Add missed .c for the file name in Changelog
[official-gcc.git] / gcc / ChangeLog
blob91dba09a86624356c45e32f48f108bccb2a47808
1 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
3         PR middle-end/58335
4         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
6         pointer elimination with using remove_reg_equal_offset_note.
8 2013-09-12  DJ Delorie  <dj@redhat.com>
10         * config/msp430/: New port.
11         * config.gcc (msp430): Added.
12         * doc/invoke.texi: Document MSP430 options.
13         * doc/install.texi: Document msp430-elf
14         * doc/md.texi: Document msp430-elf
15         * doc/contrib.texi: Document msp430-elf
17         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
18         PSImode.
20 2013-09-12  Martin Jambor  <mjambor@suse.cz>
22         PR ipa/58389
23         * ipa-prop.c (remove_described_reference): Give up if the edge in the
24         reference descriptor is NULL.
25         (ipa_edge_removal_hook): If owning a reference descriptor, set its
26         edge to NULL.
28 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
30         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
31         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
32         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
33         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
34         make_temp_ssa_name): move to tree-ssanames.h
35         * tree-ssa-alias.h: Move prototype.
36         * tree-ssa.h: Include tree-ssanames.h.
37         * tree-ssanames.c (FREE_SSANAMES): Move to here.
38         * tree-ssanames.h: New.  Move items from tree-flow*.h
39         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
41 2013-09-12  Richard Biener  <rguenther@suse.de>
43         PR tree-optimization/58404
44         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
45         propagate non-invariant addresses into dereferences wrapped
46         in component references.
48 2013-09-12  Richard Biener  <rguenther@suse.de>
50         PR tree-optimization/58402
51         * passes.def: Move pass_late_warn_uninitialized later.
53 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
55         * tree-ssa.h: New.  Move content from tree-flow.h and
56         tree-flow-inline.h.
57         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
58         Move prototypes belonging to tree-ssa.c.
59         * tree-flow-inline.h (redirect_edge_var_map_def, 
60         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
61         tree-ssa.h.
62         * gimple.h: Adjust prototypes.
63         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move to...
64         * gimple.c (useless_type_conversion_p, types_compatible_p):  Here.
65         * tree.h: Move prototype to tree-ssa.h.
66         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
67         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
68         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
69         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
70         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
71         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
72         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
73         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
74         graphite-blocking.c, graphite-clast-to-gimple.c, graphite-dependences.c,
75         graphite-interchange.c, graphite-optimize-isl.c, graphite-poly.c,
76         graphite-scop-detection.c, graphite-sese-to-poly.c, graphite.c,
77         ipa-cp.c, ipa-inline-analysis.c, ipa-inline-transform.c, ipa-inline.c,
78         ipa-prop.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa-utils.c,
79         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
80         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
81         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
82         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
83         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
84         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
85         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
86         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
87         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
88         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
89         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
90         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
91         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
92         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
93         value-prof.c, var-tracking.c,
94         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
96 2013-09-12  Richard Biener  <rguenther@suse.de>
98         PR tree-optimization/58396
99         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
100         (build_rdg): Take a loop-nest parameter, fix memleaks.
101         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
103 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
105         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
106         for SLM.
108 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
110         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
111         The FMA instruction names should have a 'v' prefix.
113 2013-09-12  Richard Biener  <rguenther@suse.de>
115         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
116         (dot_rdg): Use popen instead of system in optional code.
117         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
118         (already_processed_vertex_p): Adjust.
119         (has_anti_or_output_dependence, predecessor_has_mem_write,
120         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
121         rdg_flag_uses): Remove.
122         (rdg_flag_vertex): Simplify.
123         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
124         remove recursion.
125         (build_rdg_partition_for_component): Process the first vertex
126         of a component only.
127         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
129 2013-09-12  Alan Modra  <amodra@gmail.com>
131         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
133 2013-09-11  DJ Delorie  <dj@redhat.com>
134             Nick Clifton  <nickc@redhat.com>
136         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
137         (rl78_stack_based_mem): New.
138         * config/rl78/constraints.md (Iv08): New.
139         (Iv16): New.
140         (Iv24): New.
141         (Is09): New.
142         (Is17): New.
143         (Is25): New.
144         (ISsi): New.
145         (IShi): New.
146         (ISqi): New.
147         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
148         (movhi): Likewise.
149         (movsi): Change from expand to insn-and-split.
150         (ashrsi3): Clobber AX.
151         (lshrsi3): New.
152         (ashlsi3): New.
153         (cbranchsi4): New.
154         * config/rl78/rl78.md (CC_REG): Fix.
155         (addsi3): Allow memory and immediate operands.
156         (addsi3_internal): Split into...
157         (addsi3_internal_virt): ...new, and ...
158         (addsi3_internal_real): ...new.
159         (subsi): New.
160         (subsi3_internal_virt): New.
161         (subsi3_internal_real): New.
162         (mulsi3): Add memory operand.
163         (mulsi3_rl78): Likewise.
164         (mulsi3_g13): Likewise.
165         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
166         (cbranchqi4_real): Add more constraint options.
167         (cbranchhi4_real): Expand pattern.
168         (cbranchhi4_real_signed): New.
169         (cbranchhi4_real_inverted): New.
170         (cbranchsi4_real_lt): New.
171         (cbranchsi4_real_ge): New.
172         (cbranchsi4_real_signed): New.
173         (cbranchsi4_real): New.
174         (peephole2): New.
175         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
176         constant shifts.
177         (lshrsi3_virt): Likewise.
178         (ashlsi3_virt): Likewise.
179         (cbranchqi4_virt_signed): New.
180         (cbranchhi4_virt_signed): New.
181         (cbranchsi4_virt): New.
182         * config/rl78/rl78.c: Whitespace fixes throughout.
183         (move_elim_pass): New.
184         (pass_data_rl78_move_elim): New.
185         (pass_rl78_move_elim): New.
186         (make_pass_rl78_move_elim): New.
187         (rl78_devirt_info): Run devirt earlier.
188         (rl78_move_elim_info): New.
189         (rl78_asm_file_start): Register it.
190         (rl78_split_movsi): New.
191         (rl78_as_legitimate_address): Allow virtual base registers when
192         appropriate.
193         (rl78_addr_space_convert): Remove spurious debug stuff.
194         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
195         (rl78_print_operand): More cases for not printing '#'.
196         (rl78_expand_compare): Remove most of the logic.
197         (content_memory): New.
198         (clear_content_memory): New.
199         (get_content_index): New.
200         (get_content_name): New.
201         (display_content_memory): New.
202         (update_content): New.
203         (record_content): New.
204         (already_contains): New.
205         (insn_ok_now): Re-recog insns with virtual registers.
206         (add_postponed_content_update): New.
207         (process_postponed_content_update): New.
208         (gen_and_emit_move): New.
209         (transcode_memory_rtx): Record new location content.
210         Use gen_and_emit_move.
211         (force_into_acc): New.
212         (move_to_acc): Use gen_and_emit_move.
213         (move_from_acc): Likewise.
214         (move_acc_to_reg): Likewise.
215         (move_to_x): Likewise.
216         (move_to_hl): Likewise.
217         (move_to_de): Likewise.
218         (rl78_alloc_physical_registers_op1): Record location content.
219         (has_constraint): New.
220         (rl78_alloc_physical_registers_op2): Record location content.
221         Optimize use of HL.
222         (rl78_alloc_physical_registers_ro1): Likewise.
223         (rl78_alloc_physical_registers_cmp): Likewise.
224         (rl78_alloc_physical_registers_umul): Likewise.
225         (rl78_alloc_address_registers_macax): New.
226         (rl78_alloc_physical_registers): Initialize and set location
227         content memory as needed.
228         (rl78_reorg): Make sure split2 is called.
229         (rl78_rtx_costs): New.
231 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
233         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
234         for (not (neg ...)) and (neg (not ...)) cases.
236 2013-09-11  Richard Biener  <rguenther@suse.de>
238         PR middle-end/58377
239         * passes.def: Split critical edges before late uninit warning passes.
240         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
242 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
244         PR tree-optimization/58385
245         * fold-const.c (build_range_check): If both low and high are NULL,
246         use omit_one_operand_loc to preserve exp side-effects.
248 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
250         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
252 2013-09-11  Richard Biener  <rguenther@suse.de>
254         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
255         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
256         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
257         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
258         stmts_from_loop, known_dependences_p, build_empty_rdg,
259         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
260         * tree-loop-distribution.c: ... here.
261         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
262         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
263         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
264         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
265         * tree-loop-distribution.c: ... here.
266         * tree-loop-distribution.c: Include gimple-pretty-print.h.
267         (struct partition_s): Add loops member.
268         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
269         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
270         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
272 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
273             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
274             Sergey Lega  <sergey.s.lega@intel.com>
275             Anna Tikhonova  <anna.tikhonova@intel.com>
276             Ilya Tocar  <ilya.tocar@intel.com>
277             Andrey Turetskiy  <andrey.turetskiy@intel.com>
278             Ilya Verbin  <ilya.verbin@intel.com>
279             Kirill Yukhin  <kirill.yukhin@intel.com>
280             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
282         * config/i386/constraints.md (k): New.
283         (Yk): Ditto.
284         * config/i386/i386.c (const regclass_map): Add new mask registers.
285         (dbx_register_map): Ditto.
286         (dbx64_register_map): Ditto.
287         (svr4_dbx_register_map): Ditto.
288         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
289         disabled.
290         (ix86_preferred_reload_class): Disable constants for mask registers.
291         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
292         (ix86_hard_regno_mode_ok): Support new mask registers.
293         (x86_order_regs_for_local_alloc): Ditto.
294         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
295         (FIXED_REGISTERS): Add new mask registers.
296         (CALL_USED_REGISTERS): Ditto.
297         (REG_ALLOC_ORDER): Ditto.
298         (VALID_MASK_REG_MODE): New.
299         (FIRST_MASK_REG): Ditto.
300         (LAST_MASK_REG): Ditto.
301         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
302         (MAYBE_MASK_CLASS_P): New.
303         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
304         (REG_CLASS_CONTENTS): Ditto.
305         (MASK_REGNO_P): New.
306         (ANY_MASK_REG_P): Ditto.
307         (HI_REGISTER_NAMES): Add new mask registers.
308         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
309         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
310         mask registers.
311         (attribute "type"): Add mskmov, msklog.
312         (attribute "length_immediate"): Support them.
313         (attribute "memory"): Ditto.
314         (attribute "prefix_0f"): Ditto.
315         (*movhi_internal): Support new mask registers.
316         (*movqi_internal): Ditto.
317         (define_split): Split out clobber pattern is a logic
318         insn on mask registers.
319         (*k<logic><mode>): New.
320         (*andhi_1): Extend to support mask regs.
321         (*andqi_1): Extend to support mask regs.
322         (kandn<mode>): New.
323         (define_split): Split and-not to and and not if operands
324         are not mask regs.
325         (*<code><mode>_1): Separate HI mode to new pattern...
326         (*<code>hi_1): This.
327         (*<code>qi_1): Extend to support mask regs.
328         (kxnor<mode>): New.
329         (kortestzhi): Ditto.
330         (kortestchi): Ditto.
331         (kunpckhi): Ditto.
332         (*one_cmpl<mode>2_1): Remove HImode and handle it...
333         (*one_cmplhi2_1): ...Here, now with mask registers support.
334         (*one_cmplqi2_1): Support new mask registers.
335         (HI/QImode arithmetics splitter): Don't split if mask registers
336         are used.
337         (HI/QImode not splitter): Ditto.
338         * config/i386/predicated.md (mask_reg_operand): New.
339         (general_reg_operand): Ditto.
341 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
343         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
344         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
346 2013-09-10  Jeff Law  <law@redhat.com>
348         PR tree-optimization/58380
349         * tree-ssa-threadupdate.c (thread_block): Recognize another case
350         of threading through a buried loop header.
352         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
353         return value for single successor case.
355 2013-09-10  Jan Hubicka  <jh@suse.cz>
357         * ipa-devirt.c (ipa_devirt): Enable with LTO.
359 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
361         PR target/58361
362         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
363         support conditional execution.
364         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
366 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
368         * lra.c (lra): Clear lra_optional_reload_pseudos before every
369         constraint pass.
370         * lra-constraints.c (curr_insn_transform): Switch on optional
371         reloads.  Check destination too to check move insn.
372         (undo_optional_reloads): Add check that the original peudo did not
373         changed its allocation and the optional reload was inherited on
374         last inheritance pass.  Break loop after deciding to keep optional
375         reload.
376         (lra_undo_inheritance): Add check that inherited pseudo still in
377         memory.
379 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
381         * config/aarch64/aarch64.md (generic_sched): New.
382         * config/aarch64/aarch64-generic.md (load): Make conditional
383         on generic_sched attribute.
384         (nonload): Likewise.
386 2013-09-10  Jan Hubicka  <jh@suse.cz>
388         * lto-cgraph.c: Include ipa-utils.h.
389         (compute_ltrans_boundary): Also add possible targets into the boundary.
391 2013-09-10  Jan Hubicka  <jh@suse.cz>
393         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
394         VAR_DECL of vtable rather than full expression.
396 2013-09-10  Jan Hubicka  <jh@suse.cz>
397             Paolo Carlini  <paolo.carlini@oracle.com>
399         * cgraphunit.c (analyze_functions): Save input_location, set it
400         to UNKNOWN_LOCATION and restore it at the end.
402 2013-09-10  Martin Jambor  <mjambor@suse.cz>
404         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
405         represented by a thunk.
407 2013-09-10  Jeff Law  <law@redhat.com>
409         PR tree-optimization/58343
410         * tree-ssa-threadupdate.c (thread_block): Identify and disable
411         jump threading requests through loop headers buried in the middle
412         of a jump threading path.
414         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
415         in return value/type.
417 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
419         PR rtl-optimization/58365
420         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
421         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
422         it differs.
424 2013-09-10  Richard Biener  <rguenther@suse.de>
426         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
427         * tree-data-ref.c (create_rdg_vertices): Collect all data
428         references, signal failure to the caller, use data-ref API.
429         (build_rdg): Compute data references only once.  Maintain lifetime
430         of data references and data dependences from within RDG.
431         (free_rdg): Free dependence relations.
432         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
433         inventing extra dependences.
434         (distribute_loop): Update for RDG API changes.
436 2013-09-10  Kai Tietz  <ktietz@redhat.com>
438         * doc/invoke.texi (fms-extensions): Document changed
439         behavior for ms-abi targets.
440         * config/i386/i386.c (ix86_option_override_internal):
441         Set default value of option -fms-extension for ms-abi targets.
443 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
445         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
447 2013-09-10  Alan Modra  <amodra@gmail.com>
449         PR target/58330
450         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
452 2013-09-10  Alan Modra  <amodra@gmail.com>
454         * config/rs6000/predicates.md (add_cint_operand): New.
455         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
456         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
457         using add_cint_operand.
458         (largetoc_high_plus_aix): Likewise.
460 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
462         PR tree-optimization/58364
463         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
464         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
465         the current range can't be an unconditional true or false.
467 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
469         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
471 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
473         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
475 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
477         PR c++/43452
478         * doc/invoke.texi (-Wdelete-incomplete): Document it.
480 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
482         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
483         NO_REGS for immediate that can't be moved directly into FP_REGS.
485 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
487         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
488         comparison with negated operand.
489         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
490         RTL form.
492 2013-09-09  Richard Biener  <rguenther@suse.de>
494         PR middle-end/58326
495         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
496         of a subloop record all its block as affecting loop-closed SSA form.
498 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
500         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
501         of an rtx/bitpos pair.
502         (store_fixed_bit_field): Update accordingly.
504 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
506         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
507         GEN_INT.
508         * builtins.c (expand_errno_check): Likewise.
509         * dwarf2cfi.c (init_return_column_size): Likewise.
510         * except.c (sjlj_mark_call_sites): Likewise.
511         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
512         * lra-constraints.c (emit_inc): Likewise.
513         * ree.c (combine_set_extension): Likewise.
514         * regmove.c (fixup_match_2): Likewise.
515         * reload1.c (inc_for_reload): Likewise.
517 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
519         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
520         (force_to_mode, simplify_shift_const_1, simplify_comparison):
521         Use gen_int_mode with the mode of the associated simplify_* call.
522         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
523         * expmed.c (expand_shift_1): Likewise.
524         * function.c (instantiate_virtual_regs_in_insn): Likewise.
525         * loop-iv.c (iv_number_of_iterations): Likewise.
526         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
527         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
529 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
531         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
532         of the associated expand_* call.
533         (asan_emit_stack_protection): Likewise.
534         * builtins.c (round_trampoline_addr): Likewise.
535         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
536         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
537         (emit_store_flag): Likewise.
538         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
539         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
540         Likewise.
541         * function.c (instantiate_virtual_regs_in_insn): Likewise.
542         * ifcvt.c (noce_try_store_flag_constants): Likewise.
543         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
544         * modulo-sched.c (generate_prolog_epilog): Likewise.
545         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
546         (expand_ctz, expand_ffs, expand_unop): Likewise.
548 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
550         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
551         of the associated gen_rtx_* call.
552         * caller-save.c (init_caller_save): Likewise.
553         * combine.c (find_split_point, make_extraction): Likewise.
554         (make_compound_operation): Likewise.
555         * dwarf2out.c (mem_loc_descriptor): Likewise.
556         * explow.c (plus_constant, probe_stack_range): Likewise.
557         * expmed.c (expand_mult_const): Likewise.
558         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
559         * reload1.c (init_reload): Likewise.
560         * valtrack.c (cleanup_auto_inc_dec): Likewise.
561         * var-tracking.c (adjust_mems): Likewise.
562         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
563         rather than gen_rtx_fmt_ee.
565 2013-09-09  Jan Hubicka  <jh@suse.cz>
567         PR middle-end/58294
568         * value-prof.c (gimple_ic): Copy also abnormal edges.
570 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
572         * asan.c (asan_shadow_cst): Use gen_int_mode.
574 2013-09-08  Jan Hubicka  <jh@suse.cz>
576         * ipa-profile.c: Add toplevel comment.
577         (ipa_propagate_frequency_1): Be more conservative when profile is read.
578         (contains_hot_call_p): New function.
579         (ipa_propagate_frequency): Set frequencies based on counts when
580         profile is read.
581         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
582         profile; do not tamper with profile after inlining if it is read.
584 2013-09-08  Jan Hubicka  <jh@suse.cz>
586         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
587         speculative edges.
589 2013-09-08  Jan Hubicka  <jh@suse.cz>
591         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
592         summary generation.
594 2013-09-08  Jeff Law  <law@redhat.com>
596         PR bootstrap/58340
597         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
598         of 'found'.
600 2013-09-08  Andi Kleen  <ak@linux.intel.com>
602         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
604 2013-09-08  Jan Hubicka  <jh@suse.cz>
606         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
608 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
610         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
611         for non-debug insns.
612         * lra.c (new_insn_reg): Take the containing insn as a parameter.
613         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
614         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
615         accordingly.
617 2013-09-08  Jan Hubicka  <jh@suse.cz>
619         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
620         targets and devirtualize to BUILT_IN_UNREACHABLE.
621         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
622         * ipa.c (walk_polymorphic_call_targets): New function.
623         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
624         functions; use the new timevar.
625         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
626         was removed from the program.
627         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
628         not consider it in the walk when its vtable is dead.
629         (possible_polymorphic_call_targets_1): Pass anonymous flag to
630         record_binfo.
631         (devirt_variable_node_removal_hook): New function.
632         (possible_polymorphic_call_targets): Also register
633         devirt_variable_node_removal_hook.
634         (ipa_devirt): Do not do non-speculative devirtualization.
635         (gate_ipa_devirt): One execute if devirtualizing speculatively.
637 2013-09-08  Jan Hubicka  <jh@suse.cz>
639         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
640         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
641         varpool_remove_variable_insertion_hook): Declare.
642         * varpool.c (varpool_node_hook_list): New structure.
643         (first_varpool_node_removal_hook,
644         first_varpool_variable_insertion_hook): New variables.
645         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
646         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
647         varpool_remove_variable_insertion_hook,
648         varpool_call_variable_insertion_hooks): New functions.
649         (varpool_remove_node): Use it.
651 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
653         PR c++/54941
654         * diagnostic.c (diagnostic_build_prefix): When s.file is
655         "<built-in>" don't output line and column numbers.
657 2013-09-06  Jan Hubicka  <jh@suse.cz>
659         * cgraphunit.c (expand_thunk): Get body before touching arguments.
660         * lto-streamer-out.c: Stream thunks, too.
661         * lto-streamer-in.c (input_function): Pop cfun here
662         (lto_read_body): Instead of here.
664 2013-09-06  Caroline Tice  <cmtice@google.com>
666         * doc/install.texi: Add documentation for the --enable-vtable-verify
667         and the --disable-libvtv configure options.
669 2013-09-06  Jeff Law  <law@redhat.com>
671         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
672         edge implied equivalences into successor phis.
674 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
676         * resource.c (mark_referenced_resources): Handle COND_EXEC.
678 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
680         * resource.c (mark_target_live_regs): Compute resources taking
681         into account if a call is predicated or not.
683 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
685         * toplev.c (output_stack_usage): Be prepared for suffixes created by
686         the compiler in the function names.
688 2013-09-06  Jan Hubicka  <jh@suse.cz>
690         PR middle-end/58094
691         * ipa-inline.c (has_caller_p): New function.
692         (want_inline_function_to_all_callers_p): Use it.
693         (sum_callers, inline_to_all_callers): Break out from ...
694         (ipa_inline): ... here.
696 2013-09-06  Jan Hubicka  <jh@suse.cz>
698         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
699         only when AVX is enabled.
701 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
703         * config/aarch64/aarch64.md
704         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
705         is fpsimd_<load/store>2.
706         (load_pair<mode>): Likewise.
707         (store_pair<mode>): Likewise.
709 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
711         * config/arm/types.md (type): Add "mrs" type.
712         * config/aarch64/aarch64.md
713         (aarch64_load_tp_hard): Make type "mrs".
714         * config/arm/arm.md
715         (load_tp_hard): Make type "mrs".
716         * config/arm/cortex-a15.md: Update with new attributes.
717         * config/arm/cortex-a5.md: Update with new attributes.
718         * config/arm/cortex-a53.md: Update with new attributes.
719         * config/arm/cortex-a7.md: Update with new attributes.
720         * config/arm/cortex-a8.md: Update with new attributes.
721         * config/arm/cortex-a9.md: Update with new attributes.
722         * config/arm/cortex-m4.md: Update with new attributes.
723         * config/arm/cortex-r4.md: Update with new attributes.
724         * config/arm/fa526.md: Update with new attributes.
725         * config/arm/fa606te.md: Update with new attributes.
726         * config/arm/fa626te.md: Update with new attributes.
727         * config/arm/fa726te.md: Update with new attributes.
729 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
731         * config/aarch64/aarch64.md
732         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
733         (*movtf_aarch64): Likewise.
734         * config/arm/arm.md
735         (thumb1_movdi_insn): Use "multiple" for type where more than one
736         instruction is used for a move.
737         (*arm32_movhf): Likewise.
738         (*thumb_movdf_insn): Likewise.
740 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
742         * config/arm/types.md (type): Rename fcpys to fmov.
743         * config/arm/vfp.md
744         (*arm_movsi_vfp): Rename type fcpys as fmov.
745         (*thumb2_movsi_vfp): Likewise
746         (*movhf_vfp_neon): Likewise
747         (*movhf_vfp): Likewise
748         (*movsf_vfp): Likewise
749         (*thumb2_movsf_vfp): Likewise
750         (*movsfcc_vfp): Likewise
751         (*thumb2_movsfcc_vfp): Likewise
752         * config/aarch64/aarch64-simd.md
753         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
754         * config/aarch64/aarch64.md
755         (*movsi_aarch64): Replace type mov_reg with fmovs.
