Rename to...
12841         (recog_data_d): ...this.
12842         (recog_data): Update accordingly.
12843         * recog.c (recog_data): Likewise.
12844         * reload.c (save_recog_data): Likewise.
12845         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
12846         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
12848 2013-05-17  Julian Brown  <julian@codesourcery.com>
12850         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
12851         found in a REG_EQUAL note, invalidate it.
12853 2013-05-17  Easwaran Raman  <eraman@google.com>
12855         * tree-ssa-reassoc.c (find_insert_point): New function.
12856         (insert_stmt_after): Likewise.
12857         (get_def_stmt): Likewise.
12858         (ensure_ops_are_available): Likewise.
12859         (not_dominated_by): Likewise.
12860         (rewrite_expr_tree): Do not move statements beyond what is
12861         necessary. Remove call to swap_ops_for_binary_stmt...
12862         (reassociate_bb): ... and move it here.
12863         (build_and_add_sum): Assign UIDs for new statements.
12864         (linearize_expr): Likewise.
12865         (do_reassoc): Renumber gimple statement UIDs.
12867 2013-05-17  Jan Hubicka  <jh@suse.cz>
12869         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
12870         weakrefs.
12871         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
12872         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
12873         weakrefs.
12874         (output_weakrefs): Update.
12876 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
12877             Martin Jambor  <mjambor@suse.cz>
12879         PR middle-end/57276
12880         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
12881         value that corresponds to the given aggval is found in values vector.
12883 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
12885         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
12886         sse, sse2, sse3, ssse3 and sse4a flags to options.
12888 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
12890         * gengtype-state.c: (s_expr_writer): New class, to handle
12891         prettifying of output layout of s-expressions.
12892         (state_writer): New class, to write out gtype.state.
12893         (state_written_type_count): Move this variable into member data of
12894         state_writer.
12895         (s_expr_writer::s_expr_writer): New code: constructor for new class
12896         (state_writer::state_writer(): ditto
12897         (s_expr_writer::write_new_line): New function
12898         (s_expr_writer::write_any_indent): ditto
12899         (s_expr_writer::begin_s_expr): ditto
12900         (s_expr_writer::end_s_expr): ditto
12901         (write_state_fileloc): convert to method of state_writer...
12902         (state_writer:: write_state_fileloc): ...and use methods of
12903         s_expr_writer to write indentation into the gtype.state output file
12904         to visually represent the hierarchical structure of the list
12905         structures
12906         (write_state_fields): ditto, renaming to...
12907         (state_writer::write_state_fields)
12908         (write_state_a_string): ditto, renaming to...
12909         (state_writer::write_state_a_string)
12910         (write_state_string_option): ditto, renaming to...
12911         (state_writer::write_state_string_option)
12912         (write_state_type_option): ditto, renaming to...
12913         (state_writer::write_state_type_option)
12914         (write_state_nested_option): ditto, renaming to...
12915         (state_writer::write_state_nested_option)
12916         (write_state_option): ditto, renaming to...
12917         (state_writer::write_state_option)
12918         (write_state_options): ditto, renaming to...
12919         (state_writer::write_state_options)
12920         (write_state_lang_bitmap): ditto, renaming to...
12921         (state_writer::write_state_lang_bitmap)
12922         (write_state_version): ditto, renaming to...
12923         (state_writer::write_state_version)
12924         (write_state_scalar_type): ditto, renaming to...
12925         (state_writer::write_state_scalar_type)
12926         (write_state_string_type): ditto, renaming to...
12927         (state_writer::write_state_string_type)
12928         (write_state_undefined_type): ditto, renaming to...
12929         (state_writer::write_state_undefined_type)
12930         (write_state_struct_union_type): ditto, renaming to...
12931         (state_writer::write_state_struct_union_type)
12932         (write_state_struct_type): ditto, renaming to...
12933         (state_writer::write_state_struct_type)
12934         (write_state_user_struct_type): ditto, renaming to...
12935         (state_writer::write_state_user_struct_type)
12936         (write_state_lang_struct_type): ditto, renaming to...
12937         (state_writer::write_state_lang_struct_type)
12938         (write_state_param_struct_type): ditto, renaming to...
12939         (state_writer::write_state_param_struct_type)
12940         (write_state_pointer_type): ditto, renaming to...
12941         (state_writer::write_state_pointer_type)
12942         (write_state_array_type): ditto, renaming to...
12943         (state_writer::write_state_array_type)
12944         (write_state_gc_used): ditto, renaming to...
12945         (state_writer::write_state_gc_used)
12946         (write_state_common_type_content): ditto, renaming to...
12947         (state_writer::write_state_common_type_content)
12948         (write_state_type): ditto, renaming to...
12949         (state_writer::write_state_type)
12950         (write_state_pair_list): ditto, renaming to...
12951         (state_writer::write_state_pair_list)
12952         (write_state_pair): ditto, renaming to...
12953         (state_writer::write_state_pair)
12954         (write_state_typedefs): ditto, renaming to...
12955         (state_writer::write_state_typedefs)
12956         (write_state_structures): ditto, renaming to...
12957         (state_writer::write_state_structures)
12958         (write_state_param_structs): ditto, renaming to...
12959         (state_writer::write_state_param_structs)
12960         (write_state_variables): ditto, renaming to...
12961         (state_writer::write_state_variables)
12962         (write_state_srcdir): ditto, renaming to...
12963         (state_writer::write_state_srcdir)
12964         (write_state_files_list): ditto, renaming to...
12965         (state_writer::write_state_files_list)
12966         (write_state_languages): ditto, renaming to...
12967         (state_writer::write_state_languages)
12968         (write_state): create a state_writer instance and use it when
12969         writing out the state file
12971 2013-05-17  Mike Stump  <mikestump@comcast.net>
12973         PR rtl-optimization/57304
12974         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
12975         accessing DF_REF_REAL_LOC.
12977 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12979         PR rtl-optimization/57281
12980         PR rtl-optimization/57300
12981         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
12982         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
12983         what the other splitter did if the registers are dead.
12985 2013-05-17  Richard Biener  <rguenther@suse.de>
12987         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
12988         MEM_REF offsets.
12990 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12992         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
12993         linking.
12995 2013-05-17  Marek Polacek  <polacek@redhat.com>
12997         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
12998         length when doing non-zero store of storing '\0' to '\0'.
13000 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
13002         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
13003         vect_external_def oprnd1 with loop_vinfo, try to emit
13004         optional cast, negation and and stmts on the loop preheader
13005         edge instead of into the pattern def seq.
13007         PR tree-optimization/57051
13008         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
13009         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
13011 2013-05-16  Nick Clifton  <nickc@redhat.com>
13013         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
13014         (rl78_is_naked_func): New function.
13015         (rl78_expand_prologue): Skip prologue generation for naked functions.
13016         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
13017         * doc/extend.texi (naked): Add RL78 to the list of processors
13018         that supports this attribute.
13020 2013-05-16  Jeff Law  <law@redhat.com>
13022         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
13024 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
13026         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
13027         cache parameters using detect_caches_amd also for CYRIX,
13028         NSC and TM2 signatures.
13030 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
13031             Dzianis Kahanovich  <mahatma@eu.by>
13033         PR target/45359
13034         PR target/46396
13035         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13036         VIA/Centaur processors and determine their cache parameters
13037         using detect_caches_amd.
13039 2013-05-16  Teresa Johnson  <tejohnson@google.com>
13041         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
13042         (rtl_verify_edges): New function.
13043         (rtl_verify_bb_insns): Ditto.
13044         (rtl_verify_bb_pointers): Ditto.
13045         (rtl_verify_bb_insn_chain): Ditto.
13046         (rtl_verify_fallthru): Ditto.
13047         (rtl_verify_bb_layout): Ditto.
13048         (rtl_verify_flow_info_1): Outline checks into new functions.
13049         (rtl_verify_flow_info): Ditto.
13051 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
13053         * cfghooks.c (copy_bbs): Add update_dominance argument.
13054         * cfghooks.h (copy_bbs): Update prototype.
13055         * tree-cfg.c (gimple_duplicate_sese_region):
13056         Add update_dominance argument.
13057         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
13058         * tree-ssa-loop-ch.c (copy_loop_headers): Update
13059         gimple_duplicate_sese_region call.
13060         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
13061         Update copy_bbs call.
13062         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
13063         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
13065 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
13067         * tree-vectorizer.h (NUM_PATTERNS): Increment.
13068         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
13069         vect_recog_rotate_pattern.
13070         (vect_recog_rotate_pattern): New function.
13072 2013-05-16  Jason Merrill  <jason@redhat.com>
13074         * Makefile.in (LLINKER): New variable.
13075         (mostlyclean): Remove link mutex.
13076         * configure.ac: Handle --enable-link-mutex.
13077         * lock-and-run.sh: New script.
13079 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13081         PR target/19599
13082         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
13083         for NULL decl.
13085 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13087         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
13089 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
13091         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
13092         * config/arm/arm.c (next_consecutive_mem): New function.
13093         (gen_movmem_ldrd_strd): Likewise.
13094         * config/arm/arm.md (movmemqi): Update condition and code.
13095         (unaligned_loaddi, unaligned_storedi): New patterns.
13097 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13099         * config.gcc: Obsolete *-*-solaris2.9*.
13100         * doc/install.texi (Specific, *-*-solaris2*): Document it.
13102 2013-05-16  Richard Biener  <rguenther@suse.de>
13104         * passes.c (init_optimization_passes): Move pass_parallelize_loops
13105         earlier, after GRAPHITE transforms and IV canonicalization.
13107 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
13109         * omp-low.c (extract_omp_for_data): For collapsed loops,
13110         if at least one of the loops is known at compile time to
13111         iterate zero times, set count to 0.
13112         (expand_omp_regimplify_p): New function.
13113         (expand_omp_for_generic): For collapsed loops, if at least
13114         one of the loops isn't known to iterate at least once,
13115         add runtime check with setting count to 0.
13116         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
13117         For unsigned types if it isn't known at compile time that
13118         the loop will iterate at least once, add runtime check to bypass
13119         the whole loop if initial condition isn't true.
13121 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
13123         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
13125 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
13127         PR middle-end/57286
13128         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
13129         transformations to avoid an infinite loop.
13131 2013-05-16  Marek Polacek  <polacek@redhat.com>
13133         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
13135 2013-05-15  Leif Ekblad  <leif@rdos.net>
13137         * config/i386/i386.c (ix86_decompose_address): Use
13138         DEFAULT_TLS_SEG_REG to access TLS segment register.
13139         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
13140         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
13141         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
13143 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
13145         PR target/57260
13146         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
13147         sibling calls to functions that would normally be lazily bound,
13148         unless $gp is call-clobbered.
13150 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
13152         * config/i386/i386.c (ix86_option_override_internal): Update
13153         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
13154         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
13155         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
13156         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
13157         of TARGET_3DNOW.
13158         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
13160 2013-05-15  Andreas Schwab  <schwab@suse.de>
13162         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
13163         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
13164         third operand.
13166 2013-05-15  Teresa Johnson  <tejohnson@google.com>
13168         * loop-unroll.c (report_unroll_peel): Check decision before
13169         emitting unroll/peel message.
13171 2013-05-15  Teresa Johnson  <tejohnson@google.com>
13173         * function.h (has_bb_partition): New rtl_data flag.
13174         (bb_reorder_complete): Ditto.
13175         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
13176         instead of flag_reorder_blocks_and_partition.
13177         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
13178         with some enhancements.
13179         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
13180         * bb-reorder.c (connect_traces): Check for has_bb_partition
13181         instead of flag_reorder_blocks_and_partition.
13182         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
13183         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
13184         verify_hot_cold_block_grouping.
13185         (partition_hot_cold_basic_blocks): Set has_bb_partition.
13187 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13189         PR target/19599
13190         * config/arm/predicates.md (call_insn_operand): New predicate.
13191         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
13192         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
13193         if insn is not a tail call.
13194         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
13195         registers.
13196         * config/arm/arm.h (enum reg_class): New caller save register class.
13197         (REG_CLASS_NAMES): Likewise.
13198         (REG_CLASS_CONTENTS): Likewise.
13199         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
13200         without decls.
13202 2013-05-15  Richard Biener  <rguenther@suse.de>
13204         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
13205         of MSG_OPTIMIZED_LOCATIONS.
13206         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
13207         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
13208         message.
13209         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
13210         of MSG_OPTIMIZED_LOCATIONS.
13211         (execute_vect_slp): Likewise.
13212         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
13213         (vect_create_cond_for_alias_checks): Likewise.
13214         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
13215         (vect_recog_widen_mult_pattern): Likewise.
13216         (vect_recog_widen_sum_pattern): Likewise.
13217         (vect_recog_over_widening_pattern): Likewise.
13218         (vect_recog_widen_shift_pattern): Likewise.
13219         (vect_recog_vector_vector_shift_pattern): Likewise.
13220         (vect_recog_divmod_pattern): Likewise.
13221         (vect_recog_mixed_size_cond_pattern): Likewise.
13222         (vect_recog_bool_pattern): Likewise.
13223         (vect_pattern_recog_1): Likewise.
13225 2013-05-15  Martin Jambor  <mjambor@suse.cz>
13227         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
13228         non-functions to builtin_unreachable.
13229         * ipa-inline-transform.c (inline_call): Do not assert estimates were
13230         correct when new direct edges were discovered.
13232 2013-05-15  Martin Jambor  <mjambor@suse.cz>
13234         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
13235         header, print symbol order instead of node uid, print more information
13236         about indirect edge targets.
13237         (ipa_make_edge_direct_to_target): Print symbol order instead of node
13238         uids.
13239         (ipa_make_edge_direct_to_target): Likewise.
13240         (remove_described_reference): Likewise.
13241         (propagate_controlled_uses): Likewise.
13242         (ipa_print_node_params): Also print symbol order.
13243         (ipcp_transform_function): Print symbol order instead of node uids.
13244         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
13245         (cgraph_get_create_real_symbol_node): Likewise.
13246         * ipa-cp.c (print_lattice): Likewise.
13247         (print_all_lattices): Likewise.
13248         (determine_versionability): Likewise.
13249         (initialize_node_lattices): Likewise.
13250         (estimate_local_effects): Likewise.
13251         (update_profiling_info): Likewise.
13252         (create_specialized_node): Likewise.
13253         (perhaps_add_new_callers): Likewise.
13254         (decide_about_value): Likewise.
13255         (decide_whether_version_node): Likewise.
13256         (identify_dead_nodes): Likewise.
13257         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
13258         (dump_inline_summary): Likewise.
13259         (estimate_node_size_and_time): Likewise.
13260         (inline_analyze_function): Likewise.
13261         * ipa-inline.c (report_inline_failed_reason): Likewise.
13262         (want_early_inline_function_p): Likewise.
13263         (edge_badness): Likewise.
13264         (update_edge_key): Likewise.
13265         (inline_small_functions): Likewise.  Add dumping of order to two other
13266         dumps.
13267         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
13268         instead of node uids.
13269         (propagate_pure_const): Likewise.
13270         (propagate_pure_const): Likewise.
13271         * ipa-utils.c (dump_cgraph_node_set): Likewise.
13272         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
13273         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
13274         of node uids.
13275         * tree-pretty-print.c (dump_function_header): Likewise.
13276         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
13277         Print symbol order instead of node uids.
13279 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13281         * config/s390/s390.c (s390_register_move_cost): Don't impose the
13282         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
13284 2013-05-15  Richard Biener  <rguenther@suse.de>
13286         PR tree-optimization/57275
13287         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
13288         return value for fail to do runtime alias checks for gather loads.
13290 2013-05-15  Jan Hubicka  <jh@suse.cz>
13292         PR lto/57038
13293         PR lto/47375
13294         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
13295         weakrefs are not external.
13296         (lto_symtab_merge_decls): Fix thinko when dealing with
13297         non-lto_symtab decls.
13298         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
13299         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
13300         * varpool.c (dump_varpool_node): Dump more flags.
13302 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
13304         * config/i386/i386.c (processor_alias_table): Add instruction
13305         FSGSBASE for AMD bdver3 architecture.
13307 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
13309         * tree.c (warn_deprecated_use): Print file:line using locus color.
13310         * diagnostic.c (diagnostic_report_current_module): Print file:line
13311         and file:line:column using locus color.
13313 2013-05-14  Mike Stump  <mikestump@comcast.net>
13315         * gdbinit.in: Add __null.
13317 2013-05-14  Mike Stump  <mikestump@comcast.net>
13319         * recog.h: Rename struct recog_data to Recog_data.
13320         * recog.c: Likewise.
13321         * reload.c (can_reload_into): Likewise.
13322         * config/picochip/picochip.c: Likewise.
13324 2013-05-14  Mike Stump  <mikestump@comcast.net>
13326         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
13328 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
13330         * resource.h (struct resources): Remove unch_memory member.
13331         (CLEAR_RESOURCE): Don't clear unch_memory.
13332         * resource.c (mark_referenced_resources): Don't set it.
13333         (mark_set_resources): Likewise.
13334         (mark_target_live_regs): Don't clear it.
13335         (init_resource_info): Likewise.
13336         * reorg.c (resource_conflicts_p): Don't compare it.
13337         (redundant_insn): Don't set it.
13339         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
13340         Remove prototypes.
13341         * emit-rtl.c (next_label): Remove unused function.
13342         (skip_consecutive_labels, link_cc0_insns): Move to ...
13343         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
13344         only place where these functions are used, and make them static.
13346 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
13348         * fold-const.c (fold_negate_expr): Handle vectors.
13349         (fold_truth_not_expr): Make it static.
13350         (fold_invert_truthvalue): New static function.
13351         (invert_truthvalue_loc): Handle vectors. Do not call
13352         fold_truth_not_expr directly.
13353         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
13354         <TRUTH_NOT_EXPR>: Do not cast to boolean.
13355         (fold_comparison): Handle vector constants.
13356         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
13357         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
13358         * tree.h (fold_truth_not_expr): Remove declaration.
13360 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
13362         * config/aarch64/aarch64-simd.md
13363         (aarch64_vcond_internal<mode>): Rename to...
13364         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
13365         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
13366         float modes. Clarify all iterator modes.
13367         (vcond<mode><mode>): Use new name for vcond expanders.
13368         (vcond<v_cmp_result><mode>): Likewise.
13369         (vcondu<mode><mode>: Likewise.
13370         * config/aarch64/iterators.md (VDQF_COND): New.
13372 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
13374         PR bootstrap/57266
13375         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
13376         variable for the shift amount. Check that we shift by non-negative
13377         amounts.
13379 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
13381         PR target/42017
13382         * config/arm/arm.h (EPILOGUE_USES): Only return true
13383         for LR_REGNUM after epilogue_completed.
13385 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
13387         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
13388         is error_mark_node.
13390 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13392         PR target/57261
13393         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
13394         and Solaris 11+/x86 with gld.
13395         * configure: Regenerate.
13397 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
13399         * expmed.c (expand_shift_1): Canonicalize rotates by
13400         constant bitsize / 2 to bitsize - 1.
13401         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
13402         case ROTATERT>: Likewise.
13404         Revert:
13405         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
13407         * config/i386/i386.md (rotateinv): New code attr.
13408         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
13409         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
13410         roll $31, %eax, etc.
13412 2013-05-14  Richard Biener  <rguenther@suse.de>
13414         PR middle-end/57235
13415         * tree-eh.c (sink_clobbers): Give up for successors with
13416         multiple predecessors and no virtual uses.
13418 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
13420         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
13421         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
13423 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
13425         PR middle-end/57251
13426         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
13427         the case when both op0 and op1 have VOIDmode.
13429 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
13431         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
13432         in multiply-accumulate mode.
13434 2013-05-13  Guozhi Wei  <carrot@google.com>
13436         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
13438 2013-05-13  Kai Tietz  <ktietz@redhat.com>
13440         PR target/56975
13441         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
13442         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
13443         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
13444         * config/i386/i386.c (ix86_option_override_internal): Likewise.
13445         (ix86_expand_prologue): Likewise.
13446         (ix86_expand_split_stack_prologue): Likewise.
13447         (legitimate_pic_address_disp_p): Likewise.
13448         (legitimize_pic_address): Likewise.
13449         (legitimize_tls_address): Likewise.
13450         (legitimize_pe_coff_symbol): Likewise.
13451         (output_pic_addr_const): Likewise.
13452         (construct_plt_address): Likewise.
13453         (ix86_expand_call): Likewise.
13454         (x86_output_mi_thunk): Likewise.
13455         (x86_function_profiler): Likewise.
13457 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
13459         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
13460         similar switch cases.
13461         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
13462         (aarch64_simd_mov_to_<mode>low): Delete.
13463         (aarch64_simd_mov_to_<mode>high): Delete.
13464         (move_lo_quad_<mode>): Add w<-r alternative.
13465         (aarch64_simd_move_hi_quad_<mode>): Likewise.
13466         (aarch64_simd_mov_from_*): Update type attribute.
13467         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
13468         statement.
13470 2013-05-13  Jan Hubicka  <jh@suse.cz>
13472         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
13473         * config/i386/i386.c (ix86_compute_frame_layout,
13474         ix86_expand_epilogue, emit_i387_cw_initialization,
13475         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
13476         ix86_local_alignment): Fix use of size/speed predicates.
13478 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
13480         PR tree-optimization/45216
13481         PR tree-optimization/57157
13482         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
13483         the (-Y) & (B - 1) variant if OP is |.
13484         * expmed.c (expand_shift_1): For rotations by const0_rtx just
13485         return shifted.  Use (-op1) & (prec - 1) as other_amount
13486         instead of prec - op1.
13488 2013-05-13  Martin Jambor  <mjambor@suse.cz>
13490         PR middle-end/42371
13491         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
13492         (ipa_constant_data): New type.
13493         (ipa_jump_func): Use ipa_constant_data to hold information about
13494         constant jump functions.
13495         (ipa_get_jf_constant): Adjust to jump function type changes.
13496         (ipa_get_jf_constant_rdesc): New function.
13497         (ipa_param_descriptor): New field controlled_uses.
13498         (ipa_get_controlled_uses): New function.
13499         (ipa_set_controlled_uses): Likewise.
13500         * ipa-ref.h (ipa_find_reference): Declare.
13501         * ipa-prop.c (ipa_cst_ref_desc): New type.
13502         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
13503         changes.
13504         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
13505         New parameter cs.  Adjust all callers.
13506         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
13507         (remove_described_reference): New function.
13508         (jfunc_rdesc_usable): Likewise.
13509         (try_make_edge_direct_simple_call): Decrement controlled use count,
13510         attempt to remove reference if it hits zero.
13511         (combine_controlled_uses_counters): New function.
13512         (propagate_controlled_uses): Likewise.
13513         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
13514         (ipa_edge_duplication_hook): Duplicate reference descriptions.
13515         (ipa_print_node_params): Print described use counter.
13516         (ipa_write_jump_function): Adjust to jump function type changes.
13517         (ipa_read_jump_function): New parameter CS, pass it to
13518         ipa_set_jf_constant.  Adjust caller.
13519         (ipa_write_node_info): Stream controlled use count
13520         (ipa_read_node_info): Likewise.
13521         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
13522         asserting.
13523         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
13524         count.  Remove cloning-added reference if it reaches zero.
13525         * ipa-ref.c (ipa_find_reference): New function.
13527 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
13529         * config/i386/i386.c (processor_target_table): Modified default
13530         alignment values for AMD BD and BT architectures.
13532 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
13534         * tree-vect-generic.c (uniform_vector_p): Move ...
13535         * tree.c (uniform_vector_p): ... here.
13536         * tree.h (uniform_vector_p): Declare it.
13537         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
13538         into a scalar.
13540 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
13542         PR tree-optimization/57230
13543         * tree-ssa-strlen.c (handle_char_store): Record length for
13544         array store from STRING_CST.
13546         PR tree-optimization/57230
13547         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
13548         check.
13550 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
13552         * config/epiphany/epiphany.c (epiphany_init): Check size of
13553         NUM_MODES_FOR_MODE_SWITCHING.
13554         (epiphany_expand_prologue):
13555         Remove CONFIG_REGNUM initial value handling code.
13556         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
13557         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
13558         (emit_set_fp_mode, epiphany_mode_after): Likewise.
13559         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
13560         Don't return 1 for FP_MODE_NONE.
13561         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
13562         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
13563         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
13564         * config/epiphany/epiphany.md (save_config): New pattern.
13566 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
13568         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
13570 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
13572         * config/i386/i386.md (memory): Handle sseishft1.
13573         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
13574         (*vec_extractv2di_1): Ditto.
13576 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
13578         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
13579         saved registers.
13581 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13583         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
13584         Add mthumb/march=armv7-a multilib.
13585         Add mthumb/march=armv7-r multilib.
13586         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
13588 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
13590         * config/v850/t-rtems: Add more multilibs.
13592 2013-05-10  Richard Biener  <rguenther@suse.de>
13594         PR tree-optimization/57214
13595         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
13596         not propagate from SSA names that occur in abnormal PHI nodes.
13598 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
13600         * stor-layout.c (element_precision): New function.
13601         * machmode.h (element_precision): Declare it.
13602         * tree.c (build_minus_one_cst): New function.
13603         (element_precision): Likewise.
13604         * tree.h (build_minus_one_cst): Declare new function.
13605         (element_precision): Likewise.
13606         * fold-const.c (operand_equal_p): Use element_precision.
13607         (fold_binary_loc): Handle vector types.
13608         * convert.c (convert_to_integer): Use element_precision.
13609         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
13610         separately.
13612 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
13614         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
13615         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
13616         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
13617         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
13618         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
13619         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
13620         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
13621         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
13622         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
13623         (Uuw8): New constraints.
13624         (Usb4): Move into alphabetical order.
13625         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
13626         (sd8_operand, ub8_operand, uw8_operand): New predicates.
13627         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
13628         previously unnamed patterns.
13629         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
13630         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
13631         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
13632         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
13633         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
13634         of specific lengths.
13636 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
13638         * config/i386/i386.md (rotateinv): New code attr.
13639         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
13640         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
13641         roll $31, %eax, etc.
13643         PR tree-optimization/45216
13644         PR tree-optimization/57157
13645         * tree-ssa-forwprop.c (simplify_rotate): New function.
13646         (ssa_forward_propagate_and_combine): Call it.
13648 2013-05-10  Richard Biener  <rguenther@suse.de>
13650         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
13651         disable peeling when we version for aliasing.
13652         (vector_alignment_reachable_p): Honor explicit user alignment.
13653         (vect_supportable_dr_alignment): Likewise.
13654         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
13655         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
13656         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
13657         then peeling to arrange for the cost-model check to come first.
13659 2013-05-10  Alan Modra  <amodra@gmail.com>
13661         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
13662         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
13663         * configure: Regenerate.
13665 2013-05-10  Alan Modra  <amodra@gmail.com>
13667         PR target/55033
13668         * varasm.c (default_elf_select_section): Move !DECL_P check..
13669         (get_named_section): ..to here before calling get_section_name.
13670         Adjust assertion.
13671         (default_section_type_flags): Add DECL_P check.
13672         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
13673         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
13675 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
13677         * config/epiphany/epiphany.c (epiphany_expand_prologue):
13678         When using gen_stack_adjust_str with a register offset, add a
13679         REG_FRAME_RELATED_EXPR note.
13681 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
13683         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
13684         (*vec_extractv4si_zext_mem): Ditto.
13685         (*vec_extractv2di): Add 0->x and x->x alternatives.
13686         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
13687         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
13689 2013-05-09  Jason Merrill  <jason@redhat.com>
13691         N3639 C++1y VLA support
13692         * gimplify.c (gimplify_vla_decl): Don't touch an existing
13693         DECL_VALUE_EXPR.
13695         * tree.c (build_constructor_va): New.
13696         * tree.h: Declare it.
13698 2013-05-09  Martin Jambor  <mjambor@suse.cz>
13700         PR lto/57084
13701         * gimple-fold.c (canonicalize_constructor_val): Call
13702         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
13704 2013-05-09  Jan Hubicka  <jh@suse.cz>
13705             Richard Biener  <rguenther@suse.de>
13707         PR lto/54095
13708         * symtab.c (symtab_make_decl_local): Do not add private names.
13710 2013-05-09  Jan Hubicka  <jh@suse.cz>
13712         PR lto/54095
13713         * symtab.c (insert_to_assembler_name_hash): Handle clones.
13714         (unlink_from_assembler_name_hash): Likewise.
13715         (symtab_prevail_in_asm_name_hash, symtab_register_node,
13716         symtab_unregister_node, symtab_initialize_asm_name_hash,
13717         change_decl_assembler_name): Update.
13719 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
13721         * config/aarch64/aarch64.md: New movtf split.
13722         (*movtf_aarch64): Update.
13723         (aarch64_movdi_tilow): Handle TF modes and rename to
13724         aarch64_movdi_<mode>low.
13725         (aarch64_movdi_tihigh): Handle TF modes and rename to
13726         aarch64_movdi_<mode>high
13727         (aarch64_movtihigh_di): Handle TF modes and rename to
13728         aarch64_mov<mode>high_di
13729         (aarch64_movtilow_di): Handle TF modes and rename to
13730         aarch64_mov<mode>low_di
13731         (aarch64_movtilow_tilow): Remove spurious whitespace.
13732         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
13733         splits.
13734         (aarch64_print_operand): Update.
13736 2013-05-09  Alan Modra  <amodra@gmail.com>
13738         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
13739         powerpc64le.
13740         * configure: Regenerate.
13742 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
13744         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
13745         splitter preparation statements.
13746         * config/i386/sse.md (*vec_extract* splitters): Ditto.
13747         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
13748         adjust_address_nv.
13750 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13752         * gimple-ssa-strength-reduction.c (count_candidates): Change
13753         return value to int.
13754         (analyze_candidates_and_replace): Change type of length to int.
13756 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
13758         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
13759         (*vec_extract<mode>): Use VI12_128 mode iterator.
13760         (*vec_extract<mode>_mem): Ditto.
13761         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
13762         attribute.
13764 2013-05-08  Diego Novillo  <dnovillo@google.com>
13766         PR bootstrap/54659
13768         Revert:
13769         2012-08-17  Diego Novillo  <dnovillo@google.com>
13771         PR bootstrap/54281
13772         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
13773         * config.in: Regenerate.
13774         * configure: Regenerate.
13775         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
13777 2013-05-08  Jan Hubicka  <jh@suse.cz>
13779         PR lto/54095
13780         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
13781         * cgraph.h (symtab_node_base): Add unique_name.
13782         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
13783         input_overwrite_node, input_varpool_node): Stream unique_name.
13784         * cgraphclones.c (cgraph_create_virtual_clone,
13785         cgraph_function_versioning): Set unique_name.
13786         * ipa.c (function_and_variable_visibility): Set unique_name.
13788 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13790         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
13791         (alloc_cand_and_find_basis): Restrict conditional candidate
13792         processing to CAND_MULTs.
13794 2013-05-08  Jan Hubicka  <jh@suse.cz>
13796         PR lto/54095
13797         lto-symtab.c (lto_symtab_symbol_p): New function.
13798         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
13799         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
13800         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
13801         Skip static symbols.
13803 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
13805         PR tree-optimization/57200
13806         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
13807         Only call inform if the preceding warning_at returns true.
13809 2013-05-07  Han Shen  <shenhan@google.com>
13811         * cfgexpand.c (record_or_union_type_has_array_p): New function.
13812         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
13813         * common.opt (fstack-protector-strong): New option.
13814         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
13815         * doc/invoke.texi (Optimization Options): Document
13816         "-fstack-protector-strong".
13817         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
13819 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
13821         * config/mips/mips.c (mips_machine_reorg2): Return 0.
13823 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
13825         * ira.c (update_equiv_regs): Add insn having equiv memory even if
13826         it is not lhs of the insn.
13827         (setup_reg_equiv): Remove insn having equiv memory which it is not
13828         lhs of the insn.
13829         * lra-constraints.c (process_address): Try to improve generation
13830         code for address base + disp.
13831         (lra_constraints): Make correct the code for checking insn setting
13832         up backward equivalence.  Remove insn only if it is in the init
13833         insn list.
13834         * lra-eliminations.c (update_reg_eliminate): Change return value.
13835         (lra_eliminate): Use the result.
13837 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
13839         * config/i386/sse.md (ssescalarnummask): New mode attribute.
13840         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
13841         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
13842         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
13843         register target operands.
13844         (*vec_extractv8hi_sse2): New pattern.
13845         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
13846         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
13847         (*vec_extract<mode>_mem): New insn and split pattern.
13849 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
13851         * config/arm/arm.c (arm_asan_shadow_offset): New function.
13852         (TARGET_ASAN_SHADOW_OFFSET): Define.
13853         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
13854         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
13856 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13858         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
13859         (incr_vec_index): Return -1 if increment not found.
13860         (create_add_on_incoming_edge): Assert if increment not found.
13861         (record_increment): Limit number of increments recorded.
13862         (all_phi_incrs_profitable): Return false if an increment not found.
13863         (replace_profitable_candidates): Don't process increments that were
13864         not recorded.
13865         (analyze_candidates_and_replace): Limit size of incr_vec.
13867 2013-05-07  Richard Biener  <rguenther@suse.de>
13869         * calls.c (special_function_p): setjmp-like functions are leaf.
13870         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
13871         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
13873 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
13875         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
13876         (aarch64_simd_mov<mode>): New expander.
13877         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
13878         (aarch64_simd_mov_to_<mode>high): Likewise.
13879         (aarch64_simd_mov_from_<mode>low): Likewise.
13880         (aarch64_simd_mov_from_<mode>high): Likewise.
13881         (aarch64_dup_lane<mode>): Update.
13882         (aarch64_dup_lanedi): New instruction pattern.
13883         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
13884         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
13886 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13888         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
13889         (replace_mult_candidate): Remove unnecessary argument; remove
13890         unnecessary parameter from call to introduce_cast_before_cand.
13891         (replace_unconditional_candidate): Remove unnecessary parameter
13892         from call to replace_mult_candidate.
13893         (replace_conditional_candidate): Likewise.
13894         (insert_initializers): Use make_temp_ssa_name.
13895         (introduce_cast_before_cand): Remove unnecessary argument; use
13896         make_temp_ssa_name.
13897         (replace_one_candidate): Remove unnecessary argument; remove
13898         unnecessary parameter from calls to introduce_cast_before_cand.
13899         (replace_profitable_candidates): Remove unnecessary parameters
13900         from calls to replace_one_candidate.
13902 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13904         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13905         phi def as possibly hiding a basis for a CAND_ADD whose operands
13906         have been commuted in the analysis.
13907         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
13909 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13911         * config/aarch64/aarch64.md
13912         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
13913         shift value between 0-4.
13915 2013-05-07  Richard Biener  <rguenther@suse.de>
13917         * double-int.h (rshift): New overload.
13918         * double-int.c (rshift): New function.
13919         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
13920         (create_reference_ops_from_ref): Remove.
13921         (vn_reference_insert): Use shared ops for constructing the
13922         reference and copy it.
13924 2013-05-07  Richard Biener  <rguenther@suse.de>
13926         PR middle-end/57190
13927         * tree-eh.c (sink_clobbers): Properly propagate
13928         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
13930 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
13932         PR tree-optimization/57149
13933         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
13934         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
13935         collect_phi_def_edges, execute_late_warn_uninitialized): Use
13936         uninit_undefined_value_p instead of ssa_undefined_value_p.
13938         PR debug/57184
13939         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
13940         for modifier == EXPAND_INITIALIZER.
13942 2013-05-07  Anton Blanchard  <anton@samba.org>
13944         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
13945         for powerpc64 little endian.
13946         * configure: Regenerate.
13948 2013-05-06  Graham Stott  <grahams@btinternet.com>
13950         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
13951         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
13952         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
13953         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
13955 2013-05-06  Graham Stott  <grahams@btinternet.com>
13957         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
13958         codes which allow non-lvalues.
13960 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
13962         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
13963         components are all 1s.
13964         (integer_minus_onep): New function.
13965         * tree.h (integer_minus_onep): Declare it.
13966         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
13967         integer_minus_onep instead of integer_all_onesp.
13969 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13971         PR target/52933
13972         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
13973         variations of these patterns.
13975 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
13977         * config/i386/i386.md (isa): Add x64_sse4 member.
13978         (enabled): Handle x64_sse4.
13979         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
13980         instruction for 64bit SSE4_1 targets.  Update insn attributes.
13981         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
13982         instruction for SSE4_1 targets.  Update insn attributes.
13983         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13984         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
13985         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
13986         const_1 selector.
13987         (*vec_extractv4si): Rename from *sse4_1_pextrd.
13988         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
13989         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
13991 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13993         PR target/57108
13994         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
13996 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
13998         * final.c (do_assembler_dialects): Don't handle curly braces and
13999         vertical bar escaped by % as dialect delimiters.
14000         (output_asm_insn): Print curly braces and vertical bar if escaped
14001         by % and ASSEMBLER_DIALECT defined.
14002         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
14003         * doc/tm.texi: Regenerated.
14005 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
14007         * config/mips/mips.c: Include tree-pass.h.
14008         (mips_reorg): Split in pre- and post-dbr_schedule parts.
14009         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
14010         (pass_mips_machine_reorg2): New machine specific pass.
14011         (insert_pass_mips_machine_reorg2): New pass plugin definition.
14012         (mips_option_override): Register the new pass.
14013         * rtl.h (cleanup_barriers): Remove prototype.
14014         (dbr_schedule): Likewise.
14015         * jump.c (cleanup_barriers): Make static.
14016         * reorg.c (dbr_schedule): Likewise.
14018 2013-05-06  Richard Biener  <rguenther@suse.de>
14020         PR tree-optimization/57185
14021         * tree-parloops.c (add_field_for_reduction): Handle anonymous
14022         SSA names properly.
14024 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
14026         PR target/57106
14027         * config/i386/i386.c (add_parameter_dependencies): Add dependence
14028         between "first_arg" and "insn", not "last" and "insn".
14030 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14032         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
14033         (find_candidates_in_block): Re-enable slsr_process_phi.
14034         (create_phi_basis): Fix double counting of candidate adjustment.
14036 2013-05-06  Richard Biener  <rguenther@suse.de>
14038         PR middle-end/57147
14039         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
14040         the edge is also fallthru, preserve it and just clear the
14041         abnormal flag.
14042         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
14043         also complex, preserve that and just clear the fallthru flag.
14044         * tree-inline.c (update_ssa_across_abnormal_edges): Also
14045         update virtual operands.
14047 2013-05-06  Alan Modra  <amodra@gmail.com>
14049         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
14050         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
14051         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
14052         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
14053         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
14054         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
14056 2013-05-06  Alan Modra  <amodra@gmail.com>
14058         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
14059         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
14060         (DEFAULT_ASM_ENDIAN): Define.
14061         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
14062         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
14063         Update -K PIC clause from sysv4.h.
14064         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
14065         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
14067 2013-05-06  Alan Modra  <amodra@gmail.com>
14069         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
14070         twice for little-endian.
14071         (ashrdi3_no_power, ashrdi3): Support little-endian.
14073 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
14075         PR target/55303
14076         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
14077         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
14078         related expanders.
14079         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
14080         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
14081         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
14082         New predicates.
14084 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
14085             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14087         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
14088         * config/pa/pa.opt: Make mbig-switch a no-op.
14089         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
14090         (CASE_VECTOR_MODE): Always return SImode.
14091         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
14092         for the !TARGET_BIG_SWITCH case.
14093         * config/pa/pa-linux.h: Likewise.
14094         * config/pa/pa-openbsd.h: Likewise.
14095         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
14096         * config/pa/pa.md (short_jump): Remove define_insn.
14097         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
14098         (casesi0): Remove define_insn.
14099         (type): Remove btable_branch.
14100         (pa_combine_type): Likewise.
14101         (in_nullified_branch_delay): Likewise.
14102         (in_call_delay): Likewise.
14103         (define_delay): Likewise.
14104         (define_insn_reservation "Z3"): Likewise.
14105         (define_insn_reservation "Z4"): Likewise.
14106         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
14107         (pa_adjust_insn_length): Remove adjustment for btable branches.
14108         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
14109         and mno-big-switch
14111 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
14113         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
14114         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
14115         Add m->r,x alternatives.
14116         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
14117         splitters using SWI48x mode iterator.
14118         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
14119         TARGET_64BIT.  Add m->x alternative.
14120         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
14121         Add o->x alternative.  Enable for TARGET_SSE.
14122         (sse_storeq): Remove expander.
14123         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
14124         with memory input operand.
14125         (*vec_extractv2di_1 splitter): New.
14126         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
14127         * config/i386/i386.md (ssevecmodelower): New mode attribute.
14129 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
14131         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
14132         (INT_LOWPART): Delete.
14133         (extract_MB): Adjust.
14134         (extract_ME): Adjust.
14135         (print_operand): Adjust.
14137 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
14139         * config/rs6000/predicates.md (reg_or_add_cint_operand,
14140         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
14141         (reg_or_logical_cint_operand, easy_fp_constant,
14142         logical_const_operand): Delete "CONST_DOUBLE" case.
14143         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
14144         "HOST_BITS_PER_WIDE_INT == 64" test.
14145         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
14146         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
14147         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
14148         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
14149         test.