756         (*movdi_aarch64): Likewise
757         (*movsf_aarch64): Likewise
758         (*movdf_aarch64): Likewise
759         * config/arm/arm.c
760         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
761         * config/arm/iwmmxt.md
762         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
763         * config/arm/arm1020e.md: Update with new attributes.
764         * config/arm/cortex-a15-neon.md: Update with new attributes.
765         * config/arm/cortex-a5.md: Update with new attributes.
766         * config/arm/cortex-a53.md: Update with new attributes.
767         * config/arm/cortex-a7.md: Update with new attributes.
768         * config/arm/cortex-a8-neon.md: Update with new attributes.
769         * config/arm/cortex-a9.md: Update with new attributes.
770         * config/arm/cortex-m4-fpu.md: Update with new attributes.
771         * config/arm/cortex-r4f.md: Update with new attributes.
772         * config/arm/marvell-pj4.md: Update with new attributes.
773         * config/arm/vfp11.md: Update with new attributes.
775 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
777         * config/aarch64/aarch64.md
778         (*madd<mode>): Fix type attribute.
779         (*maddsi_uxtw): Likewise.
780         (*msub<mode>): Likewise.
781         (*msubsi_uxtw): Likewise.
782         (<su_optab>maddsidi4): Likewise.
783         (<su_optab>msubsidi4): Likewise.
785 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
787         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
788         * config/arm/arm.md (core_cycles): Remove fdiv.
789         * config/arm/vfp.md:
790         (*sqrtsf2_vfp): Update for attribute changes.
791         (*sqrtdf2_vfp): Likewise.
792         * config/aarch64/aarch64.md:
793         (sqrt<mode>2): Update for attribute changes.
794         * config/arm/arm1020e.md: Update with new attributes.
795         * config/arm/cortex-a15-neon.md: Update with new attributes.
796         * config/arm/cortex-a5.md: Update with new attributes.
797         * config/arm/cortex-a53.md: Update with new attributes.
798         * config/arm/cortex-a7.md: Update with new attributes.
799         * config/arm/cortex-a8-neon.md: Update with new attributes.
800         * config/arm/cortex-a9.md: Update with new attributes.
801         * config/arm/cortex-m4-fpu.md: Update with new attributes.
802         * config/arm/cortex-r4f.md: Update with new attributes.
803         * config/arm/marvell-pj4.md: Update with new attributes.
804         * config/arm/vfp11.md: Update with new attributes.
806 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
808         * config/arm/types.md
809         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
810         * config/aarch64/aarch64.md
811         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
812         new attributes.
813         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
814         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
815         (float<GPI:mode><GPF:mode>2): Likewise.
816         * config/arm/vfp.md
817         (*truncsisf2_vfp): Update with new attributes.
818         (*truncsidf2_vfp): Likewise.
819         (fixuns_truncsfsi2): Likewise.
820         (fixuns_truncdfsi2): Likewise.
821         (*floatsisf2_vfp): Likewise.
822         (*floatsidf2_vfp): Likewise.
823         (floatunssisf2): Likewise.
824         (floatunssidf2): Likewise.
825         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
826         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
827         * config/arm/arm1020e.md: Update with new attributes.
828         * config/arm/cortex-a15-neon.md: Update with new attributes.
829         * config/arm/cortex-a5.md: Update with new attributes.
830         * config/arm/cortex-a53.md: Update with new attributes.
831         * config/arm/cortex-a7.md: Update with new attributes.
832         * config/arm/cortex-a8-neon.md: Update with new attributes.
833         * config/arm/cortex-a9.md: Update with new attributes.
834         * config/arm/cortex-m4-fpu.md: Update with new attributes.
835         * config/arm/cortex-r4f.md: Update with new attributes.
836         * config/arm/marvell-pj4.md: Update with new attributes.
837         * config/arm/vfp11.md: Update with new attributes.
839 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
841         * config/aarch64/arm_neon.h
842         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
843         (vqtbx<1,2,3,4><q>_s8): Likewise.
845 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
847         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
848         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
849         (add<mode>3): Add type attribute.
850         (add<mode>3): Likewise.
851         (usadd<mode>3): Likewise.
852         (ssadd<mode>3): Likewise.
853         (sub<mode>3): Likewise.
854         (sub<mode>3): Likewise.
855         (ussub<mode>3): Likewise.
856         (sssub<mode>3): Likewise.
857         (ssmulsa3): Likewise.
858         (usmulusa3): Likewise.
859         (arm_usatsihi): Likewise.
860         * config/arm/vfp.md
861         (*movdi_vfp): Add types for all instructions.
862         (*movdi_vfp_cortexa8): Likewise.
863         (*movhf_vfp_neon): Likewise.
864         (*movhf_vfp): Likewise.
865         (*movdf_vfp): Likewise.
866         (*thumb2_movdf_vfp): Likewise.
867         (*thumb2_movdfcc_vfp): Likewise.
868         * config/arm/arm.md: Add type attribute to all insn patterns.
869         (*thumb1_adddi3): Add type attribute.
870         (*arm_adddi3): Likewise.
871         (*adddi_sesidi_di): Likewise.
872         (*adddi_zesidi_di): Likewise.
873         (*thumb1_addsi3): Likewise.
874         (addsi3_compare0): Likewise.
875         (*addsi3_compare0_scratch): Likewise.
876         (*compare_negsi_si): Likewise.
877         (cmpsi2_addneg): Likewise.
878         (*addsi3_carryin_<optab>): Likewise.
879         (*addsi3_carryin_alt2_<optab>): Likewise.
880         (*addsi3_carryin_clobercc_<optab>): Likewise.
881         (*subsi3_carryin): Likewise.
882         (*subsi3_carryin_const): Likewise.
883         (*subsi3_carryin_compare): Likewise.
884         (*subsi3_carryin_compare_const): Likewise.
885         (*arm_subdi3): Likewise.
886         (*thumb_subdi3): Likewise.
887         (*subdi_di_zesidi): Likewise.
888         (*subdi_di_sesidi): Likewise.
889         (*subdi_zesidi_di): Likewise.
890         (*subdi_sesidi_di): Likewise.
891         (*subdi_zesidi_ze): Likewise.
892         (thumb1_subsi3_insn): Likewise.
893         (*arm_subsi3_insn): Likewise.
894         (*anddi3_insn): Likewise.
895         (*anddi_zesidi_di): Likewise.
896         (*anddi_sesdi_di): Likewise.
897         (*ne_zeroextracts): Likewise.
898         (*ne_zeroextracts): Likewise.
899         (*ite_ne_zeroextr): Likewise.
900         (*ite_ne_zeroextr): Likewise.
901         (*anddi_notdi_di): Likewise.
902         (*anddi_notzesidi): Likewise.
903         (*anddi_notsesidi): Likewise.
904         (andsi_notsi_si): Likewise.
905         (thumb1_bicsi3): Likewise.
906         (*iordi3_insn): Likewise.
907         (*iordi_zesidi_di): Likewise.
908         (*iordi_sesidi_di): Likewise.
909         (*thumb1_iorsi3_insn): Likewise.
910         (*xordi3_insn): Likewise.
911         (*xordi_zesidi_di): Likewise.
912         (*xordi_sesidi_di): Likewise.
913         (*arm_xorsi3): Likewise.
914         (*andsi_iorsi3_no): Likewise.
915         (*smax_0): Likewise.
916         (*smax_m1): Likewise.
917         (*arm_smax_insn): Likewise.
918         (*smin_0): Likewise.
919         (*arm_smin_insn): Likewise.
920         (*arm_umaxsi3): Likewise.
921         (*arm_uminsi3): Likewise.
922         (*minmax_arithsi): Likewise.
923         (*minmax_arithsi_): Likewise.
924         (*satsi_<SAT:code>): Likewise.
925         (arm_ashldi3_1bit): Likewise.
926         (arm_ashrdi3_1bit): Likewise.
927         (arm_lshrdi3_1bit): Likewise.
928         (*arm_negdi2): Likewise.
929         (*thumb1_negdi2): Likewise.
930         (*arm_negsi2): Likewise.
931         (*thumb1_negsi2): Likewise.
932         (*negdi_extendsid): Likewise.
933         (*negdi_zero_extend): Likewise.
934         (*arm_abssi2): Likewise.
935         (*thumb1_abssi2): Likewise.
936         (*arm_neg_abssi2): Likewise.
937         (*thumb1_neg_abss): Likewise.
938         (one_cmpldi2): Likewise.
939         (extend<mode>di2): Likewise.
940         (*compareqi_eq0): Likewise.
941         (*arm_extendhisi2addsi): Likewise.
942         (*arm_movdi): Likewise.
943         (*thumb1_movdi_insn): Likewise.
944         (*arm_movt): Likewise.
945         (*thumb1_movsi_insn): Likewise.
946         (pic_add_dot_plus_four): Likewise.
947         (pic_add_dot_plus_eight): Likewise.
948         (tls_load_dot_plus_eight): Likewise.
949         (*thumb1_movhi_insn): Likewise.
950         (*thumb1_movsf_insn): Likewise.
951         (*movdf_soft_insn): Likewise.
952         (*thumb_movdf_insn): Likewise.
953         (cbranchsi4_insn): Likewise.
954         (cbranchsi4_scratch): Likewise.
955         (*negated_cbranchsi4): Likewise.
956         (*tbit_cbranch): Likewise.
957         (*tlobits_cbranch): Likewise.
958         (*tstsi3_cbranch): Likewise.
959         (*cbranchne_decr1): Likewise.
960         (*addsi3_cbranch): Likewise.
961         (*addsi3_cbranch_scratch): Likewise.
962         (*arm_cmpdi_insn): Likewise.
963         (*arm_cmpdi_unsig): Likewise.
964         (*arm_cmpdi_zero): Likewise.
965         (*thumb_cmpdi_zero): Likewise.
966         (*deleted_compare): Likewise.
967         (*mov_scc): Likewise.
968         (*mov_negscc): Likewise.
969         (*mov_notscc): Likewise.
970         (*cstoresi_eq0_thumb1_insn): Likewise.
971         (cstoresi_nltu_thumb1): Likewise.
972         (cstoresi_ltu_thu): Likewise.
973         (thumb1_addsi3_addgeu): Likewise.
974         (*arm_jump): Likewise.
975         (*thumb_jump): Likewise.
976         (*check_arch2): Likewise.
977         (arm_casesi_internal): Likewise.
978         (thumb1_casesi_dispatch): Likewise.
979         (*arm_indirect_jump): Likewise.
980         (*thumb1_indirect_jump): Likewise.
981         (nop): Likewise.
982         (*and_scc): Likewise.
983         (*ior_scc): Likewise.
984         (*compare_scc): Likewise.
985         (*cond_move): Likewise.
986         (*cond_arith): Likewise.
987         (*cond_sub): Likewise.
988         (*cmp_ite0): Likewise.
989         (*cmp_ite1): Likewise.
990         (*cmp_and): Likewise.
991         (*cmp_ior): Likewise.
992         (*ior_scc_scc): Likewise.
993         (*ior_scc_scc_cmp): Likewise.
994         (*and_scc_scc): Likewise.
995         (*and_scc_scc_cmp): Likewise.
996         (*and_scc_scc_nod): Likewise.
997         (*negscc): Likewise.
998         (movcond_addsi): Likewise.
999         (movcond): Likewise.
1000         (*ifcompare_plus_move): Likewise.
1001         (*if_plus_move): Likewise.
1002         (*ifcompare_move_plus): Likewise.
1003         (*if_move_plus): Likewise.
1004         (*ifcompare_arith_arith): Likewise.
1005         (*if_arith_arith): Likewise.
1006         (*ifcompare_arith_move): Likewise.
1007         (*if_arith_move): Likewise.
1008         (*ifcompare_move_arith): Likewise.
1009         (*if_move_arith): Likewise.
1010         (*ifcompare_move_not): Likewise.
1011         (*if_move_not): Likewise.
1012         (*ifcompare_not_move): Likewise.
1013         (*if_not_move): Likewise.
1014         (*ifcompare_shift_move): Likewise.
1015         (*if_shift_move): Likewise.
1016         (*ifcompare_move_shift): Likewise.
1017         (*if_move_shift): Likewise.
1018         (*ifcompare_shift_shift): Likewise.
1019         (*ifcompare_not_arith): Likewise.
1020         (*ifcompare_arith_not): Likewise.
1021         (*if_arith_not): Likewise.
1022         (*ifcompare_neg_move): Likewise.
1023         (*if_neg_move): Likewise.
1024         (*ifcompare_move_neg): Likewise.
1025         (*if_move_neg): Likewise.
1026         (prologue_thumb1_interwork): Likewise.
1027         (*cond_move_not): Likewise.
1028         (*sign_extract_onebit): Likewise.
1029         (*not_signextract_onebit): Likewise.
1030         (stack_tie): Likewise.
1031         (align_4): Likewise.
1032         (align_8): Likewise.
1033         (consttable_end): Likewise.
1034         (consttable_1): Likewise.
1035         (consttable_2): Likewise.
1036         (consttable_4): Likewise.
1037         (consttable_8): Likewise.
1038         (consttable_16): Likewise.
1039         (*thumb1_tablejump): Likewise.
1040         (prefetch): Likewise.
1041         (force_register_use): Likewise.
1042         (thumb_eh_return): Likewise.
1043         (load_tp_hard): Likewise.
1044         (load_tp_soft): Likewise.
1045         (tlscall): Likewise.
1046         (*arm_movtas_ze): Likewise.
1047         (*arm_rev): Likewise.
1048         (*arm_revsh): Likewise.
1049         (*arm_rev16): Likewise.
1050         * config/arm/thumb2.md
1051         (*thumb2_smaxsi3): Likewise.
1052         (*thumb2_sminsi3): Likewise.
1053         (*thumb32_umaxsi3): Likewise.
1054         (*thumb2_uminsi3): Likewise.
1055         (*thumb2_negdi2): Likewise.
1056         (*thumb2_abssi2): Likewise.
1057         (*thumb2_neg_abss): Likewise.
1058         (*thumb2_movsi_insn): Likewise.
1059         (tls_load_dot_plus_four): Likewise.
1060         (*thumb2_movhi_insn): Likewise.
1061         (*thumb2_mov_scc): Likewise.
1062         (*thumb2_mov_negs): Likewise.
1063         (*thumb2_mov_negs): Likewise.
1064         (*thumb2_mov_nots): Likewise.
1065         (*thumb2_mov_nots): Likewise.
1066         (*thumb2_movsicc_): Likewise.
1067         (*thumb2_movsfcc_soft_insn): Likewise.
1068         (*thumb2_indirect_jump): Likewise.
1069         (*thumb2_and_scc): Likewise.
1070         (*thumb2_ior_scc): Likewise.
1071         (*thumb2_ior_scc_strict_it): Likewise.
1072         (*thumb2_cond_move): Likewise.
1073         (*thumb2_cond_arith): Likewise.
1074         (*thumb2_cond_ari): Likewise.
1075         (*thumb2_cond_sub): Likewise.
1076         (*thumb2_negscc): Likewise.
1077         (*thumb2_movcond): Likewise.
1078         (thumb2_casesi_internal): Likewise.
1079         (thumb2_casesi_internal_pic): Likewise.
1080         (*thumb2_alusi3_short): Likewise.
1081         (*thumb2_mov<mode>_shortim): Likewise.
1082         (*thumb2_addsi_short): Likewise.
1083         (*thumb2_subsi_short): Likewise.
1084         (thumb2_addsi3_compare0): Likewise.
1085         (*thumb2_cbz): Likewise.
1086         (*thumb2_cbnz): Likewise.
1087         (*thumb2_one_cmplsi2_short): Likewise.
1088         (*thumb2_negsi2_short): Likewise.
1089         (*orsi_notsi_si): Likewise.
1090         * config/arm/arm1020e.md: Update with new attributes.
1091         * config/arm/arm1026ejs.md: Update with new attributes.
1092         * config/arm/arm1136jfs.md: Update with new attributes.
1093         * config/arm/arm926ejs.md: Update with new attributes.
1094         * config/arm/cortex-a15.md: Update with new attributes.
1095         * config/arm/cortex-a5.md: Update with new attributes.
1096         * config/arm/cortex-a53.md: Update with new attributes.
1097         * config/arm/cortex-a7.md: Update with new attributes.
1098         * config/arm/cortex-a8.md: Update with new attributes.
1099         * config/arm/cortex-a9.md: Update with new attributes.
1100         * config/arm/cortex-m4.md: Update with new attributes.
1101         * config/arm/cortex-r4.md: Update with new attributes.
1102         * config/arm/fa526.md: Update with new attributes.
1103         * config/arm/fa606te.md: Update with new attributes.
1104         * config/arm/fa626te.md: Update with new attributes.
1105         * config/arm/fa726te.md: Update with new attributes.
1107 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1109         * config/aarch64/aarch64-simd.md
1110         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1111         <vwx> iterator to ensure correct register choice.
1112         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1113         (aarch64_sqdmull_n<mode>): Likewise.
1114         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1115         * config/aarch64/arm_neon.h
1116         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1117         (vml<as><q>_n_<su>16): Likewise.
1118         (vml<as>l_high_lane<q>_<su>16): Likewise.
1119         (vml<as>l_high_n_<su>16): Likewise.
1120         (vml<as>l_lane<q>_<su>16): Likewise.
1121         (vml<as>l_n_<su>16): Likewise.
1122         (vmul<q>_lane<q>_<su>16): Likewise.
1123         (vmul<q>_n_<su>16): Likewise.
1124         (vmull_lane<q>_<su>16): Likewise.
1125         (vmull_n_<su>16): Likewise.
1126         (vmull_high_lane<q>_<su>16): Likewise.
1127         (vmull_high_n_<su>16): Likewise.
1128         (vqrdmulh<q>_n_s16): Likewise.
1130 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
1132         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1133         have the correct lane parameter.
1135 2013-09-06  Richard Biener <rguenther@suse.de>
1137         * cfganal.c (control_dependences::~control_dependences):
1138         Properly free all of the vector.
1140 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
1142         PR target/58269
1143         * config/i386/i386.c (ix86_conditional_register_usage):
1144         Proper initialize extended SSE registers.
1146 2013-09-06  Jan Hubicka  <jh@suse.cz>
1148         PR tree-optimization/58311
1149         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1151 2013-09-06  Jan Hubicka  <jh@suse.cz>
1153         * Makefile.in (tree-sra.o): Update dependencies.
1154         * tree-sra.c: Include ipa-utils.h
1155         (scan_function): Use recursive_call_p.
1156         (has_caller_p): New function.
1157         (cgraph_for_node_and_aliases): Count also callers of aliases.
1159 2013-09-06  Jan Hubicka  <jh@suse.cz>
1161         PR middle-end/58094
1162         * cgraph.h (symtab_semantically_equivalent_p): Declare.
1163         * tree-tailcall.c: Include ipa-utils.h.
1164         (find_tail_calls): Use it.
1165         * ipa-pure-const.c (check_call): Likewise.
1166         * ipa-utils.c (recursive_call_p): New function.
1167         * ipa-utils.h (recursive_call_p): Dclare.
1168         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1169         (symtab_semantically_equivalent_p): New function.
1170         * Makefile.in (tree-tailcall.o): Update dependencies.
1172 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1174         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1175         non-inlinable part.
1177 2013-09-06  Richard Biener  <rguenther@suse.de>
1179         * lto-streamer.h (lto_global_var_decls): Remove.
1180         * Makefile.in (OBJS): Remove lto-symtab.o.
1181         (lto-symtab.o): Remove.
1182         (GTFILES): Remove lto-symtab.c
1183         * lto-symtab.c: Move to lto/
1185 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1187         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1188         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1189         (UNSPEC_FPINT_RINT): New constant definitions.
1190         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1191         definition with 2 attributes.
1192         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1193         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1194         definitions.
1196 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1198         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1199         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1200         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
1201         attribute to "z196_alone".
1202         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1203         "zEC12_simple".
1205 2013-09-06  Richard Biener  <rguenther@suse.de>
1207         * basic-block.h (class control_dependences): New.
1208         * tree-ssa-dce.c (control_dependence_map): Remove.
1209         (cd): New global.
1210         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1211         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1212         find_pdom, find_control_dependence, find_all_control_dependences):
1213         Move to cfganal.c.
1214         (mark_control_dependent_edges_necessary,
1215         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1216         tree_dce_done, perform_tree_ssa_dce): Adjust.
1217         * cfganal.c (set_control_dependence_map_bit,
1218         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1219         find_all_control_dependences): Move from tree-ssa-dce.c and
1220         implement as methods of control_dependences class.
1221         (control_dependences::control_dependences): New.
1222         (control_dependences::~control_dependences): Likewise.
1223         (control_dependences::get_edges_dependent_on): Likewise.
1224         (control_dependences::get_edge): Likewise.
1226 2013-09-04  Jan Hubicka  <jh@suse.cz>
1228         * tree.c (types_same_for_odr): Drop overactive check.
1229         * ipa-devirt.c (hash_type_name): Likewise.
1231 2013-09-04  Jan Hubicka  <jh@suse.cz>
1233         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
1234         (analyze_functions): ... here.
1236 2013-09-04  Jan Hubicka  <jh@suse.cz>
1238         PR middle-end/58201
1239         * cgraphunit.c (analyze_functions): Clear AUX fields
1240         after processing; initialize assembler name has.
1242 2013-09-05  Jeff Law  <law@redhat.com>
1244         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
1245         from thread_around_empty_block.  Record threading path into PATH.
1246         Recurse if threading through the initial block is successful.
1247         (thread_across_edge): Corresponding changes to slightly simplify.
1249 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1251         * config/aarch64/aarch64.md
1252         (type): Remove frecpe, frecps, frecpx.
1253         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
1254         fix to be a TARGET_SIMD instruction.
1255         (aarch64_frecps): Remove.
1256         * config/aarch64/aarch64-simd.md
1257         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
1258         (aarch64_frecps<mode>): Handle all float/vector of float modes.
1260 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1261             Sofiane Naci  <sofiane.naci@arm.com>
1263         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
1264         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
1265         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
1266         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
1267         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
1268         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
1269         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
1270         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
1271         into "clz, "rbit".  Rename "shift" to "shift_imm".
1272         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
1273         changes.  Update for attribute changes all occurrences of arlo_* and
1274         shift* types.
1275         * config/arm/arm-fixed.md: Update for attribute changes
1276         all occurrences of arlo_* types.
1277         * config/arm/thumb2.md: Update for attribute changes all occurrences
1278         of arlo_* types.
1279         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
1280         (cortexa7_older_only): Likewise.
1281         (cortexa7_younger):  Likewise.
1282         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
1283         (1020alu_shift_op): Likewise.
1284         (1020alu_shift_reg_op): Likewise.
1285         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
1286         (alu_shift_op): Likewise.
1287         (alu_shift_reg_op): Likewise.
1288         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
1289         (11_alu_shift_op): Likewise.
1290         (11_alu_shift_reg_op): Likewise.
1291         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
1292         (9_alu_shift_reg_op): Likewise.
1293         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
1294         attribute changes.
1295         (cortex_a15_alu_shift): Likewise.
1296         (cortex_a15_alu_shift_reg): Likewise.
1297         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
1298         attribute changes.
1299         (cortex_a5_alu_shift): Likewise.
1300         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
1301         attribute changes.
1302         (cortex_a53_alu_shift): Likewise.
1303         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
1304         attribute changes.
1305         (cortex_a7_alu_reg): Likewise.
1306         (cortex_a7_alu_shift): Likewise.
1307         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
1308         attribute changes.
1309         (cortex_a8_alu_shift): Likewise.
1310         (cortex_a8_alu_shift_reg): Likewise.
1311         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
1312         (cortex_a9_dp_shift): Likewise.
1313         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
1314         attribute changes.
1315         * config/arm/cortex-r4.md
1316         (cortex_r4_alu): Update for attribute changes.