14150         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
14151         CONST_DOUBLE DImode/VOIDmode case.
14152         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
14153         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
14154         CONST_DOUBLE VOIDmode case.
14155         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
14156         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
14157         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
14158         Delete CONST_DOUBLE case.
14159         (splitters for mov FMOVE64 const_double): Delete
14160         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
14161         "HOST_BITS_PER_WIDE_INT >= 64" test.
14162         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
14163         case.
14164         (mov DI const_double): Delete.
14166 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
14168         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
14169         on op shows all bits zero in mode of a lowpart subreg, return zero.
14171 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
14173         PR target/57150
14174         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
14175         to save TFmode registers and DImode to save TImode registers for
14176         caller save operations.
14177         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
14178         mark being partially clobbered since they only use the first
14179         double word.
14181         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
14182         and TDmode only use the upper 64-bits of each VSX register.
14184 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14186         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
14187         (find_candidates_in_block): Disable slsr_process_phi.
14189 2013-05-03  Guozhi Wei  <carrot@google.com>
14191         * coverage.c (coverage_obj_init): Move the construction of gcov
14192         constructor to ...
14193         (build_init_ctor): ... here.
14195 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14197         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
14198         (slsr_cand_d): Redefine def_phi.
14199         (stride_status, phi_adjust_status, count_phis_status): New enums.
14200         (find_phi_def): New.
14201         (find_basis_for_base_expr): New.
14202         (find_basis_for_candidate): Handle hidden bases.
14203         (alloc_cand_and_find_basis): Handle phi candidates.
14204         (slsr_process_phi): New.
14205         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
14206         (create_mul_imm_cand): Likewise.
14207         (create_add_ssa_cand): Exclude phi base candidates.
14208         (create_add_imm_cand): Likewise.
14209         (slsr_process_cast): Likewise.
14210         (slsr_process_copy): Likewise.
14211         (find_candidates_in_block): Handle phi candidates.
14212         (dump_candidate): Likewise.
14213         (unconditional_cands): Delete.
14214         (unconditional_cands_with_known_stride_p): Delete.
14215         (phi_dependent_cand_p): New.
14216         (cand_increment): Handle phi-dependent candidates.
14217         (replace_dependent): Delete.
14218         (replace_mult_candidate): New.
14219         (replace_unconditional_candidate): New.
14220         (incr_vec_index): Move to avoid forward reference.
14221         (create_add_on_incoming_edge): New.
14222         (create_phi_basis): New.
14223         (replace_dependents): Delete.
14224         (replace_conditional_candidate): New.
14225         (phi_add_costs): New.
14226         (replace_uncond_cands_and_profitable_phis): New.
14227         (record_increment): Handle phi adjustments.
14228         (record_phi_increments): New.
14229         (record_increments): Handle phi adjustments.
14230         (phi_incr_cost): New.
14231         (lowest_cost_path): Handle phis.
14232         (total_savings): Likewise.
14233         (analyze_increments): Likewise.
14234         (ncd_with_phi): New.
14235         (ncd_of_cand_and_phis): New.
14236         (nearest_common_dominator_for_cands): Handle phi increments.
14237         (all_phi_incrs_profitable): New.
14238         (replace_profitable_candidates): Handle phi-dependent candidates.
14239         (analyze_candidates_and_replace): Likewise.
14241 2013-05-03  Teresa Johnson  <tejohnson@google.com>
14243         PR bootstrap/57154
14244         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
14245         do not exceed REG_BR_PROB_BASE.
14247 2013-05-03  Jeff Law  <law@redhat.com>
14249         PR tree-optimization/57144
14250         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
14251         operand of the condition will bit into the new type when eliminating
14252         a cast feeding a condition.
14254 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
14256         PR rtl-optimization/57130
14257         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
14258         of COMPARE as in_code to the recursive call if needed.
14260 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
14262         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
14263         (enabled): Handle new members.
14264         * config/i386/sse.md (*vec_concatv2si): Merge from
14265         *vec_concatv2si_sse2 and vec_concatv2si_sse.
14266         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
14268 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
14270         PR tree-optimization/57027
14271         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
14272         for fnms opportunity, check we got the prerequisite kind
14273         of tree / gimple before using accessor functions.
14275 2013-05-03  Richard Biener  <rguenther@suse.de>
14277         * double-int.h (lshift): New overload without precision
14278         and arith argument.
14279         (operator *=, operator +=, operator -=): Move ...
14280         * double-int.c (operator *=, operator +=, operator -=): ... here
14281         and implement more efficiently.
14282         (mul_double_with_sign): Remove.
14283         (lshift_double): Adjust to take unsinged shift argument, push
14284         dispatching code to callers.
14285         (mul_double_wide_with_sign): Add early out for callers that
14286         are not interested in high parts or overflow.
14287         (lshift): New function.
14288         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
14289         dispatch code here.
14290         (lrotate, rrotate): Use logical shifts.
14291         * expr.c (get_inner_reference): Use lshift.
14292         * fixed-value.c (do_fixed_divide): Likewise.
14293         * tree-dfa.c (get_ref_base_and_extent): Likewise.
14294         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
14295         (indirect_refs_may_alias_p): Likewise.
14296         (stmt_kills_ref_p_1): Likewise.
14298 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
14300         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
14302 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
14304         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
14305         scalar form of FABD instruction.
14307 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
14309         * lra-constraints.c (process_alt_operands): Add checking alt
14310         number to choose the best alternative.
14312 2013-05-02  Richard Biener  <rguenther@suse.de>
14314         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
14315         bitmap and its handling.
14316         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
14318 2013-05-02  Richard Biener  <rguenther@suse.de>
14320         PR middle-end/57140
14321         * tree-inline.c (copy_loops): Properly handle removed loops.
14322         (copy_cfg_body): Mark destination loops for fixup if source
14323         loops needed fixup.
14325 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
14327         PR target/56732
14328         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
14329         generating simple_return for naked functions.
14331 2013-05-02  Martin Jambor  <mjambor@suse.cz>
14333         PR middle-end/56988
14334         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
14335         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
14336         flags match.
14337         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
14338         ipa_agg_replacement_value structures.
14339         (known_aggs_to_agg_replacement_list): Likewise.
14340         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
14341         (read_agg_replacement_chain): Likewise.
14342         (ipcp_transform_function): Also check that by_ref flags match.
14344 2013-05-02  Richard Biener  <rguenther@suse.de>
14346         * graphds.h (struct graph): Add obstack member.
14347         * graphds.c (new_graph): Initialize obstack and allocate
14348         vertices from it.
14349         (add_edge): Allocate edge from the obstack.
14350         (free_graph): Free the obstack instead of all edges and vertices.
14352 2013-05-02  Teresa Johnson  <tejohnson@google.com>
14354         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
14355         divides.
14356         * cfg.c (update_bb_profile_for_threading): Ditto.
14357         * tree-inline.c (copy_bb): Ditto.
14358         (copy_edges_for_bb): Ditto.
14359         (initialize_cfun): Ditto.
14360         (copy_cfg_body): Ditto.
14361         (expand_call_inline): Ditto.
14362         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
14363         (estimate_node_size_and_time): Ditto.
14364         (inline_merge_summary): Ditto.
14365         * cgraphclones.c (cgraph_clone_edge): Ditto.
14366         (cgraph_clone_node): Ditto.
14367         * sched-rgn.c (compute_dom_prob_ps): Ditto.
14368         (compute_trg_info): Ditto.
14370 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
14372         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
14373         S reg when fp attribute set.
14374         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
14376 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
14378         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
14379         New pattern.
14380         (*and_one_cmplsi3_compare0_uxtw): Likewise.
14381         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
14382         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
14384 2013-05-02  Richard Biener  <rguenther@suse.de>
14386         * tree-scalar-evolution.c (scev_info_hasher): Remove.
14387         (struct instantiate_cache_entry): New type.
14388         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
14389         (struct instantiate_cache_type): New type.
14390         (set_instantiated_value, get_instantiated_value): Remove.
14391         (get_instantiated_value_entry): New function.
14392         (instantiate_scev_name): Use the new cache and adjust.
14393         (instantiate_scev_poly): Adjust.
14394         (instantiate_scev_binary): Likewise.
14395         (instantiate_array_ref): Likewise.
14396         (instantiate_scev_convert): Likewise.
14397         (instantiate_scev_not): Likewise.
14398         (instantiate_scev_3): Likewise.
14399         (instantiate_scev_2): Likewise.
14400         (instantiate_scev_r): Likewise.
14401         (instantiate_scev): Likewise.
14402         (resolve_mixers): Likewise.
14404 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
14406         PR target/57091
14407         * lra-constraints.c (best_small_class_operands_num): Remove.
14408         (process_alt_operands): Remove small_class_operands_num.  Take
14409         small classes operands into losers and only if the operand is not
14410         matched.  Modify debugging output.
14411         (curr_insn_transform): Remove best_small_class_operands_num.
14412         Print insn name.
14414 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14416         * config/aarch64/aarch64-builtins.c
14417         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
14418         * config/aarch64/aarch64-simd-builtins.def
14419         (reduc_splus_): Add new modes.
14420         (reduc_uplus_): New.
14421         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
14422         (reduc_uplus_v4sf): Likewise.
14423         (reduc_splus_v4sf): Likewise.
14424         (aarch64_addv<mode>): Likewise.
14425         (reduc_uplus_<mode>): Likewise.
14426         (reduc_splus_<mode>): Likewise.
14427         (aarch64_addvv2di): Likewise.
14428         (reduc_uplus_v2di): Likewise.
14429         (reduc_splus_v2di): Likewise.
14430         (aarch64_addvv2si): Likewise.
14431         (reduc_uplus_v2si): Likewise.
14432         (reduc_splus_v2si): Likewise.
14433         (reduc_<sur>plus_<mode>): New.
14434         (reduc_<sur>plus_v2di): Likewise.
14435         (reduc_<sur>plus_v2si): Likewise.
14436         (reduc_<sur>plus_v4sf): Likewise.
14437         (aarch64_addpv4sf): Likewise.
14438         * config/aarch64/arm_neon.h
14439         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
14440         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
14441         add UNSPEC_SADDV, UNSPEC_UADDV.
14442         (SUADDV): New.
14443         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
14445 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14447         * config/aarch64/arm_neon.h
14448         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
14450 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14452         * config/aarch64/aarch64-builtins
14453         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
14455 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14457         * config/aarch64/aarch64-simd-builtins.def
14458         (reduc_smax_): New.
14459         (reduc_smin_): Likewise.
14460         (reduc_umax_): Likewise.
14461         (reduc_umin_): Likewise.
14462         (reduc_smax_nan_): Likewise.
14463         (reduc_smin_nan_): Likewise.
14464         (fmax): Remove.
14465         (fmin): Likewise.
14466         (smax): Update for V2SF, V4SF and V2DF modes.
14467         (smin): Likewise.
14468         (smax_nan): New.
14469         (smin_nan): Likewise.
14470         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
14471         (<su><maxmin><mode>3): ...This, refactor.
14472         (s<maxmin><mode>3): New.
14473         (<maxmin_uns><mode>3): Likewise.
14474         (reduc_<maxmin_uns>_<mode>): Refactor.
14475         (reduc_<maxmin_uns>_v4sf): Likewise.
14476         (reduc_<maxmin_uns>_v2si): Likewise.
14477         (aarch64_<fmaxmin><mode>: Remove.
14478         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
14479         new builtin names.
14480         (vmin<q>_f<32,64>): Likewise.
14481         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
14482         (FMAXMIN): New.
14483         (su): Add mappings for smax, smin, umax, umin.
14484         (maxmin): New.
14485         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
14486         (FMAXMIN): Rename as...
14487         (FMAXMIN_UNS): ...This.
14488         (maxminv): Remove.
14489         (fmaxminv): Likewise.
14490         (fmaxmin): Likewise.
14491         (maxmin_uns): New.
14492         (maxmin_uns_op): Likewise.
14494 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14496         * config/aarch64/arm_neon.h
14497         (vac<ge, gt><sd>_f<32, 64>): Rename to...
14498         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
14499         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
14501 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14503         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
14504         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
14506 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14508         * config/aarch64/aarch64-simd.md
14509         (vcond<mode>_internal): Handle special cases for constant masks.
14510         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
14511         (vcondu<mode><mode>): Likewise.
14512         (vcond<v_cmp_result><mode>): New.
14514 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14516         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
14517         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
14518         * config/aarch64/aarch64-simd-builtins.def
14519         (cmeq): Update to BUILTIN_VALLDI.
14520         (cmgt): Likewise.
14521         (cmge): Likewise.
14522         (cmle): Likewise.
14523         (cmlt): Likewise.
14524         * config/aarch64/arm_neon.h
14525         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
14526         to builtins or C as appropriate.
14528 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
14530         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
14531         (cmgeu): ...This.
14532         (cmhi): Rename to...
14533         (cmgtu): ...This.
14534         * config/aarch64/aarch64-simd.md
14535         (simd_mode): Add SF.
14536         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
14537         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
14538         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
14539         (cstore<mode>_neg): ...This.
14540         * config/aarch64/iterators.md
14541         (VALLF): new.
14542         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
14543         (COMPARISONS): New.
14544         (UCOMPARISONS): Likewise.
14545         (optab): Add missing comparisons.
14546         (n_optab): New.
14547         (cmp_1): Likewise.
14548         (cmp_2): Likewise.
14549         (CMP): Likewise.
14550         (cmp): Remove.
14551         (VCMP_S): Likewise.
14552         (VCMP_U): Likewise.
14553         (V_cmp_result): Add DF, SF modes.
14554         (v_cmp_result): Likewise.
14555         (v): Likewise.
14556         (vmtype): Likewise.
14557         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
14559 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
14561         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
14562         define_insn to define_insn_and_split.
14563         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
14564         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
14565         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
14566         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
14567         (thumb2_negscc): Likewise.
14569 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
14571         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
14573 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
14575         * config/arm/thumb2.md: Remove trailing whitespaces.
14577 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
14579         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
14580         Use gen_int_mode rather than GEN_INT.
14582 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
14584         * value-prof.c (stream_in_histogram_value): Remove the strayed
14585         debug_gimple_stmt.
14587 2013-04-30  Richard Biener  <rguenther@suse.de>
14589         PR middle-end/57122
14590         * cfghooks.c (split_edge): Properly check for the loop latch edge.
14592 2013-04-30  Richard Biener  <rguenther@suse.de>
14594         PR middle-end/57107
14595         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
14597 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
14599         PR rtl-optimization/56957
14600         PR rtl-optimization/57105
14601         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
14602         variable.  Use just INSN_UID for determining whether an insn
14603         should be only disconnected from the insn stream.
14604         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
14606 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
14608         PR tree-optimization/57104
14609         * tsan.c (instrument_expr): Don't instrument accesses to
14610         DECL_HARD_REGISTER VAR_DECLs.
14612 2013-04-30  Richard Biener  <rguenther@suse.de>
14614         * function.h (loops_for_fn): New inline function.
14615         (set_loops_for_fn): Likewise.
14616         * cfgloop.h (place_new_loop): Add struct function parameter.
14617         (get_loop): Likewise.
14618         (get_loops): Likewise.
14619         (number_of_loops): Likewise.
14620         (fel_next): Adjust.
14621         (fel_init): Likewise.
14622         * cfg.c (get_loop_copy): Adjust.
14623         * cfgloop.c (flow_loops_dump): Likewise.
14624         (record_loop_exits): Likewise.
14625         (verify_loop_structure): Likewise.
14626         * cfgloopanal.c (mark_irreducible_loops): Likewise.
14627         (estimate_reg_pressure_cost): Likewise.
14628         (mark_loop_exit_edges): Likewise.
14629         * cfgloopmanip.c (place_new_loop): Likewise.
14630         (add_loop): Likewise.
14631         (duplicate_loop): Likewise.
14632         * graph.c (draw_cfg_nodes): Likewise.
14633         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
14634         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
14635         (extract_affine_chrec): Likewise.
14636         (build_scop_iteration_domain): Likewise.
14637         * graphite.c (graphite_initialize): Likewise.
14638         * ira-build.c (create_loop_tree_nodes): Likewise.
14639         (more_one_region_p): Likewise.
14640         (rebuild_regno_allocno_maps): Likewise.
14641         (mark_loops_for_removal): Likewise.
14642         (mark_all_loops_for_removal): Likewise.
14643         (remove_unnecessary_regions): Likewise.
14644         (ira_build): Likewise.
14645         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
14646         * loop-init.c (fix_loop_structure): Likewise.
14647         (gate_rtl_move_loop_invariants): Likewise.
14648         (gate_rtl_unswitch): Likewise.
14649         (gate_rtl_unroll_and_peel_loops): Likewise.
14650         (rtl_doloop): Likewise.
14651         * lto-streamer-in.c (input_cfg): Likewise.
14652         * lto-streamer-out.c (output_cfg): Likewise.
14653         * modulo-sched.c (sms_schedule): Likewise.
14654         * predict.c (tree_estimate_probability): Likewise.
14655         (tree_estimate_probability_driver): Likewise.
14656         (estimate_loops): Likewise.
14657         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
14658         (move_sese_region_to_fn): Likewise.
14659         (debug_loop_num): Likewise.
14660         * tree-chrec.c (chrec_evaluate): Likewise.
14661         (hide_evolution_in_other_loops_than_loop): Likewise.
14662         (chrec_component_in_loop_num): Likewise.
14663         (reset_evolution_in_loop): Likewise.
14664         (evolution_function_is_invariant_rec_p): Likewise.
14665         * tree-if-conv.c (main_tree_if_conversion): Likewise.
14666         * tree-inline.c (copy_loops): Likewise.
14667         (copy_cfg_body): Likewise.
14668         (tree_function_versioning): Likewise.
14669         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
14670         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
14671         Likewise.
14672         (add_to_evolution_1): Likewise.
14673         (scev_const_prop): Likewise.
14674         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
14675         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
14676         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
14677         (tree_ssa_lim_initialize): Likewise.
14678         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
14679         (verify_loop_closed_ssa): Likewise.
14680         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
14681         (tree_ssa_loop_im): Likewise.
14682         (tree_ssa_loop_unswitch): Likewise.
14683         (tree_vectorize): Likewise.
14684         (check_data_deps): Likewise.
14685         (tree_ssa_loop_ivcanon): Likewise.
14686         (tree_ssa_loop_bounds): Likewise.
14687         (tree_complete_unroll): Likewise.
14688         (tree_complete_unroll_inner): Likewise.
14689         (tree_parallelize_loops): Likewise.
14690         (tree_ssa_loop_prefetch): Likewise.
14691         (tree_ssa_loop_ivopts): Likewise.
14692         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
14693         * tree-vectorizer.c (vectorize_loops): Likewise.
14695 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
14697         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
14698         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
14699         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
14700         with EABI_LINK_SPEC.
14702 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14704         PR target/44578
14705         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
14706         alternative.
14708 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
14710         PR target/57097
14711         * lra-constraints.c (process_alt_operands): Discourage a bit more
14712         using memory for pseudos.  Print cost dump for alternatives.
14713         Modify cost values for conflicts with early clobbers.
14714         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
14716 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14718         PR target/57098
14719         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
14721 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
14723         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
14724         from/to S register.
14725         (movdi_aarch64): Support LDR/STR from/to D register.
14727 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
14729         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
14730         or higher by default.
14732 2013-04-29  Richard Biener  <rguenther@suse.de>
14734         PR middle-end/57075
14735         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
14736         even if not adding abnormal edges for calls that can make
14737         abnormal gotos.
14739 2013-04-29  Richard Biener  <rguenther@suse.de>
14741         PR middle-end/57103
14742         * tree-cfg.c (move_stmt_op): Fix condition under which to update
14743         TREE_BLOCK.
14744         (move_stmt_r): Remove redundant checking.
14746 2013-04-29  Teresa Johnson  <tejohnson@google.com>
14748         PR bootstrap/57077
14749         * basic-block.h (apply_scale): New function.
14750         (apply_probability): Use apply_scale.
14751         * gimple-streamer-in.c (input_bb): Ditto.
14752         * lto-streamer-in.c (input_cfg): Ditto.
14753         * lto-cgraph.c (merge_profile_summaries): Ditto.
14754         * tree-optimize.c (execute_fixup_cfg): Ditto.
14755         * tree-inline.c (copy_bb): Update comment to use apply_scale.
14756         (copy_edges_for_bb): Ditto.
14757         (copy_cfg_body): Ditto.
14759 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14761         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
14762         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
14763         (tail_merge_optimize): Handle current_loops == NULL.
14765 2013-04-26  Jeff Law  <law@redhat.com>
14767         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
14768         (simplify_cond_using_ranges): Generalize code to simplify
14769         COND_EXPRs where one argument is a constant and the other
14770         is an SSA_NAME created by an integral type conversion.
14772 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14774         * config/arm/arm.md (store_minmaxsi): Use only when
14775         optimize_insn_for_size_p.
14777 2013-04-29  Christian Bruel  <christian.bruel@st.com>
14779         PR target/57108
14780         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
14782 2013-04-29  Richard Biener  <rguenther@suse.de>
14784         PR middle-end/57089
14785         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
14786         loop tree make sure to schedule a fixup for the child as well.
14787         (expand_omp_for_generic): Properly add loops.
14788         (expand_omp_for_static_nochunk): Likewise.
14789         (expand_omp_for_static_chunk): Likewise.
14790         (expand_omp_for): For the degenerate case fixup loops.
14791         (expand_omp_sections): Fix default bb placement in loops.
14792         (expand_omp_atomic_pipeline): Properly add loops.
14794 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14796         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
14798 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14800         * tree-ssa-tail-merge.c: Update header comment.
14802 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14804         * config/aarch64/arm_neon.h
14805         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
14806         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
14807         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
14808         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
14809         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
14810         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
14811         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
14812         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
14814 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14816         * config/aarch64/aarch64-simd.md
14817         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
14818         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
14819         fix_trunc, fixuns_trunc.
14820         (ftrunc<VDQF:mode>2): New.
14821         * config/aarch64/iterators.md (optab): Add fix, fixuns.
14822         (fix_trunc_optab): New.
14824 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14826         * config/aarch64/aarch64-builtins.c
14827         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
14828         iceilf, lround, iroundf.
14830 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14832         PR target/54349
14833         * config/i386/i386.h (enum ix86_tune_indices)
14834         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
14835         New, split from X86_TUNE_INTER_UNIT_MOVES.
14836         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
14837         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
14838         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
14839         (TARGET_INTER_UNIT_MOVES): Remove.
14840         * config/i386/i386.c (initial_ix86_tune_features): Update.
14841         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
14842         (ix86_expand_convert_uns_didf_sse): Use
14843         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14844         (ix86_expand_vector_init_one_nonzero): Ditto.
14845         (ix86_expand_vector_init_interleave): Ditto.
14846         (inline_secondary_memory_needed): Return true for moves from SSE class
14847         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
14848         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14849         * config/i386/constraints.md (Yi, Ym): Depend on
14850         TARGET_INTER_UNIT_MOVES_TO_VEC.
14851         (Yj, Yn): New constraints.
14852         * config/i386/i386.md (*movdi_internal): Change constraints of
14853         operand 1 from Yi to Yj and from Ym to Yn.
14854         (*movsi_internal): Ditto.
14855         (*movdf_internal): Ditto.
14856         (*movsf_internal): Ditto.
14857         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
14858         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14859         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
14860         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
14861         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
14862         * config/i386/sse.md (movdi_to_sse): Ditto.
14863         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
14864         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
14865         TARGET_INTER_UNIT_MOVES.
14866         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
14867         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
14868         instead of TARGET_INTER_UNIT_MOVES.
14869         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
14870         operand 1 from Yi to Yj and from Ym to Yn.
14872 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14874         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
14875         (float_truncate_hi_): Likewise.
14876         (float_extend_lo_): Likewise.
14877         (float_truncate_lo_): Likewise.
14878         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
14879         (aarch64_float_extend_lo_v2df): Likewise.
14880         (vec_unpacks_hi_v4sf): Likewise.
14881         (aarch64_float_truncate_lo_v2sf): Likewise.
14882         (aarch64_float_truncate_hi_v4sf): Likewise.
14883         (vec_pack_trunc_v2df): Likewise.
14884         (vec_pack_trunc_df): Likewise.
14886 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14888         * config/aarch64/aarch64-builtins.c
14889         (aarch64_fold_builtin): Fold float conversions.
14890         * config/aarch64/aarch64-simd-builtins.def
14891         (floatv2si, floatv4si, floatv2di): New.
14892         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
14893         * config/aarch64/aarch64-simd.md
14894         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
14895         * config/aarch64/iterators.md (FLOATUORS): New.
14896         (optab): Add float, floatuns.
14897         (su_optab): Likewise.
14899 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14901         * config/aarch64/aarch64-builtins.c
14902         (aarch64_builtin_vectorized_function): Use new names for
14903         fcvt builtins.
14904         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
14905         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
14906         (fcvtzu): Split as...
14907         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
14908         (fcvtas): Split as...
14909         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
14910         (fcvtau): Split as...
14911         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
14912         (fcvtps): Split as...
14913         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
14914         (fcvtpu): Split as...
14915         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
14916         (fcvtms): Split as...
14917         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
14918         (fcvtmu): Split as...
14919         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
14920         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
14921         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
14922         (lfrintnusf, lfrintnudf): Likewise.
14923         * config/aarch64/aarch64-simd.md
14924         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
14925         define_insn.
14926         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
14927         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
14928         (fcvt_pattern): Likewise.
14930 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14932         * config/aarch64/aarch64-simd.md
14933         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
14934         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
14936 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14938         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
14939         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
14940         (vrnd<a,m,n,p>_f32): Implement using builtins.
14941         (vrnd<i,x><q>_f<32, 64>): New.
14943 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14945         * config/aarch64/aarch64-builtins.c
14946         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
14947         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
14948         (frintz): Rename to...
14949         (btrunc): ...this.
14950         (frintp): Rename to...
14951         (ceil): ...this.
14952         (frintm): Rename to...
14953         (floor): ...this.
14954         (frinti): Rename to...
14955         (nearbyint): ...this.
14956         (frintx): Rename to...
14957         (rint): ...this.
14958         (frinta): Rename to...
14959         (round): ...this.
14960         * config/aarch64/aarch64-simd.md
14961         (aarch64_frint<frint_suffix><mode>): Delete.
14962         (<frint_pattern><mode>2): Convert to insn.
14963         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
14964         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
14965         (frint_pattern): Likewise.
14966         (frint_suffix): Likewise.
14968 2013-04-29  Richard Biener  <rguenther@suse.de>
14970         PR tree-optimization/57081
14971         * loop-init.c: Include tree-flow.h.
14972         (loop_optimizer_finalize): Free number of iteration estimates.
14973         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
14975 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
14977         PR tree-optimization/57083
14978         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
14979         non-singleton shift count range, zero extend low_bound for uns case.
14981         * config/i386/predicates.md (general_vector_operand): New predicate.
14982         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
14983         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
14984         if they aren't nonimmediate operands.  If their original values
14985         satisfy const_vector_equal_evenodd_p, don't shift them.
14986         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
14987         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
14988         if not nonimmediate_operand.
14989         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
14990         instead of register_operand.
14991         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
14993 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
14995         * stor-layout.c (finalize_size_functions): Allocate a structure and
14996         reset cfun before dumping the functions.
14998 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
15000         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
15002         PR target/56866
15003         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
15004         use xop_pmacsdqh if uns_p.
15005         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
15006         the immediate rotate count.
15008 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
15010         * rtl.h (struct rtx_def): Add comment for field jump.
15011         (LRA_SUBREG_P): New macro.
15012         * recog.c (register_operand): Check LRA_SUBREG_P.
15013         * lra.c (lra): Add note at the end of RTL code. Align non-empty
15014         stack frame.
15015         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
15016         (lra_final_code_change): Skip subreg change for operators.
15017         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
15018         if there are no operand changes.
15019         * lra-constraints.c (curr_insn_set): New.
15020         (match_reload): Set LRA_SUBREG_P.
15021         (emit_spill_move): Ditto.
15022         (check_and_process_move): Use curr_insn_set. Process only single
15023         set insns.  Don't initialize sec_mem_p and change_p.
15024         (simplify_operand_subreg): Use LRA_SUBREG_P.
15025         (reg_in_class_p): New function.
15026         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
15027         of #ifdef.  Add code to remove cycling.
15028         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
15029         non-null disp.  Reload inner instead of disp when base and index
15030         are null.  Try to put lo_sum into register.
15031         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
15032         (check_and_process_move): Move code for move cost check to
15033         simple_move_p.  Remove equiv_substitution.
15034         (simple_move_p): New function.
15035         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
15036         curr_insn_set.  Call check_and_process_move only for single set
15037         insns.  Use the new function.  Move call of check_and_process_move
15038         after operand equiv substitution and address process.
15040 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
15042         PR go/57045
15043         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
15044         with nonlocal goto receivers or returns twice calls, ignore
15045         unininitialized values from abnormal edges to nl goto receiver
15046         or returns twice call.
15048 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
15050         PR tree-optimization/57051
15051         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
15052         and VEC_RSHIFT_EXPR if shift count is a multiple of element
15053         bitsize.
15055 2013-04-26  Richard Biener  <rguenther@suse.de>
15057         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
15058         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
15059         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
15060         (fixup_loop_arrays_after_move): New function.
15061         (move_sese_region_to_fn): Properly outline the loop tree parts
15062         of the SESE region.
15064 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
15066         * config/i386/i386.md (type, unit): Fix long lines.
15068 2013-04-26  Richard Biener  <rguenther@suse.de>
15070         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
15071         (lto-streamer-out.o): Likewise.
15072         * cfgloop.c (init_loops_structure): Export, add struct function
15073         argument and adjust.
15074         (flow_loops_find): Adjust.
15075         * cfgloop.h (enum loop_estimation): Add EST_LAST.
15076         (init_loops_structure): Declare.
15077         * lto-streamer-in.c: Include cfgloop.h.
15078         (input_cfg): Input the loop tree.
15079         * lto-streamer-out.c: Include cfgloop.h.
15080         (output_cfg): Output the loop tree.
15081         (output_struct_function_base): Do not drop PROP_loops.
15083 2013-03-26  Richard Biener  <rguenther@suse.de>
15085         * tree-cfg.c (execute_build_cfg): Build the loop tree.
15086         (pass_build_cfg): Provide PROP_loops.
15087         (move_sese_region_to_fn): Remove loops that are outlined into fn
15088         for now.
15089         * tree-inline.c: Include cfgloop.h.
15090         (initialize_cfun): Do not drop PROP_loops.
15091         (copy_loops): New function.
15092         (copy_cfg_body): Copy loop structure.
15093         (tree_function_versioning): Initialize destination loop tree.
15094         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
15095         (pass_parallelize_loops): Do IL verification.
15096         * loop-init.c (loop_optimizer_init): Fixup loops if required.
15097         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
15098         the CFG make sure we fixup loops as well.
15099         * tree-ssa-tail-merge.c: Include cfgloop.h.
15100         (replace_block_by): When merging loop latches mark loops for fixup.
15101         * lto-streamer-out.c (output_struct_function_base): Drop
15102         PROP_loops for now.
15103         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
15104         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
15105         * ipa-split.c: Include cfgloop.h.
15106         (split_function): Add the new return block to the loop tree root.
15107         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
15108         whether we have removed the forwarder block.
15109         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
15110         * cfgloop.h (place_new_loop): Declare.
15111         * cfgloopmanip.c (place_new_loop): Export.
15112         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
15113         (tree-switch-conversion.o): Likewise.
15114         (tree-complex.o): Likewise.
15115         (tree-inline.o): Likewise.
15116         (tree-ssa-tailmerge.o): Likewise.
15117         (ipa-split.o): Likewise.
15118         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
15119         (tree-ssa-copy.o): Likewise.
15120         * tree-switch-conversion.c: Include cfgloop.h
15121         (process_switch): If we emit a bit-test cascade, schedule loops
15122         for fixup.
15123         * tree-complex.c: Include cfgloop.h.
15124         (expand_complex_div_wide): Properly add new basic-blocks to loops.
15125         * asan.c: Include cfgloop.h.
15126         (create_cond_insert_point): Properly add new basic-blocks to
15127         loops, schedule loop fixup.
15128         * cfgloop.c (verify_loop_structure): Check that looks are not
15129         marked for fixup.
15130         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
15131         to loops.
15132         (expand_omp_for_generic): Likewise.
15133         (expand_omp_sections): Likewise.
15134         (expand_omp_atomic_pipeline): Schedule loops for fixup.
15135         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
15136         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
15137         is initialized, not when loops are present.
15138         * tree-parloops.c (parallelize_loops): Remove checking here.
15139         * passes.c (init_optimization_passes): Schedule a copy-propagation
15140         pass before complete unrolling of inner loops.
15142 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
15144         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
15145         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
15146         (colorize_init): Add argument to _WIN32 version.
15147         * toplev.c: Include diagnostic-color.h.
15148         (process_options): Default to -fdiagnostics-color=auto if
15149         GCC_COLORS env var is in the environment.
15150         * common.opt (fdiagnostics-color=): Add Var and Init.
15151         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
15152         env var is in the environment, the default is auto rather than never.
15154         * diagnostic.h (file_name_as_prefix): Add context argument.
15155         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
15156         the string as locus.
15157         * langhooks.c (lhd_print_error_function): Adjust caller.
15159 2013-04-25  Lawrence Crowl  <crowl@google.com>
15161         * var-tracking.c (shared_hash_def::htab):
15162         Change type to hash_table.  Update dependent calls and types.
15164 2013-04-25  Lawrence Crowl  <crowl@google.com>
15166         * Makefile.in: Update as needed below.
15168         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
15169         Move declaration to after the type's method definitons.
15171         * attribs.c (htab_t scoped_attributes::attribute_hash):
15172         Change type to hash_table.  Update dependent calls and types.
15174         * bitmap.c (htab_t bitmap_desc_hash):
15175         Change type to hash_table.  Update dependent calls and types.
15177         * cselib.c (htab_t cselib_hash_table):
15178         Change type to hash_table.  Update dependent calls and types.
15180         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
15181         (hash_string_slot_node): Move implementation into lto-streamer.h
15182         struct string_slot_hasher.
15183         (eq_string_slot_node): Likewise.
15185         * data-streamer-out.c: Update output_block::string_hash_table
15186         dependent calls and types.
15188         * dwarf2cfi.c (htab_t trace_index):
15189         Change type to hash_table.  Update dependent calls and types.
15191         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
15192         Change type to hash_table.  Update dependent calls and types.
15193         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
15194         (htab_t optimize_external_refs::map): Likewise.
15195         (htab_t output_comp_unit::extern_map): Likewise.
15196         (htab_t output_comdat_type_unit::extern_map): Likewise.
15197         (htab_t output_macinfo::macinfo_htab): Likewise.
15198         (htab_t optimize_location_lists::htab): Likewise.
15199         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
15201         * except.c (htab_t ehspec_hash_type):
15202         Change type to hash_table.  Update dependent calls and types.
15203         (assign_filter_values::ttypes): Likewise.
15204         (assign_filter_values::ehspec): Likewise.
15205         (sjlj_assign_call_site_values::ar_hash): Likewise.
15206         (convert_to_eh_region_ranges::ar_hash): Likewise.
15208         * gcse.c (htab_t pre_ldst_table):
15209         Change type to hash_table.  Update dependent calls and types.
15211         * ggc-common.c (htab_t saving_htab):
15212         Change type to hash_table.  Update dependent calls and types.
15213         (htab_t loc_hash): Likewise.
15214         (htab_t ptr_hash): Likewise.
15215         (call_count): Rename ggc_call_count.
15216         (call_alloc): Rename ggc_call_alloc.
15217         (loc_descriptor): Rename make_loc_descriptor.
15218         (add_statistics): Rename ggc_add_statistics.
15220         * ggc-common.c (saving_htab):
15221         Change type to hash_table.  Update dependent calls and types.
15223         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
15224         (push_gimplify_context): Likewise.
15225         (pop_gimplify_context): Likewise.
15226         (struct gimple_temp_hash_elt): Added.
15227         (struct gimplify_hasher): Likewise.
15228         (struct gimplify_ctx.temp_htab):
15229         Change type to hash_table.  Update dependent calls and types.
15231         * gimple-fold.c: Include gimplify-ctx.h.
15233         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
15234         Change type to hash_table.  Update dependent calls and types.
15235         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
15236         avoid potential global name collision.
15238         * gimplify.c: Include gimplify-ctx.h.
15239         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
15240         (htab_t gimplify_ctx::temp_htab):
15241         Update dependent calls and types for new type hash_table.
15242         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
15243         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
15245         * gimplify-ctx.h: New.
15246         (struct gimple_temp_hash_elt): Move from gimplify.c.
15247         (class gimplify_hasher): New.
15248         (struct gimplify_ctx): Move from gimple.h.
15249         (htab_t gimplify_ctx::temp_htab):
15250         Change type to hash_table.  Update dependent calls and types.
15252         * graphite-clast-to-gimple.c: Include graphite-htab.h.
15253         (htab_t ivs_params::newivs_index):
15254         Change type to hash_table.  Update dependent calls and types.
15255         (htab_t ivs_params::params_index): Likewise.
15256         (htab_t print_generated_program::params_index): Likewise.
15257         (htab_t gloog::newivs_index): Likewise.
15258         (htab_t gloog::params_index): Likewise.
15260         * graphite.c: Include graphite-htab.h.
15261         4htab_t graphite_transform_loops::bb_pbb_mapping):
15262         Change type to hash_table.  Update dependent calls and types.
15264         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
15265         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
15266         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
15268         * graphite-dependences.c: Include graphite-htab.h.
15269         (loop_is_parallel_p): Change hash table type of parameter.
15271         * graphite-htab.h: New.
15272         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
15273         (extern find_pbb_via_hash): Move from graphite-poly.h.
15274         (extern loop_is_parallel_p): Move from graphite-poly.h.
15275         (extern get_loop_body_pbbs): Move from graphite-poly.h.
15277         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
15278         (extern loop_is_parallel_p): Move to graphite-htab.h.
15279         (extern get_loop_body_pbbs): Move to graphite-htab.h.
15281         * haifa-sched.c (htab_t delay_htab):
15282         Change type to hash_table.  Update dependent calls and types.
15283         (htab_t delay_htab_i2): Likewise.
15285         * ira-color.c (htab_t allocno_hard_regs_htab):
15286         Change type to hash_table.  Update dependent calls and types.
15288         * ira-costs.c (htab_t cost_classes_htab):
15289         Change type to hash_table.  Update dependent calls and types.
15291         * loop-invariant.c (htab_t merge_identical_invariants::eq):
15292         Change type to hash_table.  Update dependent calls and types.
15294         * loop-iv.c (htab_t bivs):
15295         Change type to hash_table.  Update dependent calls and types.
15297         * loop-unroll.c (htab_t opt_info::insns_to_split):
15298         Change type to hash_table.  Update dependent calls and types.
15299         (htab_t opt_info::insns_with_var_to_expand): Likewise.
15301         * lto-streamer.h (struct string_slot): Move from data-streamer.h
15302         (struct string_slot_hasher): New.
15303         (htab_t output_block::string_hash_table):
15304         Change type to hash_table.  Update dependent calls and types.
15306         * lto-streamer-in.c (freeing_string_slot_hasher): New.
15307         (htab_t file_name_hash_table):
15308         Change type to hash_table.  Update dependent calls and types.
15310         * lto-streamer-out.c: Update output_block::string_hash_table dependent
15311         calls and types.
15313         * lto-streamer.c (htab_t tree_htab):
15314         Change type to hash_table.  Update dependent calls and types.
15316         * omp-low.c: Include gimplify-ctx.h.
15318         * passes.c (htab_t name_to_pass_map):
15319         Change type to hash_table.  Update dependent calls and types.
15320         (pass_traverse): Rename to passes_pass_traverse.
15322         * plugin.c (htab_t event_tab):
15323         Change type to hash_table.  Update dependent calls and types.
15325         * postreload-gcse.c (htab_t expr_table):
15326         Change type to hash_table.  Update dependent calls and types.
15327         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
15329         * sese.c (debug_rename_map_1): Make extern.
15330         (htab_t copy_bb_and_scalar_dependences::rename_map):
15331         Change type to hash_table.  Update dependent calls and types.
15333         * sese.h (extern debug_rename_map): Move to .c file.
15335         * store-motion.c (htab_t store_motion_mems_table):
15336         Change type to hash_table.  Update dependent calls and types.
15338         * trans-mem.c (htab_t tm_new_mem_hash):
15339         Change type to hash_table.  Update dependent calls and types.
15341         * tree-browser.c (htab_t TB_up_ht):
15342         Change type to hash_table.  Update dependent calls and types.
15344         * tree-cfg.c (htab_t discriminator_per_locus):
15345         Change type to hash_table.  Update dependent calls and types.
15347         * tree-complex.c: Include tree-hasher.h
15348         (htab_t complex_variable_components):
15349         Change type to hash_table.  Update dependent calls and types.
15351         * tree-eh.c (htab_t finally_tree):
15352         Change type to hash_table.  Update dependent calls and types.
15354         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
15355         struct int_tree_hasher.