1317         (cortex_r4_mov): Likewise.
1318         (cortex_r4_alu_shift_reg): Likewise.
1319         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
1320         (526_alu_shift_op): Likewise.
1321         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
1322         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
1323         (626te_alu_shift_op): Likewise.
1324         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
1325         (726te_alu_shift_op): Likewise.
1326         (726te_alu_shift_reg_op): Likewise.
1327         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
1328         (mp626_alu_shift_op): Likewise.
1329         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
1330         (pj4_alu_conds): Likewise.
1331         (pj4_shift): Likewise.
1332         (pj4_shift_conds): Likewise.
1333         (pj4_alu_shift): Likewise.
1334         (pj4_alu_shift_conds): Likewise.
1335         * config/aarch64/aarch64.md: Update for attribute change
1336         all occurrences of arlo_* and shift* types.
1338 2013-09-05  Mike Stump  <mikestump@comcast.net>
1340         * tree.h: Move documentation for tree_function_decl to tree-core.h
1341         with the declaration.
1343 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
1345         PR target/58139
1346         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
1347         looking for widest mode.
1349 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
1351         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
1353 2013-09-05  Richard Biener  <rguenther@suse.de>
1355         PR tree-optimization/58137
1356         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
1357         Do not create vectors of pointers.
1358         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
1359         types for the components of the vector initializer.
1360         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
1361         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
1363 2013-09-05  Martin Jambor  <mjambor@suse.cz>
1365         * ipa-prop.c (remove_described_reference): Accept missing references,
1366         return false if that hppens, otherwise return true.
1367         (cgraph_node_for_jfunc): New function.
1368         (try_decrement_rdesc_refcount): Likewise.
1369         (try_make_edge_direct_simple_call): Use them.
1370         (ipa_edge_removal_hook): Remove references from rdescs.
1371         (ipa_edge_duplication_hook): Clone rdescs and their references
1372         when the new edge has the same caller as the old one.
1373         * cgraph.c (cgraph_resolve_speculation): Remove speculative
1374         reference before removing any edges.
1376 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
1378         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
1379         initial store.
1380         * thumb2.md (thumb2_storewb_parisi): New pattern.
1382 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
1384         * config/aarch64/aarch64-option-extensions.def: Add
1385         AARCH64_OPT_EXTENSION of 'crc'.
1386         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
1387         (AARCH64_ISA_CRC): Ditto.
1388         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
1389         description of the CRC extension.
1391 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1393         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
1394         * config/rs6000/linux.h: Ditto.
1395         * alpha/linux.h: Ditto.
1396         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
1397         no_c99_libc_has_function.
1398         * config/c6x/uclinux-elf.h: Ditto.
1399         * config/lm32/uclinux-elf.h: Ditto.
1400         * config/m68k/uclinux.h: Ditto.
1401         * config/moxie/uclinux.h: Ditto.
1402         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
1403         (crisv32-*-linux*, cris-*-linux*): Ditto.
1404         * config/bfin/bfin.c: Include "tm_p.h".
1406 2013-09-05  Richard Biener  <rguenther@suse.de>
1408         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
1409         check for a definition without a basic-block.
1411 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1412             Sofiane Naci <sofiane.naci@arm.com>
1414         * config/aarch64/aarch64.md
1415         (*movti_aarch64): Rename r_2_f and f_2_r.
1416         (*movsf_aarch64): Likewise.
1417         (*movdf_aarch64): Likewise.
1418         (*movtf_aarch64): Likewise.
1419         (aarch64_movdi_<mode>low): Likewise.
1420         (aarch64_movdi_<mode>high): Likewise.
1421         (aarch64_mov<mode>high_di): Likewise.
1422         (aarch64_mov<mode>low_di): Likewise.
1423         (aarch64_movtilow_tilow): Likewise.
1424         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
1425         values to config/arm/types.md
1426         (attribute "conds"): Update for attribute change.
1427         (anddi3_insn): Likewise.
1428         (iordi3_insn): Likewise.
1429         (xordi3_insn): Likewise.
1430         (one_cmpldi2): Likewise.
1431         * config/arm/types.md (type): Add Neon types.
1432         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
1433         use "type" attribute.
1434         (movmisalign<mode>_neon_store): Likewise.
1435         (movmisalign<mode>_neon_load): Likewise.
1436         (vec_set<mode>_internal): Likewise.
1437         (vec_setv2di_internal): Likewise.
1438         (vec_extract<mode>): Likewise.
1439         (vec_extractv2di): Likewise.
1440         (add<mode>3_neon): Likewise.
1441         (adddi3_neon): Likewise.
1442         (sub<mode>3_neon): Likewise.
1443         (subdi3_neon): Likewise.
1444         (mul<mode>3_neon): Likewise.
1445         (mul<mode>3add<mode>_neon): Likewise.
1446         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
1447         (fma<VCVTF:mode>4)): Likewise.
1448         (fma<VCVTF:mode>4_intrinsic): Likewise.
1449         (fmsub<VCVTF:mode>4)): Likewise.
1450         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1451         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1452         (ior<mode>3): Likewise.
1453         (and<mode>3): Likewise.
1454         (anddi3_neon): Likewise.
1455         (orn<mode>3_neon): Likewise.
1456         (orndi3_neon): Likewise.
1457         (bic<mode>3_neon): Likewise.
1458         (bicdi3_neon): Likewise.
1459         (xor<mode>3): Likewise.
1460         (one_cmpl<mode>2): Likewise.
1461         (abs<mode>2): Likewise.
1462         (neg<mode>2): Likewise.
1463         (umin<mode>3_neon): Likewise.
1464         (umax<mode>3_neon): Likewise.
1465         (smin<mode>3_neon): Likewise.
1466         (smax<mode>3_neon): Likewise.
1467         (vashl<mode>3): Likewise.
1468         (vashr<mode>3_imm): Likewise.
1469         (vlshr<mode>3_imm): Likewise.
1470         (ashl<mode>3_signed): Likewise.
1471         (ashl<mode>3_unsigned): Likewise.
1472         (neon_load_count): Likewise.
1473         (ashldi3_neon_noclobber): Likewise.
1474         (signed_shift_di3_neon): Likewise.
1475         (unsigned_shift_di3_neon): Likewise.
1476         (ashrdi3_neon_imm_noclobber): Likewise.
1477         (lshrdi3_neon_imm_noclobber): Likewise.
1478         (widen_ssum<mode>3): Likewise.
1479         (widen_usum<mode>3): Likewise.
1480         (quad_halves_<code>v4si): Likewise.
1481         (quad_halves_<code>v4sf): Likewise.
1482         (quad_halves_<code>v8hi): Likewise.
1483         (quad_halves_<code>v16qi): Likewise.
1484         (reduc_splus_v2di): Likewise.
1485         (neon_vpadd_internal<mode>): Likewise.
1486         (neon_vpsmin<mode>): Likewise.
1487         (neon_vpsmax<mode>): Likewise.
1488         (neon_vpumin<mode>): Likewise.
1489         (neon_vpumax<mode>): Likewise.
1490         (ss_add<mode>_neon): Likewise.
1491         (us_add<mode>_neon): Likewise.
1492         (ss_sub<mode>_neon): Likewise.
1493         (us_sub<mode>_neon): Likewise.
1494         (neon_vadd<mode>_unspec): Likewise.
1495         (neon_vaddl<mode>): Likewise.
1496         (neon_vaddw<mode>): Likewise.
1497         (neon_vhadd<mode>): Likewise.
1498         (neon_vqadd<mode>): Likewise.
1499         (neon_vaddhn<mode>): Likewise.
1500         (neon_vmul<mode>): Likewise.
1501         (neon_vmla<mode>): Likewise.
1502         (neon_vmlal<mode>): Likewise.
1503         (neon_vmls<mode>): Likewise.
1504         (neon_vmlsl<mode>): Likewise.
1505         (neon_vqdmulh<mode>): Likewise.
1506         (neon_vqdmlal<mode>): Likewise.
1507         (neon_vqdmlsl<mode>): Likewise.
1508         (neon_vmull<mode>): Likewise.
1509         (neon_vqdmull<mode>): Likewise.
1510         (neon_vsub<mode>_unspec): Likewise.
1511         (neon_vsubl<mode>): Likewise.
1512         (neon_vsubw<mode>): Likewise.
1513         (neon_vqsub<mode>): Likewise.
1514         (neon_vhsub<mode>): Likewise.
1515         (neon_vsubhn<mode>): Likewise.
1516         (neon_vceq<mode>): Likewise.
1517         (neon_vcge<mode>): Likewise.
1518         (neon_vcgeu<mode>): Likewise.
1519         (neon_vcgt<mode>): Likewise.
1520         (neon_vcgtu<mode>): Likewise.
1521         (neon_vcle<mode>): Likewise.
1522         (neon_vclt<mode>): Likewise.
1523         (neon_vcage<mode>): Likewise.
1524         (neon_vcagt<mode>): Likewise.
1525         (neon_vtst<mode>): Likewise.
1526         (neon_vabd<mode>): Likewise.
1527         (neon_vabdl<mode>): Likewise.
1528         (neon_vaba<mode>): Likewise.
1529         (neon_vabal<mode>): Likewise.
1530         (neon_vmax<mode>): Likewise.
1531         (neon_vmin<mode>): Likewise.
1532         (neon_vpaddl<mode>): Likewise.
1533         (neon_vpadal<mode>): Likewise.
1534         (neon_vpmax<mode>): Likewise.
1535         (neon_vpmin<mode>): Likewise.
1536         (neon_vrecps<mode>): Likewise.
1537         (neon_vrsqrts<mode>): Likewise.
1538         (neon_vqabs<mode>): Likewise.
1539         (neon_vqneg<mode>): Likewise.
1540         (neon_vcls<mode>): Likewise.
1541         (clz<mode>2): Likewise.
1542         (popcount<mode>2): Likewise.
1543         (neon_vrecpe): Likewise.
1544         (neon_vrsqrte): Likewise.
1545         (neon_vget_lane<mode>_sext_internal): Likewise.
1546         (neon_vget_lane<mode>_zext_internal): Likewise.
1547         (neon_vdup_n<mode>): Likewise.
1548         (neon_vdup_nv2di): Likewise.
1549         (neon_vdpu_lane<mode>_internal): Likewise.
1550         (neon_vswp<mode>): Likewise.
1551         (float<mode><V_cvtto>2): Likewise.
1552         (floatuns<mode><V_cvtto>2): Likewise.
1553         (fix_trunc<mode><V_cvtto>)2): Likewise
1554         (fixuns_trunc<mode><V_cvtto)2): Likewise.
1555         (neon_vcvt<mode>): Likewise.
1556         (neon_vcvtv4sfv4hf): Likewise.
1557         (neon_vcvtv4hfv4sf): Likewise.
1558         (neon_vcvt_n<mode>): Likewise.
1559         (neon_vmovn<mode>): Likewise.
1560         (neon_vqmovn<mode>): Likewise.
1561         (neon_vqmovun<mode>): Likewise.
1562         (neon_vmovl<mode>): Likewise.
1563         (neon_vmul_lane<mode>): Likewise.
1564         (neon_vmull_lane<mode>): Likewise.
1565         (neon_vqdmull_lane<mode>): Likewise.
1566         (neon_vqdmulh_lane<mode>): Likewise.
1567         (neon_vmla_lane<mode>): Likewise.
1568         (neon_vmlal_lane<mode>): Likewise.
1569         (neon_vqdmlal_lane<mode>): Likewise.
1570         (neon_vmls_lane<mode>): Likewise.
1571         (neon_vmlsl_lane<mode>): Likewise.
1572         (neon_vqdmlsl_lane<mode>): Likewise.
1573         (neon_vext<mode>): Likewise.
1574         (neon_vrev64<mode>): Likewise.
1575         (neon_vrev32<mode>): Likewise.
1576         (neon_vrev16<mode>): Likewise.
1577         (neon_vbsl<mode>_internal): Likewise.
1578         (neon_vshl<mode>): Likewise.
1579         (neon_vqshl<mode>): Likewise.
1580         (neon_vshr_n<mode>): Likewise.
1581         (neon_vshrn_n<mode>): Likewise.
1582         (neon_vqshrn_n<mode>): Likewise.
1583         (neon_vqshrun_n<mode>): Likewise.
1584         (neon_vshl_n<mode>): Likewise.
1585         (neon_vqshl_n<mode>): Likewise.
1586         (neon_vqshlu_n<mode>): Likewise.
1587         (neon_vshll_n<mode>): Likewise.
1588         (neon_vsra_n<mode>): Likewise.
1589         (neon_vsri_n<mode>): Likewise.
1590         (neon_vsli_n<mode>): Likewise.
1591         (neon_vtbl1v8qi): Likewise.
1592         (neon_vtbl2v8qi): Likewise.
1593         (neon_vtbl3v8qi): Likewise.
1594         (neon_vtbl4v8qi): Likewise.
1595         (neon_vtbx1v8qi): Likewise.
1596         (neon_vtbx2v8qi): Likewise.
1597         (neon_vtbx3v8qi): Likewise.
1598         (neon_vtbx4v8qi): Likewise.
1599         (neon_vtrn<mode>_internal): Likewise.
1600         (neon_vzip<mode>_internal): Likewise.
1601         (neon_vuzp<mode>_internal): Likewise.
1602         (neon_vld1<mode>): Likewise.
1603         (neon_vld1_lane<mode>): Likewise.
1604         (neon_vld1_dup<mode>): Likewise.
1605         (neon_vld1_dupv2di): Likewise.
1606         (neon_vst1<mode>): Likewise.
1607         (neon_vst1_lane<mode>): Likewise.
1608         (neon_vld2<mode>): Likewise.
1609         (neon_vld2_lane<mode>): Likewise.
1610         (neon_vld2_dup<mode>): Likewise.
1611         (neon_vst2<mode>): Likewise.
1612         (neon_vst2_lane<mode>): Likewise.
1613         (neon_vld3<mode>): Likewise.
1614         (neon_vld3qa<mode>): Likewise.
1615         (neon_vld3qb<mode>): Likewise.
1616         (neon_vld3_lane<mode>): Likewise.
1617         (neon_vld3_dup<mode>): Likewise.
1618         (neon_vst3<mode>): Likewise.
1619         (neon_vst3qa<mode>): Likewise.
1620         (neon_vst3qb<mode>): Likewise.
1621         (neon_vst3_lane<mode>): Likewise.
1622         (neon_vld4<mode>): Likewise.
1623         (neon_vld4qa<mode>): Likewise.
1624         (neon_vld4qb<mode>): Likewise.
1625         (neon_vld4_lane<mode>): Likewise.
1626         (neon_vld4_dup<mode>): Likewise.
1627         (neon_vst4<mode>): Likewise.
1628         (neon_vst4qa<mode>): Likewise.
1629         (neon_vst4qb<mode>): Likewise.
1630         (neon_vst4_lane<mode>): Likewise.
1631         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1632         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1633         (neon_vec_<US>mult_lo_<mode>): Likewise.
1634         (neon_vec_<US>mult_hi_<mode>): Likewise.
1635         (neon_vec_<US>shiftl_<mode>): Likewise.
1636         (neon_unpack<US>_<mode>): Likewise.
1637         (neon_vec_<US>mult_<mode>): Likewise.
1638         (vec_pack_trunc_<mode>): Likewise.
1639         (neon_vec_pack_trunk_<mode>): Likewise.
1640         (neon_vabd<mode>_2): Likewise.
1641         (neon_vabd<mode>_3): Likewise.
1642         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
1643         (thumb2_movsi_vfp): Likewise.
1644         (movdi_vfp): Likewise.
1645         (movdi_vfp_cortexa8): Likewise.
1646         (movhf_vfp_neon): Likewise.
1647         (movhf_vfp): Likewiwse.
1648         (movsf_vfp): Likewiwse.
1649         (thumb2_movsf_vfp): Likewiwse.
1650         (movdf_vfp): Likewise.
1651         (thumb2_movdf_vfp): Likewise.
1652         (movsfcc_vfp): Likewise.
1653         (thumb2_movsfcc_vfp): Likewise.
1654         (movdfcc_vfp): Likewise.
1655         (thumb2_movdfcc_vfp): Likewise.
1656         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
1657         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
1658         (v10_v2c): Likewise.
1659         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
1660         attribute change.
1661         (cortex_a15_neon_int_2): Likewise.
1662         (cortex_a15_neon_int_3): Likewise.
1663         (cortex_a15_neon_int_4): Likewise.
1664         (cortex_a15_neon_int_5): Likewise.
1665         (cortex_a15_neon_vqneg_vqabs): Likewise.
1666         (cortex_a15_neon_vmov): Likewise.
1667         (cortex_a15_neon_vaba): Likewise.
1668         (cortex_a15_neon_vaba_qqq): Likewise.
1669         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1670         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1671         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
1672         scalar_64_32_long_scalar): Likewise.
1673         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1674         (cortex_a15_neon_mla_qqq_8_16): Likewise.
1675         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1676         lotype_qdd_64_32_long): Likewise.
1677         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1678         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1679         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
1680         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1681         (cortex_a15_neon_shift_1): Likewise.
1682         (cortex_a15_neon_shift_2): Likewise.
1683         (cortex_a15_neon_shift_3): Likewise.
1684         (cortex_a15_neon_vshl_ddd): Likewise.
1685         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1686         (cortex_a15_neon_vsra_vrsra): Likewise.
1687         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
1688         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
1689         (cortex_a15_neon_fp_vmul_ddd): Likewise.
1690         (cortex_a15_neon_fp_vmul_qqd): Likewise.
1691         (cortex_a15_neon_fp_vmla_ddd): Likewise.
1692         (cortex_a15_neon_fp_vmla_qqq): Likewise.
1693         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
1694         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
1695         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1696         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1697         (cortex_a15_neon_bp_simple): Likewise.
1698         (cortex_a15_neon_bp_2cycle): Likewise.
1699         (cortex_a15_neon_bp_3cycle): Likewise.
1700         (cortex_a15_neon_vld1_1_2_regs): Likewise.
1701         (cortex_a15_neon_vld1_3_4_regs): Likewise.
1702         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1703         (cortex_a15_neon_vld2_4_regs): Likewise.
1704         (cortex_a15_neon_vld3_vld4): Likewise.
1705         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1706         (cortex_a15_neon_vst1_3_4_regs): Likewise.
1707         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
1708         (cortex_a15_neon_vst3_vst4): Likewise.
1709         (cortex_a15_neon_vld1_vld2_lane): Likewise.
1710         (cortex_a15_neon_vld3_vld4_lane" 10
1711         (cortex_a15_neon_vst1_vst2_lane): Likewise.
1712         (cortex_a15_neon_vst3_vst4_lane): Likewise.
1713         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
1714         (cortex_a15_neon_ldm_2): Likewise.0
1715         (cortex_a15_neon_stm_2): Likewise.
1716         (cortex_a15_neon_mcr): Likewise.
1717         (cortex_a15_neon_mcr_2_mcrr): Likewise.
1718         (cortex_a15_neon_mrc): Likewise.
1719         (cortex_a15_neon_mrrc): Likewise.
1720         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
1721         change.
1722         (cortex_a15_alu_shift): Likewise.
1723         (cortex_a15_alu_shift_reg): Likewise.
1724         (cortex_a15_mult32): Likewise.
1725         (cortex_a15_mult64): Likewise.
1726         (cortex_a15_block): Likewise.
1727         (cortex_a15_branch): Likewise.
1728         (cortex_a15_load1): Likewise.
1729         (cortex_a15_load3): Likewise.
1730         (cortex_a15_store1): Likewise.
1731         (cortex_a15_store3): Likewise.
1732         (cortex_a15_call): Likewise.
1733         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
1734         (cortex_a5_f2r): Likewise.
1735         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
1736         change.
1737         (cortex_a53_f2r): Likewise.
1738         * config/arm/cortex-a7.md
1739         (cortex_a7_branch): Update for attribute change.
1740         (cortex_a7_call): Likewise.
1741         (cortex_a7_alu_imm): Likewise.
1742         (cortex_a7_alu_reg): Likewise.
1743         (cortex_a7_alu_shift): Likewise.
1744         (cortex_a7_mul): Likewise.
1745         (cortex_a7_load1): Likewise.
1746         (cortex_a7_store1): Likewise.
1747         (cortex_a7_load2): Likewise.
1748         (cortex_a7_store2): Likewise.
1749         (cortex_a7_load3): Likewise.
1750         (cortex_a7_store3): Likewise.
1751         (cortex_a7_load4): Likewise.
1752         (cortex_a7_store4): Likewise.
1753         (cortex_a7_fpalu): Likewise.
1754         (cortex_a7_fconst): Likewise.
1755         (cortex_a7_fpmuls): Likewise.
1756         (cortex_a7_neon_mul): Likewise.
1757         (cortex_a7_fpmacs): Likewise.
1758         (cortex_a7_neon_mla: Likewise.
1759         (cortex_a7_fpmuld: Likewise.
1760         (cortex_a7_fpmacd: Likewise.
1761         (cortex_a7_fpfmad: Likewise.
1762         (cortex_a7_fdivs: Likewise.
1763         (cortex_a7_fdivd: Likewise.
1764         (cortex_a7_r2f: Likewise.
1765         (cortex_a7_f2r: Likewise.
1766         (cortex_a7_f_flags: Likewise.
1767         (cortex_a7_f_loads: Likewise.
1768         (cortex_a7_f_loadd: Likewise.
1769         (cortex_a7_f_stores: Likewise.
1770         (cortex_a7_f_stored: Likewise.
1771         (cortex_a7_neon): Likewise.
1772         * config/arm/cortex-a8-neon.md
1773         (cortex_a8_neon_mrc): Update for attribute change.
1774         (cortex_a8_neon_mrrc): Likewise.
1775         (cortex_a8_neon_int_1): Likewise.
1776         (cortex_a8_neon_int_2): Likewise.
1777         (cortex_a8_neon_int_3): Likewise.
1778         (cortex_a8_neon_int_4): Likewise.
1779         (cortex_a8_neon_int_5): Likewise.
1780         (cortex_a8_neon_vqneg_vqabs): Likewise.
1781         (cortex_a8_neon_vmov): Likewise.
1782         (cortex_a8_neon_vaba): Likewise.
1783         (cortex_a8_neon_vaba_qqq): Likewise.
1784         (cortex_a8_neon_vsma): Likewise.
1785         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1786         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1787         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1788         long_scalar): Likewise.
1789         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1790         (cortex_a8_neon_mla_qqq_8_16): Likewise.
1791         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1792         long_scalar_qdd_64_32_long): Likewise.
1793         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1794         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1795         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
1796         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1797         (cortex_a8_neon_shift_1): Likewise.
1798         (cortex_a8_neon_shift_2): Likewise.
1799         (cortex_a8_neon_shift_3): Likewise.
1800         (cortex_a8_neon_vshl_ddd): Likewise.
1801         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1802         (cortex_a8_neon_vsra_vrsra): Likewise.
1803         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
1804         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
1805         (cortex_a8_neon_fp_vsum): Likewise.
1806         (cortex_a8_neon_fp_vmul_ddd): Likewise.
1807         (cortex_a8_neon_fp_vmul_qqd): Likewise.
1808         (cortex_a8_neon_fp_vmla_ddd): Likewise.
1809         (cortex_a8_neon_fp_vmla_qqq): Likewise.
1810         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
1811         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
1812         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1813         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1814         (cortex_a8_neon_bp_simple): Likewise.
1815         (cortex_a8_neon_bp_2cycle): Likewise.
1816         (cortex_a8_neon_bp_3cycle): Likewise.
1817         (cortex_a8_neon_ldr): Likewise.
1818         (cortex_a8_neon_str): Likewise.
1819         (cortex_a8_neon_vld1_1_2_regs): Likewise.
1820         (cortex_a8_neon_vld1_3_4_regs): Likewise.
1821         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1822         (cortex_a8_neon_vld2_4_regs): Likewise.
1823         (cortex_a8_neon_vld3_vld4): Likewise.
1824         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1825         (cortex_a8_neon_vst1_3_4_regs): Likewise.
1826         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
1827         (cortex_a8_neon_vst3_vst4): Likewise.
1828         (cortex_a8_neon_vld1_vld2_lane): Likewise.
1829         (cortex_a8_neon_vld3_vld4_lane): Likewise.
1830         (cortex_a8_neon_vst1_vst2_lane): Likewise.
1831         (cortex_a8_neon_vst3_vst4_lane): Likewise.
1832         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
1833         (cortex_a8_neon_mcr): Likewise.
1834         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1835         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
1836         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
1837         change.
1838         (ca9_neon_mrrc): Likewise.
1839         (cortex_a9_neon_int_1): Likewise.
1840         (cortex_a9_neon_int_2): Likewise.
1841         (cortex_a9_neon_int_3): Likewise.
1842         (cortex_a9_neon_int_4): Likewise.
1843         (cortex_a9_neon_int_5): Likewise.
1844         (cortex_a9_neon_vqneg_vqabs): Likewise.
1845         (cortex_a9_neon_vmov): Likewise.
1846         (cortex_a9_neon_vaba): Likewise.
1847         (cortex_a9_neon_vaba_qqq): Likewise.
1848         (cortex_a9_neon_vsma): Likewise.
1849         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1850         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
1851         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
1852         long_scalar): Likewise.
1853         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
1854         (cortex_a9_neon_mla_qqq_8_16): Likewise.
1855         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
1856         long_scalar_qdd_64_32_long): Likewise.
1857         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
1858         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
1859         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
1860         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
1861         (cortex_a9_neon_shift_1): Likewise.
1862         (cortex_a9_neon_shift_2): Likewise.
1863         (cortex_a9_neon_shift_3): Likewise.
1864         (cortex_a9_neon_vshl_ddd): Likewise.
1865         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
1866         (cortex_a9_neon_vsra_vrsra): Likewise.
1867         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
1868         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
1869         (cortex_a9_neon_fp_vsum): Likewise.
1870         (cortex_a9_neon_fp_vmul_ddd): Likewise.
1871         (cortex_a9_neon_fp_vmul_qqd): Likewise.
1872         (cortex_a9_neon_fp_vmla_ddd): Likewise.
1873         (cortex_a9_neon_fp_vmla_qqq): Likewise.
1874         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
1875         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
1876         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
1877         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
1878         (cortex_a9_neon_bp_simple): Likewise.
1879         (cortex_a9_neon_bp_2cycle): Likewise.
1880         (cortex_a9_neon_bp_3cycle): Likewise.
1881         (cortex_a9_neon_ldr): Likewise.
1882         (cortex_a9_neon_str): Likewise.
1883         (cortex_a9_neon_vld1_1_2_regs): Likewise.
1884         (cortex_a9_neon_vld1_3_4_regs): Likewise.
1885         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
1886         (cortex_a9_neon_vld2_4_regs): Likewise.
1887         (cortex_a9_neon_vld3_vld4): Likewise.
1888         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
1889         (cortex_a9_neon_vst1_3_4_regs): Likewise.
1890         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
1891         (cortex_a9_neon_vst3_vst4): Likewise.
1892         (cortex_a9_neon_vld1_vld2_lane): Likewise.
1893         (cortex_a9_neon_vld3_vld4_lane): Likewise.
1894         (cortex_a9_neon_vst1_vst2_lane): Likewise.
1895         (cortex_a9_neon_vst3_vst4_lane): Likewise.
1896         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
1897         (cortex_a9_neon_mcr): Likewise.
1898         (cortex_a9_neon_mcr_2_mcrr): Likewise.
1899         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
1900         (cortex_a9_fps): Likewise.
1901         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
1902         change.
1903         (cortex_m4_fmuls): Likewise.
1904         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
1905         change.
1906         (cortex_r4_mrc): Likewise.
1907         * config/arm/iterators.md: Update comment referring to neon_type.
1908         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
1909         (iwmmxt_movsi_insn): Likewise.
1910         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
1911         attribute change.
1912         (pj4_core_to_vfp): Likewise.
1913         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
1914         attribute change.
1915         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
1916         (vfp_fstore): Likewise.
1917         * doc/md.texi: Change references to neon_type to refer to type.
1919 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1921         * tree.h (DECL_BUILT_IN): Fix typo in comment.
1923 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
1925         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
1926         lglobl if not weak.
1928 2013-09-04  Easwaran Raman  <eraman@google.com>
1930         PR middle-end/57370
1931         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
1932         (build_and_add_sum): Use it.
1933         (appears_later_in_bb): Simplify code.
1935 2013-09-04  Teresa Johnson  <tejohnson@google.com>
1937         * dumpfile.c (dump_finish): Don't close stderr/stdout.
1939 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
1941         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
1943 2013-09-04  Jan Hubicka  <jh@suse.cz>
1945         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
1946         * ipa-devirt.c: Include diganostic.h
1947         (odr_type_d): Add types and types_set.
1948         (hash_type_name): Work for types with vtables during LTO.
1949         (odr_hasher::remove): Fix comment; destroy types_set.
1950         (add_type_duplicate): New function,
1951         (get_odr_type): Use it.
1952         (dump_type_inheritance_graph): Dump type duplicates.
1953         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
1954         graph.
1955         * tree.c (types_same_for_odr): Give exact answers on types with
1956         virtual tables.
1958 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
1960         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
1961         explaining their differences.
1963 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
1965         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
1967 2013-09-03  Jeff Law  <law@redhat.com>
1969         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
1970         when not threading through a joiner block.  Pass joiner/no joiner
1971         state to register_jump_thread.
1972         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
1973         state from argument rather than implying on path length.
1974         Dump the entire jump thread path into debugging dump.
1975         * tree-flow.h (register_jump_thread): Update prototype.
1977 2013-08-29  Xinliang David Li  <davidxl@google.com>
1979         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1980         Remove a trivial gcc_assert.
1982 2013-08-29  Xinliang David Li  <davidxl@google.com>
1984         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
1985         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
1986         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1987         Delay base decl alignment adjustment.
1988         * tree-vectorizer.c (vect_destroy_datarefs): New function.
1989         * tree-vectorizer.h: New data structure.
1990         (set_dr_misalignment): New function.
1991         (dr_misalignment): Ditto.
1992         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
1993         (vectorizable_load): Ditto.
1994         (ensure_base_align): New function.
1995         (vectorize_loops): Add dbg_cnt support.
1996         (execute_vect_slp): Ditto.
1997         * dbgcnt.def: New debug counter.
1998         * Makefile: New dependency.
2000 2013-09-03  Meador Inge  <meadori@codesourcery.com>
2002         Revert:
2004         2013-08-30  Meador Inge  <meadori@codesourcery.com>
2006         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2008 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
2010         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
2011         function descriptor.
2013 2013-09-03  Richard Biener  <rguenther@suse.de>
2015         * tree-affine.c (add_elt_to_tree): Fix association issue,
2016         avoid useless converts and make sure to always return a
2017         properly typed result.
2019 2013-09-03  Richard Biener  <rguenther@suse.de>
2021         PR middle-end/57656
2022         * fold-const.c (negate_expr_p): Fix division case.
2023         (negate_expr): Likewise.
2025 2013-09-03  Richard Biener  <rguenther@suse.de>
2027         PR lto/58285
2028         * tree-streamer-out.c: Include tm.h.
2029         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2031 2013-09-03  Jan Hubicka  <jh@suse.cz>
2033         * tree-profile.c (tree_profiling): Cleanup CFG when done.
2035 2013-09-03  Alan Modra  <amodra@gmail.com>
2037         * config.gcc (powerpc*-*-linux*): Add support for little-endian
2038         multilibs to big-endian target and vice versa.
2039         * config/rs6000/t-linux64: Use := assignment on all vars.
2040         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2041         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2042         * config/rs6000/t-linux64le: New file.
2043         * config/rs6000/t-linux64bele: New file.
2044         * config/rs6000/t-linux64lebe: New file.
2046 2013-09-02  Jan Hubicka  <jh@suse.cz>
2048         * ipa-inline-transform.c (inline_transform): Do not
2049         optimize_inline_calls when not optimizing.
2051 2013-09-02  Jan Hubicka  <jh@suse.cz>
2053         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2054         duplicated nodes for assembler names.
2055         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2056         hard registers from assembler name hash.
2058 2013-09-02  Jan Hubicka  <jh@suse.cz>
2060         * ipa-split.c (execute_split_functions): Split externally visible
2061         functions called once.
2063 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2065         PR ipa/58106
2066         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2067         linked list.  When finding the correct duplicate, also consider also
2068         the caller in additon to its inlined_to node.
2070 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2072         * config/aarch64/aarch64-simd-builtins.def
2073         (dup_lane_scalar): Remove.
2074         * config/aarch64/aarch64-simd.md
2075         (aarch64_simd_dup): Add 'w->w' alternative.
2076         (aarch64_dup_lane<mode>): Allow for VALL.
2077         (aarch64_dup_lane_scalar<mode>): Remove.
2078         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2079         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2080         (aarch64_get_lane_unsigned<mode>): Likewise.
2081         (aarch64_get_lane<mode>): Likewise.
2082         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2083         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2084         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2085         (VCON): Change container of V2SF.
2086         (vswap_width_name): Likewise.
2087         * config/aarch64/arm_neon.h
2088         (__aarch64_vdup_lane_any): New.
2089         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2090         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2091         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2093 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2095         PR middle-end/56382
2096         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2097         the source or the destination is a single hard register.
2099 2013-09-02  Richard Biener  <rguenther@suse.de>
2101         PR middle-end/57511
2102         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2103         non-linear SCEVs.
2105 2013-09-02  Richard Biener  <rguenther@suse.de>
2107         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2108         arithmetic to sizetype.
2110 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2112         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2113         Find auto-increment use both before and after candidate.
2115 2013-09-02  Marek Polacek  <polacek@redhat.com>
2117         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2119 2013-09-01  Jan Hubicka  <jh@suse.cz>
2121         * Makefile.in: Add ipa-profile.o
2122         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2123         * cgraph.c (struct cgraph_propagate_frequency_data,
2124         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2125         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2126         * cgraph.h (cgraph_propagate_frequency): Remove.
2127         * ipa-inline-analysis.c: Include ipa-utils.h;
2128         drop duplicated cfgloop.h.
2129         (inline_update_callee_summaries): Update.
2130         * ipa-profile.c: New file.
2131         * ipa-utils.h (ipa_propagate_frequency): Declare.
2132         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2133         data-streamer.h, value-prof.h.
2134         (symtab_remove_unreachable_nodes): Update profile.
2135         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2136         account_time_size, cmp_counts, dump_histogram,
2137         ipa_profile_generate_summary, ipa_profile_write_summary,
2138         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2139         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2140         Move to ipa-profile.c.
2142 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
2144         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2146 2013-09-01  Jan Hubicka  <jh@suse.cz>
2148         * common.opt (fdevirtualize-speculatively): New function.
2149         * invoke.texi (fdevirtualize-speculatively): Document.
2150         * ipa-devirt.c: Include ipa-inline.h
2151         (likely_target_p): New function.
2152         (ipa_devirt): New function.
2153         (gate_ipa_devirt): New function.
2154         (pass_data_ipa_devirt): New static var.
2155         (pass_ipa_devirt): Likewise.
2156         (make_pass_ipa_devirt): New function.
2157         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2158         (common_handle_option): Disable devirtualization when
2159         value range profiling is available.
2160         * passes.def (pass_ipa_devirt): Add.
2161         * timever.def (TV_IPA_DEVIRT): New timevar.
2162         * tree-pass.h (make_pass_ipa_devirt):
2164 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
2166         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2167         include sanitize(undefined).
2169 2013-08-31  Diego Novillo  <dnovillo@google.com>
2171         * Makefile.in (TREE_CORE_H): Define.
2172         (TREE_H): Use.
2173         (GTFILES): Add tree-core.h.
2174         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2175         size the array.
2176         * tree-core.h: New file.
2177         Move all data structures, enum, typedefs, global
2178         declarations and constants from ...
2179         * tree.h: ... here.
2181 2013-08-31  Jan Hubicka  <jh@suse.cz>
2183         * bulitins.c (expand_builtin): Do not early exit for gcov
2184         instrumented functions.
2186 2013-08-31  Marek Polacek  <polacek@redhat.com>
2188         * ubsan.c: Include tm_p.h.
2190 2013-08-31  Jan Hubicka  <jh@suse.cz>
2192         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2193         warning.
2195         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2196         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2197         * tree-cfg.c (verify_gimple_label): ... here.
2198         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2199         (ipa_merge_profiles): New function.
2200         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2201         (lto_input_function_body): Likewise.
2202         * ipa-utils.h (ipa_merge_profiles): Declare.
2203         * lto-streamer.h (lto_input_function_body): Update prototype.
2204         (emit_label_in_global_context_p): Remove.
2205         * lto-symtab.c: Include ipa-utils.h
2206         (lto_cgraph_replace_node): Use ipa_merge_profiles.
2208 2013-08-31  Jan Hubicka  <jh@suse.cz>
2210         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2212 2013-08-31  Jan Hubicka  <jh@suse.cz>
2214         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2216 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
2218         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2219         "cmp" RTX before signed_comparison_operator check to account
2220         for "code" changes.
2222 2013-08-30  Jan Hubicka  <jh@suse.cz>
2224         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2225         (detect_type_change_1): Rename to ...
2226         (detect_type_change): ... this one; early return on non-polymorphic
2227         types.
2228         (detect_type_change_ssa): Add comp_type parameter; update
2229         use of detect_type_change.
2230         (compute_complex_assign_jump_func): Add param_type parameter;
2231         update use of detect_type_change_ssa.
2232         (compute_complex_ancestor_jump_func): Likewise.
2233         (ipa_get_callee_param_type): New function.
2234         (ipa_compute_jump_functions_for_edge): Compute parameter type;
2235         update calls to the jump function computation functions.
2237 2013-08-30  Teresa Johnson  <tejohnson@google.com>
2238             Steven Bosscher  <steven@gcc.gnu.org>
2240         * cfgrtl.c (fixup_new_cold_bb): New routine.
2241         (commit_edge_insertions): Invoke fixup_partitions.
2242         (find_partition_fixes): New routine.
2243         (fixup_partitions): Ditto.
2244         (verify_hot_cold_block_grouping): Update comments.
2245         (rtl_verify_edges): Invoke find_partition_fixes.
2246         (rtl_verify_bb_pointers): Update comments.
2247         (rtl_verify_bb_layout): Ditto.
2248         * basic-block.h (probably_never_executed_edge_p): Declare.
2249         (fixup_partitions): Ditto.
2250         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
2251         * bb-reorder.c (sanitize_hot_paths): New function.
2252         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
2253         sanitize_hot_paths.
2254         * predict.c (probably_never_executed_edge_p): New routine.
2255         * cfg.c (check_bb_profile): Add partition insanity warnings.
2257 2013-08-30  Meador Inge  <meadori@codesourcery.com>
2259         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2261 2013-08-30  Marek Polacek  <polacek@redhat.com>
2263         * Makefile.in (ubsan.o): Add.
2264         (c-family/c-ubsan.o): Add.
2265         (builtins.o): Add ubsan.h dependency.
2266         * ubsan.h: New file.
2267         * ubsan.c: New file.
2268         * common.opt: Add -fsanitize=undefined option.
2269         (flag_sanitize): Add variable.
2270         (fsanitize=): Add option.  Add Driver.
2271         (fsanitize=thread): Remove option.
2272         (fsanitize=address): Likewise.
2273         (static-libubsan): New option.
2274         * doc/invoke.texi: Document the new flag and -static-libubsan.
2275         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
2276         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
2277         * builtin-attrs.def (ATTR_COLD): Define.
2278         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2279         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
2280         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
2281         * flag-types.h (sanitize_code): New enum.
2282         * opts.c (common_handle_option): Parse command line arguments
2283         of -fsanitize=.  Add -fsanitize=unreachable option.
2284         * varasm.c (get_variable_section): Adjust.
2285         (assemble_noswitch_variable): Likewise.
2286         (assemble_variable): Likewise.
2287         (output_constant_def_contents): Likewise.
2288         (categorize_decl_for_section): Likewise.
2289         (place_block_symbol): Likewise.
2290         (output_object_block): Likewise.
2291         * builtins.def: Likewise.
2292         * toplev.c (compile_file): Likewise.
2293         (process_options): Likewise.
2294         * cppbuiltin.c: Likewise.
2295         * tsan.c (tsan_pass): Likewise.
2296         (tsan_gate): Likewise.
2297         (tsan_gate_O0): Likewise.
2298         * cfgexpand.c (partition_stack_vars): Likewise.
2299         (expand_stack_vars): Likewise.
2300         (defer_stack_allocation): Likewise.
2301         (expand_used_vars): Likewise.
2302         * cfgcleanup.c (old_insns_match_p): Likewise.
2303         * asan.c (asan_finish_file): Likewise.
2304         (asan_instrument): Likewise.
2305         (gate_asan): Likewise.
2306         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
2307         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2308         (asan_global_struct): Use pointer_sized_int_node instead
2309         calling build_nonstandard_integer_type.
2310         (initialize_sanitizer_builtins): Likewise.
2311         (asan_finish_file): Likewise.
2312         * gcc.c: Document %{%:function(args):X}.
2313         (static_spec_functions): Add sanitize.
2314         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
2315         store funcval != NULL there.
2316         (do_spec_1): Adjust handle_spec_function caller.
2317         (handle_braces): Allow %:function(args) as condition.
2318         (sanitize_spec_function): New function.
2319         (ADD_STATIC_LIBUBSAN_LIBS): Define.
2320         (LIBUBSAN_SPEC): Likewise.
2321         (LIBUBSAN_EARLY_SPEC): Likewise.
2322         (SANITIZER_SPEC): Handle libubsan.
2323         (SANITIZER_EARLY_SPEC): Likewise.
2324         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
2325         instead of fsanitize=address.
2326         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
2327         instead of fsanitize=address*.
2328         * builtins.c: Include ubsan.h.
2329         (fold_builtin_0): Instrument __builtin_unreachable.
2330         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
2331         instead of flag_asan.
2332         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
2333         (pointer_sized_int_node): Define.
2334         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
2336 2013-08-30  Mike Stump  <mikestump@comcast.net>
2338         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
2339         with RE patterns.
2341 2013-08-29  Jan Hubicka  <jh@suse.cz>
2343         * cgraph.c (cgraph_function_body_availability): Handle weakref
2344         correctly.
2345         * passes.def: Remove pass_fixup_cfg.
2346         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
2347         track when we need to remove functions.
2348         (gate_ipa_inline): Execute inlining always; add comment why.
2349         (pass_data_ipa_inline): Remove TODO_remove_functions.
2350         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
2351         do not produce summaries.
2352         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
2353         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
2354         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
2355         constant pool and vtable.
2357 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
2359         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
2360         New arm_neon.h's internal macros to specify 64-bit constants.
2361         Avoid using stdint.h's macros.
2363 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2365         * recog.c (verify_changes): Verify that changes[i].old is non-zero
2366         before applying REG_P.
2368 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
2370         PR tree-optimization/58277
2371         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
2372         after seeing too many stmts with vdef in between dombb and current
2373         bb, invalidate everything.
2375 2013-08-30  Richard Biener  <rguenther@suse.de>
2377         * fold-const.c (fold_single_bit_test): Fix overflow test.
2379 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
2381         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
2382         and which can live in a register, always retrieve the value on entry.
2383         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
2384         passed by invisible reference specially.
2385         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
2386         (vt_add_function_parameter): Correctly deal with a parameter passed by
2387         invisible reference.
2389 2013-08-30  Jan Hubicka  <jh@suse.cz>
2391         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
2393 2013-08-30  Richard Biener  <rguenther@suse.de>
2395         PR tree-optimization/58228
2396         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
2397         allow invariant loads in nested loop vectorization.
2399 2013-08-30  Richard Biener  <rguenther@suse.de>
2401         PR tree-optimization/58223
2402         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
2403         (has_anti_or_output_dependence): ... this and adjust to also
2404         look for output dependences.
2405         (mark_nodes_having_upstream_mem_writes): Adjust.
2406         (rdg_flag_uses): Likewise.
2408 2013-08-30  Richard Biener  <rguenther@suse.de>
2410         PR tree-optimization/58010
2411         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
2412         assert that we have a loop-closed PHI.
2414 2013-08-29  Jan Hubicka  <jh@suse.cz>
2416         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
2417         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
2418         * lto-section-in.c (lto_free_function_in_decl_state): New function.
2419         (lto_free_function_in_decl_state_for_node): New function.
2421 2013-08-29  Xinliang David Li  <davidxl@google.com>
2423         * loop-unroll.c (report_unroll_peel): Minor message change.
2424         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
2425         Emit alignment peeling message with default -fopt-info.
2426         (vect_loop_versioning): Emit loop version info message.
2427         * tree-vectorizer.c (vectorize_loops): Minor message change.
2428         (execute_vect_slp): Ditto.
2430 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
2432         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
2433         of the clone from the DECL_NAME of the original function.
2435 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
2437         * passes.c (register_pass): Add overload.
2438         * tree-pass.h (register_pass): Forward declare it.  Add comment.
2440 2013-08-29  Jan Hubicka  <jh@suse.cz>
2442         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
2443         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
2444         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
2445         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2446         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
2447         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
2448         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2449         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
2451 2013-08-29  Teresa Johnson  <tejohnson@google.com>
2453         * dumpfile.c (dump_loc): Output column number.
2454         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
2455         * doc/invoke.texi: Document optall -fopt-info flag.
2456         * profile.c (read_profile_edge_counts): Use new dump framework.
2457         (compute_branch_probabilities): Ditto.
2458         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
2459         when pass not in any opt group.
2460         * pass_manager.h (pass_manager::get_pass_profile): New method.
2461         * value-prof.c (check_counter): Use new dump framework.
2462         (check_ic_target): Ditto.
2463         * coverage.c (get_coverage_counts): Ditto.
2464         (coverage_init): Setup new dump framework.
2466 2013-08-29  Richard Biener  <rguenther@suse.de>
2468         PR tree-optimization/58246
2469         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
2470         handle the dominance check inside a basic-block.
2472 2013-08-29  Richard Biener  <rguenther@suse.de>
2474         PR middle-end/57287
2475         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
2476         of default defs that appear in abnormal PHI nodes.
2478 2013-08-29  Richard Biener  <rguenther@suse.de>
2480         PR tree-optimization/57685
2481         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
2482         single-use operands to avoid exponential complexity.
2484 2013-08-28  Dehao Chen  <dehao@google.com>
2486         * ipa-inline.c (edge_badness): Fix integer underflow.
2488 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
2490         * gtm-builtins.def (_ITM_free): Declare leaf.
2492 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2494         PR target/58067
2495         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
2496         (*tls_local_dynamic_base_64_largepic): Likewise.
2497         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
2498         Remove predicate from call operand.
2499         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
2500         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
2502 2013-08-28  Jeff Law  <law@redhat.com>
2504         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
2505         checks for the number of predecessors and successors allowed.
2506         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
2507         which require copying a joiner block if there is a request which
2508         is a subpath that requires no joiner block copying.
2510 2013-08-28  Jan Hubicka  <jh@suse.cz>
2512         * lto-streamer-out.c (DFS_write_tree_body): Drop
2513         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
2514         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
2515         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
2516         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
2517         Do not read DECL_ERROR_ISSUED.
2518         (unpack_ts_decl_with_vis_value_fields): Do not read
2519         DECL_DEFER_OUTPUT.
2520         (lto_input_ts_binfo_tree_pointers): Do not read
2521         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
2522         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
2523         write DECL_ERROR_ISSUED..