15356         (extern int_tree_map_eq): Likewise.
15357         (uid_decl_map_hash): Removed.
15358         (extern decl_tree_map_eq): Likewise.
15360         * tree-hasher.h: New.
15361         (struct int_tree_hasher): New.
15362         (typedef int_tree_htab_type): New.
15364         * tree-inline.c: Include gimplify-ctx.h.
15366         * tree-mudflap.c: Include gimplify-ctx.h.
15368         * tree-parloops.c: Include tree-hasher.h.
15369         (htab_t eliminate_local_variables_stmt::decl_address):
15370         Change type to hash_table.  Update dependent calls and types.
15371         (htab_t separate_decls_in_region::decl_copies): Likewise.
15373         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
15374         Change type to hash_table.  Update dependent calls and types.
15376         * tree-sra.c (candidates):
15377         Change type to hash_table.  Update dependent calls and types.
15379         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
15380         in tree-flow.h.
15381         (int_tree_map_hash): Likewise.
15383         * tree-ssa-dom.c (htab_t avail_exprs):
15384         Change type to hash_table.  Update dependent calls and types.
15386         * tree-ssa-live.c (var_map_base_init::tree_to_index):
15387         Change type to hash_table.  Update dependent calls and types.
15389         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
15390         Change type to hash_table.  Update dependent calls and types.
15392         * tree-ssa-phiopt.c (seen_ssa_names):
15393         Change type to hash_table.  Update dependent calls and types.
15395         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
15396         Change type to hash_table.  Update dependent calls and types.
15398         * tree-ssa-uncprop.c (equiv):
15399         Change type to hash_table.  Update dependent calls and types.
15401 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
15403         PR rtl-optimization/57003
15404         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
15405         call note_stores with kill_clobbered_value callback again after
15406         killing regs_invalidated_by_call.
15408 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15410         * config/aarch64/aarch64-simd.md
15411         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
15412         (aarch64_simd_bsl<mode>): Likewise.
15413         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
15415 2013-04-25  Marek Polacek  <polacek@redhat.com>
15417         PR tree-optimization/57066
15418         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
15420 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
15422         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
15424 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15426         * config/aarch64/aarch64-builtins.c
15427         (aarch64_fold_builtin): New.
15428         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
15429         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
15430         * config/aarch64/aarch64-simd-builtins.def (abs): New.
15431         * config/aarch64/arm_neon.h
15432         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
15434 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15435             Tejas Belagod  <tejas.belagod@arm.com>
15437         * config/aarch64/aarch64-builtins.c
15438         (aarch64_gimple_fold_builtin): New.
15439         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
15440         * config/aarch64/aarch64-simd-builtins.def (addv): New.
15441         * config/aarch64/aarch64-simd.md (addpv4sf): New.
15442         (addvv4sf): Update.
15443         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
15445 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15447         * config/aarch64/aarch64.md
15448         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
15450 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15452         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
15453         (*ngcsi_uxtw): New pattern.
15455 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15456             Julian Brown  <julian@codesourcery.com>
15458         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
15459         (TB_DREG): Add T_V4HF.
15460         (v4hf_UP): New macro.
15461         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15462         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15463         Handle initialisation of V4HF. Adjust initialisation of reinterpret
15464         built-ins.
15465         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
15466         (arm_vector_mode_supported_p): Handle V4HF.
15467         (arm_mangle_map): Handle V4HFmode.
15468         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
15469         * config/arm/arm_neon_builtins.def: Add entries for
15470         vcvtv4hfv4sf, vcvtv4sfv4hf.
15471         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
15472         (neon_vcvtv4hfv4sf): Likewise.
15473         * config/arm/neon-gen.ml: Handle half-precision floating point
15474         features.
15475         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
15476         * config/arm/arm_neon.h: Regenerate.
15477         * config/arm/neon.ml (type elts): Add F16.
15478         (type vectype): Add T_float16x4, T_floatHF.
15479         (type vecmode): Add V4HF.
15480         (type features): Add Requires_FP_bit feature.
15481         (elt_width): Handle F16.
15482         (elt_class): Likewise.
15483         (elt_of_class_width): Likewise.
15484         (mode_of_elt): Refactor.
15485         (type_for_elt): Handle F16, fix error messages.
15486         (vectype_size): Handle T_float16x4.
15487         (vcvt_sh): New function.
15488         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
15489         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
15490         (string_of_mode): Handle V4HF.
15491         * doc/arm-neon-intrinsics.texi: Regenerate.
15493 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
15495         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
15496         format specifier in 'X' case.
15498 2013-04-25  Alan Modra  <amodra@gmail.com>
15500         PR target/57052
15501         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
15502         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
15503         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
15504         Repeat for many other rotate/shift and mask patterns using subregs.
15505         Name lshiftrt insns.
15506         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
15507         on WORDS_BIG_ENDIAN.
15509 2013-04-25  Alan Modra  <amodra@gmail.com>
15511         * config.gcc: Support little-endian powerpc-linux targets.
15512         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
15513         (LINK_OS_LINUX_SPEC): Define.
15514         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
15515         Preserve MASK_LITTLE_ENDIAN.
15516         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
15517         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
15518         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
15519         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
15520         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
15521         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
15522         Correct fp word order for little-endian.  Don't shift toc entries
15523         smaller than a word for little-endian.
15524         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
15525         (bswapdi2 splits): Correct low-part subreg for little-endian.
15526         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
15527         low/high where such is correct only for be.
15528         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
15529         little-endian for -mcall-aixdesc.
15531 2013-04-25  Alan Modra  <amodra@gmail.com>
15533         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
15534         replace_equiv_address_nv.
15536 2013-04-25  Alan Modra  <amodra@gmail.com>
15538         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
15540 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15542         Revert:
15543         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15544         * rtl.h (struct rtx_def): ...
15546 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15548         PR rtl-optimizations/57046
15549         * lra-constraints (split_reg): Set up lra_risky_transformations_p
15550         for multi-reg splits.
15552 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
15554         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
15556 2013-04-24  Sterling Augustine  <saugustine@google.com>
15558         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
15559         (comp_dir_string, debug_str_dwo_section): New.
15560         (DEBUG_STR_DWO_SECTION): Rename to ...
15561         (DEBUG_DWO_STR_SECTION): ... this.
15562         (DEBUG_NORM_STR_SECTION): Delete.
15563         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
15564         (DEBUG_STR_DWO_SECTION_FLAGS): New.
15565         (find_AT_string): Move most logic to ...
15566         (find_AT_string_in_table): ... here.  New.
15567         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
15568         add_skeleton_AT_string.  Delete logic.
15569         (output_skeleton_debug_sections): Remove call to
15570         add_top_level_skeleton_die_attrs.
15571         (add_comp_dir_attribute): Move logic to comp_dir_string.
15572         (dwarf2out_init): Initialize debug_str_dwo_section.
15573         (output_indirect_string): Call find_string_form.
15574         (output_indirect_strings): Rewrite.
15575         (prune_unused_types): Empty skeleton_debug_str_hash.
15576         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
15577         (dwarf2out_finish):  Call output_indirect_strings.
15579 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
15581         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
15583 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
15585         * rtl.h (struct rtx_def): Add comment for field jump.
15586         (LRA_SUBREG_P): New macro.
15587         * recog.c (register_operand): Check LRA_SUBREG_P.
15588         * lra.c (lra): Add note at the end of RTL code. Align non-empty
15589         stack frame.
15590         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
15591         (lra_final_code_change): Skip subreg change for operators.
15592         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
15593         if there are no operand changes.
15594         * lra-constraints.c (curr_insn_set): New.
15595         (match_reload): Set LRA_SUBREG_P.
15596         (emit_spill_move): Ditto.
15597         (check_and_process_move): Use curr_insn_set. Process only single
15598         set insns.  Don't initialize sec_mem_p and change_p.
15599         (simplify_operand_subreg): Use LRA_SUBREG_P.
15600         (reg_in_class_p): New function.
15601         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
15602         of #ifdef.  Add code to remove cycling.
15603         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
15604         non-null disp.  Reload inner instead of disp when base and index
15605         are null.  Try to put lo_sum into register.
15606         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
15607         (check_and_process_move): Move code for move cost check to
15608         simple_move_p.  Remove equiv_substitution.
15609         (simple_move_p): New function.
15610         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
15611         curr_insn_set.  Call check_and_process_move only for single set
15612         insns.  Use the new function.  Move call of check_and_process_move
15613         after operand equiv substitution and address process.
15615 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
15617         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
15618         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
15619         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
15621 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
15623         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
15625 2013-04-24  Marek Polacek  <polacek@redhat.com>
15627         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
15628         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
15629         (select_loops_exit_conditions): Likewise.
15630         (number_of_iterations_for_all_loops): Likewise.
15631         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
15632         (scev_analysis): Likewise.
15634 2013-04-02  Catherine Moore  <clm@codesourcery.com>
15635             Chao-ying Fu  <fu@mips.com>
15637         * config/mips/micromips.md (jraddiusp): New pattern.
15638         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
15639         instruction if possible.
15641 2013-04-24  Alan Modra  <amodra@gmail.com>
15643         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
15645 2013-04-24  Julian Brown  <julian@codesourcery.com>
15646             Chung-Lin Tang  <cltang@codesourcery.com>
15648         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
15649         dependency behavior in enumeration type DIE generation. Add TODO note
15650         to comments about future DW_FORM_sdata/udata re-work of related code.
15652 2013-04-23  Lawrence Crowl  <crowl@google.com>
15654         * Makefile.in: Update as needed below.
15656         * hash-table.h (class hash_table):
15657         Correct many methods with parameter types compare_type to the correct
15658         value_type.  (Correct code was unlikely to notice the change.)
15659         (hash_table::elements_with_deleted) New.
15660         (class hashtable::iterator): New.
15661         (hashtable::begin()): New.
15662         (hashtable::end()): New.
15663         (FOR_EACH_HASH_TABLE_ELEMENT): New.
15665         * statistics.c (statistics_hashes):
15666         Change type to hash_table.  Update dependent calls and types.
15668         * tree-into-ssa.c (var_infos):
15669         Change type to hash_table.  Update dependent calls and types.
15671         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
15672         Change type to hash_table.  Update dependent calls and types.
15674         * tree-ssa-loop-im.c (struct mem_ref.refs):
15675         Change type to hash_table.  Update dependent calls and types.
15677         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
15678         Change type to hash_table.  Update dependent calls and types.
15680         * tree-ssa-sccvn.c (vn_tables_s::nary):
15681         Change type to hash_table.  Update dependent calls and types.
15682         (vn_tables_s::phis): Likewise.
15683         (vn_tables_s::references): Likewise.
15685         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
15686         (vn_reference_eq): Update parameter and return types.
15688         * tree-ssa-structalias.c (pointer_equiv_class_table):
15689         Change type to hash_table.  Update dependent calls and types.
15690         (location_equiv_class_table): Likewise.
15692         * tree-vect-data-refs.c: Consequential changes for making
15693         peeling a hash_table.
15695         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
15696         (destroy_loop_vec_info): Dependent hash_table update.
15698         * tree-vectorizer.h (peeling_htab):
15699         Change type to hash_table.  Update dependent calls and types.
15701 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
15703         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
15704         to check the register content is equal or not.
15705         * lra-constraints.c (match_reload): Use lra_assign_reg_val
15706         to assign register content record.
15707         * lra-eliminations.c (update_reg_eliminate): Use
15708         lra_update_reg_val_offset to update register content offset.
15709         * lra-int.h (struct lra_reg): Add offset member.
15710         (lra_reg_val_equal_p): New static inline function.
15711         (lra_update_reg_val_offset): New static inline function.
15712         (lra_assign_reg_val): New static inline function.
15713         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
15714         to assign register content record.
15715         (initialize_lra_reg_info_element): Initial offset to zero.
15717 2013-04-23  Catherine Moore  <clm@codesourcery.com>
15719         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
15720         operands.  Record compression.
15722 2013-04-23  Xinliang David Li  <davidxl@google.com>
15724         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
15726 2013-04-23  Richard Biener  <rguenther@suse.de>
15728         PR middle-end/57036
15729         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
15730         parameter, only add abnormal goto edges from the copied body
15731         if the call could perform abnormal gotos.
15732         (copy_cfg_body): Adjust.
15734 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
15736         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
15738 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
15740         * coretypes.h (gimple_stmt_iterator): Add struct to make
15741         compatible with C.
15743 2013-04-23  Richard Biener  <rguenther@suse.de>
15745         PR tree-optimization/57026
15746         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
15747         from SSA names occuring in abnormal PHI nodes.
15749 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15751         * lto/lto.c (print_lto_report_1): Fix LTO report names.
15753 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15755         * lto/lto.c (print_lto_report_1): Declare early.
15756         (read_cgraph_and_symbols): Call print_lto_report_1 early.
15758 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15760         * common.opt (-flto-report-wpa): Add.
15761         * doc/invoke.texi (-flto-report-wpa): Add.
15762         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
15763         (lto_main): dito.
15765 2013-04-22  Xinliang David Li  <davidxl@google.com>
15767         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
15768         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
15769         * Makefile.in: New dependency
15771             David Daney <ddaney.cavm@gmail.com>
15773         * configure.ac (gcc_cv_as_micromips_support):  Use the
15774         --fatal-warnings option.
15775         * configure: Regenerate.
15777 2013-04-22  Marek Polacek  <polacek@redhat.com>
15779         PR sanitizer/56990
15780         * tsan.c (instrument_expr): Don't instrument expression
15781         in case its size is zero.
15783 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
15785         PR target/57032
15786         Revert:
15787         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15789         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15791 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15793         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
15794         (gimple_stmt_iterator): New typedef.
15795         * gimple.h (gimple_stmt_iterator): Rename to...
15796         (gimple_stmt_iterator_d): ... This.
15797         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
15798         trees be valid for GIMPLE and GENERIC.
15799         (TARGET_GIMPLE_FOLD_BUILTIN): New.
15800         * gimple-fold.c (gimple_fold_call): Call target hook
15801         gimple_fold_builtin.
15802         * hooks.c (hook_bool_gsiptr_false): New.
15803         * hooks.h (hook_bool_gsiptr_false): New.
15804         * target.def (fold_stmt): New.
15805         * doc/tm.texi: Regenerate.
15807 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
15809         PR target/57018
15810         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
15811         a set sp if no stack realignment.
15813 2013-04-22  Nick Clifton  <nickc@redhat.com>
15815         * config.gcc (tilegx-linux): Extend extra_objs rather than
15816         overwriting it.
15817         (tilepro-linux): Likewise.
15819 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15821         * config/aarch64/aarch64-builtins.c
15822         (CF): Remove.
15823         (CF0, CF1, CF2, CF3, CF4, CF10): New.
15824         (VAR<1-12>): Add MAP parameter.
15825         (BUILTIN_*): Likewise.
15826         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
15827         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
15828         (aarch64_ushl_n<mode>): Likewise.
15829         (aarch64_sshr_n<mode>): Likewise.
15830         (aarch64_ushr_n<mode>): Likewise.
15831         (aarch64_<maxmin><mode>): Likewise.
15832         (aarch64_sqrt<mode>): Likewise.
15833         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
15834         (vshr<q>_n_*): Likewise.
15836 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15838         * config/aarch64/aarch64-builtins.c
15839         (aarch64_simd_builtin_type_mode): Handle SF types.
15840         (sf_UP): Define.
15841         (BUILTIN_GPF): Define.
15842         (aarch64_init_simd_builtins): Handle SF types.
15843         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
15844         (frecps): Likewise.
15845         (frecpx): Likewise.
15846         * config/aarch64/aarch64-simd.md
15847         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
15848         (aarch64_frecpe<mode>): New.
15849         (aarch64_frecps<mode>): Likewise.
15850         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
15851         (v8type): Add frecp<esx>.
15852         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
15853         (aarch64_frecps<mode>): Likewise.
15854         * config/aarch64/iterators.md (FRECP): New.
15855         (frecp_suffix): Likewise.
15856         * config/aarch64/arm_neon.h
15857         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
15859 2013-04-22  Christian Bruel  <christian.bruel@st.com>
15861         PR target/56995
15862         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
15863         (REG_CLASS_NAMES): Idem.
15864         (REG_CLASS_CONTENTS): Idem.
15865         (REGCLASS_HAS_FP_REG): Idem.
15866         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
15867         (sh_conditional_register_usage): Idem.
15869 2013-04-21  Jeff Law  <law@redhat.com>
15871         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
15872         (ssa_forward_propagate_and_combine): Use it.
15874 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15876         * lra.c: Update the flow chart diagram.
15878 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15880         PR rtl-optimization/56847
15881         * lra-constraints.c (process_alt_operands): Discourage alternative
15882         with non-matche doffsettable memory constraint fro memory with
15883         known offset.
15885 2013-04-19  Richard Biener  <rguenther@suse.de>
15887         PR tree-optimization/56982
15888         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
15889         function.
15890         * gimplify.c (gimplify_call_expr): Notice special calls.
15891         (gimplify_modify_expr): Likewise.
15892         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
15893         abnormal control flow receivers.
15894         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
15895         in the same way as cfun->has_nonlocal_labels.
15896         (gimple_purge_dead_abnormal_call_edges): Likewise.
15897         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
15898         receivers start a basic-block.
15900 2013-04-19  Richard Biener  <rguenther@suse.de>
15902         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
15903         member ...
15904         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
15905         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
15906         (SLP_TREE_LOAD_PERMUTATION): Add.
15907         (vect_transform_slp_perm_load): Adjust prototype.
15908         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
15909         (vect_free_slp_instance): Likewise.
15910         (vect_create_new_slp_node): Likewise.
15911         (vect_supported_slp_permutation_p): Remove.
15912         (vect_slp_rearrange_stmts): Adjust.
15913         (vect_supported_load_permutation_p): Likewise.  Inline
15914         vect_supported_slp_permutation_p here.
15915         (vect_analyze_slp_instance): Compute load permutations per
15916         slp node instead of per instance.
15917         (vect_get_slp_defs): Adjust.
15918         (vect_transform_slp_perm_load): Likewise.
15919         (vect_schedule_slp_instance): Remove redundant code.
15920         (vect_schedule_slp): Remove hack for PR56270, add it ...
15921         * tree-vect-stmts.c (vectorizable_load): ... here, do not
15922         CSE loads for SLP.  Adjust.
15924 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
15926         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
15927         spelling in two comments.
15929 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
15931         PR target/56797
15932         * config/arm/arm.c (load_multiple_sequence): Require SP
15933         as base register for loads if SP is in the register list.
15935 2013-04-19  Martin Jambor  <mjambor@suse.cz>
15937         PR tree-optimization/56718
15938         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
15939         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
15940         and made public.  Adjusted all callers.
15941         (ipa_intraprocedural_devirtualization): New function.
15942         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
15943         (ipa_intraprocedural_devirtualization): Likewise.
15944         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
15946 2013-04-19  Richard Biener  <rguenther@suse.de>
15948         PR tree-optimization/57000
15949         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
15951 2013-04-19  Terry Guo  <terry.guo@arm.com>
15953         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
15954         Replace with ...
15955         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
15956         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
15957         (cortex_m4_fmacs): Use new reservations.
15958         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
15960 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
15962         PR rtl-optimization/56999
15963         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
15964         related code.
15965         (lra_coalesce): Remove split_origin_bitmap and related code.
15966         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
15967         ranges if necessary.
15969 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
15971         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
15972         New array.
15973         (ix86_expand_call): Remove clobbered_registers array and use
15974         x86_64_ms_sysv_extra_clobbered_registers instead.
15975         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
15976         Declare here.
15977         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
15978         predicate.
15979         * config/i386/i386.md (*call_rex64_ms_sysv): Use
15980         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
15981         (*call_value_rex64_ms_sysv): Ditto.
15983 2013-04-18  Cary Coutant  <ccoutant@google.com>
15985         * dwarf2out.c (output_pubnames): Check die_perennial_p of
15986         parent instead of die_mark.
15988 2013-04-18  Diego Novillo  <dnovillo@google.com>
15990         * gimple.c (create_gimple_tmp): New.
15991         (get_expr_type): New.
15992         (build_assign): New.
15993         (build_type_cast): New.
15994         * gimple.h (enum ssa_mode): Define.
15995         (gimple_seq_set_location): New.
15996         * asan.c (build_check_stmt): Change some gimple_build_* calls
15997         to use build_assign and build_type_cast.
15999 2013-04-18  Richard Biener  <rguenther@suse.de>
16001         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
16002         handle negative step.  Remove redundant checks.
16003         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
16004         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
16005         for negative step and grouped loads fail to vectorize.
16007 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
16009         * emit-rtl.c (reset_insn_used_flags): New function.
16010         (reset_all_used_flags): Use it.
16011         (verify_insn_sharing): New function.
16012         (verify_rtl_sharing): Fix verification for SEQUENCEs.
16014 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
16016         PR tree-optimization/56984
16017         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
16018         and (x >> M) >= N don't register any assertion if N << M is the
16019         minimum value.
16021 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
16023         * lower-subreg.c (resolve_simple_move): If called self-recursive,
16024         do not delete_insn insns that have not yet been emitted, only
16025         unlink them with remove_insn.
16026         * df-scan.c (df_insn_delete): Revert r197492.
16028 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
16030         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
16031         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
16033 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
16035         * config/arm/arm.md (movsicc_insn): Convert define_insn into
16036         define_insn_and_split.
16037         (and_scc,ior_scc,negscc): Likewise.
16038         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
16040 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
16042         * config/arm/arm.c (use_return_insn): Return 0 for targets that
16043         can benefit from using a sequence of LDRD instructions in epilogue
16044         instead of a single LDM instruction.
16046 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
16048         PR 45688
16049         * doc/extend.texi: Fix typo.
16051 2013-04-17  Richard Biener  <rguenther@suse.de>
16053         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
16054         (vect_build_slp_tree): ... here.
16055         (vect_build_slp_tree_1): Compute which stmts of the SLP group
16056         match.  Remove special-casing of mismatched complex loads.
16057         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
16058         re-try the match with swapped commutative operands.
16059         (vect_supported_load_permutation_p): Remove special-casing of
16060         mismatched complex loads.
16061         (vect_analyze_slp_instance): Adjust.
16063 2013-04-17  Richard Biener  <rguenther@suse.de>
16065         PR rtl-optimization/56921
16066         * cfgloop.h (struct loop): Add simple_loop_desc member.
16067         (struct niter_desc): Mark with GTY(()).
16068         (simple_loop_desc): Do not use aux field but simple_loop_desc.
16069         * loop-iv.c (get_simple_loop_desc): Likewise.
16070         (free_simple_loop_desc): Likewise.
16072         Revert
16073         2013-04-16  Richard Biener  <rguenther@suse.de>
16075         PR rtl-optimization/56921
16076         * loop-init.c (pass_rtl_move_loop_invariants): Add
16077         TODO_do_not_ggc_collect to todo_flags_finish.
16078         (pass_rtl_unswitch): Same.
16079         (pass_rtl_unroll_and_peel_loops): Same.
16080         (pass_rtl_doloop): Same.
16082 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
16084         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
16085         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
16086         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
16087         references.
16088         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
16089         * tree-streamer.c (record_common_node): Adjust reference in comment.
16091 2013-04-17  Terry Guo  <terry.guo@arm.com>
16093         * config/arm/cortex-m4.md: Add a new bypass.
16095 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16097         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
16098         New pattern.
16099         (*subs_<optab><mode>_multp2): New pattern.
16100         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
16101         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
16103 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16105         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
16106         (*subs_mul_imm_<mode>): New pattern.
16108 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
16110         PR target/56948
16111         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
16112         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
16113         (vsx_movti_32bit): Change j->wa to O->wa.
16115 2013-04-16  Richard Biener  <rguenther@suse.de>
16117         PR rtl-optimization/56921
16118         * loop-init.c (pass_rtl_move_loop_invariants): Add
16119         TODO_do_not_ggc_collect to todo_flags_finish.
16120         (pass_rtl_unswitch): Same.
16121         (pass_rtl_unroll_and_peel_loops): Same.
16122         (pass_rtl_doloop): Same.
16124 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16126         * config/arm/arm.c (emit_multi_reg_push): New declaration
16127         for an existing function.
16128         (arm_emit_strd_push): New function.
16129         (arm_expand_prologue): Used here.
16130         (arm_emit_ldrd_pop): New function.
16131         (arm_expand_epilogue): Used here.
16132         (arm_get_frame_offsets): Update condition.
16133         (arm_emit_multi_reg_pop): Add a special case for load of a single
16134         register with writeback.
16136 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
16138         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
16139         description.
16141 2013-04-16  Richard Biener  <rguenther@suse.de>
16143         PR tree-optimization/56756
16144         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
16145         (first_mem_ref_loc): New.
16146         (execute_sm): Place the load temporarily before a previous
16147         access instead of in the latch edge to ensure its SSA dependencies
16148         are defined at points dominating the load.
16150 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
16152         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
16153         correct fix by moving header and footer insn to the footer of
16154         the merged basic block.  Clear BB_END of the merged-away block.
16156         PR middle-end/43631
16157         * emit-rtl.c (make_note_raw): New function.
16158         (link_insn_into_chain): New static inline function.
16159         (add_insn): Use it.
16160         (add_insn_before, add_insn_after): Factor insn chain linking code...
16161         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
16162         using link_insn_into_chain.
16163         (note_outside_basic_block_p): New helper function for emit_note_after
16164         and emit_note_before.
16165         (emit_note_after): Use nobb variant of add_insn_after if the note
16166         should not be contained in a basic block.
16167         (emit_note_before): Use nobb variant of add_insn_before if the note
16168         should not be contained in a basic block.
16169         (emit_note_copy): Use make_note_raw.
16170         (emit_note): Likewise.
16171         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
16172         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
16173         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
16174         the moved barrier the tail of the basic block it follows.
16175         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
16177 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
16179         PR tree-optimization/56962
16180         * gimple-ssa-strength-reduction.c (record_increment): Only set
16181         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
16182         either rhs1 or rhs2 is equal to c->base_expr.
16184 2013-04-15  Richard Biener  <rguenther@suse.de>
16186         PR tree-optimization/56933
16187         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
16188         member.
16189         (GROUP_READ_WRITE_DEPENDENCE): Remove.
16190         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
16191         * tree-vect-data-refs.c (vect_analyze_group_access): Move
16192         dependence check ...
16193         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
16194         ... here.
16195         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
16196         GROUP_READ_WRITE_DEPENDENCE.
16198 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16200         * emit-rtl.c (reset_all_used_flags): New function.
16201         (verify_rtl_sharing): Call reset_all_used_flags before and after
16202         performing the checks.
16204 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16206         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
16207         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
16208         * config/arm/constraints.md (De): New constraint.
16209         * config/arm/neon.md (anddi3_neon): Delete.
16210         (neon_vand<mode>): Expand to standard anddi3 pattern.
16211         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
16212         Move earlier in the file.
16213         (neon_inv_logic_op2): Likewise.
16214         (arm_anddi_operand_neon): New predicate.
16216 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16218         * configure.ac (gcc_cv_ld_as_needed): Set
16219         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
16220         Use -z ignore, -z record on *-*-solaris2*.
16221         (HAVE_LD_AS_NEEDED): Update comment.
16222         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
16223         * configure: Regenerate.
16224         * config.in: Regenerate.
16225         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
16226         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
16227         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
16228         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
16229         equivalents.  Fix markup.
16230         * doc/tm.texi: Regenerate.
16232 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
16234         * config/i386/i386.opt: New option mstack-protector-guard=.
16235         * config/i386/i386-opts.h: Add enum stack_protector_guard.
16236         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
16237         TARGET_SSP_TLS_GUARD.
16238         * config/i386/i386.c (ix86_option_override_internal): Set
16239         ix86_stack_protector_guard.
16240         * config/i386/i386.md (stack_protect_set): Enable for
16241         TARGET_SSP_TLS_GUARD only.
16242         (stack_protect_set_<mode>): Ditto.
16243         (stack_protect_test): Ditto.
16244         (stack_protect_test_<mode>): Ditto.
16245         * doc/invoke.texi (i386 Option): Document.
16247 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
16249         PR target/56890
16250         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
16251         (S_MODES): Set H_MODE bit.
16252         (SF_MODES): Set only S_MODE and SF_MODE bits.
16253         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
16254         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
16255         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
16256         <MODE_FLOAT>: Likewise.
16258 2013-04-15  Joey Ye  <joey.ye@arm.com>
16260         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
16262 2013-04-15  Joey Ye  <joey.ye@arm.com>
16264         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
16265         for real far jump.
16266         (thumb_far_jump_used_p): Count instruction size and set
16267         far_jump_used.
16269 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
16271         * reorg.c (fill_simple_delay_slots): Reindent block of code.
16272         * resource.c (mark_target_live_regs): Reformat conditional block.
16274 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
16276         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
16277         notes, they are emitted only just before final.
16278         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
16280 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
16282         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
16283         * cfgrtl.c (delete_insn): Call it here instead.
16284         * lra-spills.c (lra_final_code_change): Use delete_insn.
16285         * haifa-sched.c (sched_remove_insn): Likewise.
16286         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
16287         returning to the nop pool.
16288         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
16289         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
16291 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
16293         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
16294         * doc/tm.texi: Regenerated.
16296 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
16298         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
16299         QImode checks.
16301 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
16303         * df-core.c (df_find_def): Compare register numbers.
16304         (df_find_use): Likewise.
16306 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
16308         PR target/56903
16309         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
16310         lra_in_progress for return.
16312 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
16314         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
16315         define_insn into define_insn_and_split and emit movsicc patterns.
16317 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
16319         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
16321 2013-04-12  Richard Biener  <rguenther@suse.de>
16323         * tree-pass.h (TODO_do_not_ggc_collect): New.
16324         * passes.c (execute_one_ipa_transform_pass): Honor
16325         TODO_do_not_ggc_collect.
16326         (execute_one_pass): Likewise.
16328         Revert
16329         2013-04-10  Richard Biener  <rguenther@suse.de>
16331         * passes.c (init_optimization_passes): Remove reload pass.
16332         * ira.c (do_reload): Merge into ...
16333         (ira): ... this.
16334         (rest_of_handle_reload): Remove.
16335         (pass_reload): Likewise.
16336         * config/i386/i386.c (ix86_option_override): Refer to ira instead
16337         of reload for vzeroupper pass placement.
16339 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
16341         PR tree-optimization/56918
16342         PR tree-optimization/56920
16343         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
16344         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
16345         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
16346         use wide_mul_with_sign method.
16348 2013-04-12  Richard Biener  <rguenther@suse.de>
16350         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
16351         not be considered a gimple constant.
16353 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
16355         * fold-const.c (const_binop): Handle vector shifts by a scalar.
16356         (fold_binary_loc): Call const_binop also for mixed vector-scalar
16357         operations.
16359 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
16360             Jakub Jelinek  <jakub@redhat.com>
16362         * opts.c: Include diagnostic-color.h.
16363         (common_handle_option): Handle OPT_fdiagnostics_color_.
16364         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
16365         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
16366         (diagnostic-color.o): New.
16367         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
16368         (diagnostic_color_rule): New enum.
16369         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
16370         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
16371         the location string.
16372         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
16373         either NULL, or color kind.
16374         * diagnostic-color.c: New file.
16375         * diagnostic-color.h: New file.
16376         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
16377         arguments.
16378         * doc/invoke.texi (-fdiagnostics-color): Document.
16379         * pretty-print.h (pp_show_color): Define.
16380         (struct pretty_print_info): Add show_color field.
16381         * diagnostic.c: Include diagnostic-color.h.
16382         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
16383         macros.  Colorize error:, warning: etc. strings and also the location
16384         string.
16385         (diagnostic_show_locus): Colorize the caret line.
16386         * pretty-print.c: Include diagnostic-color.h.
16387         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
16388         inside of %< %> quotes or quoted through q format modifier.
16390 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16392         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
16394 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16396         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
16397         code in CC_NZ mode.
16398         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
16399         pattern.
16401 2013-04-11  Marek Polacek  <polacek@redhat.com>
16403         PR tree-optimization/48184
16404         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
16406 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
16408         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
16409         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
16410         (skip_simple_arithmetic): Tidy up.
16411         * tree.h (skip_simple_constant_arithmetic): Declare.
16413 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16415         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
16417 2013-04-11  Richard Biener  <rguenther@suse.de>
16419         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16420         generate vector constants.
16422 2013-04-11  Richard Biener  <rguenther@suse.de>
16424         PR tree-optimization/56878
16425         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
16426         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
16427         New function.
16428         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16429         Prefer to align the DR with the most invariant base address.
16431 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
16433         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
16434         comment.
16436 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
16438         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
16439         floating-point vector comparisons against 0.
16441 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
16443         PR tree-optimization/56899
16444         * fold-const.c (extract_muldiv_1): Apply distributive law
16445         only if TYPE_OVERFLOW_WRAPS (ctype).
16447 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
16449         PR target/56124
16450         * ira-costs.c (scan_one_insn): Check whether the source rtx of
16451         loading has side effect.
16453 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
16455         * config/sparc/sparc.c: Include tree-pass.h.
16456         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
16457         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
16458         head of file.  Change return type.  Split off gate function.
16459         (sparc_gate_work_around_errata): New function.
16460         (pass_work_around_errata): New pass definition.
16461         (insert_pass_work_around_errata) New pass insert definition to
16462         insert pass_work_around_errata just after delayed-branch scheduling.
16463         (sparc_option_override): Insert the pass.
16464         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
16466 2013-04-10  David S. Miller  <davem@davemloft.net>
16468         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
16469         or -mcpu=hypersparc.
16471         * target.def (cstore_mode): New hook.
16472         * target.h: Include insn-codes.h
16473         * targhooks.c: Likewise.
16474         (default_cstore_mode): New function.
16475         * targhooks.h: Declare it.
16476         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
16477         * doc/tm.texi: Rebuild.
16478         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
16479         target hook, rather than inspecting the insn_data.
16480         * config/sparc/sparc.c (sparc_cstore_mode): New function.
16481         (TARGET_CSTORE_MODE): Redefine.
16482         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
16483         result patterns.
16484         * config/sparc/predicates.md (cstore_result_operand): New special
16485         predicate.
16486         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
16487         Use it for operand 0.
16488         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
16489         (*snesi_special): Likewise.
16490         (*snesi_zero): Likewise.
16491         (*seqsi_zero): Likewise.
16492         (*sltu_insn): Likewise.
16493         (*sgeu_insn): Likewise.
16494         (*seqdi_special): Make operand 0 and comparison operation be of
16495         DImode.
16496         (*snedi_special): Likewise.
16497         (*snedi_special_vis3): Likewise.
16498         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
16499         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
16500         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
16501         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
16502         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
16503         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
16504         (*sltu_extend_sp64): Likewise.
16505         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
16506         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
16507         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
16508         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
16509         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
16511 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
16513         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
16514         (aarch64_start_file): Use the new function.
16516 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
16517             Jason Merrill  <jason@redhat.com>
16519         * common.opt: Add -gdwarf.
16520         * opts.c (common_handle_option): Handle it.
16521         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
16523 2013-04-10  Richard Biener  <rguenther@suse.de>
16525         * passes.c (execute_todo): Do not call ggc_collect conditional here.
16526         (execute_one_ipa_transform_pass): But unconditionally here.
16527         (execute_one_pass): And here.
16528         (init_optimization_passes): Remove reload pass.
16529         * tree-pass.h (TODO_ggc_collect): Remove.
16530         (pass_reload): Likewise.
16531         * ira.c (do_reload): Merge into ...
16532         (ira): ... this.
16533         (rest_of_handle_reload): Remove.
16534         (pass_reload): Likewise.
16535         * config/i386/i386.c (ix86_option_override): Refer to ira instead
16536         of reload for vzeroupper pass placement.
16537         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
16538         and todo_flags_finish of all passes.
16540 2013-04-10  Richard Biener  <rguenther@suse.de>
16542         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
16543         first_const_oprnd field, rename first_def_type to first_op_type.
16544         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
16545         (vect_get_and_check_slp_defs): Always use the type of the
16546         operand.  Allow mixed vect_external_def, vect_constant_def types.
16547         (vect_get_constant_vectors): Handle mixed vect_external_def,
16548         vect_constant_def types.
16550 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
16552         PR tree-optimization/55524
16553         * tree-ssa-math-opts.c
16554         (convert_mult_to_fma): Don't use an fms construct
16555         when we don't have an fms operation, but fnma, and it looks
16556         likely that we'll be able to use the latter.
16558 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
16560         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
16561         function.
16562         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
16563         inline fail caused by overwritable functions.
16565 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
16567         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
16568         unnecessary bits in the constant power of two case.
16570 2013-04-10  Richard Biener  <rguenther@suse.de>
16572         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
16573         broken code swapping operands.
16574         (vect_build_slp_tree): Do not compute load permutations here.
16575         (vect_analyze_slp_instance): Compute load permutations here,
16576         after building the SLP tree.
16578 2013-04-09  Christian Bruel  <christian.bruel@st.com>
16580         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
16581         of next/prev_real_insn.
16583 2013-04-09  Jan Hubicka  <jh@suse.cz>
16585         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
16586         Drop aliased parameter.
16587         (function_and_variable_visibility): Do not handle alias pairs.
16588         * cgraph.c (varpool_externally_visible_p): Update prototype.
16589         * varpool.c (varpool_add_new_variable): Update.
16591 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16593         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
16595 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
16597         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
16599         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
16601 2013-04-09  Marek Polacek  <polacek@redhat.com>
16603         PR tree-optimization/48762
16604         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
16606 2013-04-09  Richard Biener  <rguenther@suse.de>
16608         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
16609         dealing with cost.
16610         (vect_build_slp_tree): Likewise.
16611         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
16612         calculating the cost of a SLP instance.
16613         (vect_analyze_slp_instance): Use it from here, after building
16614         the SLP tree.
16616 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
16618         PR middle-end/56883
16619         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
16620         expand_omp_for_static_chunk): Use simple_p = true in
16621         force_gimple_operand_gsi calls when assigning to addressable decls.
16623 2013-04-09  Jeff Law  <law@redhat.com>
16625         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
16626         when the boolean was created by converting a wider object which
16627         had a boolean range.
16629 2013-04-09  Richard Biener  <rguenther@suse.de>
16631         * tree-vectorizer.h (slp_void_p): Remove.
16632         (slp_tree): Typedef before _slp_tree declaration.
16633         (struct _slp_tree): Use a vector of slp_tree as children.
16634         (vect_get_place_in_interleaving_chain): Remove.
16635         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
16636         Move ...
16637         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
16638         and make static.
16639         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
16640         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
16641         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
16642         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
16643         Use slp_node instead of slp_void_p and adjust.
16645 2013-04-09  Richard Biener  <rguenther@suse.de>
16647         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
16648         work that is not necessary.
16650 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
16652         PR tree-optimization/56854
16653         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
16654         forward into clobber stmts if it would change MEM_REF lhs into
16655         non-MEM_REF.
16657 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
16659         * tree.c (type_hash_lookup, type_hash_add): Make static.
16660         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
16662 2013-04-09  Richard Biener  <rguenther@suse.de>
16664         * tree.h (unsave_expr_now): Remove.
16665         * tree-inline.c (mark_local_for_remap_r): Remove.
16666         (unsave_expr_1): Likewise.
16667         (unsave_r): Likewise.
16668         (unsave_expr_now): Likewise.
16669         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
16670         (propagate_tree_value): Likewise.
16672 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
16674         * doc/rtl.texi (sequence): Rewrite documentation to match the
16675         current use of SEQUENCE rtl objects.
16676         * rtl.def (SEQUENCE): Likewise.
16678         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
16679         Update documentation.
16680         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
16681         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
16683         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
16685 2013-04-08  Teresa Johnson  <tejohnson@google.com>
16687         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
16688         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
16689         methods.
16690         (estimate_edge_size_and_time): Add comment to suggest using rounding
16691         methods.
16692         (estimate_node_size_and_time): Ditto.
16693         (remap_edge_change_prob): Use helper rounding divide methods.
16694         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
16695         (gimple_mod_pow2_value_transform): Ditto.
16696         (gimple_mod_subtract_transform): Ditto.
16697         (gimple_ic_transform): Ditto.
16698         (gimple_stringops_transform): Ditto.
16699         * stmt.c (conditional_probability): Ditto.
16700         (emit_case_dispatch_table): Ditto.
16701         * lto-cgraph.c (merge_profile_summaries): Ditto.
16702         * tree-optimize.c (execute_fixup_cfg): Ditto.
16703         * cfgcleanup.c (try_forward_edges): Ditto.
16704         * cfgloopmanip.c (scale_loop_profile): Ditto.
16705         (loopify): Ditto.
16706         (duplicate_loop_to_header_edge): Ditto.
16707         (lv_adjust_loop_entry_edge): Ditto.
16708         * tree-vect-loop.c (vect_transform_loop): Ditto.
16709         * profile.c (compute_branch_probabilities): Ditto.
16710         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
16711         * lto-streamer-in.c (input_cfg): Ditto.
16712         * gimple-streamer-in.c (input_bb): Ditto.
16713         * ipa-cp.c (update_profiling_info): Ditto.
16714         (update_specialized_profile): Ditto.