2524         (pack_ts_decl_with_vis_value_fields): Do not write
2525         DECL_DEFER_OUTPUT.
2526         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
2527         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
2528         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
2529         * tree.h (tree_decl_common): Update comment.
2530         (DECL_ERROR_ISSUED): Remove.
2532 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2534         PR middle-end/58257
2535         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
2537 2013-08-28  Jan Hubicka  <jh@suse.cz>
2539         * builtins.def (free): Declare leaf.
2541 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
2543         * gdbhooks.py: New.
2544         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
2545         * configure: Regenerate.
2547 2013-08-27  Martin Jambor  <mjambor@suse.cz>
2549         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
2550         (ipa_ancestor_jf_data): Likewise.
2551         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
2552         (ipa_get_jf_pass_through_type_preserved): New function.
2553         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
2554         (ipa_get_jf_ancestor_type_preserved): New function.
2555         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
2556         (ipa_get_jf_ancestor_result): Likewise.
2557         (propagate_vals_accross_pass_through): Use
2558         ipa_get_jf_pass_through_result to do all the value mappings.
2559         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
2560         type_preserved flag.
2561         (ipa_set_jf_cst_copy): New function.
2562         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
2563         (ipa_set_jf_arith_pass_through): Likewise.
2564         (ipa_set_ancestor_jf): Likewise.
2565         (compute_complex_assign_jump_func): Set type_preserved instead of
2566         punting.
2567         (ipa_compute_jump_functions_for_edge): Likewise.
2568         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
2569         (update_jump_functions_after_inlining): Update type_preserved.
2570         Explicitely create jump functions when combining one with pass_through.
2571         (ipa_write_jump_function): Stream the type_preserved flags.
2572         (ipa_read_jump_function): Likewise.
2574 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
2575             Aldy Hernandez  <aldyh@redhat.com>
2577         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
2578         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
2579         * function.h (struct function): Add has_force_vect_loops and
2580         has_simduid_loops.
2581         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
2582         * gimple.c (gimple_build_omp_critical): Add KIND argument and
2583         handle it.
2584         * gimple.def: Update CLAUSES comments.
2585         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
2586         (gimple_build_omp_for): Add argument to prototype.
2587         (gimple_omp_for_kind): New.
2588         (gimple_omp_for_set_kind): New.
2589         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
2590         GOVD_DATA_SHARE_CLASS.
2591         (enum omp_region_type): Add ORT_SIMD.
2592         (gimple_add_tmp_var): Handle ORT_SIMD.
2593         (gimplify_var_or_parm_decl): Same.
2594         (is_gimple_stmt): Same.
2595         (omp_firstprivatize_variable): Same.
2596         (omp_add_variable): Only use splay_tree_insert if lookup failed.
2597         (omp_notice_variable): Handle ORT_SIMD.
2598         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
2599         (omp_check_private): Handle ORT_SIMD.
2600         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2601         OMP_CLAUSE_SAFELEN.
2602         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
2603         Handle OMP_CLAUSE_LASTPRIVATE.
2604         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2605         OMP_CLAUSE_SAFELEN.
2606         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
2607         (gimplify_expr): Handle OMP_SIMD.
2608         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
2609         (expand_GOMP_SIMD_VF): New.
2610         (expand_GOMP_SIMD_LAST_LANE): New.
2611         * internal-fn.def (GOMP_SIMD_LANE): New.
2612         (GOMP_SIMD_VF): New.
2613         (GOMP_SIMD_LAST_LANE): New.
2614         * omp-low.c: Include target.h.
2615         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
2616         OMP_CLAUSE_SAFELEN.
2617         (check_omp_nesting_restrictions): Same.
2618         (omp_max_vf): New.
2619         (lower_rec_simd_input_clauses): New.
2620         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
2621         OMP_CLAUSE_LINEAR.
2622         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
2623         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
2624         (expand_omp_build_assign): New.
2625         (expand_omp_for_init_counts): New.
2626         (expand_omp_for_init_vars): New.
2627         (extract_omp_for_update_vars): New.
2628         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
2629         and rewrite accordingly.
2630         (expand_omp_simd): New.
2631         (expand_omp_for): Use expand_omp_simd.
2632         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
2633         (lower_omp_for): Do not lower the body.
2634         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
2635         in their own loops.
2636         * tree-flow.h (find_omp_clause): Remove prototype.
2637         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
2638         forcing vectorization of the loop, or if flag_tree_vectorize.
2639         (gate_tree_if_conversion): Similarly.
2640         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
2641         gimple_build_omp_for.
2642         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
2643         * tree-parloops (create_parallel_loop): Pass kind argument to
2644         gimple_build_omp_for.
2645         * tree-pretty-print.c (dump_omp_clause): Add cases for
2646         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
2647         OMP_CLAUSE__SIMDUID_.
2648         (dump_generic_node): Handle OMP_SIMD.
2649         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
2650         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
2651         unroll OMP_SIMD loops here.
2652         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
2653         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
2654         loop->safelen.
2655         (vect_analyze_data_refs): Handle simd loops.
2656         * tree-vect-loop.c (vectorizable_live_operation): Handle
2657         IFN_GOMP_SIMD*.
2658         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
2659         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
2660         (vectorizable_load): Same.
2661         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
2662         (struct simduid_to_vf): New.
2663         (simduid_to_vf::hash): New.
2664         (simduid_to-vf::equal): New.
2665         (struct simd_array_to_simduid): New.
2666         (simd_array_to_simduid::hash): New.
2667         (simd_array_to_simduid::equal): New.
2668         (adjust_simduid_builtins): New.
2669         (struct note_simd_array_uses_struct): New.
2670         (note_simd_array_uses_cb): New.
2671         (note_simd_array_uses): New.
2672         (vectorize_loops): Handle simd hints and adjust simd builtins
2673         accordingly.
2674         * tree-vectorizer.h (struct _stmt_vec_info): Add
2675         simd_lane_access_p field.
2676         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
2677         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
2678         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
2679         (omp_clause_code_name): Same.
2680         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
2681         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
2682         * tree.def (OMP_SIMD): New entry.
2683         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
2684         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
2685         (OMP_CLAUSE_DECL): Adjust range for new clauses.
2686         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
2687         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
2688         (OMP_CLAUSE_LINEAR_STEP): New.
2689         (OMP_CLAUSE_SAFELEN_EXPR): New.
2690         (OMP_CLAUSE__SIMDUID__DECL): New.
2691         (find_omp_clause): New prototype.
2693 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
2695         * config/i386/driver-i386.c (host_detect_local_cpu): Update
2696         Haswell processor detection.
2698 2013-08-27  Christian Widmer  <shadow@umbrox.de>
2700         PR target/57927
2701         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
2702         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
2703         AVX2 capable processors.
2705 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
2707         * config/aarch64/arm_neon.h: Replace all inline asm implementations
2708         of vget_low_* with implementations in terms of other intrinsics.
2710 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
2712         PR middle-end/57219
2713         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
2714         values to -1, 0 and 1.
2716 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
2718         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
2719         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
2720         (<optab><mode>3_insn): Remove.
2721         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
2722         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
2723         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
2724         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
2725         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
2726         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
2727         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
2728         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
2729         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
2730         (ror<mode>3_insn): Likewise.
2731         * config/aarch64/predicates.md (aarch64_simd_register): New.
2733 2013-08-27  Richard Biener  <rguenther@suse.de>
2735         PR tree-optimization/57521
2736         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
2737         one edge is non-critical.
2738         (find_phi_replacement_condition): Make sure to use a non-critical
2739         edge.  Cleanup and remove old bug workarounds.
2740         (bb_postdominates_preds): Remove.
2741         (if_convertible_loop_p_1): Do not compute post-dominators.
2742         (combine_blocks): Do not free post-dominators.
2743         (main_tree_if_conversion): Likewise.
2744         (pass_data_if_conversion): Add TODO_verify_ssa.
2746 2013-08-27  DJ Delorie  <dj@redhat.com>
2748         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
2750 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
2752         * function.c (assign_parm_find_data_types): Set passed_mode and
2753         nominal_mode to the TYPE_MODE of nominal_type for the built
2754         pointer type in case of the struct-pass-by-reference.
2756 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2758         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
2759         (UINT16_TYPE): Change default to "unsigned int".
2761         * config/avr/avr.opt (mfract-convert-truncate): New option.
2762         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
2763         is set, round negative fractional integers according to n1169
2764         when converting to integer types.
2766 2013-08-26  Jan Hubicka  <jh@suse.cz>
2768         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
2769         methods can not be called indirectly when their address is not taken.
2771 2013-08-26  Jan Hubicka  <jh@suse.cz>
2773         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
2774         ctor_for_folding.
2776 2013-08-26  Jan Hubicka  <jh@suse.cz>
2778         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
2779         can be unshared.
2781 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2783         * reload.c (find_valid_class): Allow classes that do not include
2784         FIRST_PSEUDO_REGISTER - 1.
2786 2013-08-26  Jan Hubicka  <jh@suse.cz>
2788         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
2789         fix edge count/frequency when speculation failed; fix type check
2790         for the direct call.
2792 2013-08-26  Jan Hubicka  <jh@suse.cz>
2794         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
2796 2013-08-26  Jan Hubicka  <jh@suse.cz>
2798         * ipa-inline-transform.c (inline_transform): Be ready for basic block
2799         to be changed by edge redirection.
2801 2013-08-26  Jan Hubicka  <jh@suse.cz>
2803         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
2804         formating; add sanity check.
2805         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
2806         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
2807         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
2808         (resolve_noninline_speculation): Update callee keys, too.
2810 2013-08-26  Jan Hubicka  <jh@suse.cz>
2812         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
2813         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
2815 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
2817         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
2818         into proper place.
2820 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
2822         * config/i386/i386.c (ix86_debug_options): Remove prototype.
2823         (x86_64_elf_select_section): Ditto.
2824         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
2825         arguments.
2826         (ix86_pass_by_reference): Ditto.
2827         (output_set_got): Ditto.
2828         (ix86_unary_operator_ok): Ditto.
2829         (ix86_expand_builtin): Ditto.
2831 2013-08-23  Jan Hubicka  <jh@suse.cz>
2833         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
2835 2013-08-23  Jan Hubicka  <jh@suse.cz>
2837         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
2838         (tree_decl_with_vis): Add FINAL field.
2840 2013-08-23  Jeff Law  <law@redhat.com>
2842         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
2843         the debugging dump when the expression is fully redundant.
2845 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2847         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
2848         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
2849         * pretty-print.c (pp_formatted_text_data): Likewise.
2850         (pp_write_text_to_stream): Likewise.
2851         (pp_write_text_as_dot_label_to_stream): Likewise.
2852         (pp_append_r): Likewise.
2853         (pp_format): Likewise.
2854         (pp_flush): Likewise.
2855         (pp_clear_output_area): Likewise.
2856         (pp_append_text): Likewise.
2857         (pp_formatted_text): Likewise.
2858         (pp_remaining_character_count_for_line): Likewise.
2859         (pp_newline): Likewise.
2860         (pp_character): Likewise.
2861         (output_buffer::~output_buffer): Define.
2862         (pretty_printer::~pretty_printer): Destruct output buffer.
2863         * pretty-print.h (output_buffer::~output_buffer): Declare.
2864         (pretty_printer::~pretty_printer): Declare virtual.
2866 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
2868         PR other/57324
2869         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
2870         HOST_WIDE_INT_M1U): New macros.
2871         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
2872         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
2873         unsigned -1 for lshift.
2874         * cse.c (cse_insn): Likewise.
2875         * double-int.c (rshift_double, lshift_double): Likewise.
2876         * builtins.c (fold_builtin_bitop): Likewise.
2877         * combine.c (force_to_mode): Likewise.
2878         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
2879         * simplify-rtx.c (simplify_const_unary_operation,
2880         simplify_const_binary_operation): Likewise.
2881         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
2882         check_va_list_escapes): Likewise.
2883         * rtlanal.c (nonzero_bits1): Likewise.
2884         * expmed.c (expand_smod_pow2): Likewise.
2885         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
2887 2013-08-23  Jan Hubicka  <jh@suse.cz>
2889         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
2890         as having address taken.
2892 2013-08-23  Jan Hubicka  <jh@suse.cz>
2894         * ipa-utils.h (method_class_type): Declare.
2895         * ipa-devirt.c (method_class_type): Export.
2897         * cgraphunit.c (analyze_functions): Do basic devirtualization;
2898         do not walk base classes of anonymous types.
2900 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
2902         PR rtl-optimization/58220
2903         PR regression/58221
2904         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
2905         handle SEQUENCE insns properly.
2907 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2909         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
2910         definition.
2911         (pp_newline_and_indent): Likewise.
2912         (pp_separate_with): Likewise.
2913         * pretty-print.c (pp_newline_and_flush): Define.
2914         (pp_newline_and_indent): Likewise.
2915         (pp_separate_with): Likewise.
2917 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2919         PR target/58218
2920         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
2921         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
2923 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
2925         * config/i386/predicates.md (ext_sse_reg_operand): New.
2926         * config/i386/i386.md (*movti_internal): Use
2927         predicate to determine if EVEX is needed.
2928         (*movsi_internal): Ditto.
2929         (*movdf_internal): Ditto.
2930         (*movsf_internal): Ditto.
2931         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
2933 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
2935         PR tree-optimization/58209
2936         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
2937         (find_tail_calls): Give up for pointer result types if m is non-NULL.
2938         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
2939         emit POINTER_PLUS_EXPR.
2940         (create_tailcall_accumulator): For pointer result type accumulate in
2941         sizetype type.
2943 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
2945         * configure.ac: Add backslashes missing from the last change.
2946         * configure: Regenerate.
2948 2013-08-22  Jan Hubicka  <jh@susue.cz>
2950         * ipa.c (function_and_variable_visibility): First remember function
2951         was global and then make it local.
2953 2013-08-22  Julian Brown  <julian@codesourcery.com>
2955         * configure.ac: Add aarch64 to list of arches which use "nop" in
2956         debug_line test.
2957         * configure: Regenerate.
2959 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2961         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
2962         gnu_libc_has_function.
2963         * config/s390/tpf.h: Likewise.
2965 2013-08-22  Jan Hubicka  <jh@susue.cz>
2967         * timevar.c (validate_phases): Add cast.
2969 2013-08-22  Jan Hubicka  <jh@susue.cz>
2971         * timevar.c (validate_phases): Use size_t for memory.
2972         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
2974 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2976         * pretty-print.h (output_buffer::output_buffer): Declare.
2977         (pretty_printer::pretty_printer): Likewise.
2978         (pp_construct): Remove.
2979         * pretty-print.c (output_buffer::output_buffer): Define.
2980         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
2981         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
2982         (print_gimple_expr): Likewise.
2983         (print_gimple_seq): Likewise.
2984         (gimple_dump_bb): Likewise.
2985         * sched-vis.c (dump_value_slim): Likewise.
2986         (dump_insn_slim): Likewise.
2987         (dump_rtl_slim): Likewise.
2988         (str_pattern_slim): Likewise.
2989         * tree-mudflap.c (mf_varname_tree): Likewise.
2990         * graph.c (print_graph_cfg): Likewise.
2991         (start_graph_dump): Likewise.
2992         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
2993         placement-new.
2994         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
2995         pretty printer initialization.
2996         * coretypes.h (diagnostic_context): Remove superflous type alias
2997         declaration.
2998         (pretty_printer): Likewise.  Declare directly as a class.
2999         (pretty_print_info): Remove declaration as class.
3000         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
3001         and pp_clear_output_area.
3002         (asan_add_global): Likewise.
3004 2013-08-22  Jan Hubicka  <jh@suse.cz>
3006         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
3007         * ipa-utils.h (update_type_inheritance_graph): Declare.
3008         (possible_polymorphic_call_target_p): Declare.
3009         (possible_polymorphic_call_target_p): New.
3010         * ipa-devirt.c: Update toplevel comments.
3011         (cached_polymorphic_call_targets): Move up.
3012         (odr_type_d): Move ID down.
3013         (polymorphic_type_binfo_p): Update comment.
3014         (odr_hasher::remove): Likewise;
3015         (get_odr_type): Set anonymous_namespace.
3016         (dump_odr_type): Dump it.
3017         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3018         (maybe_record_node): Record node in cached_polymorphic_call_targets.
3019         (record_binfo): Add comment.
3020         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3021         built.
3022         (devirt_node_removal_hook): Do not iCE when cache is freed.
3023         (possible_polymorphic_call_target_p): New predicate.
3024         (update_type_inheritance_graph): New function.
3026 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3027             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3028             Sergey Lega  <sergey.s.lega@intel.com>
3029             Anna Tikhonova  <anna.tikhonova@intel.com>
3030             Ilya Tocar  <ilya.tocar@intel.com>
3031             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3032             Ilya Verbin  <ilya.verbin@intel.com>
3033             Kirill Yukhin  <kirill.yukhin@intel.com>
3034             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3036         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3037         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3038         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3039         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3040         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3041         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3042         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3043         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3044         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3045         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3046         OPT_mavx512pf, OPT_mavx512er cases.
3047         * config/i386/constraints.md (v): New constraint.
3048         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3049         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3050         (bit_AVX512CD): New.
3051         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3052         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3053         * config/i386/i386-c.c (ix86_target_macros_internal):
3054         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3055         __AVX512PF__.
3056         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3057         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3058         * config/i386/i386.c (regclass_map, dbx_register_map)
3059         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3060         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3061         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3062         -mavx512pf options.
3063         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3064         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3065         -mavx512pf options.  Fix formatting.
3066         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3067         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3068         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3069         -mavx512cd, -mavx512pf options.
3070         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3071         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3072         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3073         ALL_SSE_REGS.
3074         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3075         (ix86_set_reg_reg_cost): Ditto.
3076         (x86_order_regs_for_local_alloc): Ditto.
3077         (MAX_VECT_LEN): Extend to 64-byte.
3078         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3079         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3080         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3081         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3082         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3083         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3084         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3085         (SSE_REG_MODE_P): Support new modes.
3086         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3087         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3088         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3089         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3090         (REG_CLASS_CONTENTS): Add new registers.
3091         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3092         (EXT_REX_SSE_REGNO_P): New.
3093         (HI_REGISTER_NAMES): Add new registers.
3094         * config/i386/i386.md: Define constants for new registers.
3095         (mode): Add new 512-bit modes.
3096         (prefix): Support evex prefix.
3097         (isa): Support avx512f, noavx512f, fma_avx512f.
3098         (ssemodesuffix): Add new 512-bit modes.
3099         (movxi): New.
3100         (*movxi_internal_avx512f): Ditto.
3101         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3102         Support MODE_XI.
3103         (*movsi_internal): Likewise.
3104         (*movdf_internal): Likewise.
3105         (*movsf_internal): Likewise.
3106         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3107         (<code><mode>3): Likewise.
3108         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3109         New.
3110         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3111         with the new constraint "v".
3112         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3113         modes.
3114         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3115         with the new constraint "v".
3116         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3117         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3118         (*<plusminus_insn><mode>3): Likewise.
3119         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3120         (*mul<mode>3): Likewise.
3121         (<sse>_vmmul<mode>3): Likewise.
3122         (<sse>_div<mode>3): Likewise.
3123         (<sse>_vmdiv<mode>3): Likewise.
3124         (<sse>_sqrt<mode>2): Likewise.
3125         (<sse>_vmsqrt<mode>2): Likewise.
3126         (*<code><mode>3_finite): Likewise.
3127         (*<code><mode>3) <smaxmin>: Likewise.
3128         (<sse>_vm<code><mode>3): Likewise.
3129         (*<code><mode>3) <any_logic>: Likewise.
3130         (*fma_fmadd_<mode>): Likewise.
3131         (*fma_fmsub_<mode>): Likewise.
3132         (*fma_fnmadd_<mode>): Likewise.
3133         (*fma_fnmsub_<mode>): Likewise.
3134         (*fma_fmaddsub_<mode>): Likewise.
3135         (*fma_fmsubadd_<mode>): Likewise.
3136         (*fmai_fmadd_<mode>): Likewise.
3137         (*fmai_fmsub_<mode>): Likewise.
3138         (*fmai_fnmadd_<mode>): Likewise.
3139         (*fmai_fnmsub_<mode>): Likewise.
3140         (sse_cvtsi2ss): Likewise.
3141         (sse_cvtsi2ssq): Likewise.
3142         (sse_cvtss2si): Likewise.
3143         (sse_cvtss2si_2): Likewise.
3144         (sse_cvtss2siq): Likewise.
3145         (sse_cvtss2siq_2): Likewise.
3146         (sse_cvttss2si): Likewise.
3147         (sse_cvtss2siq_2): Likewise.
3148         (float<sseintvecmodelower><mode>2): Likewise.
3149         (sse2_cvtsd2si_2): Likewise.
3150         (sse2_cvtsd2siq_2): Likewise.
3151         (*<plusminus_insn><mode>3): Likewise.
3152         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3153         (*<sse4_1_avx2>_mul<mode>3): Likewise.
3154         (ashr<mode>3): Likewise.
3155         (<shift_insn><mode>3): Likewise.
3156         (avx2_<code><mode>3): Likewise.
3157         (*avx2_<code><mode>3): Likewise.
3158         (*andnot<mode>3): Likewise.
3159         (*<code><mode>3) <any_logic>: Likewise.
3160         (abs<mode>2): Likewise.
3161         (avx2_permvar<mode>): Likewise.
3162         (avx2_perm<mode>_1): Likewise.
3163         (*avx_vpermilp<mode>): Likewise.
3164         (avx_vpermilvar<mode>3): Likewise.
3165         (avx2_ashrv<mode>): Likewise.
3166         (avx2_<shift_insn>v<mode>): Likewise.
3167         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3168         -mavx512cd.
3169         * doc/rtl.texi: Document XImode.
3171 2013-08-21  Jeff Law  <law@redhat.com>
3173         * tree-flow.h (register_jump_thread): Pass vector of edges
3174         instead of each important edge.
3175         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3176         thread path into a vector and pass that to register_jump_thread.
3177         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3178         passed in edge vector to the current 3-edge form.
3180         Revert:
3181         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3183         * dce.c (fini_dce): Call df_analyze again just in case
3184         delete_unmarked_insns removed anything.
3186 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
3188         * reload.h (struct reg_equivs): Rename to ..
3189         (struct reg_equivs_s): .. this.
3191 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
3193         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3195 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3197         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3199 2013-08-21  Jeff Law  <law@redhat.com>
3201         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3202         simplify assignments too.  If the RHS collapses to a singleton
3203         range, then return the value for the range.
3205 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
3207         * config/i386/sse.md (V16): Rename to...
3208         (VMOVE): this.
3209         (mov<mode>): Update iterator name.
3210         (*mov<mode>_internal): Ditto.
3211         (push<mode>1): Ditto.
3212         (movmisalign<mode>): Ditto.
3214 2013-08-20  Jan Hubicka  <jh@suse.cz>
3216         PR bootstrap/58186
3217         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3218         entry for direct edges.
3219         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3221 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3223         Revert my last two changes, r201865 and r201864:
3225         Revert r201865:
3226         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3228         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3229         instances can own GC refs.
3231         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3232         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3233         (gcc::context::gt_pch_nx): Likewise.
3234         (gcc::context::gt_pch_nx):  Likewise.
3235         * ggc.h (gt_ggc_mx <T>): New.
3236         (gt_pch_nx_with_op <T>): New.
3237         (gt_pch_nx <T>): New.
3238         * passes.c (opt_pass::gt_ggc_mx): New.
3239         (opt_pass::gt_pch_nx): New.
3240         (opt_pass::gt_pch_nx_with_op): New.
3241         (pass_manager::gt_ggc_mx): New.
3242         (pass_manager::gt_pch_nx): New.
3243         (pass_manager::gt_pch_nx_with_op): New.
3244         (pass_manager::operator new): Use
3245         ggc_internal_cleared_alloc_stat rather than xcalloc.