16715         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
16716         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
16717         rounding methods.
16718         * sched-rgn.c (compute_dom_prob_ps): Ditto.
16719         (compute_trg_info): Ditto.
16720         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
16721         (purge_dead_edges): Ditto.
16722         * loop-unswitch.c (unswitch_loop): Ditto.
16723         * cgraphclones.c (cgraph_clone_edge): Ditto.
16724         (cgraph_clone_node): Ditto.
16725         * tree-inline.c (copy_bb): Ditto.
16726         (copy_edges_for_bb): Ditto.
16727         (initialize_cfun): Ditto.
16728         (copy_cfg_body): Ditto.
16729         (expand_call_inline): Ditto.
16731 2013-04-08  Kai Tietz  <ktietz@redhat.com>
16733         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
16734         TARGET_CYGWIN64 by TARGET_64BIT.
16736 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16738         * config/epiphany/epiphany.md (GPR_1): New constant.
16739         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
16740         * config/epiphany/epiphany.c (gen_compare_reg):
16741         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
16742         is already in place.
16743         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
16744         Don't require being called during rtl expansion; If y operlaps r0,
16745         return 0.
16746         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
16747         (epiphany_expand_epilogue): Likewise.
16749         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
16750         Don't use CC_FPmode for ORDERED / UNORDERED.
16751         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
16753         * config/epiphany/constraints.md (CnL): New constraint.
16754         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
16755         * config/epiphany/predicates.md (add_operand): Allow 1024.
16757         * config/epiphany/epiphany.md (logical_op): New code iterator.
16758         (op_mnc): New code attribute.
16759         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
16760         (mov_f+1, mov_f+2): New peephole2 patterns.
16762         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
16763         (cstoresi4): Also allow re-use of zero result when doing a NE
16764         comparison to a non-zero operand.
16765         Use (clobber (scratch)) for first insn if the gpr output is not needed.
16767         * config/epiphany/epiphany.md (<insn_opname>v2si3):
16768         Use gen_addsi3_i / gen_subsi3_i.
16770 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16772         PR c++/34949
16773         PR c++/50243
16774         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
16775         contain anything but clobbers, at most one __builtin_stack_restore,
16776         optionally debug stmts and final resx, and if it has at least one
16777         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
16778         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
16779         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
16780         which isn't defaut definition, remove them.
16781         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
16782         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
16783         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
16784         with MEM_REF LHS with SSA_NAME address.
16786 2013-04-08  Jeff Law  <law@redhat.com>
16788         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
16790 2013-04-08  Richard Biener  <rguenther@suse.de>
16792         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
16793         extra newline.
16794         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
16795         determined vector type.
16796         (vect_analyze_data_refs): Likewise.
16797         (vect_get_new_vect_var): Adjust.
16798         (vect_create_destination_var): Preserve SSA name versions.
16799         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
16800         not dump anything here.
16802 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16804         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
16805         Add member lr_slot_known.
16806         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
16807         if necessary.
16808         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
16809         Remove code that sets lr_slot_offset according to what a previous
16810         version of epiphany_emit_save_restore used to do.
16811         (epiphany_emit_save_restore): When doing an lr save or restore,
16812         set/verify lr_slot_known and lr_slot_offset.
16814 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
16816         PR target/54338
16817         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
16818         in ALL_REGS.
16820 2013-04-08  Richard Biener  <rguenther@suse.de>
16822         * alias.c (find_base_term): Fix thinko in previous change.
16824 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16826         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
16827         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
16828         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
16829         if possible to compute val.
16830         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
16831         For QImode integers don't require anything about precision.  Use
16832         const_with_all_bytes_same to find out if the constant doesn't have
16833         repeated bytes in it.
16835 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16837         * config/s390/s390.c (s390_expand_insv): Only accept insertions
16838         within mode size.
16840 2013-04-08  Marek Polacek  <polacek@redhat.com>
16842         PR rtl-optimization/48182
16843         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
16844         value to 1.
16846 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16848         PR target/55487
16849         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
16850         nuses, make sure we have a label.
16852 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16854         PR target/56843
16855         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
16856         (rs6000_emit_swdiv_low_precision): Remove.
16857         (rs6000_emit_swdiv): Rewrite to handle between one and four
16858         iterations of Newton-Raphson generally; modify required number of
16859         iterations for some cases.
16860         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
16862 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
16864         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
16865         set-but-unused variable.
16867         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
16868         basic blocks of released function bodies garbage-collectable.
16870         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
16871         (struct rtl_opt_pass): Add TODO_df_finish.
16873         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
16875 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16877         * config/arm/constraints.md (q): New constraint.
16878         * config/arm/ldrdstrd.md: New file.
16879         * config/arm/arm.md (ldrdstrd.md) New include.
16880         (arm_movdi): Use "q" instead of "r" constraint
16881         for double-word memory access.
16882         (movdf_soft_insn): Likewise.
16883         * config/arm/vfp.md (movdi_vfp): Likewise.
16884         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
16885         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
16886         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
16887         (mem_ok_for_ldrd_strd): Likewise.
16888         (output_move_double): Update assertion.
16890 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16892         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
16894 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16896         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
16897         define_insn_and_split.
16898         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
16900 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16902         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
16903         define_insn_and_split.
16904         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
16905         (shiftsi3_compare): New pattern.
16906         (rrx): New pattern.
16907         * config/arm/unspecs.md (UNSPEC_RRX): New.
16909 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16911         * config/arm/arm.md (negdi_extendsidi): New pattern.
16912         (negdi_zero_extendsidi): Likewise.
16914 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16916         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
16917         define_insn_and_split.
16918         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
16919         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
16921 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16923         * config/arm/arm.md (arm_subdi3): Convert define_insn into
16924         define_insn_and_split.
16925         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
16926         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
16928 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16930         * config/arm/arm.md (subsi3_carryin): New pattern.
16931         (subsi3_carryin_const): Likewise.
16932         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
16933         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
16935 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16937         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
16939 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16941         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
16942         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
16944 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16946         * config/arm/arm.c (arm_expand_builtin): Change fcode
16947         type to unsigned int.
16949 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16951         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
16953 2013-04-04  Ian Lance Taylor  <iant@google.com>
16955         * doc/standards.texi (Standards): The Go frontend supports the Go 1
16956         language standard.
16958 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
16960         PR middle-end/56729
16961         * df-scan.c (df_insn_delete): Disable failing assert.
16963 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16965         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
16966         New function prototype.
16967         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
16968         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
16969         (arm_builtin_vectorized_function): New function.
16971 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16973         * config/arm/arm_neon_builtins.def: New file.
16974         * config/arm/arm.c (neon_builtin_data): Move contents to
16975         arm_neon_builtins.def.
16976         (enum arm_builtins): Include neon builtin definitions.
16977         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
16978         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
16980 2013-04-04  Marek Polacek  <polacek@redhat.com>
16982         PR tree-optimization/48186
16983         * predict.c (maybe_hot_frequency_p): Return false if
16984         HOT_BB_FREQUENCY_FRACTION is 0.
16985         (cgraph_maybe_hot_edge_p): Likewise.
16987 2013-04-04  Richard Biener  <rguenther@suse.de>
16989         PR tree-optimization/56826
16990         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
16991         more accurately.
16993 2013-04-04  Richard Biener  <rguenther@suse.de>
16995         PR tree-optimization/56213
16996         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
16997         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
16999 2013-04-04  Richard Biener  <rguenther@suse.de>
17001         PR tree-optimization/56837
17002         * tree-loop-distribution.c (classify_partition): For non-zero
17003         values require that the value has the same precision as its
17004         mode to be useful as memset value.
17006 2013-04-03  Nick Clifton  <nickc@redhat.com>
17008         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
17009         (fmssf4): Use fmsf.s on E3V5 architectures.
17010         (fnmasf4): Use fnmaf.s on E3V5 architectures.
17011         (fnmssf4): Use fnmsf.s on E3V5 architectures.
17013 2013-04-03  Jeff Law  <law@redhat.com>
17015         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
17016         (lra-eliminations.o): Likewise.
17018 2013-04-03  Teresa Johnson  <tejohnson@google.com>
17020         * gcov-io.c (compute_working_sets): Moved most of body of old
17021         compute_working_sets here from profile.c.
17022         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
17023         (gcov_working_set_t): Moved typedef here from basic-block.h
17024         (compute_working_set): Declare.
17025         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
17026         (get_working_sets): Renamed from compute_working_set,
17027         replace most of body with call to new compute_working_sets.
17028         (get_exec_counts): Replace call to compute_working_sets
17029         to get_working_sets.
17030         * profile.h (get_working_sets): Renamed from compute_working_set.
17031         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
17032         to get_working_sets.
17033         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
17034         * gcov-dump.c (dump_working_sets): New function.
17036 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
17038         * hwint.c (sext_hwi, zext_hwi): New functions.
17039         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
17040         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
17041         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
17042         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
17043         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
17044         (sext_hwi, zext_hwi): New functions.
17046 2013-04-03  Jeff Law  <law@redhat.com>
17048         PR tree-optimization/56799
17049         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
17050         back test for widening conversion erroneously dropped in prior change.
17052 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17054         PR target/56809
17055         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
17056         instead of next_real_insn.
17058 2013-04-03  Marek Polacek  <polacek@redhat.com>
17060         PR sanitizer/55702
17061         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
17063 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17065         PR target/56809
17066         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
17067         next_real_insn.
17068         (thumb1_output_casesi): Likewise.
17069         (thumb2_output_casesi): Likewise.
17071 2013-04-03  Richard Biener  <rguenther@suse.de>
17073         PR tree-optimization/56817
17074         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
17075         Split out ...
17076         (tree_unroll_loops_completely_1): ... new function to manually
17077         walk the loop tree, properly defering outer loops of unrolled
17078         loops to later iterations.
17080 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17082         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
17083         (vectorizable_load): Likewise.
17084         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
17085         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
17087 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17089         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
17090         BIT_FIELD_REF.
17092 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
17094         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
17096 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
17098         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
17100 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17102         PR tree-optimization/56790
17103         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
17104         folding.
17106 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
17108         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
17109         Handle VEC_MERGE.
17110         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
17111         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
17112         equal arguments.
17114 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
17116         PR c/19449
17117         * tree.h (force_folding_builtin_constant_p): New decl.
17118         * builtins.c (force_folding_builtin_constant_p): New variable.
17119         (fold_builtin_constant_p): Fold immediately also if
17120         force_folding_builtin_constant_p.
17122 2013-04-03  Richard Biener  <rguenther@suse.de>
17124         PR tree-optimization/56812
17125         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
17126         DRs of the same interleaving chain are independent.
17128 2013-04-02  Jason Merrill  <jason@redhat.com>
17130         * gdbinit.in (pbb): Use debug fn.
17132 2013-04-02  Lawrence Crowl  <crowl@google.com>
17134         * sese.h (struct ivtype_map_elt_s): Remove unused.
17135         (extern debug_ivtype_map): Remove unused.
17136         (extern eq_ivtype_map_elts): Remove unused.
17137         * sese.c (debug_ivtype_map): Removed unused.
17138         (debug_ivtype_map_1): Removed unused.
17139         (debug_ivtype_elt): Remove unused.
17140         (eq_ivtype_map_elts): Remove unused.
17143 2013-04-02  Kai Tietz  <ktietz@redhat.com>
17145         PR target/52790
17146         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
17147         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
17148         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
17149         function.
17150         (legitimize_pe_coff_symbol): Likewise.
17151         (is_imported_p): New helper-function.
17152         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
17153         for Windows x64 targets.
17154         (ix86_expand_prologue): Optimize for pe-coff targets.
17155         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
17156         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
17157         medium/large code-model.
17158         (legitimize_pic_address): Likewise.
17159         (legitimize_tls_address): Likewise.
17160         (ix86_expand_call): Likewise.
17161         (x86_output_mi_thunk): Likewise.
17162         (get_dllimport_decl): Add new beimport argument.
17163         (construct_plt_address): Don't assert for x64 pe-coff targets.
17164         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
17165         targets.
17166         (SYMBOL_FLAG_STUBVAR): New macro.
17167         (SYMBOL_REF_STUBVAR_P): Likewise.
17168         * config/i386/winnt.c (stub_list): New structure.
17169         (stub_head): New local variable.
17170         (i386_pe_record_stub): New function.
17171         (i386_pe_file_end): Emit refptr-stubs.
17173 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
17175         PR rtl-optimization/56745
17176         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
17177         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
17179         PR c++/34949
17180         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
17181         and both of them are MEM_REFs, just compare first argument for
17182         equality and attempt to deal even with differing offsets.
17184         PR c++/34949
17185         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
17186         of gimple_clobber_p to be MEM_REF.
17187         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
17188         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
17189         after gimplification.
17190         * asan.c (get_mem_ref_of_assignment): Don't instrument
17191         gimple_clobber_p stmts.
17192         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
17193         gimple_clobber_p stmt if they have MEM_REF lhs and
17194         are dead because of another gimple_clobber_p stmt.
17195         * tree-ssa-live.c (clear_unused_block_pointer): Treat
17196         gimple_clobber_p stmts like debug stmts.
17197         (remove_unused_locals): Remove clobbers with MEM_REF lhs
17198         that refer to unused VAR_DECLs or uninitialized values.
17199         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
17200         gimple_clobber_p stmts if they refer to removed parameters.
17201         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
17202         formatting.
17204 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
17206         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
17207         using SWI48 mode attribute.
17209 2013-04-02  Wei Mi  <wmi@google.com>
17211         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
17212         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
17213         *<rotate_insn><mode>3_mask in i386.md.
17215 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17217         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
17219 2013-04-02  Richard Biener  <rguenther@suse.de>
17221         PR tree-optimization/56778
17222         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
17223         Runtime alias tests are not supported for gather loads.
17224         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
17225         stmts referenced from SSA operands before updating SSA form.
17227 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
17228             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17230         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
17231         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
17232         * config/arm/cortex-a53.md: New file.
17233         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
17234         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
17235         * config/arm/arm.c (arm_issue_rate): Likewise.
17236         * config/arm/arm-tune.md: Regenerate
17237         * config/arm/arm-tables.opt: Regenerate.
17238         * config/arm/arm-cores.def: Add cortex-a53.
17240 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
17242         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
17243         non-static link.
17245 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
17247         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
17248         scalar load/store operations using B/H registers.
17249         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
17251 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
17253         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
17254         scalar move.
17255         * config/aarch64/aarch64.c
17256         (aarch64_simd_scalar_immediate_valid_for_move): New.
17257         * config/aarch64/aarch64-protos.h
17258         (aarch64_simd_scalar_immediate_valid_for_move): New.
17259         * config/aarch64/constraints.md (Dh, Dq): New.
17260         * config/aarch64/iterators.md (hq): New.
17262 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
17264         * reorg.c (get_branch_condition): Deal with conditional returns.
17265         (fill_simple_delay_slots): Remove dead code dealing with jumps.
17267 2013-04-01  Wei Mi  <wmi@google.com>
17269         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
17270         Truncate operand 2 using %b asm operand modifier.
17271         (*<shift_insn><mode>3_mask): Ditto.
17272         (*<rotate_insn><mode>3_mask): Ditto.
17274 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
17276         PR middle-end/56798
17277         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
17279 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
17281         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
17282         of next_real_insn.
17283         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
17285 2013-03-30  Lawrence Crowl  <crowl@google.com>
17287         * dse.c (clear_alias_sets): Remove never set.
17288         (disqualified_clear_alias_sets): Remove never set.
17289         (clear_alias_mode_pool): Remove never set.
17290         (dse_step0): Remove condition that is never true.
17291         (canon_address): Remove condition that is never true.
17292         (dse_step7): Remove condition that is never true.
17293         (rest_of_handle_dse): Remove condition that is never true.
17294         (rest_of_handle_dse::did_global): Remove never read from above.
17295         (dse_step2_spill): Remove never called from above.
17296         (dse_step5_spill): Remove never called from above.
17298 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
17300         * doc/md.texi (Standard Names) <casesi>: Update documentation for
17301         JUMP_TABLE_DATA changes.
17302         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
17303         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
17304         (Insns) <jump_table_data>: New entry.
17305         * doc/tm.texi: Regenerate.
17307         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
17309         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
17310         for table jump at the end of a basic block using tablejump_p.
17311         * targhooks.c (default_invalid_within_doloop): Likewise.
17312         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
17313         target hook implementation that is identical to the default hook.
17314         (rs6000_invalid_within_doloop): Remove.
17316         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
17317         unused variable from tablejump_p call.
17319         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
17320         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
17321         (INSN_DELETED_P): Likewise.
17322         (emit_jump_table_data): New prototype.
17323         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
17324         after 4th as unused.
17325         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
17326         * sched-vis.c (print_insn): Likewise.
17327         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
17328         insn for compatibility with back ends that use next_active_insn to
17329         identify jump table data.
17330         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
17331         (remove_insn): Likewise.
17332         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
17333         to be emitted.
17334         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
17335         (emit_jump_table_data): New function.
17337         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
17338         basic block, a JUMP_TABLE_DATA never is.
17339         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
17340         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
17341         off from code handling real insns.
17342         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
17343         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
17344         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
17345         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
17346         is not a NONDEBUG_INSN_P.
17347         * ira-costs.c (scan_one_insn): Likewise.
17348         * jump.c (mark_all_labels): Likewise.
17349         (mark_jump_label_1): Likewise.
17350         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
17351         * lra.c (get_insn_freq): Expect all insns reaching here to be in
17352         a basic block.
17353         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
17354         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
17355         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
17356         JUMP_TABLE_DATA_P insns.
17357         (calculate_elim_costs_all_insns): Likewise.
17358         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
17359         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
17360         (delete_output_reload): Code style fixups.
17361         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
17362         insn flags on this non-insn.
17363         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
17364         as scheduling barriers, for pre-change compatibility.
17365         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
17366         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
17368         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
17369         redundant JUMP_TABLE_DATA_P test.
17370         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
17371         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
17372         (frv_for_each_packet): Likewise.
17373         * config/i386/i386.c (min_insn_size): Likewise.
17374         (ix86_avoid_jump_mispredicts): Likewise.
17375         * config/m32r/m32r.c (m32r_is_insn): Likewise.
17376         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
17377         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
17378         (mips16_insn_length): Robustify.
17379         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
17380         (mips16_split_long_branches): Likewise.
17381         * config/pa/pa.c (pa_combine_instructions): Likewise.
17382         * config/rs6000/rs6000.c (get_next_active_insn): Treat
17383         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
17384         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
17385         as contributing to pool range lengths.
17386         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
17387         Remove redundant JUMP_TABLE_DATA_P test.
17388         (sh_loop_align): Likewise.
17389         (split_branches): Likewise.
17390         (sh_insn_length_adjustment): Likewise.
17391         * config/spu/spu.c (get_branch_target): Likewise.
17393 2013-03-29  Jan Hubicka  <jh@suse.cz>
17395         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
17396         gcov streaming; stream hot bb threshold to ltrans.
17397         * predict.c (get_hot_bb_threshold): Break out from ....
17398         (maybe_hot_count_p): ... here.
17399         (set_hot_bb_threshold): New function.
17400         * lto-section-in.c (lto_section_name): Add profile.
17401         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
17402         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
17403         and data-streamer.h
17404         (histogram_entry): New structure.
17405         (histogram, histogram_pool): New global vars.
17406         (histogram_hash): New structure.
17407         (histogram_hash::hash): New method.
17408         (histogram_hash::equal): Likewise.
17409         (account_time_size): New function.
17410         (cmp_counts): New function.
17411         (dump_histogram): New function.
17412         (ipa_profile_generate_summary): New function.
17413         (ipa_profile_write_summary): New function.
17414         (ipa_profile_read_summary): New function.
17415         (ipa_profile): Decide on threshold.
17416         (pass_ipa_profile): Add ipa_profile_write_summary and
17417         ipa_profile_read_summary.
17418         * Makefile.in (ipa.o): Update dependencies.
17419         * lto-streamer.h (LTO_section_ipa_profile): New section.
17421 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
17423         * tree.h (VAR_P): New.
17425 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
17427         PR lto/56777
17428         * doc/invoke.texi ([-fwhole-program]): Fix typo.
17430 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
17432         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
17433         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
17434         (control_flow_insn_p): Likewise.
17435         * cfgrtl.c (duplicate_insn_chain): Likewise.
17436         * final.c (get_attr_length_1): Likewise.
17437         (shorten_branches): Likewise.
17438         (final_scan_insn): Likewise.
17439         * function.c (instantiate_virtual_regs): Likewise.
17440         * gcse.c (insert_insn_end_basic_block): Likewise.
17441         * ira-costs.c (scan_one_insn): Likewise.
17442         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
17443         * lra.c (check_rtl): Likewise.
17444         * reload1.c (elimination_costs_in_insn): Likewise.
17445         * reorg.c (follow_jumps): Likewise.
17447         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
17448         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
17449         (thumb_far_jump_used_p): Likewise.
17450         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
17451         (workaround_speculation): Likewise.
17452         (add_sched_insns_for_speculation): Likewise.
17453         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
17454         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
17455         (frv_for_each_packet): Likewise.
17456         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
17457         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
17458         (final_emit_insn_group_barriers): Likewise.
17459         * config/m32r/m32r.c (m32r_is_insn): Likewise.
17460         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
17461         (mips16_insn_length): Likewise.
17462         * config/pa/pa.c (pa_reorg): Likewise.
17463         (pa_combine_instructions): Likewise.
17464         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
17465         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
17466         (sh_reorg): Likewise.
17467         (split_branches): Likewise.
17468         * config/spu/spu.c (get_branch_target): Likewise.
17470         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
17471         JUMP_TABLE_DATA_P.
17473 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
17475         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
17476         Fix declaration name.
17478 2013-03-28  Lawrence Crowl  <crowl@google.com>
17480         * graphds.h (struct graph.indicies): Remove unused.
17481         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
17482         (SCOP_ORIGINAL_PDDRS): Remove unused.
17483         * sese.h (extern insert_loop_close_phis): Removed unused.
17484         (extern insert_guard_phis): Removed unused.
17485         (extern ivtype_map_elt_info): Removed unused.
17486         (new_ivtype_map_elt): Removed unused.
17487         * sese.c (ivtype_map_elt_info): Removed unused.
17489 2013-03-28  Lawrence Crowl  <crowl@google.com>
17491         * Makefile.in: Add several missing include dependences.
17492         (DUMPFILE_H): New.
17493         (test-dump.o): New.  This object is not added to any executable,
17494         but is present for ad-hoc testing.
17495         * bitmap.c
17496         (debug (const bitmap_head_def &)): New.
17497         (debug (const bitmap_head_def *)): New.
17498         * bitmap.h
17499         (extern debug (const bitmap_head_def &)): New.
17500         (extern debug (const bitmap_head_def *)): New.
17501         * cfg.c
17502         (debug (edge_def &)): New.
17503         (debug (edge_def *)): New.
17504         * cfghooks.c
17505         (debug (basic_block_def &)): New.
17506         (debug (basic_block_def *)): New.
17507         * dumpfile.h
17508         (dump_node (const_tree, int, FILE *)): Correct source file.
17509         * dwarf2out.c
17510         (debug (die_struct &)): New.
17511         (debug (die_struct *)): New.
17512         * dwarf2out.h
17513         (extern debug (die_struct &)): New.
17514         (extern debug (die_struct *)): New.
17515         * gimple-pretty-print.c
17516         (debug (gimple_statement_d &)): New.
17517         (debug (gimple_statement_d *)): New.
17518         * gimple-pretty-print.h
17519         (extern debug (gimple_statement_d &)): New.
17520         (extern debug (gimple_statement_d *)): New.
17521         * ira-build.c
17522         (debug (ira_allocno_copy &)): New.
17523         (debug (ira_allocno_copy *)): New.
17524         (debug (ira_allocno &)): New.
17525         (debug (ira_allocno *)): New.
17526         * ira-int.h
17527         (extern debug (ira_allocno_copy &)): New.
17528         (extern debug (ira_allocno_copy *)): New.
17529         (extern debug (ira_allocno &)): New.
17530         (extern debug (ira_allocno *)): New.
17531         * ira-lives.c
17532         (debug (live_range &)): New.
17533         (debug (live_range *)): New.
17534         * lra-int.h
17535         (debug (lra_live_range &)): New.
17536         (debug (lra_live_range *)): New.
17537         * lra-lives.c
17538         (debug (lra_live_range &)): New.
17539         (debug (lra_live_range *)): New.
17540         * omega.c
17541         (debug (omega_pb_d &)): New.
17542         (debug (omega_pb_d *)): New.
17543         * omega.h
17544         (extern debug (omega_pb_d &)): New.
17545         (extern debug (omega_pb_d *)): New.
17546         * print-rtl.c
17547         (debug (const rtx_def &)): New.
17548         (debug (const rtx_def *)): New.
17549         * print-tree.c
17550         (debug_tree (tree): Move within file.
17551         (debug_raw (const tree_node &)): New.
17552         (debug_raw (const tree_node *)): New.
17553         (dump_tree_via_hooks (const tree_node *, int)): New.
17554         (debug (const tree_node &)): New.
17555         (debug (const tree_node *)): New.
17556         (debug_verbose (const tree_node &)): New.
17557         (debug_verbose (const tree_node *)): New.
17558         (debug_head (const tree_node &)): New.
17559         (debug_head (const tree_node *)): New.
17560         (debug_body (const tree_node &)): New.
17561         (debug_body (const tree_node *)): New.
17562         (debug_vec_tree (tree): Move and reimplement in terms of dump.
17563         (debug (vec<tree, va_gc> &)): New.
17564         (debug (vec<tree, va_gc> *)): New.
17565         * rtl.h
17566         (extern debug (const rtx_def &)): New.
17567         (extern debug (const rtx_def *)): New.
17568         * sbitmap.c
17569         (debug_raw (simple_bitmap_def &)): New.
17570         (debug_raw (simple_bitmap_def *)): New.
17571         (debug (simple_bitmap_def &)): New.
17572         (debug (simple_bitmap_def *)): New.
17573         * sbitmap.h
17574         (extern debug (simple_bitmap_def &)): New.
17575         (extern debug (simple_bitmap_def *)): New.
17576         (extern debug_raw (simple_bitmap_def &)): New.
17577         (extern debug_raw (simple_bitmap_def *)): New.
17578         * sel-sched-dump.c
17579         (debug (vinsn_def &)): New.
17580         (debug (vinsn_def *)): New.
17581         (debug_verbose (vinsn_def &)): New.
17582         (debug_verbose (vinsn_def *)): New.
17583         (debug (expr_def &)): New.
17584         (debug (expr_def *)): New.
17585         (debug_verbose (expr_def &)): New.
17586         (debug_verbose (expr_def *)): New.
17587         (debug (vec<rtx> &)): New.
17588         (debug (vec<rtx> *)): New.
17589         * sel-sched-dump.h
17590         (extern debug (vinsn_def &)): New.
17591         (extern debug (vinsn_def *)): New.
17592         (extern debug_verbose (vinsn_def &)): New.
17593         (extern debug_verbose (vinsn_def *)): New.
17594         (extern debug (expr_def &)): New.
17595         (extern debug (expr_def *)): New.
17596         (extern debug_verbose (expr_def &)): New.
17597         (extern debug_verbose (expr_def *)): New.
17598         (extern debug (vec<rtx> &)): New.
17599         (extern debug (vec<rtx> *)): New.
17600         * sel-sched-ir.h
17601         (_list_iter_cond_expr): Make inline instead of static.
17602         * sreal.c
17603         (debug (sreal &)): New.
17604         (debug (sreal *)): New.
17605         * sreal.h
17606         (extern debug (sreal &)): New.
17607         (extern debug (sreal *)): New.
17608         * tree.h
17609         (extern debug_raw (const tree_node &)): New.
17610         (extern debug_raw (const tree_node *)): New.
17611         (extern debug (const tree_node &)): New.
17612         (extern debug (const tree_node *)): New.
17613         (extern debug_verbose (const tree_node &)): New.
17614         (extern debug_verbose (const tree_node *)): New.
17615         (extern debug_head (const tree_node &)): New.
17616         (extern debug_head (const tree_node *)): New.
17617         (extern debug_body (const tree_node &)): New.
17618         (extern debug_body (const tree_node *)): New.
17619         (extern debug (vec<tree, va_gc> &)): New.
17620         (extern debug (vec<tree, va_gc> *)): New.
17621         * tree-cfg.c
17622         (debug (struct loop &)): New.
17623         (debug (struct loop *)): New.
17624         (debug_verbose (struct loop &)): New.
17625         (debug_verbose (struct loop *)): New.
17626         * tree-dump.c: Add header dependence.
17627         * tree-flow.h
17628         (extern debug (struct loop &)): New.
17629         (extern debug (struct loop *)): New.
17630         (extern debug_verbose (struct loop &)): New.
17631         (extern debug_verbose (struct loop *)): New.
17632         * tree-data-ref.c
17633         (debug (data_reference &)): New.
17634         (debug (data_reference *)): New.
17635         (debug (vec<data_reference_p> &)): New.
17636         (debug (vec<data_reference_p> *)): New.
17637         (debug (vec<ddr_p> &)): New.
17638         (debug (vec<ddr_p> *)): New.
17639         * tree-data-ref.h
17640         (extern debug (data_reference &)): New.
17641         (extern debug (data_reference *)): New.
17642         (extern debug (vec<data_reference_p> &)): New.
17643         (extern debug (vec<data_reference_p> *)): New.
17644         (extern debug (vec<ddr_p> &)): New.
17645         (extern debug (vec<ddr_p> *)): New.
17646         * tree-ssa-alias.c
17647         (debug (pt_solution &)): New.
17648         (debug (pt_solution *)): New.
17649         * tree-ssa-alias.h
17650         (extern debug (pt_solution &)): New.
17651         (extern debug (pt_solution *)): New.
17652         * tree-ssa-alias.c
17653         (debug (_var_map &)): New.
17654         (debug (_var_map *)): New.
17655         (debug (tree_live_info_d &)): New.
17656         (debug (tree_live_info_d *)): New.
17657         * tree-ssa-alias.h
17658         (extern debug (_var_map &)): New.
17659         (extern debug (_var_map *)): New.
17660         (extern debug (tree_live_info_d &)): New.
17661         (extern debug (tree_live_info_d *)): New.
17663 2013-03-28  Jan Hubicka  <jh@suse.cz>
17665         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
17667 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
17669         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
17670         record only when desired or required.
17672 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
17674         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
17675         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
17677 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17679         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
17680         (*andsi3_compare0_uxtw): New pattern.
17681         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
17682         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
17684 2013-03-28  Jan Hubicka  <jh@suse.cz>
17686         * data-streamer-in.c (streamer_read_gcov_count): New function.
17687         * gimple-streamer-out.c: Include value-prof.h.
17688         (output_gimple_stmt): Output histogram.
17689         (output_bb): Use streamer_write_gcov_count.
17690         * value-prof.c: Include data-streamer.h
17691         (dump_histogram_value): Add HIST_TYPE_MAX.
17692         (stream_out_histogram_value): New function.
17693         (stream_in_histogram_value): New function.
17694         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
17695         (stream_out_histogram_value, stream_in_histogram_value): Declare.
17696         * data-streamer-out.c (streamer_write_gcov_count): New function.
17697         (streamer_write_gcov_count_stream): New function.
17698         * lto-cgraph.c (lto_output_edge): Update counter streaming.
17699         (lto_output_node): Likewise.
17700         (input_node, input_edge): Likewise.
17701         * lto-streamer-out.c (output_cfg): Update streaming.
17702         * lto-streamer-in.c (input_cfg): Likewise.
17703         * data-streamer.h (streamer_write_gcov_count,
17704         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
17705         * gimple-streamer-in.c: Include value-prof.h
17706         (input_gimple_stmt): Input histograms.
17707         (input_bb): Update profile streaming.
17709 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
17711         * genmodes.c (emit_max_int): New function.
17712         (emit_insn_modes_h): Added call to emit_max_function.
17713         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
17714         Added doc.
17715         * machmode.def: Fixed comment.
17717 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
17719         * combine.c (try_combine): Removed useless assert.
17720         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
17722 2013-03-28  Marek Polacek  <polacek@redhat.com>
17723             Richard Biener  <rguenther@suse.de>
17725         PR tree-optimization/56695
17726         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
17727         build signed result of a vector comparison.
17728         * tree-cfg.c (verify_gimple_comparison): Check that a result
17729         of a vector comparison has signed type.
17731 2013-03-28  Richard Biener  <rguenther@suse.de>
17733         PR tree-optimization/37021
17734         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
17735         do not restrict gaps between groups.
17736         * tree-vect-stmts.c (vectorizable_load): Properly account for
17737         a gap between groups.
17739 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
17741         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
17742         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
17743         is not enabled.
17745 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
17747         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
17748         * doc/extend.texi (Named Address Spaces): Ditto.
17749         (Variable Attributes): Ditto.
17751 2013-03-27  Kai Tietz  <ktietz@redhat.com>
17753         * config.build: Add support for cygwin x64 target.
17754         * config.gcc: Likewise.
17755         * config.host: Likewise.
17756         * configure.ac: Likewise
17757         * configure: Regenerated.
17759 2013-03-27  Kai Tietz  <ktietz@redhat.com>
17761         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
17762         * config/i386/t-cygwin-w64: New file.
17763         * config/i386/cygwin-w64.h: New file.
17764         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
17765         and add support for x64-cygwin target.
17766         (CPP_SPEC): Likewise.
17767         (CXX_WRAP_SPEC_LIST): Undefine before define.
17768         (LIBGCJ_SONAME): Use 15 as version.
17770 2013-03-27  Richard Biener  <rguenther@suse.de>
17772         PR tree-optimization/56716
17773         * tree-ssa-structalias.c (perform_var_substitution): Adjust
17774         dumping for ref nodes.
17776 2013-03-27  Martin Jambor  <mjambor@suse.cz>
17778         PR tree-optimization/55334
17779         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
17780         restricted pointers to arrays.
17782 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
17784         * Makefile.in (.SUFFIXES): Add .cc.
17785         (.c.o): Apply same recipe for implicit rule .cc.o.
17787 2013-03-27  Richard Biener  <rguenther@suse.de>
17789         PR tree-optimization/37021
17790         * tree-vect-data-refs.c (vect_check_strided_load): Allow
17791         REALPART/IMAGPART_EXPRs around the supported refs.
17792         * tree-ssa-structalias.c (find_func_aliases): Assume that
17793         floating-point values are not used to transfer pointers.
17795 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17797         * target.def (TARGET_HAS_IFUNC_P): New target hook.
17798         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
17799         * doc/tm.texi: Regenerate.
17800         * targhooks.h (default_has_ifunc_p): New.
17801         * targhooks.c (default_has_ifunc_p): Ditto.
17802         * config/linux-protos.h: New file.
17803         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
17804         hook for linux which disables support of indirect functions in android.
17805         * config/linux-android.c: New file.
17806         * config/t-linux-android.c: Ditto.
17807         * config.gcc: Added new object file linux-android.o.
17808         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
17809         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
17810         * varasm.c (do_assemble_alias): Likewise.
17811         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
17812         doesn't support indirect functions.
17813         * configure: Regenerate.
17815 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
17817         PR target/56102
17818         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
17819         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
17820         mult-word mode.
17822 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17824         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
17826 2013-03-27  Terry Guo  <terry.guo@arm.com>
17828         * config/arm/arm-cores.def: Added core cortex-r7.
17829         * config/arm/arm-tune.md: Regenerated.
17830         * config/arm/arm-tables.opt: Regenerated.
17831         * doc/invoke.texi: Added entry for core cortex-r7.
17833 2013-03-27  Walter Lee  <walt@tilera.com>
17835         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
17836         double-decrement of next_scratch_regno.
17838 2013-03-27  Walter Lee  <walt@tilera.com>
17840         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
17841         input operands.
17842         (insn_v1mulus): Ditto.
17843         (insn_v2muls): Ditto.
17845 2013-03-27  Walter Lee  <walt@tilera.com>
17847         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
17848         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
17850 2013-03-27  Walter Lee  <walt@tilera.com>
17852         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
17853         (*sibcall_value): Ditto.
17855 2013-03-27  Walter Lee  <walt@tilera.com>
17857         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
17858         (insn_mnz_v8qi): ... this ...
17859         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
17860         vector equivalent.
17861         (insn_v<n>mnz): Replaced by ...
17862         (insn_v1mnz): ... this ...
17863         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
17864         equivalent.
17865         (insn_mz_<mode>): Replaced by ...
17866         (insn_mz_v8qi): ... this ...
17867         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
17868         vector equivalent.
17869         (insn_v<n>mz): Replaced by ...
17870         (insn_v1mz): ... this ...
17871         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
17872         equivalent.
17874 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
17876         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
17878 2013-03-26  Roland McGrath  <mcgrathr@google.com>
17880         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
17881         than fprintf with a non-constant, non-format string.
17883 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17885         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
17886         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
17887         operand 0 predicate.
17888         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
17889         attribute.  Use general_x64nomem_operand as operand 1 predicate.
17890         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
17891         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17892         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
17893         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17894         (mov<mode>_insv_1): Remove expander.  Merge insn with
17895         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
17896         Use general_x64nomem_operand as operand 1 predicate.
17897         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
17898         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
17899         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17900         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
17901         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
17902         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
17903         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17904         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
17905         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
17906         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
17907         (general_x64nomem_operand): Ditto.
17909 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17911         * config/rtems.opt: Add -pthread option.
17913 2013-03-26  Richard Biener  <rguenther@suse.de>
17915         * alias.c (find_base_term): Avoid redundant and not used recursion.
17916         (base_alias_check): Get the initial base term from the caller.
17917         (true_dependence_1): Compute and pass base terms to base_alias_check.
17918         (write_dependence_p): Likewise.
17919         (may_alias_p): Likewise.
17921 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
17923         * config/aarch64/aarch64.c (aarch64_classify_address): Support
17924         PC-relative load in SI modes and above only.
17926 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
17928         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
17929         * config/arm/iwmmxt.md (WCGR0): Update.
17930         (WCGR1, WCGR2, WCGR3): Likewise.
17932 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17934         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
17935         Use x64 and nox64 isa attributes.
17937 2013-03-26  Richard Biener  <rguenther@suse.de>
17939         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
17940         alignment computations and rely on get_object_alignment_1
17941         for the !TYPE_P case.
17942         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
17944 2013-03-26  Walter Lee  <walt@tilera.com>
17946         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
17947         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
17949 2013-03-25  Jeff Law  <law@redhat.com>
17951         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
17952         check for INTEGRAL_TYPE_P that was missing due to checking in
17953         wrong version of prior patch.
17955 2013-03-25  Walter Lee  <walt@tilera.com>
17957         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
17958         TILEGX_INSN_SHUFFLEBYTES1.
17959         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
17960         shufflebytes1.
17961         (tilegx_builtins): Ditto.
17962         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
17964 2013-03-25  Walter Lee  <walt@tilera.com>
17966         * config/tilegx/tilegx.md (floatsisf2): New pattern.
17967         (floatunssisf2): New pattern.
17968         (floatsidf2): New pattern.
17969         (floatunssidf2): New pattern.
17971 2013-03-25  Walter Lee  <walt@tilera.com>
17973         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
17974         tests for constraint J, K, N, P.
17976 2013-03-25  Walter Lee  <walt@tilera.com>
17978         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
17979         Use indirect/pcrel encoding.
17980         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
17981         Ditto.
17983 2013-03-25  Steve Ellcey  <sellcey@mips.com>
17985         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
17986         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
17987         * config/mips/mips.c (mips_option_override): Set IMADD default.
17988         * config/mips/mips.h (PTF_AVOID_IMADD): New.
17989         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
17990         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
17991         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
17992         * doc/invoke.texi (-mimadd/-mno-imadd): New.
17994 2013-03-25  Jeff Law  <law@redhat.com>
17996         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
17997         slightly to avoid creating and folding useless trees.  Simplify
17998         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
18000 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
18002         * config/i386/i386.md (*zero_extendsidi2): Merge with
18003         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
18004         * config/i386/predicates.md (x86_64_zext_operand): Rename from
18005         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
18006         targets.  Clarify comment.
18008 2013-03-25  Martin Jambor  <mjambor@suse.cz>
18010         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
18011         pass-through jump functions differently.
18012         (ipa_read_jump_function): Likewise.  Also use setter functions to set
18013         up jump functions.
18015 2013-03-25  Martin Jambor  <mjambor@suse.cz>
18017         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
18018         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
18019         process it.
18020         (ipa_get_indirect_edge_target): New function.
18021         (devirtualization_time_bonus): New parameter known_aggs, pass it to
18022         ipa_get_indirect_edge_target.  Update all callers.
18023         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
18024         ipa_get_indirect_edge_target_1 instead of calling
18025         ipa_get_indirect_edge_target.
18026         (create_specialized_node): Pass aggvlas to
18027         ipcp_discover_new_direct_edges.
18029 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18031         * config/arm/arm.md (f_sels, f_seld): New types.
18032         (*cmov<mode>): New pattern.
18033         * config/arm/predicates.md (arm_vsel_comparison_operator): New
18034         predicate.
18036 2013-03-25  Kai Tietz  <ktietz@redhat.com>
18038         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
18039         POSIX-printf for mingw-hosted  builds.
18041 2013-03-25  Richard Biener  <rguenther@suse.de>
18043         PR middle-end/56694
18044         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
18045         must-not-throw stmt location.
18047 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18049         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
18050         Emit load-acquire versions when acq is true.
18051         (arm_emit_store_exclusive): Add rel parameter.
18052         Emit store-release versions when rel is true.
18053         (arm_split_compare_and_swap): Use acquire-release instructions
18054         instead.
18055         of barriers when appropriate.
18056         (arm_split_atomic_op): Likewise.
18057         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
18058         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
18059         (VUNSPEC_SLX): Likewise.
18060         (VUNSPEC_LDA): Likewise.
18061         (VUNSPEC_STL): Likewise.
18062         * config/arm/sync.md (atomic_load<mode>): New pattern.
18063         (atomic_store<mode>): Likewise.
18064         (arm_load_acquire_exclusive<mode>): Likewise.
18065         (arm_load_acquire_exclusivesi): Likewise.
18066         (arm_load_acquire_exclusivedi): Likewise.
18067         (arm_store_release_exclusive<mode>): Likewise.
18069 2013-03-25  Catherine Moore  <clm@codesourcery.com>
18071         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
18072         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
18073         * config/mip/predicates.md (lwsp_swsp_operand,
18074         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
18075         sb16_operand, db4_operand, db7_operand, ib3_operand,
18076         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
18077         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
18078         andi16_operand): New predicates.
18079         * config/mips/mips.md (compression): New attribute.
18080         (enabled): New attribute.
18081         (length): Consider compression in computing length.
18082         (shift_compression): New code attribute.
18083         (*add<mode>3): New operands. Record compression.
18084         (sub<mode>3): Likewise.
18085         (one_cmpl<mode>2): Likewise.
18086         (*and<mode>3): Likewise.
18087         (*ior<mode>3): Likewise.
18088         (unnamed pattern for xor): Likewise.
18089         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
18090         (*<optab><mode>3): Likewise.
18091         (*mov<mode>_internal: Likewise.
18092         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
18093         (mips_unsigned_immediate_p): New.
18094         (umips_lwsp_swsp_address_p): New.
18095         (m16_based_address_p): New.
18096         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
18097         (mips_unsigned_immediate_p): New prototype.
18098         (lwsp_swsp_address_p): New prototype.
18099         (m16_based_address_p): New prototype.
18100         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
18101         (mips_signed_immediate_p): New function.
18102         (m16_based_address_p): New function.
18103         (lwsp_swsp_address_p): New function.
18104         (mips_print_operand_punctuation): Recognize short delay slot insns
18105         for microMIPS.add<mode>3"
18107 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18109         PR target/56720
18110         * config/arm/iterators.md (v_cmp_result): New mode attribute.
18111         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
18113 2013-03-25  Richard Biener  <rguenther@suse.de>
18115         PR tree-optimization/56689
18116         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
18117         any edge.
18119 2013-03-25  Richard Biener  <rguenther@suse.de>
18121         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
18122         of bitmap.
18123         (memory_references): Likewise.
18124         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
18125         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
18126         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
18127         (gather_mem_refs_in_loops): Fold into ...
18128         (analyze_memory_references): ... this.  Move initialization
18129         to tree_ssa_lim_initialize.
18130         (fill_always_executed_in): Rename to ...
18131         (fill_always_executed_in_1): ... this.
18132         (fill_always_executed_in): Move contains_call computation to
18133         this new function from ...
18134         (tree_ssa_lim_initialize): ... here.
18135         (tree_ssa_lim): Call fill_always_executed_in.
18137 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
18139         * postreload.c (reload_combine): Fix code detecting returns.
18141 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
18143         * function.c (emit_use_return_register_into_block): On cc0 targets,
18144         do not emit the sequence between cc0 setter and user.
18146 2013-03-25  Kai Tietz  <ktietz@redhat.com>
18148         * config/i386/predicates.md (local_symbolic_operand): Interpret
18149         dll-imported symbols as none-local.
18151 2013-03-25  Richard Biener  <rguenther@suse.de>
18153         * tree-ssa-loop-im.c (struct depend): Remove.
18154         (struct lim_aux_data): Make depends a vec of gimples.
18155         (free_lim_aux_data): Adjust.
18156         (add_dependency): Likewise.
18157         (set_level): Likewise.
18159 2013-03-25  Richard Biener  <rguenther@suse.de>
18161         PR middle-end/56434
18162         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
18163         the pointer returned by calls with ECF_MALLOC set.
18165 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
18167         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
18169 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
18171         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
18172         using MMXMODE mode iterator.
18173         (*move<mode>_internal): Merge with *movv2sf_internal and
18174         *movv2sf_internal_rex64 using MMXMODE mode iterator.
18176 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
18178         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
18179         (record_last_mem_set_info): Likewise.
18181         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
18182         of XNEWVEC followed by memset.
18183         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
18185 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
18187         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
18188         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
18189         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
18190         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
18191         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
18192         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
18193         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
18194         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
18195         BARRIER_P instead of GET_CODE.
18197 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
18199         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
18200         inaccuracy in the probing code.
18202         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
18203         (ctrapdi4): Likewise.
18205 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
18207         * calls.c (expand_call): Add missing guard to code handling return
18208         of non-BLKmode structures in MSB.
18209         * function.c (expand_function_end): Likewise.
18211 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
18213         * combine.c (try_combine): Adjust comment.  Do not add the set of
18214         insn #0 if the destination indirectly is set or dies in insn #2.
18215         Tidy up code to distribute a new note.
18217 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
18219         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
18220         also for alternatives 16 and 17.
18222 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
18224         * config/i386/sse.md (*mov<mode>_internal): Merge with
18225         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
18226         Emit insn template depending on type attribute.  Use
18227         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
18228         movd instead of movq mnemonic for interunit moves.  Rewrite mode
18229         attribute calculation.  Remove unit attribute calculation.
18230         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
18231         Set prefix_data16 attribute for DImode ssemov types.
18232         Use Ym instead of y for SSE-MMX conversion alternatives.
18233         Reorder operand constraints.
18235 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
18237         * df.h (df_insn_delete): Adjust prototype.
18238         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
18239         and let it decide whether mark the basic block dirty.
18240         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
18241         * df-scan.c (df_insn_info_delete): New helper function, split
18242         off from df_insn_delete.
18243         (df_scan_free_bb_info): Use it.
18244         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
18245         Likewise.
18246         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
18247         that the insn is actually an insn and it has a non-NULL basic block.
18248         Do not mark basic block dirty if only deleting a DEBUG_INSN.
18250 2013-03-22  Richard Biener  <rguenther@suse.de>
18252         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
18253         dep_ref members.
18254         (mem_ref_alloc): Do not allocate them.
18255         (refs_independent_p): Do not query or maintain a cache.
18257 2013-03-22  Richard Biener  <rguenther@suse.de>
18259         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
18260         (gather_mem_refs_in_loops): Do not compute it.
18261         (analyze_memory_references): Do not allocate it.
18262         (tree_ssa_lim_finalize): Do not free it.
18263         (for_all_locs_in_loop): Do not query all_refs_in_loop.
18265 2013-03-22  Richard Biener  <rguenther@suse.de>
18267         * is-a.h (as_a): Use gcc_checking_assert.
18269 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
18271         * config/aarch64/aarch64.c (aarch64_print_operand): New
18272         format specifier for printing a constant in hex.
18273         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
18274         format specifier for printing second operand.
18276 2013-03-22  Richard Biener  <rguenther@suse.de>
18278         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
18279         bitmaps.
18280         (gather_mem_refs_in_loops): Perform store accumulation here.
18281         (create_vop_ref_mapping_loop): Remove.
18282         (create_vop_ref_mapping): Likewise.
18283         (analyze_memory_references): Initialize refs_stored_in_loop.
18284         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
18285         (record_indep_loop): Remove.
18286         (record_dep_loop): New function.
18287         (ref_indep_loop_p_1): Adjust to only walk over references
18288         in the loop, not its subloops.
18289         (ref_indep_loop_p): Rename to ...
18290         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
18291         maintaining a more fine-grained cache.
18292         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
18293         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
18295 2013-03-22  Richard Biener  <rguenther@suse.de>
18297         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
18298         (struct mem_ref): Make accesses_in_loop a vec of a vec of
18299         aggregate mem_ref_loc.
18300         (free_mem_ref_locs): Inline into ...
18301         (memref_free): ... this and adjust.
18302         (mem_ref_alloc): Adjust.
18303         (mem_ref_locs_alloc): Remove.
18304         (record_mem_ref_loc): Adjust.
18305         (get_all_locs_in_loop): Rewrite into ...
18306         (for_all_locs_in_loop): ... this iterator.
18307         (rewrite_mem_ref_loc): New functor.
18308         (rewrite_mem_refs): Use for_all_locs_in_loop.
18309         (sm_set_flag_if_changed): New functor.
18310         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
18311         (ref_always_accessed): New functor.
18312         (ref_always_accessed_p): Use for_all_locs_in_loop.
18314 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
18316         * tree-pass.h (PROP_gimple_lvec): New.
18317         * passes.c (dump_properties): Handle PROP_gimple_lvec.
18318         (init_optimization_passes): Move pass_lower_vector.
18319         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
18320         PROP_gimple_lvec.
18321         (pass_lower_vector): Provide PROP_gimple_lvec.
18322         (pass_lower_vector_ssa): Likewise.
18323         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
18325 2013-03-21  Mark Wielaard  <mjw@redhat.com>
18327         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
18329 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
18331         * config/i386/i386.md (*movdi_internal): Disparage slightly
18332         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
18333         conversion alternatives.
18335 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
18337         PR middle-end/48087
18338         * diagnostic.def (DK_WERROR): New kind.
18339         * diagnostic.h (werrorcount): Define.
18340         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
18341         promoted to DK_ERROR, increment DK_WERROR counter instead of
18342         DK_ERROR counter.
18343         * toplev.c (toplev_main): Call print_ignored_options even if
18344         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
18345         even if just werrorcount is non-zero.
18347         PR debug/55608
18348         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
18349         on failure.
18350         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
18351         (string_cst_pool_decl): New function.
18352         (optimize_one_addr_into_implicit_ptr): New function.
18353         (resolve_addr_in_expr): Optimize DWARF location expression
18354         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
18355         which doesn't live in memory, but has DW_AT_location or
18356         DW_AT_const_value, or refers to a string literal, into
18357         DW_OP_GNU_implicit_pointer.
18358         (optimize_location_into_implicit_ptr): New function.
18359         (resolve_addr): If removing DW_AT_location of a variable because
18360         it was DW_OP_addr of address of the variable, but the variable doesn't
18361         live in memory, try to emit const value attribute for the initializer.
18363 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
18365         * tree.h (VECTOR_TYPE_P): New macro.
18366         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
18367         TYPE_MODE): Use it.
18368         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
18369         VEC_COND_EXPR cannot be lvalues.
18370         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
18372 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
18374         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
18375         Restrict the transformation to equal modes.
18377 2013-03-21  Richard Biener  <rguenther@suse.de>
18379         PR tree-optimization/39326
18380         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
18381         (MEM_ANALYZABLE): Adjust.
18382         (record_mem_ref_loc): Move bitmap ops ...
18383         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
18384         unanalyzable refs, do not record locations for it.
18385         (analyze_memory_references): Allocate ref zero as shared
18386         unanalyzable ref.
18387         (refs_independent_p): Do not test for unanalyzed mems here.
18388         (ref_indep_loop_p_1): Special-case disambiguation against
18389         the unanalyzed ref.
18390         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
18392 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
18394         * config/arm/arm-protos.h (tune_params): Add
18395         prefer_neon_for_64bits field.
18396         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
18397         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
18398         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
18399         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
18400         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
18401         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
18402         (arm_option_override): Handle -mneon-for-64bits new option.
18403         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
18404         (prefer_neon_for_64bits): Declare new variable.
18405         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
18406         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
18407         (arch_enabled): Handle new arch types. Remove support for onlya8
18408         and nota8.
18409         (one_cmpldi2): Use new arch names.
18410         (zero_extend<mode>di2, extend<mode>di2): Ditto.
18411         * config/arm/arm.opt (mneon-for-64bits): Add option.
18412         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
18413         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
18414         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
18415         of onlya8.
18416         * doc/invoke.texi (-mneon-for-64bits): Document.
18418 2013-03-21  Richard Biener  <rguenther@suse.de>
18420         PR tree-optimization/39326
18421         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
18422         (sort_bbs_in_loop_postorder_cmp): New function.
18423         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
18425 2013-03-21  Richard Biener  <rguenther@suse.de>
18427         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
18428         (vect_insert_into_interleaving_chain): Likewise.
18429         (vect_drs_dependent_in_basic_block): Inline ...
18430         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
18431         split out from ...
18432         (vect_analyze_data_ref_dependence): ... here.  Simplify.
18433         (vect_check_interleaving): Simplify.
18434         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
18435         (vect_slp_analyze_data_ref_dependences): ... this new function.
18436         (dr_group_sort_cmp): New function.
18437         (vect_analyze_data_ref_accesses): Compute data-reference groups
18438         here instead of in vect_analyze_data_ref_dependence.  Use
18439         a more efficient algorithm.
18440         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
18441         vect_slp_analyze_data_ref_dependences.  Call
18442         vect_analyze_data_ref_accesses earlier.
18443         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
18444         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
18445         (vect_slp_analyze_data_ref_dependences): New prototype.
18447 2013-03-21  Richard Biener  <rguenther@suse.de>
18449         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
18450         ref is stored in the loop.
18451         (find_refs_for_sm): Walk only over all stores.
18452         (store_motion_loop): Allocate from lim_bitmap_obstack.
18453         (store_motion): Likewise.
18455 2013-03-21  Richard Biener  <rguenther@suse.de>
18457         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
18458         Update virtual SSA form.
18460 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18462         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
18463         * configure: Regenerate.
18464         * config.in: Regenerate.
18465         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
18466         if !HAVE_LD_EH_FRAME_CIEV3.
18468 2013-03-21  Richard Biener  <rguenther@suse.de>
18470         * tree-cfg.c (verify_expr_no_block): New function.
18471         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
18472         nor DECL_VALUE_EXPR have locations with associated blocks.
18473         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
18474         (clear_unused_block_pointer): Remove code dealing with
18475         blocks in DECL_DEBUG_EXPR locations.
18477 2013-03-21  Richard Biener  <rguenther@suse.de>
18479         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
18480         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
18481         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
18482         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
18483         instead of DECL_DEBUG_EXPR_IS_FROM.
18484         * gimplify.c (gimplify_modify_expr): Likewise.
18485         * tree-cfg.c (verify_expr_location_1): Likewise.
18486         * tree-complex.c (create_one_component_var): Likewise.
18487         * tree-sra.c (create_access_replacement): Likewise.
18488         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
18489         (clear_unused_block_pointer): Likewise.
18490         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
18491         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
18492         * var-tracking.c (var_debug_decl): Likewise.
18493         (track_expr_p): Likewise.
18494         * tree-inline.c (add_local_variables): Likewise.  Set
18495         DECL_HAS_DEBUG_EXPR_P after copying it.
18496         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
18497         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
18499 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
18501         PR bootstrap/56656
18502         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
18503         * configure: Regenerate.
18504         * config.in: Regenerate.
18505         * config/i386/i386.md (*movdf_internal): Use
18506         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
18507         movd instead of movq mnemonic for interunit moves.
18508         (*movdi_internal): Ditto.
18510 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18512         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
18513         (abd<mode>_3): New pattern.
18514         (aba<mode>_3): New pattern.
18515         (fabd<mode>_3): New pattern.
18517 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18519         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
18520         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
18521         occurrence of REGISTER_PREFIX as its empty string.
18523 2013-03-20  Jeff Law  <law@redhat.com>
18525         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
18526         addititional equivalences for equality comparisons between an SSA_NAME
18527         and a constant where the SSA_NAME was set from a widening conversion.
18529 2013-03-20  Walter Lee  <walt@tilera.com>
18531         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
18533 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
18535         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
18536         depending on type attribute.
18537         (*movti_internal): Ditto.
18538         (*movtf_internal): Ditto.
18539         (*movxf_internal): Ditto.
18540         (*movdf_internal): Ditto.
18541         (*movsf_internal): Ditto.
18543 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
18545         * config/i386/i386.md (*movti_internal): Set prefix attribute to
18546         maybe_vex for sselog1 and ssemov types.
18547         (*movdi_internal): Reorder operand constraints.
18548         (*movsi_internal): Ditto.  Set prefix attribute to
18549         maybe_vex for sselog1 and ssemov types.
18550         (*movtf_internal): Set prefix attribute to maybe_vex
18551         for sselog1 and ssemov types.
18552         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
18553         DImode ssemov types.  Reorder operand constraints.
18554         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
18555         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
18556         attribute for SImode ssemov types.  Reorder operand constraints.
18558 2013-03-20  Martin Jambor  <mjambor@suse.cz>
18560         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
18561         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
18563 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
18565         * config/rs6000/predicates.md (indexed_address, update_address_mem
18566         update_indexed_address_mem): New predicates.
18567         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
18568         attribute for load/store instructions.
18569         * config/rs6000/dfp.md (movsd_store): Likewise.
18570         (movsd_load): Likewise.
18571         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
18572         (unnamed HI->DI extend define_insn): Likewise.
18573         (unnamed SI->DI extend define_insn): Likewise.
18574         (unnamed QI->SI extend define_insn): Likewise.
18575         (unnamed QI->HI extend define_insn): Likewise.
18576         (unnamed HI->SI extend define_insn): Likewise.
18577         (unnamed HI->SI extend define_insn): Likewise.
18578         (extendsfdf2_fpr): Likewise.
18579         (movsi_internal1): Likewise.
18580         (movsi_internal1_single): Likewise.
18581         (movhi_internal): Likewise.
18582         (movqi_internal): Likewise.
18583         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
18584         attribute for load/store instructions.
18585         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
18586         instructions.
18587         (mov<mode>_softfloat): Likewise.
18588         (mov<mode>_hardfloat32): Likewise.
18589         (mov<mode>_hardfloat64): Likewise.
18590         (mov<mode>_softfloat64): Likewise.
18591         (movdi_internal32): Likewise.
18592         (movdi_internal64): Likewise.
18593         (probe_stack_<mode>): Likewise.
18595 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
18597         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
18598         floating point, and decimal floating point to reload iterator.
18600         * config/rs6000/constraints.md (wl constraint): New constraints to
18601         return FLOAT_REGS if certain options are used to reduce the number
18602         of separate patterns that exist in the file.
18603         (wx constraint): Likewise.
18604         (wz constraint): Likewise.
18606         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18607         -mdebug=reg, print wg, wl, wx, and wz constraints.
18608         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
18609         Initialize the reload functions for 64-bit binary/decimal floating
18610         point types.
18611         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
18612         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
18613         create the buffer on the stack to overcome not having a 32-bit
18614         load and store.
18615         (rs6000_emit_move): Likewise.
18616         (rs6000_secondary_memory_needed_rtx): Likewise.
18617         (rs6000_alloc_sdmode_stack_slot): Likewise.
18618         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
18619         via xxlxor, just like DFmode 0.0.
18621         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
18622         define as 1 if we are running on a power7 or newer.
18623         (enum r6000_reg_class_enum): Add new constraints.
18625         * config/rs6000/dfp.md (movsd): Delete, combine with binary
18626         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
18627         with other moves by using conditional constraits (wg).  Use LFIWZX
18628         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
18629         (movsd splitter): Likewise.
18630         (movsd_hardfloat): Likewise.
18631         (movsd_softfloat): Likewise.
18633         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
18634         binary and decimal floating point moves.
18635         (fmove_ok): New attributes to combine binary and decimal floating
18636         point moves, and to combine power6x (mfpgpr) moves along normal
18637         floating moves.
18638         (real_value_to_target): Likewise.
18639         (f32_lr): Likewise.
18640         (f32_lm): Likewise.
18641         (f32_li): Likewise.
18642         (f32_sr): Likewise.
18643         (f32_sm): Likewise.
18644         (f32_si): Likewise.
18645         (movsf): Combine binary and decimal floating point moves.  Combine
18646         power6x (mfpgpr) moves with other moves by using conditional
18647         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
18648         (mov<mode> for SFmode/SDmode); Likewise.
18649         (SFmode/SDmode splitters): Likewise.
18650         (movsf_hardfloat): Likewise.
18651         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
18652         (movsf_softfloat): Likewise.
18653         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
18655         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
18656         wx and wz constraints.
18658         * config/rs6000/constraints.md (wg constraint): New constraint to
18659         return FLOAT_REGS if -mmfpgpr (power6x) was used.
18661         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
18662         constraint.
18664         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18665         -mdebug=reg, print wg, wl, wx, and wz constraints.
18666         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
18667         Initialize the reload functions for 64-bit binary/decimal floating
18668         point types.
18669         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
18670         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
18671         create the buffer on the stack to overcome not having a 32-bit
18672         load and store.
18673         (rs6000_emit_move): Likewise.
18674         (rs6000_secondary_memory_needed_rtx): Likewise.
18675         (rs6000_alloc_sdmode_stack_slot): Likewise.
18676         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
18677         via xxlxor, just like DFmode 0.0.
18679         * config/rs6000/dfp.md (movdd): Delete, combine with binary
18680         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
18681         with other moves by using conditional constraits (wg).  Use LFIWZX
18682         and STFIWX for loading SDmode on power7.
18683         (movdd splitters): Likewise.
18684         (movdd_hardfloat32): Likewise.
18685         (movdd_softfloat32): Likewise.
18686         (movdd_hardfloat64_mfpgpr): Likewise.
18687         (movdd_hardfloat64): Likewise.
18688         (movdd_softfloat64): Likewise.
18690         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
18691         64-bit binary and decimal floating point moves.
18692         (FMOVE64X): Likewise.
18693         (movdf): Combine 64-bit binary and decimal floating point moves.
18694         Combine power6x (mfpgpr) moves with other moves by using
18695         conditional constraits (wg).
18696         (mov<mode> for DFmode/DDmode): Likewise.
18697         (DFmode/DDmode splitters): Likewise.
18698         (movdf_hardfloat32): Likewise.
18699         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
18700         (movdf_softfloat32): Likewise.
18701         (movdf_hardfloat64_mfpgpr): Likewise.
18702         (movdf_hardfloat64): Likewise.
18703         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
18704         (movdf_softfloat64): Likewise.
18705         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
18706         (reload_<mode>_load): Move to later in the file so they aren't in
18707         the middle of the floating point move insns.
18708         (reload_<mode>_store): Likewise.
18710         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
18711         constraint.
18713         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
18714         constraint if -mdebug=reg.
18715         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
18716         Enable using dd reload support if needed.
18718         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
18719         binary and decimal floating point moves in rs6000.md.
18720         (movtd_internal): Likewise.
18722         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
18723         decimal floating point moves.
18724         (movtf): Likewise.
18725         (movtf_internal): Likewise.
18726         (mov<mode>_internal, TDmode/TFmode): Likewise.
18727         (movtf_softfloat): Likewise.
18728         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
18730         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
18731         movdi_internal64, using wg constraint for move direct operations.
18732         (movdi_internal64): Likewise.
18734         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
18735         MODES_TIEABLE_P for selected modes.  Print the numerical value of
18736         the various virtual registers. Use GPR/FPR first/last values,
18737         instead of hard coding the register numbers.  Print which modes
18738         have reload functions registered.
18739         (rs6000_option_override_internal): If -mdebug=reg, trace the options
18740         settings before/after setting cpu, target and subtarget settings.
18741         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
18742         and for secondary reload failures in rs6000_secondary_reload_inner.
18743         (rs6000_secondary_reload_fail): Likewise.
18744         (rs6000_secondary_reload_inner): Likewise.
18746         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
18747         macros for first/last GPR and FPR registers.
18748         (LAST_GPR_REGNO): Likewise.
18749         (FIRST_FPR_REGNO): Likewise.
18750         (LAST_FPR_REGNO): Likewise.
18752         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
18753         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
18754         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
18755         (vcond<mode><mode>): Likewise.
18756         (vcondu<mode><mode>): Likewise.
18757         (vector_gtu<mode>): Likewise.
18758         (vector_gte<mode>): Likewise.
18759         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
18760         to prevent the compiler from converting DImode operations to TImode.
18761         (ior<mode>3): Likewise.
18762         (and<mode>3): Likewise.
18763         (one_cmpl<mode>2): Likewise.
18764         (nor<mode>3): Likewise.
18765         (andc<mode>3): Likewise.
18767         * config/rs6000/constraints.md (wt constraint): New constraint
18768         that returns VSX_REGS if TImode is allowed in VSX registers.
18770         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
18771         constant under VSX.
18773         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
18774         similar to TImode, but it is restricted to being in the GPRs.
18776         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
18777         TImode to occupy a single VSX register.
18779         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
18780         -mvsx-timode for power7/power8.
18781         (power7 cpu): Likewise.
18782         (power8 cpu): Likewise.
18784         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
18785         sure that TFmode/TDmode take up two registers if they are ever
18786         allowed in the upper VSX registers.
18787         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
18788         registers.
18789         (rs6000_init_hard_regno_mode_ok): Likewise.
18790         (rs6000_debug_reg_global): Add debugging for PTImode and wt
18791         constraint.  Print if LRA is turned on.
18792         (rs6000_option_override_internal): Give an error if -mvsx-timode
18793         and VSX is not enabled.
18794         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
18795         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
18796         to reg+offset addressing.  Use PTImode when checking offset
18797         addresses for validity.
18798         (reg_offset_addressing_ok_p): Likewise.
18799         (rs6000_legitimate_offset_address_p): Likewise.
18800         (rs6000_legitimize_address): Likewise.
18801         (rs6000_legitimize_reload_address): Likewise.
18802         (rs6000_legitimate_address_p): Likewise.
18803         (rs6000_eliminate_indexed_memrefs): Likewise.
18804         (rs6000_emit_move): Likewise.
18805         (rs6000_secondary_reload): Likewise.
18806         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
18807         reloads to fpr registers to continue to use reg+offset addressing,
18808         but 64-bit reloads to altivec registers need reg+reg addressing.
18809         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
18810         it.  Treat LO_SUM like a PLUS operation.
18811         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
18812         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
18813         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
18814         registers to share a register with a smaller sized type, since VSX
18815         puts scalars in the upper 64-bits.
18816         (print_operand): Add support for PTImode.
18817         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
18818         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
18819         registers, but don't have arithmetic support.
18820         (rs6000_memory_move_cost): Add test for VSX.
18821         (rs6000_opt_masks): Add -mvsx-timode.
18823         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
18824         for TImode.
18825         (VSs): Likewise.
18826         (VSr): Use wt constraint for TImode.
18827         (VSv): Drop TImode support.
18828         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
18829         (vsx_movti_64bit): Likewise.
18830         (vsx_movti_32bit): Likewise.
18831         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
18832         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
18833         one '?' on the appropriate output constraint.  Do not allow TImode
18834         logical operations on 32-bit systems.
18835         (vsx_ior<mode>3): Likewise.
18836         (vsx_xor<mode>3): Likewise.
18837         (vsx_one_cmpl<mode>2): Likewise.
18838         (vsx_nor<mode>3): Likewise.
18839         (vsx_andc<mode>3): Likewise.
18840         (vsx_concat_<mode>): Likewise.
18841         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
18843         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
18844         OPTION_MASK_VSX_TIMODE.
18845         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
18846         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
18848         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
18849         (TI2 iterator): New iterator for TImode, PTImode.
18850         (wd mode attribute): Add values for vector types.
18851         (movti_string): Replace TI move operations with operations for TImode
18852         and PTImode.  Add support for TImode being allowed in VSX registers.
18853         (mov<mode>_string, TImode/PTImode): Likewise.
18854         (movti_ppc64): Likewise.
18855         (mov<mode>_ppc64, TImode/PTImode): Likewise.
18856         (TI mode splitters): Likewise.
18858         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
18859         constraint.
18861 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
18863         PR tree-optimization/56355
18864         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
18865         Also handle integers with undefined overflow.
18867 2013-03-20  Catherine Moore  <clm@codesourcery.com>
18868             Maciej W. Rozycki  <macro@codesourcery.com>
18869             Tom de Vries  <tom@codesourcery.com>
18870             Nathan Sidwell  <nathan@codesourcery.com>
18871             Iain Sandoe  <iain@codesourcery.com>
18872             Nathan Froyd  <froydnj@codesourcery.com>
18873             Chao-ying Fu  <fu@mips.com>
18875         * doc/extend.texi: (micromips, nomicromips, nocompression):
18876         Document new function attributes.
18877         * doc/invoke.texi (minterlink-compressed, mmicromips,
18878         m14k, m14ke, m14kec): Document new options.
18879         (minterlink-mips16): Update documentation.
18880         * doc/md.texi (ZC, ZD): Document new constraints.
18881         * configure.ac (gcc_cv_as_micromips): Check if linker
18882         supports the .set micromips directive.
18883         * configure: Regenerate.
18884         * config.in: Regenerate.
18885         * config/mips/mips-tables.opt: Regenerate.
18886         * config/mips/micromips.md: New file.
18887         * constraints.md (ZC, ZD): New constraints.
18888         * config/mips/predicates.md (movep_src_register): New predicate.
18889         (movep_src_operand): New predicate.
18890         (non_volatile_mem_operand): New predicate.
18891         * config/mips/mips.md (multimem): New type.
18892         (length): Differentiate between 17-bit and 18-bit branch offsets.
18893         (MOVEP1, MOVEP2): New mode iterator.
18894         (mov_<load>l): Use ZC constraint.
18895         (mov_<load>r): Likewise.
18896         (mov_<store>l): Likewise.
18897         (mov_<store>r): Likewise.
18898         (*branch_equality<mode>_inverted): Add microMIPS support.
18899         (*branch_equality<mode>): Likewise.
18900         (*jump_absolute): Likewise.
18901         (indirect_jump_<mode>): Likewise.
18902         (tablejump_<mode>): Likewise.
18903         (<optab>_internal): Likewise.
18904         (sibcall_internal): Likewise.
18905         (sibcall_value_internal): Likewise.
18906         (prefetch): Use constraint ZD.
18907         * config/mips/mips.opt (minterlink-compressed): New option.
18908         (minterlink-mips16): Now an alias for minterlink-compressed.
18909         (mmicromips): New option.
18910         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18911         (compare_and_swap_12): Likewise.
18912         (sync_add<mode>): Likewise.
18913         (sync_<optab>_12): Likewise.
18914         (sync_old_<optab>_12): Likewise.
18915         (sync_new_<optab>_12): Likewise.
18916         (sync_nand_12): Likewise.
18917         (sync_old_nand_12): Likewise.
18918         (sync_new_nand_12): Likewise.
18919         (sync_sub<mode>): Likewise.
18920         (sync_old_add<mode>): Likewise.
18921         (sync_old_sub<mode>): Likewise.
18922         (sync_new_add<mode>): Likewise.
18923         (sync_new_sub<mode>): Likewise.
18924         (sync_<optab><mode>): Likewise.
18925         (sync_old_<optab><mode>): Likewise.
18926         (sync_new_<optab><mode>): Likewise.
18927         (sync_nand<mode>): Likewise.
18928         (sync_old_nand<mode>): Likewise.
18929         (sync_new_nand<mode>): Likewise.
18930         (sync_lock_test_and_set<mode>): Likewise.
18931         (test_and_set_12): Likewise.
18932         (atomic_compare_and_swap<mode>): Likewise.
18933         (atomic_exchange<mode>_llsc): Likewise.
18934         (atomic_fetch_add<mode>_llsc): Likewise.
18935         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18936         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18937         (umips_save_restore_pattern_p): Likewise.
18938         (umips_load_store_pair_p): Likewise.
18939         (umips_output_load_store_pair): Likewise.
18940         (umips_movep_target_p): Likewise.
18941         (umips_12bit_offset_address_p): Likewise.
18942         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18943         (mips_base_mips16): Rename this...
18944         (mips_base_compression_flags): ...to this. Update all uses.
18945         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18946         (mips_mips16_decl_p): Delete.
18947         (mips_nomips16_decl_p): Delete.
18948         (mips_get_compress_on_flags): New function.
18949         (mips_get_compress_off_flags): New function.
18950         (mips_get_compress_mode): New function.
18951         (mips_get_compress_on_name): New function.
18952         (mips_get_compress_off_name): New function.
18953         (mips_insert_attributes): Support multiple compression types.
18954         (mips_merge_decl_attributes): Likewise.
18955         (umips_12bit_offset_address_p): New function.
18956         (mips_start_function_definition): Emit .set micromips directive.
18957         (mips_call_may_need_jalx_p): New function.
18958         (mips_function_ok_for_sibcall): Add microMIPS support.
18959         (mips_print_operand_punctuation): Support short delay slots and
18960         compact jumps.
18961         (umips_swm_mask, umips_swm_encoding): New.
18962         (umips_build_save_restore): New function.
18963         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18964         (was_mips16_p): Remove.
18965         (old_compression_mode): New.
18966         (mips_set_compression_mode): New function.
18967         (mips_set_current_function): Add microMIPS support.
18968         (mips_option_override): Likewise.
18969         (umips_save_restore_pattern_p): New function.
18970         (umips_output_save_restore): New function.
18971         (umips_load_store_pair_p_1): New function.
18972         (umips_load_store_pair_p): New function.
18973         (umips_output_load_store_pair_1): New function.
18974         (umips_output_load_store_pair): New function.
18975         (umips_movep_target_p) New function.
18976         (mips_prepare_pch_save): Add microMIPS support.
18977         * config/mips/mips.h (TARGET_COMPRESSION): New.
18978         (TARGET_CPU_CPP_BUILTINS): Update macro
18979         to use new compression flags and to support microMIPS.
18980         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18981         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18982         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18983         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18984         (ASM_SPEC): Support mmicromips and mno-micromips.
18985         (M16STORE_REG_P): New macro.
18986         (MIPS_CALL): Support TARGET_MICROMIPS.
18987         (MICROMIPS_J): New macro.
18988         (mips_base_mips16): Rename this...
18989         (mips_base_compression_flags): ...to this.
18990         (UMIPS_12BIT_OFFSET_P): New macro.
18991         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18992         (MULTILIB_DIRNAMES): Likewise.
18993 2013-03-20  Richard Biener  <rguenther@suse.de>
18995         PR tree-optimization/56661
18996         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
18997         the result does not have to be distinct.
18999 2013-03-20  Richard Biener  <rguenther@suse.de>
19001         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
19002         remap_gimple_op_r.
19004 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19005             Steven Bosscher  <steven@gcc.gnu.org>
19007         PR rtl-optimization/56605
19008         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
19010 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
19012         PR bootstrap/56656
19013         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
19014         that require movd instead of movq.
19016 2013-03-20  Richard Biener  <rguenther@suse.de>
19018         * tree-ssa-structalias.c (struct variable_info): Add pointer
19019         to the first field of an aggregate with sub-vars.  Make
19020         this and the pointer to the next subfield its ID.
19021         (vi_next): New function.
19022         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
19023         storedanything_id, integer_id): Increment by one.
19024         (new_var_info, get_call_vi, lookup_call_clobber_vi,
19025         get_call_clobber_vi): Adjust.
19026         (solution_set_expand): Simplify and speedup.
19027         (solution_set_add): Inline into ...
19028         (set_union_with_increment): ... this.  Adjust accordingly.
19029         (do_sd_constraint): Likewise.
19030         (do_ds_constraint): Likewise.
19031         (do_complex_constraint): Simplify.
19032         (build_pred_graph): Adjust.
19033         (solve_graph): Likewise.  Simplify and speedup.
19034         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
19035         get_constraint_for_component_ref, get_constraint_for_1,
19036         first_vi_for_offset, first_or_preceding_vi_for_offset,
19037         create_function_info_for, create_variable_info_for_1,
19038         create_variable_info_for, intra_create_variable_infos): Adjust.
19039         (init_base_vars): Push NULL for ID zero.
19040         (compute_points_to_sets): Adjust.
19042 2013-03-20  Richard Biener  <rguenther@suse.de>
19044         * cfgloop.c (verify_loop_structure): Streamline and avoid
19045         ICEing on corrupt loop tree.
19046         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
19047         loop tree.
19049 2013-03-20  Richard Biener  <rguenther@suse.de>
19051         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
19052         check whether an SSA update is needed.
19054 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
19056         * config/mips/constraints.md (T): Rename to...
19057         (Yf): ...this.
19058         (U): Rename to...
19059         (Yd): ...this.
19060         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
19061         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
19063 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
19065         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
19066         (*subsi3_carryin_uxtw): Likewise.
19068 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
19070         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
19071         (*rorsi3_insn_uxtw): Likewise.
19073 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
19075         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
19076         (*extrsi5_insn_uxtw): Likewise.
19078 2013-03-19  Richard Biener  <rguenther@suse.de>
19080         PR tree-optimization/56273
19081         * passes.c (init_optimization_passes): Move second VRP after DOM.
19083 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
19085         * config/i386/i386.md (*movti_internal): Merge from
19086         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
19087         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
19088         nox64 isa attributes.
19090 2013-03-18  Richard Biener  <rguenther@suse.de>
19092         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
19093         (unite): Likewise.
19094         (merge_node_constraints): Likewise.
19095         (build_succ_graph): Likewise.
19096         (valid_graph_edge): Inline into single caller.
19097         (unify_nodes): Likewise.  Use bitmap_set_bit return value
19098         and cache varinfo.
19099         (scc_visit): Fix formatting and variable use.
19100         (do_sd_constraint): Use gcc_checking_assert.
19101         (do_ds_constraint): Likewise.
19102         (do_complex_constraint): Likewise.
19103         (condense_visit): Likewise.  Cleanup.
19104         (dump_pred_graph): New function.
19105         (perform_var_substitution): Dump the pred-graph before
19106         variable substitution.
19107         (find_equivalent_node): Use gcc_checking_assert.
19108         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
19110 2013-03-18  Richard Biener  <rguenther@suse.de>
19112         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
19113         Remove cond_expr_stmt_list argument and do not gimplify the
19114         built expression.
19115         (vect_loop_versioning): Adjust.
19116         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
19117         Cleanup to use less temporaries.
19118         (vect_create_data_ref_ptr): Cleanup.
19120 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
19122         PR tree-optimization/56635
19123         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
19124         require types_compatible_p types.
19126 2013-03-18  Nick Clifton  <nickc@redhat.com>
19128         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
19129         spurious backslash.
19131         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
19132         Add missing line to comment describing function.
19134 2013-03-18  Richard Biener  <rguenther@suse.de>
19136         PR tree-optimization/56210
19137         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19138         Handle string / character search functions.
19139         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
19141 2013-03-18  Richard Biener  <rguenther@suse.de>
19143         PR middle-end/56483
19144         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
19145         and implement properly.
19146         * gimple.h (gimple_cond_single_var_p): Remove.
19148 2013-03-18  Richard Biener  <rguenther@suse.de>
19150         * tree-data-ref.h (find_data_references_in_loop): Declare.
19151         * tree-data-ref.c (get_references_in_stmt): Use a stack
19152         vector pre-allocated in the callers.
19153         (find_data_references_in_stmt): Adjust.
19154         (graphite_find_data_references_in_stmt): Likewise.
19155         (create_rdg_vertices): Likewise.
19156         (find_data_references_in_loop): Export.
19157         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
19158         Compute dependences here...
19159         (vect_analyze_data_refs): ...not here.  When we encounter
19160         a non-vectorizable data reference in basic-block vectorization
19161         truncate the data reference vector.  Do not bother to
19162         fixup data-dependence information for gather loads.
19163         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
19164         of data references, as reported.
19166 2013-03-18  Richard Biener  <rguenther@suse.de>
19168         PR tree-optimization/3713
19169         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
19170         has_constants and expr.
19171         (stmt_has_constants): Properly valueize SSA names when deciding
19172         whether the stmt has constants.
19174 2013-03-18  Richard Biener  <rguenther@suse.de>
19176         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
19177         whole function when there is nothing to do.
19178         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
19179         * tree-vectorizer.c (vectorize_loops): Update virtual and
19180         loop-closed SSA once.
19181         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
19183 2013-03-18  Richard Biener  <rguenther@suse.de>
19185         PR middle-end/56113
19186         * domwalk.c (bb_postorder): New global static.
19187         (cmp_bb_postorder): New function.
19188         (walk_dominator_tree): Replace scheme imposing an order for
19189         visiting dominator sons by one sorting them at the time they
19190         are pushed on the stack.
19192 2013-03-18  Richard Biener  <rguenther@suse.de>
19194         PR tree-optimization/39326
19195         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
19196         (struct mem_ref): Replace mem member with ao_ref typed member.
19197         (MEM_ANALYZABLE): Adjust.
19198         (memref_eq): Likewise.
19199         (mem_ref_alloc): Likewise.
19200         (gather_mem_refs_stmt): Likewise.
19201         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
19202         (execute_sm_if_changed_flag_set): Adjust.
19203         (execute_sm): Likewise.
19204         (ref_always_accessed_p): Likewise.
19205         (refs_independent_p): Likewise.
19206         (can_sm_ref_p): Likewise.