3246         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3247         (pass_manager::gt_ggc_mx): New.
3248         (pass_manager::gt_pch_nx): New.
3249         (pass_manager::gt_pch_nx_with_op): New.
3250         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3251         (opt_pass::operator new): New.
3252         (opt_pass::gt_ggc_mx): New.
3253         (opt_pass::gt_pch_nx): New.
3254         (opt_pass::gt_pch_nx_with_op): New.
3256         Revert r201864:
3257         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3259         * Makefile.in (GTFILES): Add context.h.
3260         * context.c (gcc::context::operator new): New.
3261         (gcc::context::gt_ggc_mx): New.
3262         (gcc::context::gt_pch_nx): New.
3263         (gcc::context::gt_pch_nx): New.
3264         * context.h (gcc::context): Add GTY((user)) marking.
3265         (gcc::context::operator new): New.
3266         (gcc::context::gt_ggc_mx): New.
3267         (gcc::context::gt_pch_nx): New.
3268         (gcc::context::gt_pch_nx): New.
3269         (g): Add GTY marking.
3270         (gt_ggc_mx (gcc::context *)): New.
3271         (gt_pch_nx (gcc::context *)): New.
3272         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3273         void *cookie)): New.
3274         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3276 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3278         * dce.c (fini_dce): Call df_analyze again just in case
3279         delete_unmarked_insns removed anything.
3281 2013-08-20  Teresa Johnson  <tejohnson@google.com>
3283         PR rtl-optimizations/57451
3284         * final.c (reemit_insn_block_notes): Prevent lexical blocks
3285         from crossing split section boundaries.
3287 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
3289         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
3290         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
3291         with MULTLIB_DEFAULTS.
3293 2013-08-20  Nick Clifton  <nickc@redhat.com>
3295         * target.def (narrow_volatile_bitfield): Note that the default
3296         value is false, not !TARGET_STRICT_ALIGN.
3297         * doc/tm.texi: Regenerate.
3299 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
3301         Fix LIB_SPEC for systems without libpthread.
3303         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
3304         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
3305         for Android.
3306         * config/i386/linux-common.h: Likewise.
3307         * config/mips/linux-common.h: Likewise.
3309 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
3311         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
3312         checks.
3314 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3316         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3317         instances can own GC refs.
3319         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3320         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3321         (gcc::context::gt_pch_nx): Likewise.
3322         (gcc::context::gt_pch_nx):  Likewise.
3323         * ggc.h (gt_ggc_mx <T>): New.
3324         (gt_pch_nx_with_op <T>): New.
3325         (gt_pch_nx <T>): New.
3326         * passes.c (opt_pass::gt_ggc_mx): New.
3327         (opt_pass::gt_pch_nx): New.
3328         (opt_pass::gt_pch_nx_with_op): New.
3329         (pass_manager::gt_ggc_mx): New.
3330         (pass_manager::gt_pch_nx): New.
3331         (pass_manager::gt_pch_nx_with_op): New.
3332         (pass_manager::operator new): Use
3333         ggc_internal_cleared_alloc_stat rather than xcalloc.
3334         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3335         (pass_manager::gt_ggc_mx): New.
3336         (pass_manager::gt_pch_nx): New.
3337         (pass_manager::gt_pch_nx_with_op): New.
3338         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3339         (opt_pass::operator new): New.
3340         (opt_pass::gt_ggc_mx): New.
3341         (opt_pass::gt_pch_nx): New.
3342         (opt_pass::gt_pch_nx_with_op): New.
3344 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3346         * Makefile.in (GTFILES): Add context.h.
3347         * context.c (gcc::context::operator new): New.
3348         (gcc::context::gt_ggc_mx): New.
3349         (gcc::context::gt_pch_nx): New.
3350         (gcc::context::gt_pch_nx): New.
3351         * context.h (gcc::context): Add GTY((user)) marking.
3352         (gcc::context::operator new): New.
3353         (gcc::context::gt_ggc_mx): New.
3354         (gcc::context::gt_pch_nx): New.
3355         (gcc::context::gt_pch_nx): New.
3356         (g): Add GTY marking.
3357         (gt_ggc_mx (gcc::context *)): New.
3358         (gt_pch_nx (gcc::context *)): New.
3359         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3360         void *cookie)): New.
3361         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3363 2013-08-20  Alan Modra  <amodra@gmail.com>
3365         PR target/57865
3366         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
3367         (rs6000_emit_epilogue): Likewise.
3369 2013-08-19  Dehao Chen  <dehao@google.com>
3371         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
3373 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
3374             Jakub Jelinek  <jakub@redhat.com>
3376         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
3377         (BUILT_IN_FABSD64): Likewise.
3378         (BUILT_IN_FABSD128): Likewise.
3379         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
3380         (fold_builtin_1): Likewise.
3381         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
3382         destination and source operands.
3383         (*abstd2_fpr): Likewise.
3384         (*nabstd2_fpr): Likewise.
3386 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
3388         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
3389         JUMP_P and INSN_P.
3391 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
3393         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
3394         incomplete.
3396 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3398         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
3399         * builtins.c (default_libc_has_function): New.
3400         (gnu_libc_has_function): Ditto.
3401         (no_c99_libc_has_function): Ditto.
3402         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
3403         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
3404         (fold_builtin_sincos): Likewise.
3405         (fold_builtin_cexp): Likewise.
3406         * builtins.def (DEF_C94_BUILTIN): Likewise.
3407         (DEF_C99_BUILTIN): Likewise.
3408         (DEF_C99_C90RES_BUILTIN): Likewise.
3409         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
3410         definitions to using this define.
3411         * config/darwin-protos.h (darwin_libc_has_function): New.
3412         * config/darwin.c (darwin_libc_has_function): Ditto.
3413         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
3414         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
3415         * config/darwin.h: Ditto.
3416         * config/elfos.h: Ditto.
3417         * config/freebsd.h: Ditto.
3418         * config/i386/cygming.h: Ditto.
3419         * config/i386/djgpp.h: Ditto.
3420         * config/i386/i386-interix.h: Ditto.
3421         * config/microblaze/microblaze.h: Ditto.
3422         * config/mmix/mmix.h: Ditto.
3423         * config/gnu-user.h: Ditto.
3424         * config/ia64/hpux.h: Ditto.
3425         * config/pa/pa-hpux.h: Ditto.
3426         * config/pdp11/pdp11.h: Ditto.
3427         * config/picochip/picochip.h: Ditto.
3428         * config/linux.h: Ditto.
3429         * config/netbsd.h: Ditto.
3430         * config/openbsd.h: Ditto.
3431         * config/rs6000/aix43.h: Ditto.
3432         * config/rs6000/aix51.h: Ditto.
3433         * config/rs6000/aix52.h: Ditto.
3434         * config/rs6000/aix53.h: Ditto.
3435         * config/rs6000/aix61.h: Ditto.
3436         * config/rs6000/darwin.h: Ditto.
3437         * config/rs6000/linux.h: Ditto.
3438         * config/rs6000/linux64.h: Ditto.
3439         * config/s390/tpf.h: Ditto.
3440         * config/sol2-10.h: Ditto.
3441         * config/sol2.h: Ditto.
3442         * config/vms/vms.h: Ditto.
3443         * config/vxworks.h: Ditto.
3444         * config/linux-android.c (linux_android_libc_has_function):
3445         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
3446         * config/linux-protos.h (linux_android_libc_has_function):
3447         New declaration.
3448         * config/i386/i386.c (ix86_libc_has_function): New.
3449         * config/i386/i386-protos.h
3450         (ix86_libc_has_function): New declaration.
3451         * config/i386/i386.md
3452         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
3453         ("isinf<mode>2): Likewise.
3454         * convert.c (convert_to_integer): Using new target hook
3455         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
3456         TARGET_C99_FUNCTIONS.
3457         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
3458         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
3459         * coretypes.h (function_class): New enum for different
3460         classes of functions.
3461         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
3462         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
3463         (TARGET_HAS_SINCOS): Likewise.
3464         (TARGET_LIBC_HAS_FUNCTION): New.
3465         * doc/tm.texi: Regenerated.
3466         * targhooks.h (default_libc_has_function): New declaration.
3467         (no_c99_libc_has_function): Ditto.
3468         (gnu_libc_has_function): Ditto.
3469         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
3470         and TARGET_HAS_SINCOS.
3472 2013-08-18  Jan Hubicka  <jh@suse.cz>
3474         * Makeifle-in (ipa-devirt.o): New.
3475         (GTFILES): Add ipa-utils.h and ipa-devirt.c
3476         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
3477         (analyze_functions): Look into possible targets of polymorphic call.
3478         * dumpfile.c (dump_files): Add type-inheritance dump.
3479         * dumpfile.h (TDI_inheritance): New.
3480         * ipa-devirt.c: New file.
3481         * ipa-utils.h (odr_type_d): Forward declare.
3482         (odr_type): New type.
3483         (build_type_inheritance_graph): Declare.
3484         (possible_polymorphic_call_targets): Declare and introduce inline
3485         variant when only edge is pased.
3486         (dump_possible_polymorphic_call_targets): Likewise.
3487         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
3488         * tree.c (type_in_anonymous_namespace_p): Break out from ...
3489         (types_same_for_odr): ... here.
3490         * tree.h (type_in_anonymous_namespace_p): Declare.
3492 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
3494         PR tree-optimization/58006
3495         * tree-parloops.c (take_address_of): Don't ICE if get_name
3496         returns NULL.
3497         (eliminate_local_variables_stmt): Remove clobber stmts.
3499 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
3501         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
3502         error message is issued for an alias to undefined symbol.
3504 2013-08-18  Jan Hubicka  <jh@suse.cz>
3506         * cgraph.c (cgraph_create_indirect_edge): Discover
3507         polymorphic calls and record basic info into indirect_info.
3508         * gimple-fold.c (gimple_fold_call): When doing BINFO based
3509         devirtualization, ignore objc function calls.
3510         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
3511         call with no parm index info.
3512         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
3513         * tree.c (virtual_method_call_p): New function.
3514         * tree.h (virtual_method_call_p): Declare.
3516 2013-08-16  Jan Hubicka  <jh@suse.cz>
3518         PR middle-end/58179
3519         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
3521 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
3523         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
3524         attribute.
3526 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3528         * gengtype.c (type_for_name): Add special-case support for
3529         locating types within the "gcc::" namespace.
3530         (open_base_files): Emit a "using namespace gcc" directive.
3532 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3534         PR target/58160
3535         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
3536         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
3538         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
3539         array instead of each individual operand as a separate argument.
3540         (emit_fusion_gpr_load): Likewise.
3541         (expand_fusion_gpr_load): Add new function declaration.
3543         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
3544         signature to have the operands passed as an array, instead of as
3545         separate arguments.  Allow ZERO_EXTEND to be in the memory
3546         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
3547         depend on the register live/dead flags when peepholes are run.
3548         (expand_fusion_gpr_load): New function to be called from the
3549         peephole2 pass, to change the register that addis sets to be the
3550         target register.
3551         (emit_fusion_gpr_load): Change the calling signature to have the
3552         operands passed as an array, instead of as separate arguments.
3553         Allow ZERO_EXTEND to be in the memory address, and also
3554         SIGN_EXTEND if -mpower8-fusion-sign.
3556         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
3557         unspec enumeration.
3558         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
3559         adjust the register addis loads up in the peephole2 pass.  Do not
3560         depend on the register live/dead state when the peephole pass is done.
3562 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3564         * gengtype.c (create_user_defined_type): Ensure that the kind
3565         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
3566         declaration is seen before the GTY((user)) marking.
3568 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3570         PR target/58105
3571         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
3573 2013-08-16  Jan Hubicka  <jh@suse.cz>
3575         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
3576         arugment expected_type.
3577         (gimple_fold_call): Use it.
3578         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
3579         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
3580         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
3581         (try_make_edge_direct_virtual_call): Likewise.
3582         * tree.c (obj_type_ref_class): New.
3583         * tree.h (obj_type_ref_class): Use it.
3585 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3587         * sched-vis.c (rtl_slim_pp_initialized): Remove.
3588         (rtl_slim_pp): Likewise.
3589         (init_rtl_slim_pretty_print): Likewise.
3590         (dump_value_slim):  Don't call it.  Use local pretty printer.
3591         (dump_insn_slim): Likewise.
3592         (dump_rtl_slim): Likewise.
3593         (str_pattern_slim): Likewise.
3594         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
3595         Simplify.
3597 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
3599         PR tree-optimization/58164
3600         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
3601         walk gimple_goto_dest of GIMPLE_GOTO.
3603         PR tree-optimization/58165
3604         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
3605         bi_call must be the last stmt in a bb, don't split_block, instead
3606         use fallthru edge from it and give up if there is none.
3607         Release conds vector when returning early.
3609 2013-08-14  Xinliang David Li  <davidxl@google.com>
3611         * config/i386/i386.c (ix86_option_override_internal):
3612         Remove unused variable and field.
3614 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3616         PR target/57949
3617         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
3618         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
3619         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
3620         and Linux, correct BLKmode alignment when 128-bit alignment is
3621         required and compatibility flag is not set.
3622         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
3623         for zero-size arguments when compatibility flag is not set.
3625 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3627         PR tree-optimization/58145
3628         * tree-sra.c (build_ref_for_offset): If prev_base has
3629         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
3631 2013-08-14  Xinliang David Li  <davidxl@google.com>
3633         * config/i386/i386.c (ix86_option_override_internal):
3634         Fix uninitialized variable error.
3636 2013-08-14  Xinliang David Li  <davidxl@google.com>
3638         * config/i386/i386.opt: Define two new options.
3639         * config/i386/x86-tune.def: Add arch selector field in macros.
3640         * config/i386/i386.h: Adjust macro definition.
3641         * config/i386/i386.c (ix86_option_override_internal):
3642         Refactor the code.
3643         (parse_mtune_ctrl_str): New function.
3644         (set_ix86_tune_features): New function.
3645         (ix86_function_specific_restore): Call the new helper function.
3647 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
3649         PR rtl-optimization/57662
3650         * sel-sched.c (code_motion_process_successors): When the current insn
3651         is removed after the recursive traversal, break from the loop.
3652         Add comments and debug printouts.
3654 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
3655             Alexandre Oliva  <aoliva@redhat.com>
3657         PR target/58067
3658         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
3659         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
3660         there also UNSPEC_PLTOFF.
3662 2013-08-14  Marek Polacek  <polacek@redhat.com>
3664         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
3665         NUM_CONDITIONS bit positions.
3667 2013-08-13  Cary Coutant  <ccoutant@google.com>
3669         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
3670         (attr_checksum): Hash vector contents instead of pointer.
3671         (attr_checksum_ordered): Likewise.
3673 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
3675         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
3676         when Pmode != word_mode.  Add length_address attribute.
3677         (sse3_monitor_<mode>): Merge from sse3_monitor and
3678         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
3679         Pmode != word_mode.  Update insn length attribute.
3680         * config/i386/i386.c (ix86_option_override_internal): Update
3681         ix86_gen_monitor selection for merged sse3_monitor insn.
3683 2013-08-13  Julian Brown  <julian@codesourcery.com>
3685         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
3686         perform invalid legitimization on greater-than-word-size modes for
3687         TARGET_E500_DOUBLE.
3689 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
3691         * ira.c (setup_class_translate_array): Use aclass instead of cl
3692         for classes not fully covered by allocno classes.
3694 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
3696         PR tree-optimization/57661
3697         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
3698         * tree-inline.c (tree_function_versioning): Initialize it.
3699         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
3700         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
3701         that is not being copied.
3703         PR sanitizer/56417
3704         * asan.c (instrument_strlen_call): Fix typo in comment.
3705         Use char * type even for the lhs of POINTER_PLUS_EXPR.
3707 2013-08-13  Steve Ellcey  <sellcey@mips.com>
3709         * config/mips/mips.md (prefetch): Use lw instead of ld on
3710         loongson in 32bit mode.
3712 2013-08-13  Nick Clifton  <nickc@redhat.com>
3714         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
3716 2013-08-13  Jan Hubicka  <jh@suse.cz>
3718         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
3719         introduced edge; fix typo in sanity check.
3720         (cgraph_resolve_speculation): Export; improve diagnostic.
3721         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
3722         speculation at type mismatch.
3723         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
3724         (cgraph_resolve_speculation): Declare.
3725         (symtab_can_be_discarded): New function.
3726         * value-prof.c (gimple_ic_transform): Remove actual transform code.
3727         * ipa-inline-transform.c (speculation_removed): New global var.
3728         (clone_inlined_nodes): See if speculation can be removed.
3729         (inline_call): If speculations was removed, we growths may not match.
3730         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
3731         (speculation_useful_p): New function.
3732         (resolve_noninline_speculation): New function.
3733         (inline_small_functions): Resolve useless speculations.
3734         * ipa-inline.h (speculation_useful_p): Declare
3735         * ipa.c (can_replace_by_local_alias): Simplify.
3736         (ipa_profile): Produce speculative calls in non-lto, too;
3737         add simple cost model; produce local aliases.
3739 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
3741         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
3742         PASS_MANAGER_H.
3744 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
3746         * config/i386/i386.c (ix86_function_versions): Use error + inform.
3748 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
3750         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
3751         iterator instead of X87MODEF.
3753 2013-08-12  Perez Read  <netfirewall@gmail.com>
3755         PR target/58132
3756         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
3757         operand 0 for intel asm alternative.
3758         (*movabs<mode>_2): Ditto for operand 1.
3760 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
3762         * config/aarch64/arm_none.h
3763         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
3765 2013-08-12  Nick Clifton  <nickc@redhat.com>
3767         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
3769 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
3771         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
3772         expand for QI/HImode operand to produce more effictive code for
3773         unsigned char(short) --> float(double) conversion.
3775 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
3777         * doc/invoke.texi: Mention that -ftls-model does not force the final
3778         model.
3780 2013-08-12  Marek Polacek  <polacek@redhat.com>
3781             Marc Glisse  <marc.glisse@inria.fr>
3783         PR tree-optimization/57980
3784         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
3785         when creating -1 constant.
3787 2013-08-10  Jan Hubicka  <jh@suse.cz>
3789         Workaround binutils PR14342.
3790         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
3791         (gimple_init_edge_profiler): Likewise.
3792         (gimple_gen_ic_func_profiler): Likewise.
3794 2013-08-09  Jan Hubicka  <jh@suse.cz>
3796         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
3798 2013-08-09  Xinliang David Li  <davidxl@google.com>
3800         * config/i386/stringop.def: New file.
3801         * config/i386/stringop.opt: New file.
3802         * config/i386/i386-opts.h: Include stringopt.def.
3803         * config/i386/i386.opt: Include stringopt.opt.
3804         * config/i386/i386.c (ix86_option_override_internal):
3805         Override default size based stringop inline strategies with options.
3806         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
3807         New function.
3809 2013-08-09  Jan Hubicka  <jh@suse.cz>
3811         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
3813 2013-08-09  Jan Hubicka  <jh@suse.cz>
3815         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
3816         CGRAPH_FREQ_MAX.
3817         (dump_cgraph_node): Dump profile-id.
3818         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
3819         and common_target_probability.
3820         * lto-cgraph.c (lto_output_edge): Stream common targets.
3821         (lto_output_node): Stream profile ids.
3822         (input_node): Stream profile ids.
3823         (input_edge): Stream common targets.
3824         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
3825         * ipa.c: Include value-prof.h
3826         (ipa_profile_generate_summary): Turn indirect call statement histograms
3827         into common targets.
3828         (ipa_profile): Turn common targets into speculative edges.
3830 2013-08-09  Jan Hubicka  <jh@suse.cz>
3832         * cgraph.h (cgraph_node): Add profile_id.
3833         * value-prof.c (cgraph_node_map): Turn into pointer_map.
3834         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
3835         (del_node_map): Update.
3836         (find_func_by_funcdef_no): Replace by ...
3837         (find_func_by_profile_id): ... this one.
3838         (gimple_ic_transform): Do not remove useful histograms when
3839         speculation is not done; dump info when indirect call removal
3840         can happen at LTO.
3841         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
3842         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
3843         (__gcov_indirect_call_profiler_v2): .. this one.
3844         * profile.h (init_node_map): Update.
3845         * coverage.c (coverage_compute_profile_id): New function.
3846         * coverage.h (coverage_compute_profile_id): Declare.
3847         * tree-profile.c (init_ic_make_global_vars): Make
3848         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
3849         (gimple_init_edge_profiler): Update prototype of
3850         __gcov_indirect_call_profiler.
3851         (gimple_gen_ic_func_profiler): Simplify.
3852         (tree_profiling): Use init_node_map
3854 2013-08-09  Jan Hubicka  <jh@suse.cz>
3856         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
3857         non-speculative refs.
3858         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
3859         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
3860         (cgraph_set_call_stmt): Likewise.
3861         (cgraph_create_edge_1): Fix release checking compilatoin;
3862         clear lto_stmt_uid.
3863         (cgraph_free_edge): Free indirect info.
3864         (cgraph_turn_edge_to_speculative): New function.
3865         (cgraph_speculative_call_info): New function.
3866         (cgraph_make_edge_direct): Return direct edge; handle speculation.
3867         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
3868         (dump_cgraph_node): Dump speculation.
3869         (verify_edge_count_and_frequency): Accept speculative edges.
3870         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
3871         (verify_cgraph_node): Handle speculation.
3872         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
3873         (cgraph_set_call_stmt): Update prototype.
3874         (cgraph_make_edge_direct): Update prototype.
3875         (cgraph_speculative_call_info): Declare.
3876         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
3877         to change; update call of ipa_find_references.
3878         * ipa-ref.c (ipa_record_reference): Fix return value; clear
3879         lto_stmt_uid and speculative flags.
3880         (ipa_dump_references): Dump speculation.
3881         (ipa_clone_references): Clone speculative flag.
3882         (ipa_clone_referring): Likewise.
3883         (ipa_clone_ref): New function.
3884         (ipa_find_reference): Look into lto_stmt_uids
3885         (ipa_clear_stmts_in_references): Do not clear speculative calls.
3886         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
3887         (ipa_find_reference): Update declaration.
3888         (ipa_clone_ref): Declare.
3889         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
3890         stream speculative flag.
3891         (lto_output_ref): Stream statements uids and speculation.
3892         (input_ref): Likewise.
3893         (input_edge): Stream speuclation.
3894         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
3895         (cgraph_set_call_stmt_including_clones): Handle speculation.
3896         * ipa-inline.c (heap_edge_removal_hook): New function.
3897         (inline_small_functions): Register it.
3898         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
3899         also initialize refs.
3900         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
3901         edge to change.
3902         (try_make_edge_direct_simple_call): Likewise.
3903         (try_make_edge_direct_simple_call): Likewise.
3904         (update_indirect_edges_after_inlining): Likewise.
3905         (remove_described_reference): Look proper lto_stmt_uid.
3906         (propagate_controlled_uses): Likewise.
3907         (propagate_controlled_uses): Liekwise.
3908         * tree-inline.c (copy_bb): Copy speculative edges.
3909         (redirect_all_calls): New function.
3910         (copy_cfg_body): Do redirection after loop info is updated.
3911         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
3913 2013-08-09  Jan Hubicka  <jh@suse.cz>
3915         * lto-streamer-out.c (output_function): Renumber PHIs.
3916         * lto-streamer-in.c (input_function): Likewise.
3918 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
3920         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
3921         (get_lane_unsigned): Likewise.
3922         (dup_lane_scalar): Likewise.
3923         (get_lane): enable for VALL.
3924         * config/aarch64/aarch64-simd.md
3925         (aarch64_dup_lane_scalar<mode>): Remove.
3926         (aarch64_get_lane_signed<mode>): Likewise.
3927         (aarch64_get_lane_unsigned<mode>): Likewise.
3928         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
3929         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
3930         (aarch64_get_lane<mode>): Enable for all vector modes.