19208 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
19210         PR c/56566
19211         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
19212         return 1 even for !unsignedp.
19214 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19216         * config/i386/i386.md (isa): Add x64 and nox64.
19217         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
19218         (*pushtf): Enable *roF alternative for x64 isa only.
19219         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
19220         mode attribute of integer alternatives to DImode for TARGET_64BIT.
19221         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
19222         (*movtf_internal): Merge from *movtf_internal_rex64 and
19223         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
19224         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
19225         nox64 isa attributes.
19226         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
19227         nox64 isa attributes.
19228         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
19230 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19232         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19234 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
19236         PR target/56640
19237         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
19238         class names.  Remove trailing comma after "ALL_REGS".
19240 2013-03-16  Jan Hubicka  <jh@suse.cz>
19242         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
19243         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
19244         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
19245         of cgraph_get_create_node.
19246         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
19248 2013-03-16  Jason Merrill  <jason@redhat.com>
19250         PR debug/49090
19251         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
19252         with DW_AT_default_value.
19254 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
19256         * BASE-VER: Set to 4.9.0.
19258 2013-03-14  Andi Kleen  <ak@linux.intel.com>
19260         PR target/56619
19261         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
19262         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
19263         Document _x* TSX intrinsics.
19265 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
19266             David Holsgrove  <david.holsgrove@xilinx.com>
19268         * configure.ac: Add MicroBlaze TLS support detection.
19269         * configure: Regenerate.
19270         * config/microblaze/microblaze-protos.h
19271         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
19272         symbol_mentioned_p, label_mentioned_p): Add prototypes.
19273         * config/microblaze/microblaze.c (microblaze_address_type): Add
19274         ADDRESS_TLS and tls_reloc address types.
19275         (microblaze_address_info): Add tls_reloc.
19276         (TARGET_HAVE_TLS): Define.
19277         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
19278          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
19279          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
19280         load_tls_operand,  microblaze_call_tls_get_addr,
19281         microblaze_legitimize_tls_address): New functions.
19282         (microblaze_classify_unspec): Handle UNSPEC_TLS.
19283         (get_base_reg): Use microblaze_tls_symbol_p.
19284         (microblaze_classify_address): Handle TLS.
19285         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
19286         label_mentioned_p and microblaze_tls_referenced_p.
19287         (microblaze_legitimize_address): Handle TLS.
19288         (microblaze_address_insns): Handle ADDRESS_TLS.
19289         (pic_address_needs_scratch): Handle TLS.
19290         (print_operand_address): Handle TLS.
19291         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
19292         (microblaze_expand_move): Handle TLS.
19293         (microblaze_legitimate_constant_p): Check
19294         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
19295         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
19296         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
19297         (PIC_OFFSET_TABLE_REGNUM): Set.
19298         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
19299         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
19300         (addsi3, movsi_internal2, movdf_internal): Update constraints
19301         * config/microblaze/predicates.md (arith_plus_operand): Define
19302         (move_operand): Redefine as move_src_operand,
19303         check microblaze_tls_referenced_p.
19305 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
19307         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
19308         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
19310 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
19312         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
19313         CC mode for AND.
19315 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
19317         PR tree-optimization/53265
19318         * common.opt (Waggressive-loop-optimizations): New option.
19319         * tree-ssa-loop-niter.c: Include tree-pass.h.
19320         (do_warn_aggressive_loop_optimizations): New function.
19321         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
19322         if number_of_latch_executions returned constant.
19323         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
19324         early.  If number_of_latch_executions returned constant, set
19325         nb_iterations_upper_bound back to it.
19326         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
19327         field.
19328         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
19329         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
19331         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
19332         (MULTILIB_OSDIRNAMES): Set.
19333         * genmultilib: If defaultosdirname doesn't start with :: , set
19334         defaultosdirname2 instead, clear it and emit two . multilib_raw
19335         entries instead of just one.
19337 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
19339         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
19340         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
19341         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
19342         (SUBTARGET_OVERRIDE_OPTIONS): New.
19344 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
19346         PR target/49880
19347         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
19348         (musermode): Convert to Var(TARGET_USERMODE).
19349         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
19350         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
19351         * config/sh/sh.c (sh_option_override): Use
19352         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
19353         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
19354         condition.
19355         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
19356         TARGET_SH4.
19357         (udivsi3_i4_single, divsi3_i4_single): Use
19358         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
19360 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
19362         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
19363         default setting.
19365 2013-03-13  Richard Biener  <rguenther@suse.de>
19367         PR tree-optimization/56608
19368         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
19369         calls when vectorizing basic-blocks.
19371 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
19373         PR plugins/45078
19374         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
19375         tm_file.
19377 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
19379         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
19381 2013-03-11  Jan Hubicka  <jh@suse.cz>
19383         PR lto/56557
19384         * lto-streamer-out.c (output_symbol_p): Skip references from
19385         constructors of external variables.
19387 2013-03-11  Jan Hubicka  <jh@suse.cz>
19389         PR middle-end/56571
19390         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
19391         from pseudos.
19392         * emit-rtl.c (verify_rtx_sharing): Likewise.
19393         (copy_insn_1): Likewise.
19394         * rtl.c (copy_rtx): Likewise.
19396 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
19398         PR target/56591
19399         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
19400         output_operand_lossage message.
19402 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
19404         PR target/56470
19405         * arm.c (shift_op): Validate RTL pattern on the fly.
19406         (arm_print_operand, case 'S'): Don't use shift_operator to validate
19407         the RTL.
19409 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19411         PR target/56347
19412         * config/pa/pa.md (call_value): Check for calls to powf and direct to
19413         new call patterns that clobber %fr12.
19414         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
19415         split and postreload patterns.
19416         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
19417         registers %fr12 and %fr12R as call used.
19419 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
19421         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
19422         (canon_address, record_store, replace_read, check_mem_read_rtx,
19423         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
19424         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
19425         rest_of_handle_dse): Likewise.
19427 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
19429         PR middle-end/56524
19430         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
19431         Add base_optabs.
19432         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
19433         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
19434         (save_optabs_if_changed): Replace with...
19435         (init_tree_optimization_optabs): ...this.
19436         * optabs.c (save_optabs_if_changed): Rename to...
19437         (init_tree_optimization_optabs): ...this.  Take the optimization node
19438         as argument.  Do nothing if the base optabs are already correct.
19439         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
19440         to recompute optabs.
19441         * function.h (function): Remove optabs field.
19442         * function.c (invoke_set_current_function_hook): Call
19443         init_tree_optimization_optabs.  Use the result to initialize
19444         this_fn_optabs.
19446 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
19448         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
19449         if GTMA_HAS_NO_INSTRUMENTATION.
19450         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
19451         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
19452         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
19453         * gimple-pretty-print.c (dump_gimple_transaction): Handle
19454         GTMA_HAS_NO_INSTRUMENTATION.
19456 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
19458         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
19459         libasan_preinit.o.
19461 2013-03-08  Marek Polacek  <polacek@redhat.com>
19462             Jakub Jelinek  <jakub@redhat.com>
19464         PR tree-optimization/56478
19465         * predict.c (is_comparison_with_loop_invariant_p): Change the
19466         type of loop_step to tree.
19467         (predict_loops): Adjust.
19468         (predict_iv_comparison): Perform the computations on double_ints.
19470 2013-03-08  Richard Biener  <rguenther@suse.de>
19472         PR tree-optimization/56570
19473         * tree-cfg.c (verify_expr_location_1): Verify locations for
19474         DECL_DEBUG_EXPR.
19475         * tree-sra.c (create_access_replacement): Strip locations
19476         from DECL_DEBUG_EXPRs.
19478 2013-03-08  Richard Biener  <rguenther@suse.de>
19480         * tree-inline.c (expand_call_inline): Do not associate
19481         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
19482         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
19484 2013-03-08  Richard Biener  <rguenther@suse.de>
19486         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
19487         or block changes with -Og.  Fix for location / block encoding
19488         changes and PHI arguments with locations.
19490 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
19492         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
19493         for all counters.
19494         (struct output_info): Likewise.
19495         (register_overhead): Remove bad gcc_assert.
19496         (bitmap_find_bit): If there is only a single bitmap element, do not
19497         count a miss as a search.
19498         (print_statistics): Update for counter type changes.
19499         (dump_bitmap_statistics): Likewise.  Print headers such that they
19500         are properly lined up with the printed counters.
19502 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
19504         PR tree-optimization/56559
19505         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
19506         check that it has only a single use.
19508 2013-03-07  Richard Biener  <rguenther@suse.de>
19510         * doc/invoke.texi (fwhole-program): Discourage use in combination
19511         with -flto.
19513 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
19515         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
19517         PR tree-optimization/56539
19518         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
19519         instead of GSI_CONTINUE_LINKING as last argument to
19520         force_gimple_operand_gsi.  Adjust function comment.
19522         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
19523         aarch64-cores.def.
19525         PR middle-end/56548
19526         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
19527         promoted mode, convert the result back to the original mode.
19529 2013-03-06  Richard Biener  <rguenther@suse.de>
19531         PR middle-end/56294
19532         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
19533         (insert_updated_phi_nodes_compare_uids): New function.
19534         (update_ssa): Sort symbols_to_rename after UID before
19535         traversing it to insert PHI nodes.
19537 2013-03-06  Richard Biener  <rguenther@suse.de>
19539         PR middle-end/50494
19540         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
19541         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
19543         Revert
19544         2013-02-13  Richard Biener  <rguenther@suse.de>
19546         PR lto/50494
19547         * varasm.c (output_constant_def_1): Get the decl representing
19548         the constant as argument.
19549         (output_constant_def): Wrap output_constant_def_1.
19550         (make_decl_rtl): Use output_constant_def_1 with the decl
19551         representing the constant.
19552         (build_constant_desc): Optionally re-use a decl already
19553         representing the constant.
19554         (tree_output_constant_def): Adjust.
19556 2013-03-06  Joey Ye  <joey.ye@arm.com>
19558         PR lto/50293
19559         * gcc.c (convert_white_space): New function.
19560         (main): Handles white space in function name.
19562 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
19564         PR target/56529
19565         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
19566         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
19567         to SH_DIV_CALL_TABLE for TARGET_SH2.
19568         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
19569         list.
19570         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
19571         call-table options.
19573 2013-03-05  Sterling Augustine  <saugustine@google.com>
19574             Cary Coutant  <ccoutant@google.com>
19576         PR debug/55364
19577         * dwarf2out.c (resolve_addr): Don't call
19578         remove_loc_list_addr_table_entries a second time for the same
19579         expression.
19581 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19583         PR debug/56510
19584         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
19585         (avoid_complex_debug_insns): New function.
19586         (expand_debug_locations): Call it.
19588         PR rtl-optimization/56484
19589         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
19590         lifetimes of hard registers on small register class machines.
19592 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
19594         * config/microblaze/microblaze-protos.h: Rename
19595         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
19596         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
19597         fast_interrupt.
19598         (microblaze_fast_interrupt_function_p): New function.
19599         (microblaze_is_interrupt_handler): Rename to
19600         microblaze_is_interrupt_variant and add fast_interrupt check.
19601         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
19602         (save_restore_insns): Likewise.
19603         (compute_frame_size): Likewise.
19604         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
19605         (microblaze_globalize_label): Likewise.
19606         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
19607         * config/microblaze/microblaze.md: Use wrapper
19608         microblaze_is_interrupt_variant.
19610 2013-03-05  Kai Tietz  <ktietz@redhat.com>
19612         * sdbout.c (sdbout_one_type): Switch to current function's section
19613         supporting cold/hot.
19615 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
19617         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
19618         -mxl-reorder.
19620 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19622         PR middle-end/56461
19623         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
19624         if VALGRIND_GET_VBITS is defined, temporarily make object
19625         memory all defined, and restore previous valgrind addressability
19626         and definability afterwards.  Free this_object at the end.
19628         PR middle-end/56461
19629         * lra.c (lra): Call lra_clear_live_ranges if live_p,
19630         right before calling lra_create_live_ranges, also call it
19631         when clearing live_p.  Only call lra_clear_live_ranges
19632         at the end if live_p.
19634         PR middle-end/56461
19635         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
19637 2013-03-05  Richard Biener  <rguenther@suse.de>
19639         PR tree-optimization/56521
19640         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
19641         value-id.
19643 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
19645         PR c++/55135
19646         * except.h (remove_unreachable_eh_regions): New prototype.
19647         * except.c (remove_eh_handler_splicer): New function, split out
19648         of remove_eh_handler.
19649         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
19650         warning about running it on many EH regions one at a time.
19651         (remove_unreachable_eh_regions_worker): New function, walk the
19652         EH tree in depth-first order and remove non-marked regions.
19653         (remove_unreachable_eh_regions): New function.
19654         * tree-eh.c (mark_reachable_handlers): New function, split out
19655         from remove_unreachable_handlers.
19656         (remove_unreachable_handlers): Use mark_reachable_handlers and
19657         remove_unreachable_eh_regions.
19658         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
19659         and remove_unreachable_eh_regions.
19661 2013-03-05  Richard Biener  <rguenther@suse.de>
19663         PR middle-end/56525
19664         * loop-init.c (fix_loop_structure): Remove loops in two stages,
19665         not freeing them until the end.
19667 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19669         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
19671 2013-03-05  Richard Biener  <rguenther@suse.de>
19673         PR tree-optimization/56270
19674         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
19675         of loads after scheduling an SLP instance.
19677 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19679         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
19680         tic6x.exp.
19681         (check_gcc_parallelize): Run guality.exp as a separate job from
19682         vect.exp with unsorted.exp and $(dg_target_exps) separately from
19683         struct-layout-1.exp with stackalign.exp.
19685         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
19687         PR middle-end/56461
19688         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
19689         load_index sbitmap even if some bit in it isn't set.
19691         PR middle-end/56461
19692         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
19693         (discover_iteration_bound_by_body_walk): Change queues to
19694         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
19695         spelling in comment.  Call safe_push on queues[bound_index] directly.
19696         Release queues[queue_index] in every iteration unconditionally.
19697         Release bounds vector.
19699         PR middle-end/56461
19700         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
19701         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
19702         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
19703         inner_phis vector.
19705 2013-03-05  Richard Biener  <rguenther@suse.de>
19707         PR lto/56515
19708         * tree-inline.c (remap_blocks_to_null): New function.
19709         (expand_call_inline): When expanding a call stmt without
19710         an associated block inline remap all callee blocks to NULL.
19712 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19714         PR rtl-optimization/56494
19715         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
19716         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
19717         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
19719         PR middle-end/56461
19720         * sel-sched-ir.c (free_sched_pools): Release
19721         succs_info_pool.stack[succs_info_pool.max_top] vectors too
19722         if succs_info_pool.max_top isn't -1.
19724         PR bootstrap/56509
19725         * opts.c (opts_obstack, opts_concat): Moved to...
19726         * opts-common.c (opts_obstack, opts_concat): ... here.
19728 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
19730         PR middle-end/56461
19731         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
19733 2013-03-04  Martin Jambor  <mjambor@suse.cz>
19735         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
19736         all appropriate places.
19738 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
19740         PR tree-optimization/56424
19741         * ipa-split.c (split_function): Do not set the RSO flag if result is
19742         not by reference and its type is a register type.
19744 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
19746         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
19747         (microblaze_legitimate_pic_operand): Likewise
19748         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
19749         new function microblaze_legitimate_pic_operand
19750         * config/microblaze/microblaze-protos.h
19751         (microblaze_legitimate_pic_operand): Declare.
19753 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19755         * config/microblaze/predicates.md (call_insn_simple_operand):
19756         New predicate for supported rtx code types.
19757         * config/microblaze/microblaze.md (call_internal1): Use
19758         call_insn_simple_operand predicate.
19760 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
19762         PR middle-end/56461
19763         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
19764         partitions.ordered_remove.
19766         PR middle-end/56461
19767         * tree-vect-stmts.c (vectorizable_conversion): Don't call
19768         vec_oprnds0.create (1) for modifier == NONE.
19770         PR middle-end/56461
19771         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
19772         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
19773         vec_oprnds1 right before pushing anything to it for
19774         scalar_shift_arg.
19776         PR middle-end/56461
19777         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
19778         set nbbs to 0 instead of having separate code path.
19779         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
19780         instead of false as last argument if returning NULL.
19782 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
19784         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
19785         the attribute is now called "target" instead of "option".
19786         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
19787         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
19788         attribute/pragma name for TARGET_OPTION_VALID_P and
19789         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
19790         * doc/tm.texi: Regenerated.
19792 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
19794         * config/microblaze/microblaze.c:
19795         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
19796         * config/microblaze/microblaze.h: Add -mxl-reorder to
19797         DRIVER_SELF_SPECS.
19798         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
19799         instructions emitted if TARGET_REORDER.
19800         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
19801         or 0 for -m/-mno case, but initialises as 2 to detect default use case
19802         separately.
19804 2013-03-01  Xinliang David Li  <davidxl@google.com>
19806         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
19807         walk length.
19809 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
19811         PR middle-end/56461
19812         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
19813         vector even when returning true.  Fix up function comment formatting.
19815         PR middle-end/56461
19816         * ira-build.c (ira_loop_nodes_count): New variable.
19817         (create_loop_tree_nodes): Initialize it.
19818         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
19820         PR middle-end/56461
19821         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
19822         method on dr_chain and result_chain.
19823         * tree-vect-stmts.c (vectorizable_store): Only call
19824         result_chain.create if j == 0.
19826         PR middle-end/56461
19827         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
19828         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
19829         before overwriting it.
19831 2013-03-01  Tobias Burnus  <burnus@net-b.de>
19833         * doc/extended.texi (C Extensions): Change order in @menu
19834         to match @node.
19835         (Other MIPS Built-in Functions): Move last MIPS entry before
19836         "picoChip Built-in Functions".
19837         (SH Built-in Functions): Move after RX Built-in Functions.
19838         * doc/gcc.texi (Introduction): Change order in @menu
19839         to match @node.
19840         * doc/md.texi (Constraints): Ditto.
19841         * gty.texi (Type Information): Ditto.
19842         (User-provided marking routines for template types): Make
19843         subsection.
19844         * doc/invoke.texi (AArch64 Options): Move before
19845         "Adapteva Epiphany Options".
19847 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
19848             Jakub Jelinek  <jakub@redhat.com>
19850         PR sanitizer/56454
19851         * asan.c (gate_asan): Lookup no_sanitize_address instead of
19852         no_address_safety_analysis attribute.
19853         * doc/extend.texi (no_address_safety_attribute): Rename to
19854         no_sanitize_address attribute, mention no_address_safety_analysis
19855         attribute as deprecated alias.
19857 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
19859         PR middle-end/56461
19860         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
19861         type to vec<vec<tree> > *.
19862         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
19863         to be vec<tree> instead of vec<tree> *, set vec_defs
19864         to vNULL and call vec_defs.create (number_of_vects), adjust other
19865         uses of vec_defs.
19866         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
19867         vectorizable_condition): Adjust vect_get_slp_defs callers.
19869 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19871         * config/aarch64/aarch64.c
19872         (aarch64_float_const_representable): Remove unused variable.
19874 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19876         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
19878 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19880         * config/aarch64/aarch64-builtins.c
19881         (aarch64_init_simd_builtins): Make static.
19883 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19885         * config/aarch64/aarch64.c
19886         (aarch64_simd_make_constant): Make static.
19888 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19890         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
19891         with no initialization to the RHS of debug statements.
19893 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19895         PR tree-optimization/56294
19896         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
19897         Adjust dumping.
19898         (get_access_replacement): Do not call create_access_replacement.
19899         Assert a replacement exists.
19900         (get_repl_default_def_ssa_name): Create the replacement declaration
19901         itself.
19903 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19905         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
19906         final_end_function.
19908 2013-02-28  Marek Polacek  <polacek@redhat.com>
19910         PR rtl-optimization/56466
19911         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
19912         if we're changing a loop.
19913         (peel_loops_completely): Likewise.
19915 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
19917         PR c++/55813
19918         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
19920 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19922         PR target/56445
19923         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
19924         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
19925         INTX_FTYPE_FX, FX_FTYPE_INTX.
19926         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
19928 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19930         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
19931         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
19932         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
19933         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
19934         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
19935         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
19936         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
19937         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
19938         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
19939         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
19940         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
19941         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
19942         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
19943         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
19944         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
19945         (avrxmega6): Increase max flash segments from 5 to 6.
19946         * config/avr/t-multilib: Regenerate.
19947         * config/avr/avr-tables.opt: Regenerate.
19948         * doc/avr-mmcu.texi: Regenerate.
19950 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19952         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
19953         (avr_device_to_arch): Rename to avr_device_to_ld.
19954         (avr_device_to_as): New prototype.
19955         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
19956         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
19957         * config/avr/driver-avr.c (avr_device_to_as): New.
19958         (avr_device_to_arch): Rename to avr_device_to_ld.
19960 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19962         PR middle-end/56461
19963         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
19964         method on dr_chain and result_chain.
19966         PR middle-end/56461
19967         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
19968         pointer_set_destroy on not_executed_last_iteration.
19970         PR middle-end/56461
19971         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
19973         PR middle-end/56461
19974         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
19975         FOR_EACH_DEFINED_FUNCTION when freeing state.
19977         PR middle-end/56461
19978         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
19979         pool_free.
19980         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
19981         overwriting it.
19983         PR middle-end/56461
19984         * ipa-cp.c (decide_whether_version_node): Call vec_free on
19985         known_aggs[i].items and release known_aggs vector.
19987         PR middle-end/56461
19988         * ipa-reference.c (propagate): Free node_info even for alias nodes.
19990 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19992         * config/microblaze/microblaze.c (microblaze_emit_compare):
19993         Use xor for EQ/NE comparisions.
19994         * config/microblaze/microblaze.md (cstoresf4): Add constraints
19995         (cbranchsf4): Adjust operator to comparison_operator.
19997 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19999         PR middle-end/56461
20000         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
20001         vector.
20002         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
20003         vec_safe_push, always update *slot.
20004         (redirect_edge_var_map_clear): Use vec_free.
20005         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
20006         (free_var_map_entry): Use vec_free.
20007         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
20008         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
20010 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
20012         PR middle-end/45472
20013         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
20014         when the may_trap_p bit of the exprs being merged differs.
20015         Reorder tests for speculativeness in the logical and operator.
20017 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
20019         * incpath.c (add_standard_paths): Use reconcat instead of concat
20020         where appropriate and avoid leaking memory.
20022         * opts.h: Include obstack.h.
20023         (opts_concat): New prototype.
20024         (opts_obstack): New declaration.
20025         * opts.c (opts_concat): New function.
20026         (opts_obstack): New variable.
20027         (init_options_struct): Call gcc_init_obstack on opts_obstack.
20028         (finish_options): Use opts_concat instead of concat
20029         and XOBNEWVEC instead of XNEWVEC.
20030         * opts-common.c (generate_canonical_option, decode_cmdline_option,
20031         generate_option): Likewise.
20032         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
20033         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
20035         PR target/56455
20036         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
20037         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
20039 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
20041         PR middle-end/56461
20042         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
20044 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
20046         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
20047         (arm_block_move_unaligned_straight): Likewise.
20048         (arm_adjust_block_mem): Likewise.
20050 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
20052         PR target/48901
20053         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
20054         temp, cond and label.
20055         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
20057         PR target/52500
20058         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
20059         * config/c6x/c6x.h (dbx_register_map): Update declaration.
20061         PR target/52501
20062         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
20063         of prologue/epilogue functions.
20065         PR target/52550
20066         * config/tilegx/tilegx.c (tilegx_expand_prologue):
20067         Remove unused variable cfa_offset.
20068         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
20070         PR target/54639
20071         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
20072         type promotion to unsigned.
20074         PR target/54640
20075         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
20076         for HOST_WIDE_INT of 32 bit / same size as int.
20077         (arm_block_move_unaligned_straight): Likewise.
20078         (arm_adjust_block_mem): Likewise.
20080         PR target/54662
20081         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
20082         ALL_CFLAGS.
20084 2013-02-26  Marek Polacek  <polacek@redhat.com>
20086         PR tree-optimization/56426
20087         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
20089 2013-02-26  Richard Biener  <rguenther@suse.de>
20091         PR target/56444
20092         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
20093         unused variable loops.
20095 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
20097         PR tree-optimization/56448
20098         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
20099         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
20100         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
20101         later operands of the references, or even first operand for
20102         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
20104         PR tree-optimization/56443
20105         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
20106         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
20107         to type_for_mode langhook.
20109 2013-02-25  Matt Turner  <mattst88@gmail.com>
20111         * doc/invoke.texi: Document r4700.
20113 2013-02-25  Richard Biener  <rguenther@suse.de>
20115         PR tree-optimization/56175
20116         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
20117         split out from ...
20118         (simplify_bitwise_binary): ... here.  Also guard the conversion
20119         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
20121 2013-02-25  Catherine Moore  <clm@codesourcery.com>
20123         Revert:
20124         2013-02-24  Catherine Moore  <clm@codesourcery.com>
20125             Maciej W. Rozycki  <macro@codesourcery.com>
20126             Tom de Vries  <tom@codesourcery.com>
20127             Nathan Sidwell  <nathan@codesourcery.com>
20128             Iain Sandoe  <iain@codesourcery.com>
20129             Nathan Froyd  <froydnj@codesourcery.com>
20130             Chao-ying Fu  <fu@mips.com>
20132         * doc/extend.texi: (micromips, nomicromips, nocompression):
20133         Document new function attributes.
20134         * doc/invoke.texi (minterlink-compressed, mmicromips,
20135         m14k, m14ke, m14kec): Document new options.
20136         (minterlink-mips16): Update documentation.
20137         * doc/md.texi (ZC, ZD): Document new constraints.
20138         * configure.ac (gcc_cv_as_micromips): Check if linker
20139         supports the .set micromips directive.
20140         * configure: Regenerate.
20141         * config.in: Regenerate.
20142         * config/mips/mips-tables.opt: Regenerate.
20143         * config/mips/micromips.md: New file.
20144         * constraints.md (ZC, AD): New constraints.
20145         * config/mips/predicates.md (movep_src_register): New predicate.
20146         (movep_src_operand): New predicate.
20147         (non_volatile_mem_operand): New predicate.
20148         * config/mips/mips.md (multimem): New type.
20149         (length): Differentiate between 17-bit and 18-bit branch offsets.
20150         (MOVEP1, MOVEP2): New mode iterator.
20151         (mov_<load>l): Use ZC constraint.
20152         (mov_<load>r): Likewise.
20153         (mov_<store>l): Likewise.
20154         (mov_<store>r): Likewise.
20155         (*branch_equality<mode>_inverted): Add microMIPS support.
20156         (*branch_equality<mode>): Likewise.
20157         (*jump_absolute): Likewise.
20158         (indirect_jump_<mode>): Likewise.
20159         (tablejump_<mode>): Likewise.
20160         (<optab>_internal): Likewise.
20161         (sibcall_internal): Likewise.
20162         (sibcall_value_internal): Likewise.
20163         (prefetch): Use constraint ZD.
20164         * config/mips/mips.opt (minterlink-compressed): New option.
20165         (minterlink-mips16): Now an alias for minterlink-compressed.
20166         (mmicromips): New option.
20167         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
20168         (compare_and_swap_12): Likewise.
20169         (sync_add<mode>): Likewise.
20170         (sync_<optab>_12): Likewise.
20171         (sync_old_<optab>_12): Likewise.
20172         (sync_new_<optab>_12): Likewise.
20173         (sync_nand_12): Likewise.
20174         (sync_old_nand_12): Likewise.
20175         (sync_new_nand_12): Likewise.
20176         (sync_sub<mode>): Likewise.
20177         (sync_old_add<mode>): Likewise.
20178         (sync_old_sub<mode>): Likewise.
20179         (sync_new_add<mode>): Likewise.
20180         (sync_new_sub<mode>): Likewise.
20181         (sync_<optab><mode>): Likewise.
20182         (sync_old_<optab><mode>): Likewise.
20183         (sync_new_<optab><mode>): Likewise.
20184         (sync_nand<mode>): Likewise.
20185         (sync_old_nand<mode>): Likewise.
20186         (sync_new_nand<mode>): Likewise.
20187         (sync_lock_test_and_set<mode>): Likewise.
20188         (test_and_set_12): Likewise.
20189         (atomic_compare_and_swap<mode>): Likewise.
20190         (atomic_exchange<mode>_llsc): Likewise.
20191         (atomic_fetch_add<mode>_llsc): Likewise.
20192         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
20193         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
20194         (umips_save_restore_pattern_p): Likewise.
20195         (umips_load_store_pair_p): Likewise.
20196         (umips_output_load_store_pair): Likewise.
20197         (umips_movep_target_p): Likewise.
20198         (umips_12bit_offset_address_p): Likewise.
20199         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
20200         (mips_base_mips16): Rename this...
20201         (mips_base_compression_flags): ...to this. Update all uses.
20202         (mips_attribute_table): Add micromips, nomicromips and nocompression.
20203         (mips_mips16_decl_p): Delete.
20204         (mips_nomips16_decl_p): Delete.
20205         (mips_get_compress_on_flags): New function.
20206         (mips_get_compress_off_flags): New function.
20207         (mips_get_compress_mode): New function.
20208         (mips_get_compress_on_name): New function.
20209         (mips_get_compress_off_name): New function.
20210         (mips_insert_attributes): Support multiple compression types.
20211         (mips_merge_decl_attributes): Likewise.
20212         (umips_12bit_offset_address_p): New function.
20213         (mips_start_function_definition): Emit .set micromips directive.
20214         (mips_call_may_need_jalx_p): New function.
20215         (mips_function_ok_for_sibcall): Add microMIPS support.
20216         (mips_print_operand_punctuation): Support short delay slots and
20217         compact jumps.
20218         (umips_swm_mask, umips_swm_encoding): New.
20219         (umips_build_save_restore): New function.
20220         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
20221         (was_mips16_p): Remove.
20222         (old_compression_mode): New.
20223         (mips_set_compression_mode): New function.
20224         (mips_set_current_function): Add microMIPS support.
20225         (mips_option_override): Likewise.
20226         (umips_save_restore_pattern_p): New function.
20227         (umips_output_save_restore): New function.
20228         (umips_load_store_pair_p_1): New function.
20229         (umips_load_store_pair_p): New function.
20230         (umips_output_load_store_pair_1): New function.
20231         (umips_output_load_store_pair): New function.
20232         (umips_movep_target_p) New function.
20233         (mips_prepare_pch_save): Add microMIPS support.
20234         * config/mips/mips.h (TARGET_COMPRESSION): New.
20235         (TARGET_CPU_CPP_BUILTINS): Update macro
20236         to use new compression flags and to support microMIPS.
20237         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
20238         (MIPS_ARCH_FLOAT_SPEC): Likewise.
20239         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
20240         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
20241         (ASM_SPEC): Support mmicromips and mno-micromips.
20242         (M16STORE_REG_P): New macro.
20243         (MIPS_CALL): Support TARGET_MICROMIPS.
20244         (MICROMIPS_J): New macro.
20245         (mips_base_mips16): Rename this...
20246         (mips_base_compression_flags): ...to this.
20247         (UMIPS_12BIT_OFFSET_P): New macro.
20248         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
20249         (MULTILIB_DIRNAMES): Likewise.
20251 2013-02-25  Tom de Vries  <tom@codesourcery.com>
20253         PR rtl-optimization/56131
20254         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
20255         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
20256         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
20258 2013-02-25  Tobias Burnus  <burnus@net-b.de>
20260         * doc/invoke.texi (-fsanitize=): Move from optimization
20261         to debugging options.
20263 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
20265         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
20267 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
20268             Alexander Monakov  <amonakov@ispras.ru>
20270         PR middle-end/56077
20271         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
20272         flush pending lists also on non-jumps.  Adjust comment.
20274 2013-02-24  Catherine Moore  <clm@codesourcery.com>
20275             Maciej W. Rozycki  <macro@codesourcery.com>
20276             Tom de Vries  <tom@codesourcery.com>
20277             Nathan Sidwell  <nathan@codesourcery.com>
20278             Iain Sandoe  <iain@codesourcery.com>
20279             Nathan Froyd  <froydnj@codesourcery.com>
20280             Chao-ying Fu  <fu@mips.com>
20282         * doc/extend.texi: (micromips, nomicromips, nocompression):
20283         Document new function attributes.
20284         * doc/invoke.texi (minterlink-compressed, mmicromips,
20285         m14k, m14ke, m14kec): Document new options.
20286         (minterlink-mips16): Update documentation.
20287         * doc/md.texi (ZC, ZD): Document new constraints.
20288         * configure.ac (gcc_cv_as_micromips): Check if linker
20289         supports the .set micromips directive.
20290         * configure: Regenerate.
20291         * config.in: Regenerate.
20292         * config/mips/mips-tables.opt: Regenerate.
20293         * config/mips/micromips.md: New file.
20294         * constraints.md (ZC, AD): New constraints.
20295         * config/mips/predicates.md (movep_src_register): New predicate.
20296         (movep_src_operand): New predicate.
20297         (non_volatile_mem_operand): New predicate.
20298         * config/mips/mips.md (multimem): New type.
20299         (length): Differentiate between 17-bit and 18-bit branch offsets.
20300         (MOVEP1, MOVEP2): New mode iterator.
20301         (mov_<load>l): Use ZC constraint.
20302         (mov_<load>r): Likewise.
20303         (mov_<store>l): Likewise.
20304         (mov_<store>r): Likewise.
20305         (*branch_equality<mode>_inverted): Add microMIPS support.
20306         (*branch_equality<mode>): Likewise.
20307         (*jump_absolute): Likewise.
20308         (indirect_jump_<mode>): Likewise.
20309         (tablejump_<mode>): Likewise.
20310         (<optab>_internal): Likewise.
20311         (sibcall_internal): Likewise.
20312         (sibcall_value_internal): Likewise.
20313         (prefetch): Use constraint ZD.
20314         * config/mips/mips.opt (minterlink-compressed): New option.
20315         (minterlink-mips16): Now an alias for minterlink-compressed.
20316         (mmicromips): New option.
20317         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
20318         (compare_and_swap_12): Likewise.
20319         (sync_add<mode>): Likewise.
20320         (sync_<optab>_12): Likewise.
20321         (sync_old_<optab>_12): Likewise.
20322         (sync_new_<optab>_12): Likewise.
20323         (sync_nand_12): Likewise.
20324         (sync_old_nand_12): Likewise.
20325         (sync_new_nand_12): Likewise.
20326         (sync_sub<mode>): Likewise.
20327         (sync_old_add<mode>): Likewise.
20328         (sync_old_sub<mode>): Likewise.
20329         (sync_new_add<mode>): Likewise.
20330         (sync_new_sub<mode>): Likewise.
20331         (sync_<optab><mode>): Likewise.
20332         (sync_old_<optab><mode>): Likewise.
20333         (sync_new_<optab><mode>): Likewise.
20334         (sync_nand<mode>): Likewise.
20335         (sync_old_nand<mode>): Likewise.
20336         (sync_new_nand<mode>): Likewise.
20337         (sync_lock_test_and_set<mode>): Likewise.
20338         (test_and_set_12): Likewise.
20339         (atomic_compare_and_swap<mode>): Likewise.
20340         (atomic_exchange<mode>_llsc): Likewise.
20341         (atomic_fetch_add<mode>_llsc): Likewise.
20342         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
20343         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
20344         (umips_save_restore_pattern_p): Likewise.
20345         (umips_load_store_pair_p): Likewise.
20346         (umips_output_load_store_pair): Likewise.
20347         (umips_movep_target_p): Likewise.
20348         (umips_12bit_offset_address_p): Likewise.
20349         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
20350         (mips_base_mips16): Rename this...
20351         (mips_base_compression_flags): ...to this. Update all uses.
20352         (mips_attribute_table): Add micromips, nomicromips and nocompression.
20353         (mips_mips16_decl_p): Delete.
20354         (mips_nomips16_decl_p): Delete.
20355         (mips_get_compress_on_flags): New function.
20356         (mips_get_compress_off_flags): New function.
20357         (mips_get_compress_mode): New function.
20358         (mips_get_compress_on_name): New function.
20359         (mips_get_compress_off_name): New function.
20360         (mips_insert_attributes): Support multiple compression types.
20361         (mips_merge_decl_attributes): Likewise.
20362         (umips_12bit_offset_address_p): New function.
20363         (mips_start_function_definition): Emit .set micromips directive.
20364         (mips_call_may_need_jalx_p): New function.
20365         (mips_function_ok_for_sibcall): Add microMIPS support.
20366         (mips_print_operand_punctuation): Support short delay slots and
20367         compact jumps.
20368         (umips_swm_mask, umips_swm_encoding): New.
20369         (umips_build_save_restore): New function.
20370         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
20371         (was_mips16_p): Remove.
20372         (old_compression_mode): New.
20373         (mips_set_compression_mode): New function.
20374         (mips_set_current_function): Add microMIPS support.
20375         (mips_option_override): Likewise.
20376         (umips_save_restore_pattern_p): New function.
20377         (umips_output_save_restore): New function.
20378         (umips_load_store_pair_p_1): New function.
20379         (umips_load_store_pair_p): New function.
20380         (umips_output_load_store_pair_1): New function.
20381         (umips_output_load_store_pair): New function.
20382         (umips_movep_target_p) New function.
20383         (mips_prepare_pch_save): Add microMIPS support.
20384         * config/mips/mips.h (TARGET_COMPRESSION): New.
20385         (TARGET_CPU_CPP_BUILTINS): Update macro
20386         to use new compression flags and to support microMIPS.
20387         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
20388         (MIPS_ARCH_FLOAT_SPEC): Likewise.
20389         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
20390         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
20391         (ASM_SPEC): Support mmicromips and mno-micromips.
20392         (M16STORE_REG_P): New macro.
20393         (MIPS_CALL): Support TARGET_MICROMIPS.
20394         (MICROMIPS_J): New macro.
20395         (mips_base_mips16): Rename this...
20396         (mips_base_compression_flags): ...to this.
20397         (UMIPS_12BIT_OFFSET_P): New macro.
20398         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
20399         (MULTILIB_DIRNAMES): Likewise.
20401 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
20403         PR target/52555
20404         * target-globals.c (save_target_globals): For init_reg_sets and
20405         target_reinit remporarily set this_fn_optabs to this_target_optabs.
20407 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
20409         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
20410         * config/aarch64/t-aarch64
20411         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
20413 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
20415         PR inline-asm/56148
20416         * lra-constraints.c (process_alt_operands): Reload operand
20417         conflicting with earlier clobber only if no more other conflicting
20418         operands.
20420 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
20422         PR sanitizer/56393
20423         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
20424         if not linking a shared library.
20426 2013-02-22  Seth LaForge  <sethml@google.com>
20428         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
20430 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
20432         * config/arm/arm.md (split for extendsidi): Update condition.
20433         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
20434         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
20435         (qhs_zextenddi_cstr): Likewise.
20437 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
20439         PR middle-end/56420
20440         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
20441         avoid signed wrapping.
20442         (expand_mult): Handle properly multiplication by
20443         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
20444         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
20445         in the compiler if coeff is HOST_WIDE_INT_MIN.
20446         (expand_divmod): Don't make ext_op1 static, change it's type to
20447         uhwi.  Avoid undefined behavior in -INTVAL (op1).
20449         PR rtl-optimization/50339
20450         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
20451         field.
20452         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
20453         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
20454         into splitting_ashiftrt field.
20455         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
20456         ASHIFTRT.
20457         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
20458         choices.
20460 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
20462         PR middle-end/56108
20463         * trans-mem.c (execute_tm_mark): Do not expand transactions that
20464         are sure to go irrevocable.
20466 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
20468         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
20469         scalars are valid operands.
20471 2013-02-21  Martin Jambor  <mjambor@suse.cz>
20473         PR tree-optimization/56310
20474         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
20475         only matching indices and non-negative final offsets.
20476         (intersect_aggregates_with_edge): Pass src_idx to
20477         agg_replacements_to_vector.  Pass src_idx insstead of index to
20478         intersect_with_agg_replacements.
20480 2013-02-21  Martin Jambor  <mjambor@suse.cz>
20482         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
20483         instead of hard-wired defaults.
20485 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
20487         * doc/invoke.texi (MIPS Options): Update documentation of the
20488         floating-point multiply-accumulate instruction restrictions.
20490 2013-02-21  Kostya Serebryany  <kcc@google.com>
20492         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
20493         asan_shadow_offset on x86_64 linux.
20495 2013-02-21  Richard Biener  <rguenther@suse.de>
20497         PR tree-optimization/56415
20498         Revert
20499         2013-02-11  Richard Biener  <rguenther@suse.de>
20501         PR tree-optimization/56273
20502         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20503         first VRP run.
20505 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
20507         PR bootstrap/56258
20508         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
20509         instead of @itemx.
20511         PR inline-asm/56405
20512         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
20513         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
20515 2013-02-20  Jan Hubicka  <jh@suse.cz>
20517         PR tree-optimization/56265
20518         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
20519         when target is referenced for first time.
20521 2013-02-20  Richard Biener  <rguenther@suse.de>
20523         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
20524         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
20525         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
20526         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
20527         not return anything.