3931         (aarch64_get_lanedi): Remove misleading constraints.
3932         * config/aarch64/arm_neon.h
3933         (__aarch64_vget_lane_any): Define.
3934         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
3935         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
3936         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
3937         * config/aarch64/iterators.md (VDQQH): New.
3938         (VDQQHS): Likewise.
3939         (vwcore): Likewise.
3941 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
3943         * configure.ac: Add GAS check for LEON instructions on SPARC.
3944         * configure: Regenerate.
3945         * config.in: Likewise.
3946         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
3947         sparc*-*-* block.
3948         * config/sparc/sparc.opt (LEON, LEON3): New masks.
3949         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
3950         for LEON or LEON3.
3951         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
3952         (AS_LEON_FLAG): New macro.
3953         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
3954         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
3955         Deal with LEON and LEON3 for the memory model.
3956         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
3957         (atomic_compare_and_swap<mode>_1): Likewise.
3958         (*atomic_compare_and_swap<mode>_1): Likewise.
3960 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3962         * config/arm/neon.md (vcond): Fix floating-point vector
3963         comparisons against 0.
3965 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
3967         * lra-constraints.c (emit_spill_move): Remove assert.
3968         (process_alt_operands): Add more debugging
3969         output.  Increase reject for spilling into memory.  Decrease
3970         reject for reloading scratch.
3971         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
3973 2013-08-08  Steve Ellcey  <sellcey@mips.com>
3975         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
3976         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
3977         micromips incompatible.  Add nan2008.
3978         (MULTILIB_DIRNAMES): Add nan2008.
3979         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3980         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
3981         and micromips incompatible.  Add nan2008.
3982         (MULTILIB_DIRNAMES): Add nan2008.
3983         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
3985 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
3987         PR rtl-optimization/58079
3988         * combine.c (combine_simplify_rtx): Avoid using SUBST if
3989         simplify_comparison has widened a comparison with an integer.
3991 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3993         * config/arm/neon.md (movmisalign<mode>): Disable when we
3994         don't allow unaligned accesses.
3995         (*movmisalign<mode>_neon_store): Likewise.
3996         (*movmisalign<mode>_neon_load): Likewise.
3997         (*movmisalign<mode>_neon_store): Likewise.
3998         (*movmisalign<mode>_neon_load): Likewise.
4000 2013-08-08  Jan Hubicka  <jh@suse.cz>
4002         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
4003         (make_pass_rebuild_cgraph_edges): Also clear references.
4004         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
4005         * ipa-inline-transform.c (inline_transform): Remove all references
4006         after inlining.
4007         * cgraphunit.c (expand_function): Remove all references after
4008         expansion.
4009         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
4010         (ipa_find_reference): Rewrite to iterator.
4011         (remove_stmt_references): Likewise.
4012         (ipa_clear_stmts_in_references): New function.
4013         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4014         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4015         clear references.
4016         * ipa-split.c (split_function): Remove references in split function.
4018 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4020         PR target/57431
4021         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4022         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4024 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4026         PR target/56979
4027         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4028         suggested mode for the assignment isn't compatible with the
4029         registers required.
4031 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4033         PR target/58065
4034         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4036 2013-08-07  Xinliang David Li  <davidxl@google.com>
4038         * config/i386/i386.opt: New option -mtune-ctrl=.
4039         * config/i386/x86-tune.def: New file.
4040         * config/i386/i386.h: include x86-tune.def.
4041         * config/i386/i386.c (ix86_option_override_internal):
4042         Parsing -mtune-ctrl= option and set tune features.
4044 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4046         PR other/12081
4047         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4048         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4049         to gen_2arg_fn_t.
4051 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4053         * rtl.h (update_alignments): Declare.
4054         * final.c (grow_label_align): New function extracted from...
4055         (shorten_branches): ...here.  Call it.
4056         (update_alignments): New function.
4057         * reorg.c (sibling_labels): New variable.
4058         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4059         the new label along with it onto the sibling_labels vector.
4060         (fill_simple_delay_slots): Adjust call to get_label_before.
4061         (fill_slots_from_thread): Likewise.
4062         (relax_delay_slots): Likewise.
4063         (make_return_insns): Likewise.
4064         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4066 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4068         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4069         document its semantics.
4070         (diagnostic_report_diagnostic): Adjust accordingly.
4072 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4074         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4075         (sparc_option_override): ...and port to new C++ pass API.
4076         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4078 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4080         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4082 2013-08-06  Caroline Tice  <cmtice@google.com>
4084         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4085         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4086         * tree-pass.h: Add pass_vtable_verify.
4087         * varasm.c (assemble_variable): Add code to properly set the comdat
4088         section and name for the .vtable_map_vars section.
4089         (assemble_vtyv_preinit_initializer): New function.
4090         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4091         LINK_ONCE flag.
4092         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4093         * vtable-verify.c: New file.
4094         * vtable-verify.h: New file.
4095         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4096         initialiation levels.
4097         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4098         * passes.def: Insert pass_vtable_verify.
4099         * aclocal.m4: Reorder includes.
4100         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4101         -fvtv-counts options.
4102         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4103         as appropriate, if -fvtable-verify=... is used.
4104         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4105         -fvtable-verify=... is used.
4106         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4107         (vtable-verify.o): Add new build rule.
4108         (GTFILES): Add vtable-verify.c to list.
4109         * common.opt (fvtable-verify=): New flag.
4110         (vtv_priority): Values for fvtable-verify= flag.
4111         (fvtv-counts): New flag.
4112         (fvtv-debug): New flag.
4113         * tree.h (save_vtable_map_decl): New extern function decl.
4115 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4117         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4118         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4119         (pass_data_rl78_devirt): ...new pass_data instance and...
4120         (make_pass_rl78_devirt): ...new function.
4121         (rl78_asm_file_start): Port pass registration to new C++ API.
4123 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4125         * coretypes.h (rtl_opt_pass): Add.
4126         (gcc::context): Add.
4127         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4128         (epiphany_init): Port to new C++ pass API.
4129         (epiphany_optimize_mode_switching): Likewise.
4130         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4131         (pass_manager::get_pass_mode_switching): New.
4132         (pass_manager::get_pass_peephole2): New.
4133         * mode-switching.c (pass_mode_switching): Add clone method.
4134         * recog.c (pass_peephole2): Add clone method.
4135         (pass_split_all_insns): Add clone method.
4137 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
4139         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4140         (mips_option_override): ...here, porting to new C++ API for passes.
4142 2013-08-06  Jan Hubicka  <jh@suse.cz>
4144         * cgraph.c (cgraph_get_body): New function based on lto.c
4145         implementation.
4146         * cgraph.h (cgraph_get_body): Declare.
4147         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4148         LTO paths.
4149         * cgraphunit.c (expand_function): Get body prior expanding.
4150         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4151         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4152         really need.
4153         * passes.c (do_per_function_toporder): Get body.
4154         * tree-inline.c (expand_call_inline): Get body prior inlining it.
4155         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4157 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4159         PR fortran/57987
4160         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4161         is not re-finalized.  Rename second parameter to no_collect.
4163 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4165         PR middle-end/58041
4166         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4167         MEM_REF has proper alignment information.
4169 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
4171         PR other/12081
4172         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4173         class insn_gen_fn.
4174         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4175         rtx (*) (rtx, ...) with insn_gen_fn.
4176         * genoutput.c (output_insn_data): Cast gen_? function pointers to
4177         insn_gen_fn::stored_funcptr.  Add initializer braces.
4179 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4181         Rewrite how instances of passes are cloned to remove assumptions
4182         about their sizes (thus allowing pass subclasses to have
4183         additional data fields, albeit non-GC-managed ones at this point).
4185         * passes.c (make_pass_instance): Now that passes have clone
4186         methods, rewrite this function to eliminate XNEW and memcpy
4187         calls that used hardcoded sizes.  Since this function no longer
4188         creates pass instances, rename it to...
4189         (add_pass_instance): ...this.  Document the old way that passes were
4190         numbered and flagged, and rework this function to continue using it.
4191         (next_pass_1): Add an initial_pass argument for use by
4192         add_pass_instance.
4193         (position_pass): When adding multiple instances of a pass, use
4194         the pass's clone method, rather than relying on the XNEW/memcpy
4195         within the former make_pass_instance (now add_pass_instance).
4196         (pass_manager::pass_manager): When invoking next_pass_1, also supply
4197         the initial instance of the current pass within the pass manager.
4199 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4201         This is the automated part of the conversion of passes from C
4202         structs to C++ classes.
4204         Patch autogenerated by refactor_passes.py from
4205         https://github.com/davidmalcolm/gcc-refactoring-scripts
4206         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4208         * asan.c (pass_asan): Convert from a global struct to a subclass of
4209         gimple_opt_pass along with...
4210         (pass_data_asan): ...new pass_data instance and...
4211         (make_pass_asan): ...new function.
4212         (pass_asan_O0): Convert from a global struct to a subclass of
4213         gimple_opt_pass along with...
4214         (pass_data_asan_O0): ...new pass_data instance and...
4215         (make_pass_asan_O0): ...new function.
4216         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4217         subclass of rtl_opt_pass along with...
4218         (pass_data_inc_dec): ...new pass_data instance and...
4219         (make_pass_inc_dec): ...new function.
4220         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4221         a subclass of rtl_opt_pass along with...
4222         (pass_data_reorder_blocks): ...new pass_data instance and...
4223         (make_pass_reorder_blocks): ...new function.
4224         (pass_duplicate_computed_gotos): Convert from a global struct to a
4225         subclass of rtl_opt_pass along with...
4226         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4227         (make_pass_duplicate_computed_gotos): ...new function.
4228         (pass_partition_blocks): Convert from a global struct to a subclass of
4229         rtl_opt_pass along with...
4230         (pass_data_partition_blocks): ...new pass_data instance and...
4231         (make_pass_partition_blocks): ...new function.
4232         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
4233         struct to a subclass of rtl_opt_pass along with...
4234         (pass_data_branch_target_load_optimize1): ...new pass_data instance
4235         and...
4236         (make_pass_branch_target_load_optimize1): ...new function.
4237         (pass_branch_target_load_optimize2): Convert from a global struct to a
4238         subclass of rtl_opt_pass along with...
4239         (pass_data_branch_target_load_optimize2): ...new pass_data instance
4240         and...
4241         (make_pass_branch_target_load_optimize2): ...new function.
4242         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
4243         of rtl_opt_pass along with...
4244         (pass_data_jump): ...new pass_data instance and...
4245         (make_pass_jump): ...new function.
4246         (pass_jump2): Convert from a global struct to a subclass of
4247         rtl_opt_pass along with...
4248         (pass_data_jump2): ...new pass_data instance and...
4249         (make_pass_jump2): ...new function.
4250         * cfgexpand.c (pass_expand): Convert from a global struct to a
4251         subclass of rtl_opt_pass along with...
4252         (pass_data_expand): ...new pass_data instance and...
4253         (make_pass_expand): ...new function.
4254         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
4255         of rtl_opt_pass along with...
4256         (pass_data_free_cfg): ...new pass_data instance and...
4257         (make_pass_free_cfg): ...new function.
4258         (pass_into_cfg_layout_mode): Convert from a global struct to a
4259         subclass of rtl_opt_pass along with...
4260         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
4261         (make_pass_into_cfg_layout_mode): ...new function.
4262         (pass_outof_cfg_layout_mode): Convert from a global struct to a
4263         subclass of rtl_opt_pass along with...
4264         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
4265         (make_pass_outof_cfg_layout_mode): ...new function.
4266         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
4267         struct to a subclass of gimple_opt_pass along with...
4268         (pass_data_build_cgraph_edges): ...new pass_data instance and...
4269         (make_pass_build_cgraph_edges): ...new function.
4270         (pass_rebuild_cgraph_edges): Convert from a global struct to a
4271         subclass of gimple_opt_pass along with...
4272         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
4273         (make_pass_rebuild_cgraph_edges): ...new function.
4274         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
4275         subclass of gimple_opt_pass along with...
4276         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
4277         and...
4278         (make_pass_remove_cgraph_callee_edges): ...new function.
4279         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
4280         struct to a subclass of rtl_opt_pass along with...
4281         (pass_data_stack_adjustments): ...new pass_data instance and...
4282         (make_pass_stack_adjustments): ...new function.
4283         * combine.c (pass_combine): Convert from a global struct to a subclass
4284         of rtl_opt_pass along with...
4285         (pass_data_combine): ...new pass_data instance and...
4286         (make_pass_combine): ...new function.
4287         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
4288         global struct to a subclass of rtl_opt_pass along with...
4289         (pass_data_compare_elim_after_reload): ...new pass_data instance
4290         and...
4291         (make_pass_compare_elim_after_reload): ...new function.
4292         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
4293         of rtl_opt_pass along with...
4294         (pass_data_rtl_cprop): ...new pass_data instance and...
4295         (make_pass_rtl_cprop): ...new function.
4296         * cse.c (pass_cse): Convert from a global struct to a subclass of
4297         rtl_opt_pass along with...
4298         (pass_data_cse): ...new pass_data instance and...
4299         (make_pass_cse): ...new function.
4300         (pass_cse2): Convert from a global struct to a subclass of
4301         rtl_opt_pass along with...
4302         (pass_data_cse2): ...new pass_data instance and...
4303         (make_pass_cse2): ...new function.
4304         (pass_cse_after_global_opts): Convert from a global struct to a
4305         subclass of rtl_opt_pass along with...
4306         (pass_data_cse_after_global_opts): ...new pass_data instance and...
4307         (make_pass_cse_after_global_opts): ...new function.
4308         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
4309         of rtl_opt_pass along with...
4310         (pass_data_ud_rtl_dce): ...new pass_data instance and...
4311         (make_pass_ud_rtl_dce): ...new function.
4312         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
4313         rtl_opt_pass along with...
4314         (pass_data_fast_rtl_dce): ...new pass_data instance and...
4315         (make_pass_fast_rtl_dce): ...new function.
4316         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
4317         a subclass of rtl_opt_pass along with...
4318         (pass_data_df_initialize_opt): ...new pass_data instance and...
4319         (make_pass_df_initialize_opt): ...new function.
4320         (pass_df_initialize_no_opt): Convert from a global struct to a
4321         subclass of rtl_opt_pass along with...
4322         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
4323         (make_pass_df_initialize_no_opt): ...new function.
4324         (pass_df_finish): Convert from a global struct to a subclass of
4325         rtl_opt_pass along with...
4326         (pass_data_df_finish): ...new pass_data instance and...
4327         (make_pass_df_finish): ...new function.
4328         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
4329         rtl_opt_pass along with...
4330         (pass_data_rtl_dse1): ...new pass_data instance and...
4331         (make_pass_rtl_dse1): ...new function.
4332         (pass_rtl_dse2): Convert from a global struct to a subclass of
4333         rtl_opt_pass along with...
4334         (pass_data_rtl_dse2): ...new pass_data instance and...
4335         (make_pass_rtl_dse2): ...new function.
4336         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
4337         subclass of rtl_opt_pass along with...
4338         (pass_data_dwarf2_frame): ...new pass_data instance and...
4339         (make_pass_dwarf2_frame): ...new function.
4340         * except.c (pass_set_nothrow_function_flags): Convert from a global
4341         struct to a subclass of rtl_opt_pass along with...
4342         (pass_data_set_nothrow_function_flags): ...new pass_data instance
4343         and...
4344         (make_pass_set_nothrow_function_flags): ...new function.
4345         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
4346         subclass of rtl_opt_pass along with...
4347         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
4348         and...
4349         (make_pass_convert_to_eh_region_ranges): ...new function.
4350         * final.c (pass_compute_alignments): Convert from a global struct to a
4351         subclass of rtl_opt_pass along with...
4352         (pass_data_compute_alignments): ...new pass_data instance and...
4353         (make_pass_compute_alignments): ...new function.
4354         (pass_final): Convert from a global struct to a subclass of
4355         rtl_opt_pass along with...
4356         (pass_data_final): ...new pass_data instance and...
4357         (make_pass_final): ...new function.
4358         (pass_shorten_branches): Convert from a global struct to a subclass of
4359         rtl_opt_pass along with...
4360         (pass_data_shorten_branches): ...new pass_data instance and...
4361         (make_pass_shorten_branches): ...new function.
4362         (pass_clean_state): Convert from a global struct to a subclass of
4363         rtl_opt_pass along with...
4364         (pass_data_clean_state): ...new pass_data instance and...
4365         (make_pass_clean_state): ...new function.
4366         * function.c (pass_instantiate_virtual_regs): Convert from a global
4367         struct to a subclass of rtl_opt_pass along with...
4368         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
4369         (make_pass_instantiate_virtual_regs): ...new function.
4370         (pass_leaf_regs): Convert from a global struct to a subclass of
4371         rtl_opt_pass along with...
4372         (pass_data_leaf_regs): ...new pass_data instance and...
4373         (make_pass_leaf_regs): ...new function.
4374         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
4375         subclass of rtl_opt_pass along with...
4376         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
4377         and...
4378         (make_pass_thread_prologue_and_epilogue): ...new function.
4379         (pass_match_asm_constraints): Convert from a global struct to a
4380         subclass of rtl_opt_pass along with...
4381         (pass_data_match_asm_constraints): ...new pass_data instance and...
4382         (make_pass_match_asm_constraints): ...new function.
4383         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
4384         subclass of rtl_opt_pass along with...
4385         (pass_data_rtl_fwprop): ...new pass_data instance and...
4386         (make_pass_rtl_fwprop): ...new function.
4387         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
4388         rtl_opt_pass along with...
4389         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
4390         (make_pass_rtl_fwprop_addr): ...new function.
4391         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
4392         rtl_opt_pass along with...
4393         (pass_data_rtl_pre): ...new pass_data instance and...
4394         (make_pass_rtl_pre): ...new function.
4395         (pass_rtl_hoist): Convert from a global struct to a subclass of
4396         rtl_opt_pass along with...
4397         (pass_data_rtl_hoist): ...new pass_data instance and...
4398         (make_pass_rtl_hoist): ...new function.
4399         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
4400         subclass of gimple_opt_pass along with...
4401         (pass_data_lower_cf): ...new pass_data instance and...
4402         (make_pass_lower_cf): ...new function.
4403         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
4404         from a global struct to a subclass of gimple_opt_pass along with...
4405         (pass_data_strength_reduction): ...new pass_data instance and...
4406         (make_pass_strength_reduction): ...new function.
4407         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
4408         of rtl_opt_pass along with...
4409         (pass_data_rtl_ifcvt): ...new pass_data instance and...
4410         (make_pass_rtl_ifcvt): ...new function.
4411         (pass_if_after_combine): Convert from a global struct to a subclass of
4412         rtl_opt_pass along with...
4413         (pass_data_if_after_combine): ...new pass_data instance and...
4414         (make_pass_if_after_combine): ...new function.
4415         (pass_if_after_reload): Convert from a global struct to a subclass of
4416         rtl_opt_pass along with...
4417         (pass_data_if_after_reload): ...new pass_data instance and...
4418         (make_pass_if_after_reload): ...new function.
4419         * init-regs.c (pass_initialize_regs): Convert from a global struct to
4420         a subclass of rtl_opt_pass along with...
4421         (pass_data_initialize_regs): ...new pass_data instance and...
4422         (make_pass_initialize_regs): ...new function.
4423         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
4424         of ipa_opt_pass_d along with...
4425         (pass_data_ipa_cp): ...new pass_data instance and...
4426         (make_pass_ipa_cp): ...new function.
4427         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
4428         global struct to a subclass of gimple_opt_pass along with...
4429         (pass_data_inline_parameters): ...new pass_data instance and...
4430         (make_pass_inline_parameters): ...new function.
4431         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
4432         subclass of gimple_opt_pass along with...
4433         (pass_data_early_inline): ...new pass_data instance and...
4434         (make_pass_early_inline): ...new function.
4435         (pass_ipa_inline): Convert from a global struct to a subclass of
4436         ipa_opt_pass_d along with...
4437         (pass_data_ipa_inline): ...new pass_data instance and...
4438         (make_pass_ipa_inline): ...new function.
4439         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
4440         struct to a subclass of gimple_opt_pass along with...
4441         (pass_data_local_pure_const): ...new pass_data instance and...
4442         (make_pass_local_pure_const): ...new function.
4443         (pass_ipa_pure_const): Convert from a global struct to a subclass of
4444         ipa_opt_pass_d along with...
4445         (pass_data_ipa_pure_const): ...new pass_data instance and...
4446         (make_pass_ipa_pure_const): ...new function.
4447         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
4448         to a subclass of ipa_opt_pass_d along with...
4449         (pass_data_ipa_reference): ...new pass_data instance and...
4450         (make_pass_ipa_reference): ...new function.
4451         * ipa-split.c (pass_split_functions): Convert from a global struct to
4452         a subclass of gimple_opt_pass along with...
4453         (pass_data_split_functions): ...new pass_data instance and...
4454         (make_pass_split_functions): ...new function.
4455         (pass_feedback_split_functions): Convert from a global struct to a
4456         subclass of gimple_opt_pass along with...
4457         (pass_data_feedback_split_functions): ...new pass_data instance and...
4458         (make_pass_feedback_split_functions): ...new function.
4459         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
4460         global struct to a subclass of simple_ipa_opt_pass along with...
4461         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
4462         instance and...
4463         (make_pass_ipa_function_and_variable_visibility): ...new function.
4464         (pass_ipa_free_inline_summary): Convert from a global struct to a
4465         subclass of simple_ipa_opt_pass along with...
4466         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
4467         (make_pass_ipa_free_inline_summary): ...new function.
4468         (pass_ipa_whole_program_visibility): Convert from a global struct to a
4469         subclass of ipa_opt_pass_d along with...
4470         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
4471         and...
4472         (make_pass_ipa_whole_program_visibility): ...new function.
4473         (pass_ipa_profile): Convert from a global struct to a subclass of
4474         ipa_opt_pass_d along with...
4475         (pass_data_ipa_profile): ...new pass_data instance and...
4476         (make_pass_ipa_profile): ...new function.
4477         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
4478         ipa_opt_pass_d along with...
4479         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
4480         (make_pass_ipa_cdtor_merge): ...new function.
4481         * ira.c (pass_ira): Convert from a global struct to a subclass of
4482         rtl_opt_pass along with...
4483         (pass_data_ira): ...new pass_data instance and...
4484         (make_pass_ira): ...new function.
4485         (pass_reload): Convert from a global struct to a subclass of
4486         rtl_opt_pass along with...
4487         (pass_data_reload): ...new pass_data instance and...
4488         (make_pass_reload): ...new function.
4489         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
4490         subclass of rtl_opt_pass along with...
4491         (pass_data_cleanup_barriers): ...new pass_data instance and...
4492         (make_pass_cleanup_barriers): ...new function.
4493         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
4494         of rtl_opt_pass along with...
4495         (pass_data_loop2): ...new pass_data instance and...
4496         (make_pass_loop2): ...new function.
4497         (pass_rtl_loop_init): Convert from a global struct to a subclass of
4498         rtl_opt_pass along with...
4499         (pass_data_rtl_loop_init): ...new pass_data instance and...
4500         (make_pass_rtl_loop_init): ...new function.
4501         (pass_rtl_loop_done): Convert from a global struct to a subclass of
4502         rtl_opt_pass along with...
4503         (pass_data_rtl_loop_done): ...new pass_data instance and...
4504         (make_pass_rtl_loop_done): ...new function.
4505         (pass_rtl_move_loop_invariants): Convert from a global struct to a
4506         subclass of rtl_opt_pass along with...
4507         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
4508         (make_pass_rtl_move_loop_invariants): ...new function.
4509         (pass_rtl_unswitch): Convert from a global struct to a subclass of
4510         rtl_opt_pass along with...
4511         (pass_data_rtl_unswitch): ...new pass_data instance and...
4512         (make_pass_rtl_unswitch): ...new function.