20528         (rename_ssa_copies): Do not remove unused locals.
20529         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
20530         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
20531         * passes.c (execute_function_todo): Do not schedule unused locals
20532         removal if cleanup_tree_cfg did something.
20533         * tree-ssa-live.c (remove_unused_locals): Dump statistics
20534         about the number of removed locals.
20536 2013-02-20  Richard Biener  <rguenther@suse.de>
20538         PR tree-optimization/56398
20539         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
20541 2013-02-20  Martin Jambor  <mjambor@suse.cz>
20543         PR tree-optimization/55334
20544         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
20545         restricted pointers to arrays.
20547 2013-02-20  Richard Biener  <rguenther@suse.de>
20548             Jakub Jelinek  <jakub@redhat.com>
20550         PR tree-optimization/56396
20551         * tree-ssa-ccp.c (n_const_val): New static variable.
20552         (get_value): Return NULL for SSA names we don't have a lattice
20553         entry for.
20554         (ccp_initialize): Initialize n_const_val.
20555         * tree-ssa-copy.c (n_copy_of): New static variable.
20556         (init_copy_prop): Initialize n_copy_of.
20557         (get_value): Return NULL_TREE for SSA names we don't have a
20558         lattice entry for.
20560 2013-02-20  Martin Jambor  <mjambor@suse.cz>
20562         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
20564 2013-02-20  Richard Biener  <rguenther@suse.de>
20566         * genpreds.c (write_lookup_constraint): Do not compare first
20567         letter of the constraint again.
20569 2013-02-20  Richard Biener  <rguenther@suse.de>
20571         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
20572         and ceil_log2.
20573         (get_use_iv_cost): Terminate hashtable walk when coming across
20574         an empty entry.
20576 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
20578         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
20579         reassociation for avx2 targets.
20581 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20583         * config/microblaze/microblaze.c: microblaze_has_clz = 0
20584         Add version check for v8.10.a to enable microblaze_has_clz
20585         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
20586         version and TARGET_PATTERN_COMPARE check
20587         * config/microblaze/microblaze.md: New clzsi2 instruction
20589 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20591         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
20592         function before branching.
20594 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
20596         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
20597         DUMP_INSN_RTX_UID.
20598         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
20600 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
20602         PR middle-end/55889
20603         * sel-sched.c: Include ira.h.
20604         (implicit_clobber_conflict_p): New function.
20605         (moveup_expr): Use it.
20606         * Makefile.in (sel-sched.o): Depend on ira.h.
20608 2013-02-19  Richard Biener  <rguenther@suse.de>
20610         PR tree-optimization/56384
20611         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
20612         (vn_hash_type): Split out from ...
20613         (vn_hash_constant_with_type): ... here.
20614         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
20615         (vn_phi_eq): Compare types from vn_phi_s structure.
20616         (vn_phi_lookup): Populate vn_phi_s type.
20617         (vn_phi_insert): Likewise.
20619 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
20621         PR tree-optimization/56350
20622         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
20623         if haven't found reduction or nested cycle operand, rather than
20624         asserting we must find it.
20626         PR tree-optimization/56381
20627         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
20628         to fold_build3.
20630 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
20631             Jakub Jelinek  <jakub@redhat.com>
20633         PR target/52555
20634         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
20635         (swap_optab_enable): Same.
20636         (init_all_optabs): Use argument instead of global.
20637         * tree.h (struct tree_optimization_option): New field target_optabs.
20638         * expr.h (init_all_optabs): Add argument to prototype.
20639         (TREE_OPTIMIZATION_OPTABS): New.
20640         (save_optabs_if_changed): Protoize.
20641         * optabs.h: Declare this_fn_optabs.
20642         * optabs.c (save_optabs_if_changed): New.
20643         Declare this_fn_optabs.
20644         (init_optabs): Add argument to init_all_optabs() call.
20645         * function.c (invoke_set_current_function_hook): Handle per
20646         function optabs.
20647         * function.h (struct function): New field optabs.
20648         * config/mips/mips.c (mips_set_mips16_mode): Handle when
20649         optimization_current_node has changed.
20650         * target-globals.h (save_target_globals_default_opts): Protoize.
20651         * target-globals.c (save_target_globals_default_opts): New.
20653 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20655         PR target/56347
20656         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
20657         registers %fr12 and %fr12R as call used.
20659         PR target/56214
20660         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
20661         and HImode, require all displacements to be an integer multiple of
20662         their mode size.
20663         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
20664         only allow QImode and HImode when reload is in progress and strict is
20665         true.  Likewise for symbolic addresses.  Use base14_operand to check
20666         displacements in REG+BASE addresses.
20668 2013-02-18  Richard Biener  <rguenther@suse.de>
20670         PR tree-optimization/56366
20671         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20672         handle sign-conversion of outer-loop initial induction value.
20674 2013-02-18  Richard Biener  <rguenther@suse.de>
20676         PR middle-end/56349
20677         * cfghooks.c (merge_blocks): If we merge a latch into another
20678         block adjust references to it.
20679         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
20680         (verify_loop_structure): Verify that a recorded latch is in fact
20681         a latch.
20683 2013-02-18  Richard Biener  <rguenther@suse.de>
20685         PR tree-optimization/56321
20686         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
20687         order SSA name release and virtual operand unlinking.
20689 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20691         * config/microblaze/microblaze.md (save_stack_block): Define.
20692         (restore_stack_block): Likewise.
20694 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20696         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
20697         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
20698         * config/microblaze/microblaze.c (microblaze_option_override):
20699         Bail out early for PIC modes when target does not support PIC.
20701 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20703         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
20704         Replace with a microblaze version.
20705         (microblaze_trampoline_init): Adapt for microblaze.
20706         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
20707         microblaze.
20709 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
20710             Dodji Seketeli  <dodji@redhat.com>
20712         PR asan/56330
20713         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
20714         (instrument_mem_region_access): Do not forget to always put
20715         instrumentation of the of 'base' and 'base + len' in a "if (len !=
20716         0) statement, even for cases where either 'base' or 'base + len'
20717         are not instrumented -- because they have been previously
20718         instrumented.  Simplify the logic by putting all the statements
20719         instrument 'base + len' inside a sequence, and then insert that
20720         sequence right before the current insertion point.  Then, to
20721         instrument 'base + len', just get an iterator on that statement.
20722         And do not forget to update the pointer to iterator the function
20723         received as argument.
20725 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
20727         PR rtl-optimization/56348
20728         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
20730 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
20732         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
20733         (clean_graph_dump_file): Pass base to start_graph_dump.
20735 2013-02-14  Richard Henderson  <rth@redhat.com>
20737         PR target/55941
20738         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
20740 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
20742         * collect2-aix.h: Define F_LOADONLY.
20744 2013-02-14  Richard Biener  <rguenther@suse.de>
20746         PR lto/50494
20747         * varasm.c (output_constant_def_1): Get the decl representing
20748         the constant as argument.
20749         (output_constant_def): Wrap output_constant_def_1.
20750         (make_decl_rtl): Use output_constant_def_1 with the decl
20751         representing the constant.
20752         (build_constant_desc): Optionally re-use a decl already
20753         representing the constant.
20754         (tree_output_constant_def): Adjust.
20756 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
20758         Fix an asan crash
20759         * asan.c (instrument_builtin_call):  Really put the length of the
20760         second source argument into src1_len.
20762 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
20764         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
20765         argument.  If it is false, don't create edge from then_bb to
20766         fallthru_bb.
20767         (insert_if_then_before_iter): Pass true to it.
20768         (build_check_stmt): Pass false to it.
20769         (transform_statements): Flush hash table only on extended basic
20770         block boundaries, rather than at the beginning of every bb.
20771         Don't flush hash table on nonfreeing_call_p calls.
20772         * tree-flow.h (nonfreeing_call_p): New prototype.
20773         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
20775 2013-02-13  David S. Miller  <davem@davemloft.net>
20777         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
20779 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
20781         PR target/56184
20782         * ira.c (max_regno_before_ira): Move from ...
20783         (ira): ... here.
20784         (fix_reg_equiv_init): Use max_regno_before_ira instead of
20785         vec_safe_length.
20787 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
20789         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
20791 2013-02-13  Richard Biener  <rguenther@suse.de>
20793         PR lto/56295
20794         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
20795         globals in MEM_REFs.
20797 2013-02-13  Richard Biener  <rguenther@suse.de>
20799         * loop-init.c (loop_optimizer_init): Clear loop state when
20800         re-initializing preserved loops.
20801         * loop-unswitch.c (unswitch_single_loop): Return whether
20802         we unswitched the loop.  Do not verify loop state here.
20803         (unswitch_loops): When we unswitched a loop discover new loops.
20805 2013-02-13  Kostya Serebryany  <kcc@google.com>
20807         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
20808         on x86_64 linux.
20809         * sanitizer.def: Rename __asan_init to __asan_init_v1.
20811 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
20813         Avoid instrumenting duplicated memory access in the same basic block
20814         * Makefile.in (asan.o): Add new dependency on hash-table.h
20815         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
20816         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
20817         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
20818         (free_mem_ref_resources, has_mem_ref_been_instrumented)
20819         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
20820         (get_mem_ref_of_assignment): New functions.
20821         (get_mem_refs_of_builtin_call): Extract from
20822         instrument_builtin_call and tweak a little bit to make it fit with
20823         the new signature.
20824         (instrument_builtin_call): Use the new
20825         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
20826         of is_gimple_builtin_call.
20827         (instrument_derefs, instrument_mem_region_access): Insert the
20828         instrumented memory reference into the hash table.
20829         (maybe_instrument_assignment): Renamed instrument_assignment into
20830         this, and change it to advance the iterator when instrumentation
20831         actually happened and return true in that case.  This makes it
20832         homogeneous with maybe_instrument_assignment, and thus give a
20833         chance to callers to be more 'regular'.
20834         (transform_statements): Clear the memory reference hash table
20835         whenever we enter a new BB, when we cross a function call, or when
20836         we are done transforming statements.  Use
20837         maybe_instrument_assignment instead of instrumentation.  No more
20838         need to special case maybe_instrument_assignment and advance the
20839         iterator after calling it; it's now handled just like
20840         maybe_instrument_call.  Update comment.
20842 2013-02-13  Richard Biener  <rguenther@suse.de>
20844         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
20845         Fix loop discovery code.
20847 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
20849         PR inline-asm/56148
20850         * lra-constraints.c (process_alt_operands): Match early clobber
20851         operand with itself.  Check conflicts with earlyclobber only if
20852         the operand is not reloaded.  Prefer to reload conflicting operand
20853         if earlyclobber and matching operands are the same.
20855 2013-02-12  Richard Biener  <rguenther@suse.de>
20857         PR lto/56297
20858         * lto-streamer-out.c (write_symbol): Do not output symbols
20859         for hard register variables.
20861 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
20863         PR target/54222
20864         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
20865         (umulsidi3_insn, mulsidi3_insn): New insns.
20867 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
20869         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
20870         (struct tune_params): Add vec_costs field.
20871         * config/arm/arm.c (arm_builtin_vectorization_cost)
20872         (arm_add_stmt_cost): New functions.
20873         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
20874         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
20875         (arm_default_vec_cost): New struct of type cpu_vec_costs.
20876         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
20877         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
20878         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
20879         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
20881 2013-02-12  Richard Biener  <rguenther@suse.de>
20883         PR lto/56295
20884         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
20885         decls again if possible.
20887 2013-02-12  Richard Biener  <rguenther@suse.de>
20889         PR middle-end/56288
20890         * tree-ssa.c (verify_ssa_name): Fix check, move
20891         SSA_NAME_IN_FREE_LIST check up.
20893 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
20894             Steven Bosscher   <steven@gcc.gnu.org>
20896         PR rtl-optimization/56151
20897         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
20898         equal to op0 or op1, and last_insn pattern is CODE operation
20899         with MEM dest and one of the operands matches that MEM.
20901 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
20903         * doc/extend.texi: Document Function Multiversioning and "default"
20904         parameter string to target attribute.
20905         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
20906         target attribute parameter is "default".
20907         (ix86_compare_version_priority): Remove checks for target attribute.
20908         (ix86_mangle_function_version_assembler_name): Change error to sorry.
20909         Remove check for target attribute equal to NULL. Add assert.
20910         (ix86_generate_version_dispatcher_body): Change error to sorry.
20912 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
20913             Jack Howarth  <howarth@bromo.med.uc.edu>
20914             Patrick Marlier  <patrick.marlier@gmail.com>
20916         PR libitm/55693
20917         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
20918         define ENDFILE_SPEC as TM_DESTRUCTOR.
20919         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
20921 2013-02-11  Alexander Potapenko  <glider@google.com>
20922             Jack Howarth  <howarth@bromo.med.uc.edu>
20923             Jakub Jelinek  <jakub@redhat.com>
20925         PR sanitizer/55617
20926         * config/darwin.c (cdtor_record): Rename ctor_record.
20927         (sort_cdtor_records): Rename sort_ctor_records.
20928         (finalize_dtors): New routine to sort destructors by
20929         priority before use in assemble_integer.
20930         (machopic_asm_out_destructor): Use finalize_dtors if needed.
20932 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
20934         PR rtl-optimization/56275
20935         * simplify-rtx.c (avoid_constant_pool_reference): Check that
20936         offset is non-negative and less than cmode size before
20937         calling simplify_subreg.
20939 2013-02-11  Richard Biener  <rguenther@suse.de>
20941         PR tree-optimization/56264
20942         * cfgloop.h (fix_loop_structure): Adjust prototype.
20943         * loop-init.c (fix_loop_structure): Return the number of
20944         newly discovered loops.
20945         * tree-cfgcleanup.c (repair_loop_structures): When new loops
20946         are discovered, do a full loop-closed SSA rewrite.
20948 2013-02-11  Richard Biener  <rguenther@suse.de>
20950         PR tree-optimization/56273
20951         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20952         first VRP run.
20953         (check_array_ref): Fix missing newline in dumps.
20954         (search_for_addr_array): Likewise.
20956 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
20958         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
20960 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
20962         PR target/56256
20963         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
20965 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
20967         PR rtl-optimization/56246
20968         * lra-constraints.c (simplify_operand_subreg): Try to reuse
20969         reload pseudo.
20970         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
20971         constraints are satisfied.
20973 2013-02-08  Jeff Law  <law@redhat.com>
20975         PR debug/53948
20976         * emit-rtl.c (reg_is_parm_p): New function.
20977         * regs.h (reg_is_parm_p): New prototype.
20978         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
20979         callee-clobbered registers.
20981 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
20983         PR target/56043
20984         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
20985         If there is no implicit builtin declaration, just return NULL.
20987 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20989         * config/i386/sse.md (FMAMODEM): New mode iterator.
20990         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
20991         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
20993 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20995         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
20996         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
20997         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
20999 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21001         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
21002         (microblaze*-*-elf): Likewise.
21003         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
21004         LINK_SPEC.
21005         * config/microblaze/microblaze-c.c: Add builtin defines for
21006         _LITTLE_ENDIAN and _BIG_ENDIAN.
21007         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
21008         add to TARGET_DEFAULT flags.
21009         Expand ASM_SPEC and LINK_SPEC.
21010         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
21011         * config/microblaze/microblaze.md: Update extendsidi2 and
21012         movdi_internal instructions to use low-order / high-order reg
21013         print_operands.
21014         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
21015         options and inversemask / mask of LITTLE_ENDIAN.
21016         * config/microblaze/t-microblaze: Expand multilib options to
21017         include mlittle-endian (le) and update exceptions patterns.
21019 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
21021         PR rtl-optimization/56195
21022         * lra-constraints.c (get_reload_reg): Don't reuse regs
21023         if they have smaller mode than requested, if they have
21024         wider mode than requested, try to return a SUBREG.
21026         PR tree-optimization/56250
21027         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
21028         if type is unsigned and code isn't MULT_EXPR.
21030 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
21032         PR tree-optimization/56064
21033         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
21034         bits according to mode.
21035         * fixed-value.h (fixed_from_double_int)
21036         (const_fixed_from_double_int): Adjust comments.
21038 2013-02-08  Richard Biener  <rguenther@suse.de>
21040         PR lto/56231
21041         * lto-streamer.h (struct data_in): Remove current_file, current_line
21042         and current_col members.
21043         * lto-streamer-out.c (lto_output_location): Stream changed bits
21044         en-block for efficiency.
21045         * lto-streamer-in.c (clear_line_info): Remove.
21046         (lto_input_location): Cache current file, line and column
21047         globally via local statics.  Read changed bits en-block.
21048         (input_function): Do not call clear_line_info.
21049         (lto_read_body): Likewise.
21050         (lto_input_toplevel_asms): Likewise.
21052 2013-02-08  Michael Matz  <matz@suse.de>
21054         PR tree-optimization/52448
21055         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
21056         (nt_call_phase): New static.
21057         (add_or_mark_expr): Only mark accesses with newer phase than any
21058         call seen.
21059         (nonfreeing_call_p): New.
21060         (nt_init_block): Update nt_call_phase, mark blocks as visited.
21061         (nt_fini_block): Keep blocks marked as visited.
21062         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
21064 2013-02-08  Richard Biener  <rguenther@suse.de>
21066         * ira.c (ira): Free broken dominator information.
21068 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
21070         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
21072 2013-02-08  Marek Polacek  <polacek@redhat.com>
21074         * cfgloop.c (verify_loop_structure): Add more checking of headers.
21076 2013-02-08  Richard Biener  <rguenther@suse.de>
21078         PR middle-end/56181
21079         * cfgloop.h (flow_loops_find): Adjust.
21080         (bb_loop_header_p): Declare.
21081         * cfgloop.c (bb_loop_header_p): New function split out from ...
21082         (flow_loops_find): ... here.  Adjust function signature,
21083         support incremental loop structure update.
21084         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
21085         * cfgloopmanip.c (fix_loop_structure): Move ...
21086         * loop-init.c (fix_loop_structure): ... here.
21087         (apply_loop_flags): Split out from ...
21088         (loop_optimizer_init): ... here.
21089         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
21090         in incremental mode, only remove dead loops here.
21092 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
21094         PR target/54222
21095         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
21096         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
21097         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
21098         (*round<mode>3.libgcc): New insns for fixed-modes.
21099         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
21100         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
21101         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
21102         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
21103         implementations.  Define to __builtin_avr_absFX,
21104         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
21105         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
21106         __builtin_avr_countlsFX, respectively.
21107         * config/avr/avr-c.c (target.h): Include it.
21108         (enum avr_builtin_id): New enum.
21109         (avr_resolve_overloaded_builtin): New static function.
21110         (avr_register_target_pragmas): Use it to set
21111         targetm.resolve_overloaded_builtin.
21112         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
21113         tree nodes used by DEF_BUILTIN.
21114         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
21115         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
21116         <AVR_BUILTIN_xxBITS>: Same.
21118 2013-02-08  Richard Biener  <rguenther@suse.de>
21120         * cfgloop.c (verify_loop_structure): Properly handle
21121         a loop exiting to another loop header.
21122         * ira-int.h (ira_loops): Remove.
21123         * ira.c (ira_loops): Remove.
21124         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
21125         (do_reload): Use loop_optimizer_finalize.
21126         * ira-build.c (create_loop_tree_nodes): Use get_loops and
21127         number_of_loops to access the loop tree.
21128         (more_one_region_p): Likewise.
21129         (finish_loop_tree_nodes): Likewise.
21130         (rebuild_regno_allocno_maps): Likewise.
21131         (mark_loops_for_removal): Likewise.
21132         (mark_all_loops_for_removal): Likewise.
21133         (remove_unnecessary_regions): Likewise.
21134         (ira_build): Likewise.
21135         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
21137 2013-02-08  Richard Biener  <rguenther@suse.de>
21139         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
21140         * ipa-pure-const.c (analyze_function): Avoid calling
21141         mark_irreducible_loops twice.
21142         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
21144 2013-02-07  David S. Miller  <davem@davemloft.net>
21146         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
21147         on 'reg'.
21148         * var-tracking.c (vt_add_function_parameter): Test the presence of
21149         HAVE_window_save properly and do not remap argument registers when
21150         we have a leaf function.
21152 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
21154         PR bootstrap/56227
21155         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
21156         instead of "ll".
21157         * config/i386/i386.c (ix86_print_operand): Ditto.
21159 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
21161         * lra-constraints.c (process_alt_operands): Fix recently added comment.
21163 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
21165         PR rtl-optimization/56225
21166         * lra-constraints.c (process_alt_operands): Check that reload hard
21167         reg can hold value for strict_low_part.
21169 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
21171         PR debug/56154
21172         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
21173         dwarf2out_end_function.
21174         (in_first_function_p, maybe_at_text_label_p,
21175         first_loclabel_num_not_at_text_label): New variables.
21176         (dwarf2out_var_location): In the first function find out
21177         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
21178         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
21179         functions.
21181 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
21183         PR rtl-optimization/56178
21184         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
21185         SUBREG of a register.  Tidy up related block of code.
21186         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
21187         note if the source is a register or a SUBREG of a register.
21189 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
21191         PR target/56228
21192         * config/rs6000/rs6000.md (ptrm): New mode attr.
21193         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
21194         call_value_indirect_aix<pttrsize>,
21195         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
21196         m in constraints.
21198 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
21200         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
21201         if -bnortl. Convert to strcmp and strncmp.
21203 2013-02-07  Alan Modra  <amodra@gmail.com>
21205         PR target/54009
21206         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
21207         addresses won't wrap when offsetting.
21208         (rs6000_secondary_reload): Provide secondary reloads needed for
21209         wrapping LO_SUM addresses.
21211 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
21213         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
21214         MACH, just __MACH__.
21216 2013-02-06  Richard Biener  <rguenther@suse.de>
21218         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
21219         instead of calling fix_loop_structure.
21221 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
21223         PR middle-end/56217
21224         * omp-low.c (use_pointer_for_field): Return false if
21225         lower_send_shared_vars doesn't generate any copy-out code.
21227 2013-02-06  Tom de Vries  <tom@codesourcery.com>
21229         PR rtl-optimization/56131
21230         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
21231         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
21232         of the label is NULL.  Add comment.
21234 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
21236         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
21238         PR sanitizer/55374
21239         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
21240         (STATIC_LIBTSAN_LIBS): Likewise.
21241         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
21242         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
21243         is defined, don't add anything else beyond that.
21244         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
21245         (LINK_COMMAND_SPEC): Use them.
21247         PR tree-optimization/56205
21248         * tree-stdarg.c (check_all_va_list_escapes): Return true if
21249         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
21250         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
21252 2013-02-05  Richard Biener  <rguenther@suse.de>
21254         PR tree-optimization/53342
21255         PR tree-optimization/53185
21256         * tree-vectorizer.h (vect_check_strided_load): Remove.
21257         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
21258         not disallow peeling for vectorized strided loads.
21259         (vect_check_strided_load): Make static and simplify.
21260         (vect_analyze_data_refs): Adjust.
21261         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
21262         correctly when vectorizing strided loads.
21264 2013-02-05  Richard Biener  <rguenther@suse.de>
21266         * doc/install.texi: Refer to ISL, not PPL.
21268 2013-02-05  Jan Hubicka  <jh@suse.cz>
21270         PR tree-optimization/55789
21271         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
21273 2013-02-05  Jan Hubicka  <jh@suse.cz>
21275         PR tree-optimization/55789
21276         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
21277         the dead call anyway.
21279 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
21281         PR sanitizer/55374
21282         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
21284 2013-02-04  Alexander Potapenko  <glider@google.com>
21285             Jack Howarth  <howarth@bromo.med.uc.edu>
21286             Jakub Jelinek  <jakub@redhat.com>
21288         PR sanitizer/55617
21289         * config/darwin.c (sort_ctor_records): Stabilized qsort
21290         on constructor priority by using original position.
21291         (finalize_ctors): New routine to sort constructors by
21292         priority before use in assemble_integer.
21293         (machopic_asm_out_constructor): Use finalize_ctors if needed.
21295 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
21297         PR libstdc++/54314
21298         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
21299         about visibility on artificial decls.
21300         * config/sol2.c (solaris_assemble_visibility): Likewise.
21302 2013-02-04  Kai Tietz  <ktietz@redhat.com>
21304         PR target/56186
21305         * config/i386/i386.c (function_value_ms_64): Add additional valtype
21306         argument and improve checking of return-argument types for 16-byte
21307         modes.
21308         (ix86_function_value_1): Add additional valtype argument on call
21309         of function_value_64.
21310         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
21311         handling infunction_value_64 function.
21313 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
21315         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
21317 2013-02-04  Richard Biener  <rguenther@suse.de>
21319         PR tree-optimization/56188
21320         * tree-ssa-structalias.c (label_visit): Consider case with
21321         initially non-empty points-to set.
21322         (perform_var_substitution): Dump node mapping and clean up.
21324 2013-02-04  Richard Guenther  <rguenther@suse.de>
21326         PR lto/56168
21327         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
21328         node prevail as last resort.
21329         (lto_symtab_merge_decls): Remove guard on LTRANS here.
21330         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
21332 2013-02-04  Richard Biener  <rguenther@suse.de>
21334         PR tree-optimization/56113
21335         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
21336         Merge into ...
21337         (equiv_class_lookup_or_add): ... this.
21338         (label_visit): Adjust and fix error in previous patch.
21339         (perform_var_substitution): Adjust.
21341 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
21343         * config/sh/divtab.c: Fix formatting and comments throughout the file.
21344         * config/sh/sh4-300.md: Likewise.
21345         * config/sh/sh4a.md: Likewise.
21346         * config/sh/constraints.md: Likewise.
21347         * config/sh/sh.md: Likewise.
21348         * config/sh/netbsd-elf.h: Likewise.
21349         * config/sh/predicates.md: Likewise.
21350         * config/sh/sh-protos.h: Likewise.
21351         * config/sh/ushmedia.h: Likewise.
21352         * config/sh/linux.h: Likewise.
21353         * config/sh/sh.c: Likewise.
21354         * config/sh/superh.h: Likewise.
21355         * config/sh/elf.h: Likewise.
21356         * config/sh/sh4.md: Likewise.
21357         * config/sh/sh.h: Likewise.
21359 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21361         * config/pa/constraints.md: Adjust unused letters.  Change "T"
21362         constraint to match_test floating_point_store_memory_operand().
21363         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
21364         (base14_operand): New.
21365         (floating_point_store_memory_operand): New.
21366         (integer_store_memory_operand): Revise to use base14_operand and
21367         reg_plus_base_memory_operand.
21368         (move_dest_operand): Allow symbolic_memory_operands.
21369         (symbolic_memory_operand): Check for LO_SOM.
21370         (symbolic_operand): Change default case to break.
21371         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
21372         CONST_DOUBLE values to be reloaded by putting them into memory when
21373         the destination is a floating point register.
21374         (movdf): Remove code to handle CONST_DOUBLE.
21375         (movsf): Likewise.
21376         (reload_indf_r1): New.
21377         (reload_insf_r1): New.
21378         Consistently use "Q" and "T" constraints with integer and floating
21379         point move instructions, respectively.
21380         (movdi): Remove FAIL.
21381         Change predicate for source operand unamed DImode move from
21382         general_operand to move_src_operand.
21383         (umulsidi3): Change predicate for destination operand to
21384         register_operand.
21385         Likewise for similar unamed patterns.
21386         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
21387         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
21388         (hppa_legitimize_address): Simplify mask calculation.
21389         (pa_emit_move_sequence): Revised handling of secondary reloads from
21390         REG+D addresses for floating point loads and stores.  Directly handle
21391         loading CONST0_RTX (mode) to a floating point register.
21392         (pa_secondary_reload): Handle reloading DF and SFmode constant values
21393         to floating point registers.  Don't restrict secondary reloads to
21394         floating point registers to integer modes.  Revise some comments and
21395         cleanup some code.
21396         (TARGET_LEGITIMATE_ADDRESS_P): Define.
21397         (pa_legitimate_address_p): New.
21398         (pa_legitimize_reload_address): New.
21399         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
21400         (STRICT_REG_OK_FOR_BASE_P): New.
21401         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
21402         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
21404 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
21405             Andrew Dixie  <andrewd@gentrack.com>
21407         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
21408         flag set.
21410 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
21412         * expmed.c (extract_bit_field_1): Pass the full width of the
21413         structure to get_best_reg_extraction_insn.
21415 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
21417         PR target/54601
21418         * configure.ac (use_cxa_atexit): Add AIX.
21419         * configure: Regenerate.
21421         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
21423 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
21425         PR debug/54793
21426         * final.c (need_profile_function): New variable.
21427         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
21428         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
21429         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
21430         notes, targetm.asm_out.function_prologue doesn't emit anything,
21431         HAVE_prologue and profiler should be emitted before prologue,
21432         set need_profile_function instead of emitting it.
21433         (final_scan_insn): If need_profile_function, emit
21434         profile_function on the first NOTE_INSN_BASIC_BLOCK or
21435         NOTE_INSN_FUNCTION_BEG note.
21437 2013-02-01  Richard Henderson  <rth@redhat.com>
21439         * config/rs6000/rs6000.md (smulditi3): New.
21440         (umulditi3): New.
21442         * config/alpha/alpha.md (umulditi3): New.
21444 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
21446         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
21447         (ASM_OUTPUT_ALIGNED_LOCAL): New.
21449 2013-02-01  Richard Biener  <rguenther@suse.de>
21451         PR tree-optimization/56113
21452         * tree-ssa-structalias.c (label_visit): Reduce work for
21453         single-predecessor nodes.
21455 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
21457         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
21458         range isn't testing for zero.
21460 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
21462         PR middle-end/56113
21463         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
21465 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
21466             Nick Clifton  <nickc@redhat.com>
21468         * config/v850/constraints.md (Q): Define as a memory constraint.
21469         * config/v850/predicates.md (label_ref_operand): New predicate.
21470         (e3v5_shift_operand): New predicate.
21471         (ior_operator): New predicate.
21472         * config/v850/t-v850: Add e3v5 multilib.
21473         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
21474         (v850_gen_movdi): Prototype.
21475         * config/v850/v850.c: Add support for e3v5 architecture.
21476         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
21477         TARGET_V850E_UP.
21478         (construct_save_jarl): Add e3v5 long JARL support.
21479         (v850_adjust_insn_length): New function.  Adjust length of call
21480         insns when using e3v5 instructions.
21481         (v850_gen_movdi): New function: Generate instructions to move a
21482         DImode value.
21483         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
21484         (CPP_SPEC): Define __v850e3v5__ as appropriate.
21485         (TARGET_USE_FPU): Enable for e3v5.
21486         (CONST_OK_FOR_W): New macro.
21487         (ADJUST_INSN_LENGTH): Define.
21488         * config/v850/v850.md (UNSPEC_LOOP): Define.
21489         (attr cpu): Add v850e3v5.
21490         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
21491         (movdi): New pattern.
21492         (movdi_internal): New pattern.
21493         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
21494         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
21495         (cstoresf4): Likewise.
21496         (cstoredf4): Likewise.
21497         (insv): New pattern.
21498         (rotlso3_a): New pattern.
21499         (rotlsi3_b): New pattern
21500         (rotlsi3_v850e3v5): New pattern.
21501         (doloop_begin): New pattern.
21502         (fix_loop_counter): New pattern.
21503         (doloop_end): New pattern.
21504         (branch_normal): Add e3v5 long branch support.
21505         (branch_invert): Likewise.
21506         (branch_z_normal): Likewise.
21507         (branch_z_invert): Likewise.
21508         (branch_nz_normal): Likewise.
21509         (branch_nz_invert): Likewise.
21510         (call_internal_short): Add e3v5 register-indirect JARL support.
21511         (call_internal_long): Likewise.
21512         (call_value_internal_short): Likewise.
21513         (call_value_internal_long): Likewise.
21514         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
21515         (mloop): New option.
21516         * config.gcc: Add support for configuring v840e3v5 target.
21517         * doc/invoke.texi: Document new v850 specific command line options.
21519 2013-01-31  Paul Koning  <ni1d@arrl.net>
21521         PR debug/55059
21522         PR debug/54508
21523         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
21524         children if parent is a class.
21525         (prune_unused_types_prune): Don't add DW_AT_declaration.
21527 2013-01-31  Richard Biener  <rguenther@suse.de>
21529         PR tree-optimization/56157
21530         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
21531         match up operand with SLP child.
21533 2013-01-31  Jason Merrill  <jason@redhat.com>
21535         PR debug/54410
21536         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
21537         parameters the first time.
21538         (gen_scheduled_generic_parms_dies): Check completeness here.
21540 2013-01-31  Richard Biener  <rguenther@suse.de>
21542         PR middle-end/53073
21543         * common.opt (faggressive-loop-optimizations): New flag,
21544         enabled by default.
21545         * doc/invoke.texi (faggressive-loop-optimizations): Document.
21546         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
21547         infer_loop_bounds_from_undefined by it.
21549 2013-01-31  Richard Biener  <rguenther@suse.de>
21551         PR tree-optimization/56150
21552         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
21553         visit virtual operands.
21554         (find_uses_to_rename_bb): Likewise.
21556 2013-01-31  Richard Biener  <rguenther@suse.de>
21558         PR tree-optimization/56150
21559         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
21560         mixed store non-store stmts.
21562 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
21564         PR sanitizer/55374
21565         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
21566         LIBASAN_EARLY_SPEC is defined.
21567         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
21568         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
21569         before %o.
21570         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
21572         PR c++/55742
21573         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
21574         invalid args instead of ICEing on it.
21575         (ix86_valid_target_attribute_tree): Return error_mark_node if
21576         ix86_valid_target_attribute_inner_p failed.
21577         (ix86_valid_target_attribute_p): Return false only if
21578         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
21579         target("default") attribute.
21580         (sorted_attr_string): Change argument from const char * to tree,
21581         merge in all target attribute arguments rather than just one.
21582         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
21583         instead of free.  Avoid using strcat.
21584         (ix86_mangle_function_version_assembler_name): Mangle
21585         target("default") as if no target attribute is present.  Adjust
21586         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
21587         instead of xmalloc and XDELETEVEC instead of free.
21588         (ix86_function_versions): Don't return true if one of the decls
21589         doesn't have target attribute.  If they don't and one of the decls
21590         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
21591         sorted_attr_string caller.  Use XDELETEVEC instead of free.
21592         (ix86_supports_function_versions): Remove.
21593         (make_name): Fix up formatting.
21594         (make_dispatcher_decl): Remove resolver_name and its initialization.
21595         Avoid leaking memory.
21596         (is_function_default_version): Return true if there is
21597         target("default") attribute rather than no target attribute at all.
21598         (make_resolver_func): Avoid leaking memory.
21599         (ix86_generate_version_dispatcher_body): Likewise.
21600         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
21601         * target.def (supports_function_versions): Remove.
21602         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
21603         * doc/tm.texi: Regenerated.
21605 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
21607         PR rtl-optimization/56144
21608         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
21609         for values with side effects.
21611 2013-01-30  Richard Biener  <rguenther@suse.de>
21613         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
21614         (sparseset_pop): Likewise.
21615         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
21616         to be able to use quick_push in the worker loop.
21618 2013-01-30  Marek Polacek  <polacek@redhat.com>
21620         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
21622 2013-01-30  Richard Biener  <rguenther@suse.de>
21624         PR lto/56147
21625         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
21627 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
21629         PR tree-optimization/56064
21630         * fixed-value.c (fixed_from_double_int): New function.
21631         * fixed-value.h (fixed_from_double_int): New prototype.
21632         (const_fixed_from_double_int): New static inline function.
21633         * fold-const.c (native_interpret_fixed): New static function.
21634         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
21635         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
21636         (native_encode_fixed): New static function.
21637         (native_encode_expr) <FIXED_CST>: Use it.
21638         (native_interpret_int): Move double_int worker code to...
21639         * double-int.c (double_int::from_buffer): ...this new static method.
21640         * double-int.h (double_int::from_buffer): Prototype it.
21642 2013-01-30  Richard Biener  <rguenther@suse.de>
21644         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
21645         New pointer-map and obstack.
21646         (init_alias_vars): Allocate pointer-map and obstack.
21647         (delete_points_to_sets): Free them.
21648         (find_what_var_points_to): Cache result.
21649         (find_what_p_points_to): Adjust for changed interface of
21650         find_what_var_points_to.
21651         (compute_points_to_sets): Likewise.
21652         (ipa_pta_execute): Likewise.
21654 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
21656         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
21657         * configure: Regenerate.
21658         * config.in: Regenerate.
21659         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
21660         #nobits/#progbits if supported.
21662 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
21664         PR target/56121
21665         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
21666         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
21667         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
21669 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21671         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
21672         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
21674 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21676         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
21677         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
21679 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21681         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
21682         declaration.
21683         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
21684         * config/arm/cortex-a7.md: New bypasses using
21685         arm_mac_accumulator_is_result.
21687 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21689         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
21690         (cortex_a7_neon_mla): Likewise.
21691         (cortex_a7_fpfmad): New reservation.
21692         (cortex_a7_fpmacs): Use ffmas and update required units.
21693         (cortex_a7_fpmuld): Update required units and latency.
21694         (cortex_a7_fpmacd): Likewise.
21695         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
21696         (cortex_a7_neon). Likewise.
21697         (bypass) Update participating units.
21699 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21701         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
21702         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
21703         from fmac to ffma.
21704         * config/arm/vfp11.md (vfp_farith): Use ffmas.
21705         (vfp_fmul): Use ffmad.
21706         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
21707         (cortex_r4_fmacd): Use ffmad.
21708         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
21709         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
21710         (cortex_a9_fmacd): Use ffmad.
21711         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
21712         (cortex_a8_vfp_macd): Use ffmad.
21713         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
21714         (cortex_a5_fpmacd): Use ffmad.
21715         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
21716         (cortex_a15_vfp_macd): Use ffmad.
21717         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
21719 2013-01-29  Jason Merrill  <jason@redhat.com>
21721         PR libstdc++/54314
21722         * varasm.c (default_assemble_visibility): Don't warn about
21723         visibility on artificial decls.
21725 2013-01-29  Richard Biener  <rguenther@suse.de>
21727         PR tree-optimization/56113
21728         * tree-ssa-structalias.c (equiv_class_lookup): Also return
21729         the bitmap leader.
21730         (label_visit): Free duplicate bitmaps and record the leader instead.
21731         (perform_var_substitution): Adjust.
21733 2013-01-29  Richard Biener  <rguenther@suse.de>
21735         PR tree-optimization/55270
21736         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
21737         the CFG, schedule loops for fixup.
21739 2013-01-29  Nick Clifton  <nickc@redhat.com>
21741         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
21742         SP_REG.
21744 2013-01-28  Leif Ekblad  <leif@rdos.net>
21746         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
21747         * config/i386/i386.h (TARGET_RDOS): New macro.
21748         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
21749         * config/i386/i386.c (ix86_option_override_internal): For 64bit
21750         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
21751         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
21752         DEFAULT_LARGE_SECTION_THRESHOLD.
21753         * config/i386/i386.md (R14_REG, R15_REG): New constants.
21754         * config/i386/rdos.h: New file.
21755         * config/i386/rdos64.h: New file.
21757 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
21759         PR other/54814
21760         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
21761         TEST_HARD_REG_BIT.
21763 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21765         PR rtl-optimization/56117
21766         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
21767         call cselib_lookup_from_insn on the MEM before calling
21768         add_insn_mem_dependence.
21770 2013-01-28  Richard Biener  <rguenther@suse.de>
21772         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
21773         to a stmt that didn't have one.
21774         (copy_phis_for_bb): Likewise for PHI arguments.
21775         (copy_debug_stmt): Likewise for debug stmts.
21777 2013-01-28  Richard Biener  <rguenther@suse.de>
21779         PR tree-optimization/56034
21780         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
21781         (partition_builtin_p): Adjust.
21782         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
21783         it is the last partition.
21784         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
21785         up the vertex for the definition.
21786         (classify_partition): Classify whether a partition is a
21787         PKIND_REDUCTION, thus has uses outside of the loop.
21788         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
21789         Merge all PKIND_REDUCTION partitions into the last partition.
21790         (tree_loop_distribution): Seed partitions from reductions as well.
21792 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21794         PR tree-optimization/56125
21795         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
21796         pow(x,c) into sqrt(x) * powi(x, n/2) or
21797         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
21798         optimizing for size.
21799         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
21800         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
21801         integer.
21803         PR tree-optimization/56094
21804         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
21805         to UNKNOWN_LOCATION while gimplifying expr.
21807 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
21809         PR target/56114
21810         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
21811         operand 0 in movabs insn template for -masm=intel asm alternative.
21812         (*movabs<mode>_2): Ditto for operand 1.
21814 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
21816         PR target/54663
21817         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
21818         of microblaze-c.o
21820 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21822         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
21823         tm_file.
21825 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21827         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
21828         Undef to avoid warning.
21830 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
21832         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
21833         * configure: Regenerate.
21835 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
21837         PR tree-optimization/56098
21838         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
21839         for stmts with volatile ops.
21840         (cond_store_replacement): Don't optimize if assign has volatile ops.
21841         (cond_if_else_store_replacement_1): Don't optimize if either
21842         then_assign or else_assign have volatile ops.
21843         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
21844         volatile ops.
21846 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21848         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
21850 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21852         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
21853         missing ':' in asm example.
21855 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
21857         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
21858         entries into lane and laneq entries.
21859         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
21860         Remove AdvSIMD scalar modes.
21861         (aarch64_sq<r>dmulh_laneq<mode>): New.
21862         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
21863         modes.
21864         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
21865         builtin implementations to relfect changes in RTL in aarch64-simd.md.
21866         * config/aarch64/iterators.md (VCOND): New.
21867         (VCONQ): New.
21869 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21871         PR target/54222
21872         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
21873         Add NULL LIBNAME argument to existing definitions.
21874         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
21875         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
21876         * config/avr/avr.c (DEF_BUILTIN): Same.
21877         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
21878         (avr_expand_builtin): Expand to a vanilla call if a libgcc
21879         implementation is available (DECL_ASSEMBLER_NAME is set).
21880         (avr_fold_absfx): New static function.
21881         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
21882         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
21883         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
21884         AVR_BUILTIN_ABSLLK.
21885         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
21886         (abshk, absk, abslk, absllk): Provide as static inline functions.
21888 2013-01-25  Marek Polacek  <polacek@redhat.com>
21890         PR tree-optimization/56035
21891         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
21893 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
21895         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
21896         (*movtf_internal_rex64): Add (!o,C) alternative
21897         (*movxf_internal_rex64): Ditto.
21898         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
21900 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
21902         * doc/invoke.texi: fix typo.
21903         * doc/objc.texi: fix typo.
21905 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
21907         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
21908         for the first two alternatives.
21910 2013-01-24  Diego Novillo  <dnovillo@google.com>
21912         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
21913         (ggc-zone.o): Remove.
21914         * configure.ac: Remove option --with-gc.
21915         * configure: Re-generate.
21916         * doc/install.texi: Remove documentation for --with-gc.
21917         * gengtype.c (write_enum_defn): Remove.  Update all users.
21918         (write_Types_process_field): Remove generation of gt_e_* argument.
21919         (output_type_enum): Remove.  Update all users.
21920         (write_enum_defn): Remove.  Update all users.
21921         (enum alloc_zone): Remove.  Update all users.
21922         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
21923         * ggc-common.c (ggc_splay_alloc): Remove first argument.
21924         Update all callers.
21925         (struct ptr_data): Remove field TYPE.  Update all users.
21926         (gt_pch_note_object): Remove argument TYPE.  Update all users.
21927         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
21928         Update all users.
21929         * ggc-none.c (ggc_alloc_typed_stat): Remove.
21930         (struct alloc_zone): Remove.
21931         (ggc_internal_alloc_zone_stat): Remove.
21932         (ggc_internal_cleared_alloc_zone_stat): Remove.
21933         * ggc-page.c (ggc_alloc_typed_stat): Remove.
21934         (ggc_pch_count_object): Remove last argument.  Update all users.
21935         (ggc_pch_alloc_object): Remove last argument.  Update all users.
21936         (struct alloc_zone): Remove.
21937         * ggc-zone.c: Remove.
21938         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
21939         (struct alloc_zone): Remove.
21940         (ggc_alloc_typed_stat): Remove.
21941         (ggc_alloc_typed): Remove.
21942         (ggc_splay_alloc): Remove first argument.
21943         (rtl_zone): Remove.  Update all users.
21944         (tree_zone): Remove.  Update all users.
21945         (tree_id_zone): Remove.  Update all users.
21946         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
21947         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
21948         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
21949         * tree-ssanames.c: Remove references to zone allocator in comments.
21951 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
21953         * config/avr/avr.c (avr_out_fract): Make register numbers that
21954         might be outside of source operand signed.
21956 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
21958         * config/i386/constraints.md (Yf): New constraint.
21959         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
21960         of f constraint to conditionaly disable x87 register preferences.
21961         (*movdf_internal): Ditto.
21962         (*movsf_internal): Ditto.
21964 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
21966         PR inline-asm/55934
21967         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
21968         that have operands with impossible constraints.
21969         Add a FIXME for a speed-up opportunity.
21970         * lra-constraints.c (process_alt_operands): Verify that a class
21971         selected from constraints on asms is valid for the operand mode.
21972         (curr_insn_transform): Remove incorrect comment.
21974 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
21976         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
21977         TOC operand is a valid symbol ref in the constant pool.
21979 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21981         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
21983 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
21985         PR target/54222
21986         * config/avr/stdfix.h: New file.
21987         * t-avr (stdfix-gcc.h): New rule to build it.
21988         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
21990 2013-01-23  Kostya Serebryany  <kcc@google.com>
21992         * config/darwin.h: remove dependency on
21993         CoreFoundation (asan on Mac OS).
21995 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
21997         PR target/49069
21998         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
21999         instead of cmpdi_operand for first comparison operand.
22000         Don't assert that comparison operands aren't both constants.
22002 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
22004         * doc/install.texi (Downloading the Source): Update references to
22005         downloading separate components.
22007 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
22009         * doc/extend.texi (__int128): Improve grammar.
22011 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
22013         PR target/56028
22014         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
22015         alternative to (o,r).
22016         (*movdi_internal_rex64): Remove (!o,n) alternative.
22017         (DImode immediate->memory splitter): Remove.
22018         (DImode immediate->memory peephole2): Remove.
22019         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
22020         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
22021         alternative to (!o,*r).
22022         (*movtf_internal_sse): New pattern.
22023         (*movxf_internal_rex64): New pattern.
22024         (*movxf_internal): Disable for TARGET_64BIT.
22025         (*movdf_internal_rex64): Remove (!o,F) alternative.
22027 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
22029         PR middle-end/56074
22030         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
22031         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
22032         * tree-vect-loop-manip.c (find_loop_location): Also ignore
22033         stmt locations where LOCATION_LOCUS of the stmt location is
22034         UNKNOWN_LOCATION or BUILTINS_LOCATION.
22036         PR target/55686
22037         * config/i386/i386.md (UNSPEC_STOS): New.
22038         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
22039         *strsetqi_1): Add UNSPEC_STOS.
22041 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
22043         PR c++/56067
22044         * doc/invoke.texi: Remove left over -Wsynth example.
22046 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
22048         PR tree-optimization/56051
22049         * fold-const.c (fold_binary_loc): Don't fold
22050         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
22051         a narrowing conversion, or widening conversion from signed
22052         to unsigned.
22054 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
22056         PR rtl-optimization/56023
22057         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
22058         dependent on debug instruction.
22060 2013-01-21  Martin Jambor  <mjambor@suse.cz>
22062         PR middle-end/56022
22063         * function.c (allocate_struct_function): Call
22064         invoke_set_current_function_hook earlier.
22066 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
22068         * reload1.c (init_reload): Only initialize reload_obstack
22069         during the first call.
22071 2013-01-21  Marek Polacek  <polacek@redhat.com>
22073         * cfgloop.c (verify_loop_structure): Fix up grammar.
22075 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
22077         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
22078         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
22080 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
22082         PR target/56058
22083         * config/arm/marvell-pj4.md: Update copyright year.
22084         Fix up use of alu to alu_reg and simple_alu_imm.
22086 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
22088         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
22090 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
22092         PR target/55433
22093         * lra-constraints.c (curr_insn_transform): Don't reuse original
22094         insn for secondary memory move when memory mode should be different.
22096 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22098         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
22099         atomic_storedi_1): New patterns.
22101 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
22103         btver2 pipeline descriptions.
22104         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
22105         descriptions.
22106         * config/i386/i386.md (btver2_decode): New type attributes.
22107         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
22108         type attributes.
22109         * config/i386/btver2.md: New file describing btver2 pipelines.
22111 2013-01-19  Andrew Pinski  <apinski@cavium.com>
22113         PR tree-optimization/52631
22114         * tree-ssa-sccvn (visit_use): Before looking up the original
22115         statement, try looking up the simplified expression.
22117 2013-01-19  Anthony Green  <green@moxielogic.com>
22119         * config/moxie/moxie.c (moxie_expand_prologue): Set
22120         current_function_static_stack_size.
22122 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
22124         PR tree-optimization/56029
22125         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
22126         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
22128 2013-01-18  Sharad Singhai  <singhai@google.com>
22130         PR tree-optimization/55995
22131         * dumpfile.c (dump_loc): Print location only if available.
22132         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
22134 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
22136         PR target/55433
22137         * lra-constraints.c (curr_insn_transform): Reuse original insn for
22138         secondary memory move.
22139         (inherit_reload_reg): Use rclass instead of cl for
22140         check_secondary_memory_needed_p.
22142 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
22144         PR middle-end/56015
22145         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
22146         the case where writing real complex part of target modifies op1.
22148 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
22150         * config/aarch64/aarch64-simd.md
22151         (aarch64_vcond_internal<mode>): Handle unordered cases.
22152         * config/aarch64/iterators.md (v_cmp_result): New.
22154 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
22155             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
22157         * config/arm/marvell-pj4.md: New file.
22158         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
22159         * config/arm/arm.md (generic_sched): Add marvell_pj4.
22160         (generic_vfp): Likewise.
22161         * config/arm/arm-cores.def: Add marvell-pj4.
22162         * config/arm/arm-tune.md: Regenerate.
22163         * config/arm/arm-tables.opt: Regenerate.
22164         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
22165         * doc/invoke.texi: Document marvell-pj4.
22167 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
22169         * config/aarch64/arm_neon.h: Map scalar types to standard types.
22171 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
22173         PR debug/54114
22174         PR debug/54402
22175         PR debug/49888
22176         * var-tracking.c (negative_power_of_two_p): New.
22177         (global_get_addr_cache, local_get_addr_cache): New.
22178         (get_addr_from_global_cache, get_addr_from_local_cache): New.
22179         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
22180         heading comment.
22181         (vt_stack_offset_p): Remove.
22182         (vt_canon_true_dep): Always canonicalize loc's address.
22183         (clobber_overlapping_mems): Make sure we have a MEM.
22184         (local_get_addr_clear_given_value): New.
22185         (val_reset): Clear local cached entries.
22186         (compute_bb_dataflow): Create and release the local cache.
22187         Disable duplicate MEMs clobbering.
22188         (emit_notes_in_bb): Clobber MEMs likewise.
22189         (vt_emit_notes): Create and release the local cache.
22190         (vt_initialize, vt_finalize): Create and release the global
22191         cache, respectively.
22192         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
22194 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
22196         PR libmudflap/53359
22197         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
22198         not found in the symtab.
22200 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
22202         PR debug/56006
22203         PR rtl-optimization/55547
22204         PR rtl-optimization/53827
22205         PR debug/53671
22206         PR debug/49888
22207         * alias.c (offset_overlap_p): New, factored out of...
22208         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
22209         the conservative special case for symbolic constants.  Don't
22210         adjust zero sizes on alignment.
22212 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
22214         PR rtl-optimization/52573
22215         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
22216         REG_UNUSED for the same register.
22218 2013-01-17  Richard Biener  <rguenther@suse.de>
22219             Marek Polacek  <polacek@redhat.com>
22221         PR rtl-optimization/55833
22222         * loop-unswitch.c (unswitch_loops): Move loop verification...
22223         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
22224         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
22225         Set it to true when we're removing a loop from hierarchy tree in
22226         an irreducible region.
22227         (fix_bb_placements): Adjust caller.
22228         (fix_loop_placements): Likewise.
22230 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
22232         * config/avr/builtins.def (DEF_BUILTIN): Factor out
22233         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
22234         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
22235         Remove ID.  Adjust comments.
22236         * config/avr/avr-c.c (avr_builtin_name): Remove.
22237         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
22238         * config/avr/avr.c (avr_tolower): New static function.
22239         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
22240         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
22241         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
22242         default expansion.
22244 2013-01-17  Jan Hubicka  <jh@suse.cz>
22246         PR tree-optimization/55273
22247         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
22249 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
22251         PR target/55981
22252         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
22253         store through atomic_store<mode>_1.
22254         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
22256 2013-01-17  Martin Jambor  <mjambor@suse.cz>
22258         PR tree-optimizations/55264
22259         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
22260         for virtual methods.
22261         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
22262         virtual methods before inlining is over.
22263         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
22264         virtual functions.
22265         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
22266         non-virtual.
22268 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
22270         PR rtl-optimization/56005
22271         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
22272         pending reads for prefetch.
22274 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
22276         * config/aarch64/aarch64.md
22277         (*cstoresi_neg_uxtw): New pattern.
22278         (*cmovsi_insn_uxtw): New pattern.
22279         (*<optab>si3_uxtw): New pattern.
22280         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
22281         (*<optab>si3_insn_uxtw): New pattern.
22282         (*bswapsi2_uxtw): New pattern.
22284 2013-01-16  Richard Biener  <rguenther@suse.de>
22286         * tree-inline.c (tree_function_versioning): Remove set but
22287         never used variable.
22289 2013-01-16  Richard Biener  <rguenther@suse.de>
22291         PR tree-optimization/55964
22292         * tree-flow.h (rename_variables_in_loop): Remove.
22293         (rename_variables_in_bb): Likewise.
22294         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
22295         (copy_loop_before): Adjust and delete update-ssa status.
22296         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
22297         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
22298         (rename_variables_in_loop): Remove.
22299         (slpeel_update_phis_for_duplicate_loop): Likewise.
22300         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
22301         use available cfg machinery instead of duplicating it.
22302         Update PHI nodes and perform poor-mans SSA update here.
22303         (slpeel_tree_peel_loop_to_edge): Adjust.
22305 2013-01-16  Richard Biener  <rguenther@suse.de>
22307         PR tree-optimization/54767
22308         PR tree-optimization/53465
22309         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
22310         (vrp_visit_phi_node): For PHI arguments coming via backedges
22311         drop all symbolical range information.
22312         (execute_vrp): Compute backedges.
22314 2013-01-16  Richard Biener  <rguenther@suse.de>
22316         * doc/install.texi: Update CLooG and ISL requirements to
22317         0.18.0 and 0.11.1.
22319 2013-01-16  Christian Bruel  <christian.bruel@st.com>
22321         PR target/55301
22322         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
22323         (broken_move): Handle UNSPECV_SP_SWITCH_B.
22324         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
22326 2013-01-16  DJ Delorie  <dj@redhat.com>
22328         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
22329         (UNSPECV_SP_SWITCH_E): New.
22330         (sp_switch_1): Change to an unspec.
22331         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
22332         replace $r15.
22334 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
22336         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
22337         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
22338         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
22339         (expand_mem_thread_fence): Ditto.
22340         (expand_mem_signal_fence): Ditto.
22341         (expand_atomic_load): Ditto.
22342         (expand_atomic_store): Ditto.
22344 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
22346         PR rtl-optimization/55547
22347         PR rtl-optimization/53827
22348         PR debug/53671
22349         PR debug/49888
22350         * alias.c (memrefs_conflict_p): Set sizes to negative after
22351         AND adjustments.
22353 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
22355         PR target/55940
22356         * function.c (thread_prologue_and_epilogue_insns): Always
22357         add crtl->drap_reg to set_up_by_prologue.set, even if
22358         stack_realign_drap is false.
22360 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
22362         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
22363         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
22364         *call): Fix indention.
22366 2013-01-15  Tom de Vries  <tom@codesourcery.com>
22368         PR target/55876
22369         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
22370         Update comment.
22372 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
22374         PR rtl-optimization/55153
22375         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
22377 2013-01-15  Martin Jambor  <mjambor@suse.cz>
22379         PR tree-optimization/55920
22380         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
22381         accesses as grp_to_be_debug_replaced.
22383 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
22385         PR tree-optimization/55920
22386         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
22387         there is non-useless type conversion needed from debug rhs to lhs,
22388         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
22390 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
22391             Mikael Pettersson  <mikpe@it.uu.se>
22393         PR target/43961
22394         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
22395         Thumb.
22396         (ASM_OUTPUT_CASE_LABEL): Remove.
22397         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
22398         * final.c (shorten_branches): Update alignment of labels before
22399         jump tables if CASE_VECTOR_SHORTEN_MODE.
22401 2013-01-15  Richard Biener  <rguenther@suse.de>
22403         PR bootstrap/55961
22404         * system.h: Do not include gmp.h for building host tools.
22406 2013-01-15  Richard Biener  <rguenther@suse.de>
22408         PR middle-end/55882
22409         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
22410         account for bitpos when computing alignment.
22412 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
22414         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
22415         (ix86_target_macros_internal): Likewise.
22417         * config/i386/i386.c (m_CORE2I7): Removed.
22418         (m_CORE_HASWELL): New macro.
22419         (m_CORE_ALL): Likewise.
22420         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
22421         (initial_ix86_arch_features): Likewise.
22422         (processor_target_table): Initializations for Core avx2.
22423         (cpu_names): New names "core-avx2".
22424         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
22425         PROCESSOR_CORE_HASWELL.
22426         (ix86_issue_rate): New case.
22427         (ia32_multipass_dfa_lookahead): Likewise.
22428         (ix86_sched_init_global): Likewise.
22430         * config/i386/i386.h (TARGET_HASWELL): New macro.
22431         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
22432         (processor_type): New PROCESSOR_HASWELL.
22434 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
22436         PR tree-optimization/55955
22437         * tree-vect-loop.c (vectorizable_reduction): Give up early on
22438         *SHIFT_EXPR and *ROTATE_EXPR codes.
22440         PR tree-optimization/48766
22441         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
22442         -ftrapv disable -fwrapv.
22444 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22446         PR target/55974
22447         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
22448         etc. to 1 and not to __flash.
22449         Use LL suffix for __INT24_MAX__ with -mint8.
22450         Use ULL suffix for __UINT24_MAX__ with -mint8.
22452 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22454         * config/avr/avr-arch.h
22455         (struct base_arch_s): Use typedef avr_arch_t instead.
22456         (struct arch_info_s): Use typedef avr_arch_info_t instead.
22457         (struct mcu_type_s): Use typedef avr_mcu_t instead.
22458         * config/avr/avr.c: Same.
22459         * config/avr/avr-devices.c: Same.
22460         * config/avr/driver-avr.c: Same.
22461         * config/avr/gen-avr-mmcu-texi.c: Same.
22462         * config/avr/avr-mcus.def: Adjust comment.
22464 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
22466         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
22467         * config/aarch64/iterators.md (VALLDI): New.
22469 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
22470             Andi Kleen  <ak@linux.intel.com>
22472         PR target/55948
22473         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
22474         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
22475         memmodel flag.
22477 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22479         * config/avr/avr-stdint.h: Remove trailing blanks.
22480         * config/avr/avr-log.h: Same.
22481         * config/avr/avr-arch.h: Same.
22482         * config/avr/avr-devices.c: Same.
22483         * config/avr/avr-dimode.md: Same.
22484         * config/avr/predicates.md: Same.
22485         * config/avr/avr-c.c: Same.  And fix typo.
22487         * config/avr/avr-protos.h: Same.  And:
22488         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
22489         (init_cumulative_args): Rename to avr_init_cumulative_args.
22490         (expand_prologue): Rename to avr_expand_prologue.
22491         (expand_epilogue): Rename to avr_expand_epilogue.
22492         (adjust_insn_length): Rename to avr_adjust_insn_length.
22493         (notice_update_cc): Rename to avr_notice_update_cc.
22494         (final_prescan_insn): Rename to avr_final_prescan_insn.
22495         * config/avr/avr.c: Same.
22496         * config/avr/avr.h: Same.
22497         * config/avr/avr.md: Remove trailing blanks.
22498         (prologue): Use avr_expand_prologue.
22499         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
22501 2013-01-14  Richard Biener  <rguenther@suse.de>
22503         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
22504         verify_location, collect_subblocks): New functions.
22505         (verify_gimple_in_cfg): Verify that locations only reference
22506         BLOCKs in the functions BLOCK tree.
22508 2013-01-14  Richard Biener  <rguenther@suse.de>
22510         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
22511         PHI argument.
22512         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
22513         unshare reference.
22514         (insert_out_of_ssa_copy_on_edge): Likewise.
22515         (rewrite_close_phi_out_of_ssa): Likewise.
22516         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
22517         debug expressions.
22518         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
22519         propagated constants.
22520         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
22521         can not be shared.
22523 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
22525         * config/avr/avr-modes.def: Add GPL copyright notice.
22527 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
22529         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
22530         MEMMODEL_MASK to determine memory model.
22531         (atomic_store<mode>): Ditto from operands[2].
22532         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
22534 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
22536         PR fortran/55935
22537         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
22538         (fold_gimple_assign): Don't call unshare_expr here.
22539         (fold_ctor_reference): Call unshare_expr.
22541 2013-01-13  Terry Guo  <terry.guo@arm.com>
22543         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
22544         * doc/fragments.texi: Document MULTILIB_REUSE.
22545         * gcc.c (multilib_reuse): New internal spec.
22546         (set_multilib_dir): Also search multilib from multilib_reuse.
22547         * genmultilib (tmpmultilib3): Refactor code.
22548         (tmpmultilib4): Ditto.
22549         (multilib_reuse): New multilib argument.
22551 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
22553         * Makefile.in: Update copyright.
22555 2013-01-12  Tom de Vries  <tom@codesourcery.com>
22557         PR middle-end/55890
22558         * calls.c (expand_call): Check if arg_nr is valid.
22560 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
22562         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
22563         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
22564         documentation.  Add missing '__' in front of
22565         __builtin_ia32_packssdw256.
22567 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22569         PR target/55719
22570         * config/s390/s390.c (s390_preferred_reload_class): Do not return
22571         NO_REGS for larl operands.
22572         (s390_reload_larl_operand): Use s390_load_address instead of
22573         emit_move_insn.
22575 2013-01-11  Richard Biener  <rguenther@suse.de>
22577         * tree-cfg.c (verify_node_sharing_1): Split out from ...
22578         (verify_node_sharing): ... here.
22579         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
22581 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
22583         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
22584         Substitute TREECHECKING.
22585         * configure: Regenerate.
22586         * Makefile.in (TREECHECKING): New.
22588 2013-01-11  Richard Guenther  <rguenther@suse.de>
22590         PR tree-optimization/44061
22591         * tree-vrp.c (extract_range_basic): Compute zero as
22592         value-range for __builtin_constant_p of function parameters.
22594 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
22596         Update copyright years.
22598 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
22600         PR rtl-optimization/55672
22601         * lra-eliminations.c (mark_not_eliminable): Permit addition with
22602         const to be eliminable.
22604 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
22606         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
22607         * configure: Regenerate.
22609 2013-01-10  Richard Biener  <rguenther@suse.de>
22611         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
22613 2013-01-10  Richard Biener  <rguenther@suse.de>
22615         PR bootstrap/55792
22616         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
22617         locations for virtual PHI arguments.
22618         (rewrite_update_phi_arguments): Likewise.
22620 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
22622         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
22623         on to assembler.
22625 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
22627         PR tree-optimization/55921
22628         * tree-complex.c (expand_complex_asm): New function.
22629         (expand_complex_operations_1): Call it for GIMPLE_ASM.
22631 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22633         PR target/55718
22634         * config/s390/s390.c (s390_symref_operand_p)
22635         (s390_loadrelative_operand_p): Merge the two functions.
22636         (s390_check_qrst_address, print_operand_address): Add parameters
22637         to s390_loadrelative_operand_p invokation.
22638         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
22639         (s390_reload_larl_operand, s390_secondary_reload): Use
22640         s390_loadrelative_operand_p instead of s390_symref_operand_p.
22641         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
22643 2013-01-09  Mike Stump  <mikestump@comcast.net>
22645         * dse.c (record_store): Remove unnecessary assert.
22647 2013-01-09  Jan Hubicka  <jh@suse.cz>
22649         PR tree-optimization/55569
22650         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
22651         * cfgloop.h (scale_loop_profile): Likewise.
22653 2013-01-09  Jan Hubicka  <jh@suse.cz>
22655         PR lto/45375
22656         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
22657         functions.
22658         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
22660 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
22662         PR middle-end/55114
22663         * expr.h (maybe_emit_group_store): Declare.
22664         * expr.c (maybe_emit_group_store): New function.
22665         * builtins.c (expand_builtin_int_roundingfn): Call it.
22666         (expand_builtin_int_roundingfn_2): Likewise.
22668 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
22670         PR rtl-optimization/55829
22671         * lra-constraints.c (match_reload): Add code for absent output.
22672         (curr_insn_transform): Add code for reloads of matched inputs
22673         without output.
22675 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
22677         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
22678         attribute of movddup insn to DF.
22679         (*vec_interleave_lowv2df): Ditto.
22680         (vec_dupv2df): Ditto.
22682 2013-01-09  Jan Hubicka  <jh@suse.cz>
22684         PR tree-optimiation/55875
22685         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
22686         EVERY_ITERATION parameter.
22687         (number_of_iterations_exit): Check if exit is executed every iteration.
22688         (idx_infer_loop_bounds): Similarly here.
22689         (n_of_executions_at_most): Simplify
22690         to only test for cases where statement is dominated by the
22691         particular bound; handle correctly the "postdominance" test.
22692         (scev_probably_wraps_p): Use max loop iterations info
22693         as a global bound first.
22695 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
22696             Nick Clifton  <nickc@redhat.com>
22698         * config/v850/v850.md (cbranchsf4): New pattern.
22699         (cstoresf4): New pattern.
22700         (cbranchdf4): New pattern.
22701         (cstoredf4): New pattern.
22702         (movsicc): Disallow floating point comparisons.
22703         (cmpsf_le_insn): Fix order of operators.
22704         (cmpsf_lt_insn): Likewise.
22705         (cmpsf_eq_insn): Likewise.
22706         (cmpdf_le_insn): Likewise.
22707         (cmpdf_lt_insn): Likewise.
22708         (cmpdf_eq_insn): Likewise.
22709         (cmpsf_ge_insn): Use LE comparison.
22710         (cmpdf_ge_insn): Likewise.
22711         (cmpsf_gt_insn): Use LT comparison.
22712         (cmpdf_gt_insn): Likewise.
22713         (cmpsf_ne_insn): Delete pattern.
22714         (cmpdf_ne_insn): Delete pattern.
22715         * config/v850/v850.c (v850_gen_float_compare): Use
22716         gen_cmpdf_eq_insn for NE comparison.
22717         (v850_float_z_comparison_operator)
22718         (v850_float_nz_comparison_operator): Move from here ...
22719         * config/v850/predicates.md: ... to here.  Move GT and GE
22720         comparisons into v850_float_z_comparison_operator.
22721         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
22722         Delete prototype.
22723         (v850_float_nz_comparison_operator): Likewise.
22725 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22727         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
22728         with calls to gen_insvsi/gen_insvdi.
22730 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
22732         * config/i386/i386.c (initial_ix86_tune_features): Set up
22733         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
22735 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
22736             Jakub Jelinek  <jakub@redhat.com>
22738         PR tree-optimization/48189
22739         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
22740         If nitercst is 0, don't predict the exit edge.
22742 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22744         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
22745         in asm_fprintf with reg_names.
22746         (aarch64_print_operand_address): Likewise.
22747         (aarch64_return_addr): Likewise.
22748         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
22750 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22752         * config/pa/pa.h (VAL_U6_BITS_P): Define.
22753         (INT_U6_BITS): Likewise.
22754         * config/pa/predicates.md (uint6_operand): New predicate.
22755         (shift5_operand, shift6_operand): Likewise.
22756         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
22757         arith32_operand.
22758         (lshrdi3): Use shift6_operand.
22759         (shrpsi4, shrpdi4): New insn patterns.
22760         (extzv): Delete expander.
22761         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
22762         predicates in unamed zero extract patterns.  Tighten common constraint.
22763         (extv): Delete expander.
22764         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
22765         predicates in unamed sign extract patterns.  Tighten common constraint.
22766         (insv): Delete expander.
22767         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
22768         predicates in unamed insert patterns.  Tighten common constraint.
22769         Change uint32_operand predicate to uint6_operand predicate in unamed
22770         DImode pattern to insert constant values of type 1...1xxxx.
22772 2013-01-04  Jan Hubicka  <jh@suse.cz>
22774         PR tree-optimization/55823
22775         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
22776         issue.
22778 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22779             Uros Bizjak  <ubizjak@gmail.com>
22781         PR rtl-optimization/55845
22782         * df-problems.c (can_move_insns_across): Stop scanning at
22783         volatile_insn_p source instruction or give up if
22784         across_from .. across_to range contains any volatile_insn_p
22785         instructions.
22787 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22789         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
22790         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
22791         Declare.
22792         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
22793         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
22795 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22797         PR fortran/55341
22798         * asan.c (asan_clear_shadow): New function.
22799         (asan_emit_stack_protection): Use it.
22801 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22803         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
22804         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
22805         with tab instead of space.
22807 2013-01-08  Nick Clifton  <nickc@redhat.com>
22809         * config/rl78/rl78.c (rl78_expand_prologue): Always select
22810         register bank 0 at the start of an interrupt handler.
22811         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
22812         MDBH registers.
22814 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22816         * config/aarch64/aarch64-simd.md
22817         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
22818         (aarch64_simd_bsl): Likewise.
22819         (aarch64_vcond_internal<mode>): Likewise.
22820         (vcond<mode><mode>): Likewise.
22821         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
22822         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
22824 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22826         * config/aarch64/aarch64-builtins.c
22827         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
22829 2013-01-08  Martin Jambor  <mjambor@suse.cz>
22831         PR debug/55579
22832         * tree-sra.c (analyze_access_subtree): Return true also after
22833         potentially creating a debug-only replacement.
22835 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22837         PR middle-end/55890
22838         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
22840         PR tree-optimization/54120
22841         * tree-vrp.c (range_fits_type_p): Don't allow
22842         src_precision < precision from signed vr to unsigned_p
22843         if vr->min or vr->max is negative.
22844         (simplify_float_conversion_using_ranges): Test can_float_p
22845         against CODE_FOR_nothing.
22847 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22848             Richard Biener  <rguenther@suse.de>
22850         PR middle-end/55851
22851         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
22852         types instead of just INTEGER_TYPE types.
22854 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
22856         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
22857         TF_SIZE): Define.
22859 2013-01-07  Steve Ellcey  <sellcey@mips.com>
22861         PR target/42661
22862         * config/mips/mips.opt: Change mad to mmad to match documentation.
22864 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22866         PR target/55897
22867         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
22868         .progmemx.data now.
22870 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22872         PR target/55897
22873         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
22874         (avr_addrspace_t): Add .section_name field.
22875         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
22876         array size.
22877         (avr_addrspace): Same.  Initialize .section_name.  Remove last
22878         NULL entry.  Put __memx into .progmemx.data.
22879         (progmem_section_prefix): Remove.
22880         (avr_asm_init_sections): No need to initialize progmem_section.
22881         (avr_asm_named_section): Use avr_addrspace[].section_name to get
22882         section name prefix.
22883         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
22884         retrieve the progmem section.
22885         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
22886         boundary to run over avr_addrspace[].
22887         (avr_register_target_pragmas): Ditto.
22889 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
22891         * varasm.c (output_constant_def_contents): For asan_protect_global
22892         protected strings, adjust DECL_ALIGN if needed, before testing for
22893         anchored symbols.
22894         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
22895         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
22896         normal decls.
22897         (output_object_block): For asan protected decls, emit asan padding
22898         after their contents.
22899         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
22900         (asan_finish_file): Test it here instead.
22902 2013-01-07  Nick Clifton  <nickc@redhat.com>
22903             Matthias Klose  <doko@debian.org>
22904             Doug Kwan  <dougkwan@google.com>
22905             H.J. Lu  <hongjiu.lu@intel.com>
22907         PR driver/55470
22908         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
22910         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
22912         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
22914         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
22916         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
22918 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22920         PR target/54461
22921         * doc/install.texi (Cross-Compiler-Specific Options): Document
22922         --with-avrlibc.
22924 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
22926         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
22927         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
22928         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
22929         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
22930         vqmovun_high_s64): Fix source operand number and update copyright.
22932 2013-01-07  Richard Biener  <rguenther@suse.de>
22934         PR middle-end/55890
22935         * gimple.h (gimple_call_builtin_p): New overload.
22936         * gimple.c (validate_call): New function.
22937         (gimple_call_builtin_p): Likewise.
22938         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22939         Use gimple_call_builtin_p.
22940         (find_func_clobbers): Likewise.
22941         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
22942         (strlen_optimize_stmt): Likewise.
22944 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
22946         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
22947         (vld1q_dup_*): Likewise.
22948         (vld1_*): Likewise.
22949         (vld1q_*): Likewise.
22950         (vld1_lane_*): Likewise.
22951         (vld1q_lane_*): Likewise.
22953 2013-01-07  Richard Biener  <rguenther@suse.de>
22955         * lto-streamer.h (LTO_minor_version): Bump to 2.
22957 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
22959         * config/aarch64/aarch64-protos.h
22960         (aarch64_const_double_zero_rtx_p): Rename to...
22961         (aarch64_float_const_zero_rtx_p): ...this.
22962         (aarch64_float_const_representable_p): New.
22963         (aarch64_output_simd_mov_immediate): Likewise.
22964         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
22965         move immediate case.
22966         * config/aarch64/aarch64.c
22967         (aarch64_const_double_zero_rtx_p): Rename to...
22968         (aarch64_float_const_zero_rtx_p): ...this.
22969         (aarch64_print_operand): Allow printing of new constants.
22970         (aarch64_valid_floating_const): New.
22971         (aarch64_legitimate_constant_p): Check for valid floating-point
22972         constants.
22973         (aarch64_simd_valid_immediate): Likewise.
22974         (aarch64_vect_float_const_representable_p): New.
22975         (aarch64_float_const_representable_p): Likewise.
22976         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
22977         (aarch64_output_simd_mov_immediate): New.
22978         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
22979         (*movdf_aarch64): Likewise.
22980         * config/aarch64/constraints.md (Ufc): New.
22981         (Y): call aarch64_float_const_zero_rtx.
22982         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
22984 2013-01-07  Richard Biener  <rguenther@suse.de>
22986         PR tree-optimization/55888
22987         PR tree-optimization/55862
22988         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
22989         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
22990         not if it is contained therein.
22992 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22994         * config/avr/t-avr: Typo.
22996 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22998         PR55243
22999         * config/avr/t-avr: Don't automatically rebuild
23000         $(srcdir)/config/avr/t-multilib
23001         $(srcdir)/config/avr/avr-tables.opt
23002         $(srcdir)/doc/avr-mmcu.texi
23003         (avr-mcus): New phony target to build them on request.
23004         (s-avr-mlib, s-avr-mmcu-texi): Remove.
23005         * avr/avr-mcus.def: Adjust comments.
23007 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
23009         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
23011 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
23013         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
23015 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
23017         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
23019 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
23021         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
23022         to generate profiling.
23023         * config/rs6000/aix64.h (LIB_SPEC): Same.
23025 2013-01-04  Andrew Pinski  <apinski@cavium.com>
23027         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
23028         New function.
23029         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
23031 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
23033         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
23034         unconditionally.
23035         (ix86_expand_move): Ditto.
23036         (ix86_zero_extend_to_Pmode): Ditto.
23037         (ix86_expand_call): Ditto.
23038         (ix86_expand_special_args_builtin): Ditto.
23039         (ix86_expand_builtin): Ditto.
23041 2013-01-04  Richard Biener  <rguenther@suse.de>
23043         PR tree-optimization/55862
23044         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
23045         translating them through PHI nodes.
23047 2013-01-04  Martin Jambor  <mjambor@suse.cz>
23049         PR tree-optimization/55755
23050         * tree-sra.c (sra_modify_assign): Do not check that an access has no
23051         children when trying to avoid producing a VIEW_CONVERT_EXPR.
23053 2013-01-04  Marek Polacek  <polacek@redhat.com>
23055         PR middle-end/55859
23056         * opts.c (default_options_optimization): Clarify error message.
23058 2013-01-04  Richard Biener  <rguenther@suse.de>
23060         PR middle-end/55863
23061         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
23062         reassociation.
23064 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23066         PR target/53789
23067         * config/pa/pa.md (movsi): Revert previous change.
23068         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
23069         references.
23071 2013-01-03  Richard Henderson  <rth@redhat.com>
23073         * config/i386/i386.c (ix86_expand_move): Always assign to op1
23074         after eliminating TLS symbols.
23076 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
23078         PR bootstrap/50167
23079         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
23080         * graphite-poly.c (debug_gmp_value): Likewise.
23082 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
23084         PR target/55712
23085         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
23086         selected code model, define __code_mode_small__, __code_model_medium__,
23087         __code_model_large__, __code_model_32__ or __code_model_kernel__.
23088         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
23089         xchg temporary register with %k.  Declare temporary register as
23090         early clobbered.
23091         [__x86_64__]: For medium and large code models, preserve %rbx register.
23093 2013-01-03  Richard Biener  <rguenther@suse.de>
23095         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
23096         (dump_subscript): Adjust.
23097         (finalize_ddr_dependent): Do not dump redundant info.
23098         (analyze_siv_subscript): Adjust.
23099         (subscript_dependence_tester): Likewise.
23100         (compute_affine_dependence): Likewise.
23102 2013-01-03  Richard Biener  <rguenther@suse.de>
23104         Revert
23105         2013-01-03  Richard Biener  <rguenther@suse.de>
23107         PR tree-optimization/55857
23108         * tree-vect-stmts.c (vectorizable_load): Do not setup
23109         re-alignment for invariant loads.
23111         2013-01-02  Richard Biener  <rguenther@suse.de>
23113         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
23114         invariant load do not generate a vector load from the scalar location.
23116 2013-01-03  Richard Biener  <rguenther@suse.de>
23118         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
23119         for not vectorizing.
23120         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
23121         not build INDIRECT_REFs, call get_name once only.
23122         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
23123         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
23125 2013-01-03  Richard Biener  <rguenther@suse.de>
23127         PR tree-optimization/55857
23128         * tree-vect-stmts.c (vectorizable_load): Do not setup
23129         re-alignment for invariant loads.
23131 2013-01-03  Richard Biener  <rguenther@suse.de>
23133         PR lto/55848
23134         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
23135         prefer a built-in decl.
23137 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
23139         * gcc.c (process_command): Update copyright notice dates.
23140         * gcov.c (print_version): Likewise.
23141         * gcov-dump.c (print_version): Likewise.
23143         PR rtl-optimization/55838
23144         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
23145         iv0.step, iv1.step and step.
23147 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
23148             Marc Glisse  <marc.glisse@inria.fr>
23150         PR tree-optimization/55832
23151         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
23152         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
23153         integer_{one,zero}_node.
23155 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
23157         PR debug/54402
23158         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
23159         * var-tracking.c (reverse_op): Don't add reverse ops to
23160         VALUEs that have already
23161         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
23163 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
23165         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
23167 2013-01-02  Teresa Johnson  <tejohnson@google.com>
23169         * dumpfile.c (dump_loc): Print filename with location.
23170         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
23171         new location_t parameter to emit complete unroll message with
23172         new dump framework.
23173         (canonicalize_loop_induction_variables): Compute loops location
23174         and pass to try_unroll_loop_completely.
23175         * loop-unroll.c (report_unroll_peel): New function.
23176         (peel_loops_completely): Use new dump format with location
23177         for main dumpfile message, and invoke report_unroll_peel on success.
23178         (decide_unrolling_and_peeling): Ditto.
23179         (decide_peel_once_rolling): Remove old dumpfile message subsumed
23180         by report_unroll_peel.
23181         (decide_peel_completely): Ditto.
23182         (decide_unroll_constant_iterations): Ditto.
23183         (decide_unroll_runtime_iterations): Ditto.
23184         (decide_peel_simple): Ditto.
23185         (decide_unroll_stupid): Ditto.
23186         * cfgloop.c (get_loop_location): New function.
23187         * cfgloop.h (get_loop_location): Declare.
23189 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
23191         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
23192         NULL.
23194 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23196         PR middle-end/55198
23197         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
23198         BLKmode objects when EXPAND_MEMORY is specified.
23200 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
23202         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
23203         in loop predicate.
23204         (fold_builtin_cpu): Do not share cpu model decls across statements.
23206 2013-01-02  Jason Merrill  <jason@redhat.com>
23208         PR c++/55804
23209         * tree.c (build_array_type_1): Revert earlier change.
23211 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
23213         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
23214         "cortex-a57".
23215         * config/aarch64/aarch64-tune.md: Re-generate.
23217 2013-01-02  Richard Biener  <rguenther@suse.de>
23219         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
23220         invariant load do not generate a vector load from the scalar location.
23222 2013-01-02  Richard Biener  <rguenther@suse.de>
23224         PR bootstrap/55784
23225         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
23226         * configure: Regenerate.
23228 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
23230         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
23231         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
23232         (expand_builtin_int_roundingfn_2): Keep the original target around
23233         for the fallback case.
23235 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
23237         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
23238         to be clear for sign changes.
23240 2013-01-01  Jan Hubicka  <jh@suse.cz>
23242         * ipa-inline-analysis.c: Fix formatting.
23244 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
23246         PR tree-optimization/55831
23247         * tree-vect-loop.c (get_initial_def_for_induction): Use
23248         gsi_after_labels instead of gsi_start_bb.
23250 Copyright (C) 2013 Free Software Foundation, Inc.
23252 Copying and distribution of this file, with or without modification,
23253 are permitted in any medium without royalty provided the copyright
23254 notice and this notice are preserved.