4513         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
4514         subclass of rtl_opt_pass along with...
4515         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
4516         and...
4517         (make_pass_rtl_unroll_and_peel_loops): ...new function.
4518         (pass_rtl_doloop): Convert from a global struct to a subclass of
4519         rtl_opt_pass along with...
4520         (pass_data_rtl_doloop): ...new pass_data instance and...
4521         (make_pass_rtl_doloop): ...new function.
4522         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
4523         a subclass of rtl_opt_pass along with...
4524         (pass_data_lower_subreg): ...new pass_data instance and...
4525         (make_pass_lower_subreg): ...new function.
4526         (pass_lower_subreg2): Convert from a global struct to a subclass of
4527         rtl_opt_pass along with...
4528         (pass_data_lower_subreg2): ...new pass_data instance and...
4529         (make_pass_lower_subreg2): ...new function.
4530         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
4531         struct to a subclass of ipa_opt_pass_d along with...
4532         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
4533         (make_pass_ipa_lto_gimple_out): ...new function.
4534         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
4535         of ipa_opt_pass_d along with...
4536         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
4537         (make_pass_ipa_lto_finish_out): ...new function.
4538         * mode-switching.c (pass_mode_switching): Convert from a global struct
4539         to a subclass of rtl_opt_pass along with...
4540         (pass_data_mode_switching): ...new pass_data instance and...
4541         (make_pass_mode_switching): ...new function.
4542         * modulo-sched.c (pass_sms): Convert from a global struct to a
4543         subclass of rtl_opt_pass along with...
4544         (pass_data_sms): ...new pass_data instance and...
4545         (make_pass_sms): ...new function.
4546         * omp-low.c (pass_expand_omp): Convert from a global struct to a
4547         subclass of gimple_opt_pass along with...
4548         (pass_data_expand_omp): ...new pass_data instance and...
4549         (make_pass_expand_omp): ...new function.
4550         (pass_lower_omp): Convert from a global struct to a subclass of
4551         gimple_opt_pass along with...
4552         (pass_data_lower_omp): ...new pass_data instance and...
4553         (make_pass_lower_omp): ...new function.
4554         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
4555         of gimple_opt_pass along with...
4556         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
4557         (make_pass_diagnose_omp_blocks): ...new function.
4558         * passes.c (pass_early_local_passes): Convert from a global struct to
4559         a subclass of simple_ipa_opt_pass along with...
4560         (pass_data_early_local_passes): ...new pass_data instance and...
4561         (make_pass_early_local_passes): ...new function.
4562         (pass_all_early_optimizations): Convert from a global struct to a
4563         subclass of gimple_opt_pass along with...
4564         (pass_data_all_early_optimizations): ...new pass_data instance and...
4565         (make_pass_all_early_optimizations): ...new function.
4566         (pass_all_optimizations): Convert from a global struct to a subclass
4567         of gimple_opt_pass along with...
4568         (pass_data_all_optimizations): ...new pass_data instance and...
4569         (make_pass_all_optimizations): ...new function.
4570         (pass_all_optimizations_g): Convert from a global struct to a subclass
4571         of gimple_opt_pass along with...
4572         (pass_data_all_optimizations_g): ...new pass_data instance and...
4573         (make_pass_all_optimizations_g): ...new function.
4574         (pass_rest_of_compilation): Convert from a global struct to a subclass
4575         of rtl_opt_pass along with...
4576         (pass_data_rest_of_compilation): ...new pass_data instance and...
4577         (make_pass_rest_of_compilation): ...new function.
4578         (pass_postreload): Convert from a global struct to a subclass of
4579         rtl_opt_pass along with...
4580         (pass_data_postreload): ...new pass_data instance and...
4581         (make_pass_postreload): ...new function.
4582         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
4583         subclass of rtl_opt_pass along with...
4584         (pass_data_gcse2): ...new pass_data instance and...
4585         (make_pass_gcse2): ...new function.
4586         * postreload.c (pass_postreload_cse): Convert from a global struct to
4587         a subclass of rtl_opt_pass along with...
4588         (pass_data_postreload_cse): ...new pass_data instance and...
4589         (make_pass_postreload_cse): ...new function.
4590         * predict.c (pass_profile): Convert from a global struct to a subclass
4591         of gimple_opt_pass along with...
4592         (pass_data_profile): ...new pass_data instance and...
4593         (make_pass_profile): ...new function.
4594         (pass_strip_predict_hints): Convert from a global struct to a subclass
4595         of gimple_opt_pass along with...
4596         (pass_data_strip_predict_hints): ...new pass_data instance and...
4597         (make_pass_strip_predict_hints): ...new function.
4598         * recog.c (pass_peephole2): Convert from a global struct to a subclass
4599         of rtl_opt_pass along with...
4600         (pass_data_peephole2): ...new pass_data instance and...
4601         (make_pass_peephole2): ...new function.
4602         (pass_split_all_insns): Convert from a global struct to a subclass of
4603         rtl_opt_pass along with...
4604         (pass_data_split_all_insns): ...new pass_data instance and...
4605         (make_pass_split_all_insns): ...new function.
4606         (pass_split_after_reload): Convert from a global struct to a subclass
4607         of rtl_opt_pass along with...
4608         (pass_data_split_after_reload): ...new pass_data instance and...
4609         (make_pass_split_after_reload): ...new function.
4610         (pass_split_before_regstack): Convert from a global struct to a
4611         subclass of rtl_opt_pass along with...
4612         (pass_data_split_before_regstack): ...new pass_data instance and...
4613         (make_pass_split_before_regstack): ...new function.
4614         (pass_split_before_sched2): Convert from a global struct to a subclass
4615         of rtl_opt_pass along with...
4616         (pass_data_split_before_sched2): ...new pass_data instance and...
4617         (make_pass_split_before_sched2): ...new function.
4618         (pass_split_for_shorten_branches): Convert from a global struct to a
4619         subclass of rtl_opt_pass along with...
4620         (pass_data_split_for_shorten_branches): ...new pass_data instance
4621         and...
4622         (make_pass_split_for_shorten_branches): ...new function.
4623         * ree.c (pass_ree): Convert from a global struct to a subclass of
4624         rtl_opt_pass along with...
4625         (pass_data_ree): ...new pass_data instance and...
4626         (make_pass_ree): ...new function.
4627         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
4628         subclass of rtl_opt_pass along with...
4629         (pass_data_stack_regs): ...new pass_data instance and...
4630         (make_pass_stack_regs): ...new function.
4631         (pass_stack_regs_run): Convert from a global struct to a subclass of
4632         rtl_opt_pass along with...
4633         (pass_data_stack_regs_run): ...new pass_data instance and...
4634         (make_pass_stack_regs_run): ...new function.
4635         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
4636         subclass of rtl_opt_pass along with...
4637         (pass_data_cprop_hardreg): ...new pass_data instance and...
4638         (make_pass_cprop_hardreg): ...new function.
4639         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
4640         subclass of rtl_opt_pass along with...
4641         (pass_data_reginfo_init): ...new pass_data instance and...
4642         (make_pass_reginfo_init): ...new function.
4643         * regmove.c (pass_regmove): Convert from a global struct to a subclass
4644         of rtl_opt_pass along with...
4645         (pass_data_regmove): ...new pass_data instance and...
4646         (make_pass_regmove): ...new function.
4647         * regrename.c (pass_regrename): Convert from a global struct to a
4648         subclass of rtl_opt_pass along with...
4649         (pass_data_regrename): ...new pass_data instance and...
4650         (make_pass_regrename): ...new function.
4651         * reorg.c (pass_delay_slots): Convert from a global struct to a
4652         subclass of rtl_opt_pass along with...
4653         (pass_data_delay_slots): ...new pass_data instance and...
4654         (make_pass_delay_slots): ...new function.
4655         (pass_machine_reorg): Convert from a global struct to a subclass of
4656         rtl_opt_pass along with...
4657         (pass_data_machine_reorg): ...new pass_data instance and...
4658         (make_pass_machine_reorg): ...new function.
4659         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
4660         of rtl_opt_pass along with...
4661         (pass_data_sched): ...new pass_data instance and...
4662         (make_pass_sched): ...new function.
4663         (pass_sched2): Convert from a global struct to a subclass of
4664         rtl_opt_pass along with...
4665         (pass_data_sched2): ...new pass_data instance and...
4666         (make_pass_sched2): ...new function.
4667         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
4668         to a subclass of rtl_opt_pass along with...
4669         (pass_data_stack_ptr_mod): ...new pass_data instance and...
4670         (make_pass_stack_ptr_mod): ...new function.
4671         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
4672         to a subclass of rtl_opt_pass along with...
4673         (pass_data_rtl_store_motion): ...new pass_data instance and...
4674         (make_pass_rtl_store_motion): ...new function.
4675         * tracer.c (pass_tracer): Convert from a global struct to a subclass
4676         of gimple_opt_pass along with...
4677         (pass_data_tracer): ...new pass_data instance and...
4678         (make_pass_tracer): ...new function.
4679         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
4680         to a subclass of gimple_opt_pass along with...
4681         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
4682         (make_pass_diagnose_tm_blocks): ...new function.
4683         (pass_lower_tm): Convert from a global struct to a subclass of
4684         gimple_opt_pass along with...
4685         (pass_data_lower_tm): ...new pass_data instance and...
4686         (make_pass_lower_tm): ...new function.
4687         (pass_tm_init): Convert from a global struct to a subclass of
4688         gimple_opt_pass along with...
4689         (pass_data_tm_init): ...new pass_data instance and...
4690         (make_pass_tm_init): ...new function.
4691         (pass_tm_mark): Convert from a global struct to a subclass of
4692         gimple_opt_pass along with...
4693         (pass_data_tm_mark): ...new pass_data instance and...
4694         (make_pass_tm_mark): ...new function.
4695         (pass_tm_edges): Convert from a global struct to a subclass of
4696         gimple_opt_pass along with...
4697         (pass_data_tm_edges): ...new pass_data instance and...
4698         (make_pass_tm_edges): ...new function.
4699         (pass_tm_memopt): Convert from a global struct to a subclass of
4700         gimple_opt_pass along with...
4701         (pass_data_tm_memopt): ...new pass_data instance and...
4702         (make_pass_tm_memopt): ...new function.
4703         (pass_ipa_tm): Convert from a global struct to a subclass of
4704         simple_ipa_opt_pass along with...
4705         (pass_data_ipa_tm): ...new pass_data instance and...
4706         (make_pass_ipa_tm): ...new function.
4707         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
4708         subclass of gimple_opt_pass along with...
4709         (pass_data_call_cdce): ...new pass_data instance and...
4710         (make_pass_call_cdce): ...new function.
4711         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
4712         subclass of gimple_opt_pass along with...
4713         (pass_data_build_cfg): ...new pass_data instance and...
4714         (make_pass_build_cfg): ...new function.
4715         (pass_split_crit_edges): Convert from a global struct to a subclass of
4716         gimple_opt_pass along with...
4717         (pass_data_split_crit_edges): ...new pass_data instance and...
4718         (make_pass_split_crit_edges): ...new function.
4719         (pass_warn_function_return): Convert from a global struct to a
4720         subclass of gimple_opt_pass along with...
4721         (pass_data_warn_function_return): ...new pass_data instance and...
4722         (make_pass_warn_function_return): ...new function.
4723         (pass_warn_function_noreturn): Convert from a global struct to a
4724         subclass of gimple_opt_pass along with...
4725         (pass_data_warn_function_noreturn): ...new pass_data instance and...
4726         (make_pass_warn_function_noreturn): ...new function.
4727         (pass_warn_unused_result): Convert from a global struct to a subclass
4728         of gimple_opt_pass along with...
4729         (pass_data_warn_unused_result): ...new pass_data instance and...
4730         (make_pass_warn_unused_result): ...new function.
4731         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
4732         a subclass of gimple_opt_pass along with...
4733         (pass_data_merge_phi): ...new pass_data instance and...
4734         (make_pass_merge_phi): ...new function.
4735         * tree-complex.c (pass_lower_complex): Convert from a global struct to
4736         a subclass of gimple_opt_pass along with...
4737         (pass_data_lower_complex): ...new pass_data instance and...
4738         (make_pass_lower_complex): ...new function.
4739         (pass_lower_complex_O0): Convert from a global struct to a subclass of
4740         gimple_opt_pass along with...
4741         (pass_data_lower_complex_O0): ...new pass_data instance and...
4742         (make_pass_lower_complex_O0): ...new function.
4743         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
4744         subclass of gimple_opt_pass along with...
4745         (pass_data_lower_eh): ...new pass_data instance and...
4746         (make_pass_lower_eh): ...new function.
4747         (pass_refactor_eh): Convert from a global struct to a subclass of
4748         gimple_opt_pass along with...
4749         (pass_data_refactor_eh): ...new pass_data instance and...
4750         (make_pass_refactor_eh): ...new function.
4751         (pass_lower_resx): Convert from a global struct to a subclass of
4752         gimple_opt_pass along with...
4753         (pass_data_lower_resx): ...new pass_data instance and...
4754         (make_pass_lower_resx): ...new function.
4755         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
4756         of gimple_opt_pass along with...
4757         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
4758         (make_pass_lower_eh_dispatch): ...new function.
4759         (pass_cleanup_eh): Convert from a global struct to a subclass of
4760         gimple_opt_pass along with...
4761         (pass_data_cleanup_eh): ...new pass_data instance and...
4762         (make_pass_cleanup_eh): ...new function.
4763         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
4764         to a subclass of simple_ipa_opt_pass along with...
4765         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
4766         (make_pass_ipa_lower_emutls): ...new function.
4767         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
4768         a subclass of gimple_opt_pass along with...
4769         (pass_data_if_conversion): ...new pass_data instance and...
4770         (make_pass_if_conversion): ...new function.
4771         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
4772         subclass of gimple_opt_pass along with...
4773         (pass_data_build_ssa): ...new pass_data instance and...
4774         (make_pass_build_ssa): ...new function.
4775         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
4776         global struct to a subclass of gimple_opt_pass along with...
4777         (pass_data_loop_distribution): ...new pass_data instance and...
4778         (make_pass_loop_distribution): ...new function.
4779         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
4780         subclass of gimple_opt_pass along with...
4781         (pass_data_mudflap_1): ...new pass_data instance and...
4782         (make_pass_mudflap_1): ...new function.
4783         (pass_mudflap_2): Convert from a global struct to a subclass of
4784         gimple_opt_pass along with...
4785         (pass_data_mudflap_2): ...new pass_data instance and...
4786         (make_pass_mudflap_2): ...new function.
4787         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
4788         subclass of gimple_opt_pass along with...
4789         (pass_data_mudflap_1): ...new pass_data instance and...
4790         (make_pass_mudflap_1): ...new function.
4791         (pass_mudflap_2): Convert from a global struct to a subclass of
4792         gimple_opt_pass along with...
4793         (pass_data_mudflap_2): ...new pass_data instance and...
4794         (make_pass_mudflap_2): ...new function.
4795         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
4796         gimple_opt_pass along with...
4797         (pass_data_nrv): ...new pass_data instance and...
4798         (make_pass_nrv): ...new function.
4799         (pass_return_slot): Convert from a global struct to a subclass of
4800         gimple_opt_pass along with...
4801         (pass_data_return_slot): ...new pass_data instance and...
4802         (make_pass_return_slot): ...new function.
4803         * tree-object-size.c (pass_object_sizes): Convert from a global struct
4804         to a subclass of gimple_opt_pass along with...
4805         (pass_data_object_sizes): ...new pass_data instance and...
4806         (make_pass_object_sizes): ...new function.
4807         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
4808         global struct to a subclass of gimple_opt_pass along with...
4809         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
4810         and...
4811         (make_pass_cleanup_cfg_post_optimizing): ...new function.
4812         (pass_fixup_cfg): Convert from a global struct to a subclass of
4813         gimple_opt_pass along with...
4814         (pass_data_fixup_cfg): ...new pass_data instance and...
4815         (make_pass_fixup_cfg): ...new function.
4816         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
4817         (make_pass_mudflap_1): ...new function.
4818         (pass_mudflap_2): Replace declaration with that of...
4819         (make_pass_mudflap_2): ...new function.
4820         (pass_asan): Replace declaration with that of...
4821         (make_pass_asan): ...new function.
4822         (pass_asan_O0): Replace declaration with that of...
4823         (make_pass_asan_O0): ...new function.
4824         (pass_tsan): Replace declaration with that of...
4825         (make_pass_tsan): ...new function.
4826         (pass_tsan_O0): Replace declaration with that of...
4827         (make_pass_tsan_O0): ...new function.
4828         (pass_lower_cf): Replace declaration with that of...
4829         (make_pass_lower_cf): ...new function.
4830         (pass_refactor_eh): Replace declaration with that of...
4831         (make_pass_refactor_eh): ...new function.
4832         (pass_lower_eh): Replace declaration with that of...
4833         (make_pass_lower_eh): ...new function.
4834         (pass_lower_eh_dispatch): Replace declaration with that of...
4835         (make_pass_lower_eh_dispatch): ...new function.
4836         (pass_lower_resx): Replace declaration with that of...
4837         (make_pass_lower_resx): ...new function.
4838         (pass_build_cfg): Replace declaration with that of...
4839         (make_pass_build_cfg): ...new function.
4840         (pass_early_tree_profile): Replace declaration with that of...
4841         (make_pass_early_tree_profile): ...new function.
4842         (pass_cleanup_eh): Replace declaration with that of...
4843         (make_pass_cleanup_eh): ...new function.
4844         (pass_sra): Replace declaration with that of...
4845         (make_pass_sra): ...new function.
4846         (pass_sra_early): Replace declaration with that of...
4847         (make_pass_sra_early): ...new function.
4848         (pass_early_ipa_sra): Replace declaration with that of...
4849         (make_pass_early_ipa_sra): ...new function.
4850         (pass_tail_recursion): Replace declaration with that of...
4851         (make_pass_tail_recursion): ...new function.
4852         (pass_tail_calls): Replace declaration with that of...
4853         (make_pass_tail_calls): ...new function.
4854         (pass_tree_loop): Replace declaration with that of...
4855         (make_pass_tree_loop): ...new function.
4856         (pass_tree_loop_init): Replace declaration with that of...
4857         (make_pass_tree_loop_init): ...new function.
4858         (pass_lim): Replace declaration with that of...
4859         (make_pass_lim): ...new function.
4860         (pass_tree_unswitch): Replace declaration with that of...
4861         (make_pass_tree_unswitch): ...new function.
4862         (pass_predcom): Replace declaration with that of...
4863         (make_pass_predcom): ...new function.
4864         (pass_iv_canon): Replace declaration with that of...
4865         (make_pass_iv_canon): ...new function.
4866         (pass_scev_cprop): Replace declaration with that of...
4867         (make_pass_scev_cprop): ...new function.
4868         (pass_empty_loop): Replace declaration with that of...
4869         (make_pass_empty_loop): ...new function.
4870         (pass_record_bounds): Replace declaration with that of...
4871         (make_pass_record_bounds): ...new function.
4872         (pass_graphite): Replace declaration with that of...
4873         (make_pass_graphite): ...new function.
4874         (pass_graphite_transforms): Replace declaration with that of...
4875         (make_pass_graphite_transforms): ...new function.
4876         (pass_if_conversion): Replace declaration with that of...
4877         (make_pass_if_conversion): ...new function.
4878         (pass_loop_distribution): Replace declaration with that of...
4879         (make_pass_loop_distribution): ...new function.
4880         (pass_vectorize): Replace declaration with that of...
4881         (make_pass_vectorize): ...new function.
4882         (pass_slp_vectorize): Replace declaration with that of...
4883         (make_pass_slp_vectorize): ...new function.
4884         (pass_complete_unroll): Replace declaration with that of...
4885         (make_pass_complete_unroll): ...new function.
4886         (pass_complete_unrolli): Replace declaration with that of...
4887         (make_pass_complete_unrolli): ...new function.
4888         (pass_parallelize_loops): Replace declaration with that of...
4889         (make_pass_parallelize_loops): ...new function.
4890         (pass_loop_prefetch): Replace declaration with that of...
4891         (make_pass_loop_prefetch): ...new function.
4892         (pass_iv_optimize): Replace declaration with that of...
4893         (make_pass_iv_optimize): ...new function.
4894         (pass_tree_loop_done): Replace declaration with that of...
4895         (make_pass_tree_loop_done): ...new function.
4896         (pass_ch): Replace declaration with that of...
4897         (make_pass_ch): ...new function.
4898         (pass_ccp): Replace declaration with that of...
4899         (make_pass_ccp): ...new function.
4900         (pass_phi_only_cprop): Replace declaration with that of...
4901         (make_pass_phi_only_cprop): ...new function.
4902         (pass_build_ssa): Replace declaration with that of...
4903         (make_pass_build_ssa): ...new function.
4904         (pass_build_alias): Replace declaration with that of...
4905         (make_pass_build_alias): ...new function.
4906         (pass_build_ealias): Replace declaration with that of...
4907         (make_pass_build_ealias): ...new function.
4908         (pass_dominator): Replace declaration with that of...
4909         (make_pass_dominator): ...new function.
4910         (pass_dce): Replace declaration with that of...
4911         (make_pass_dce): ...new function.
4912         (pass_dce_loop): Replace declaration with that of...
4913         (make_pass_dce_loop): ...new function.
4914         (pass_cd_dce): Replace declaration with that of...
4915         (make_pass_cd_dce): ...new function.
4916         (pass_call_cdce): Replace declaration with that of...
4917         (make_pass_call_cdce): ...new function.
4918         (pass_merge_phi): Replace declaration with that of...
4919         (make_pass_merge_phi): ...new function.
4920         (pass_split_crit_edges): Replace declaration with that of...
4921         (make_pass_split_crit_edges): ...new function.
4922         (pass_pre): Replace declaration with that of...
4923         (make_pass_pre): ...new function.
4924         (pass_profile): Replace declaration with that of...
4925         (make_pass_profile): ...new function.
4926         (pass_strip_predict_hints): Replace declaration with that of...
4927         (make_pass_strip_predict_hints): ...new function.
4928         (pass_lower_complex_O0): Replace declaration with that of...
4929         (make_pass_lower_complex_O0): ...new function.
4930         (pass_lower_complex): Replace declaration with that of...
4931         (make_pass_lower_complex): ...new function.
4932         (pass_lower_vector): Replace declaration with that of...
4933         (make_pass_lower_vector): ...new function.
4934         (pass_lower_vector_ssa): Replace declaration with that of...
4935         (make_pass_lower_vector_ssa): ...new function.
4936         (pass_lower_omp): Replace declaration with that of...
4937         (make_pass_lower_omp): ...new function.
4938         (pass_diagnose_omp_blocks): Replace declaration with that of...
4939         (make_pass_diagnose_omp_blocks): ...new function.
4940         (pass_expand_omp): Replace declaration with that of...
4941         (make_pass_expand_omp): ...new function.
4942         (pass_expand_omp_ssa): Replace declaration with that of...
4943         (make_pass_expand_omp_ssa): ...new function.
4944         (pass_object_sizes): Replace declaration with that of...
4945         (make_pass_object_sizes): ...new function.
4946         (pass_strlen): Replace declaration with that of...
4947         (make_pass_strlen): ...new function.
4948         (pass_fold_builtins): Replace declaration with that of...
4949         (make_pass_fold_builtins): ...new function.
4950         (pass_stdarg): Replace declaration with that of...
4951         (make_pass_stdarg): ...new function.
4952         (pass_early_warn_uninitialized): Replace declaration with that of...
4953         (make_pass_early_warn_uninitialized): ...new function.
4954         (pass_late_warn_uninitialized): Replace declaration with that of...
4955         (make_pass_late_warn_uninitialized): ...new function.
4956         (pass_cse_reciprocals): Replace declaration with that of...
4957         (make_pass_cse_reciprocals): ...new function.
4958         (pass_cse_sincos): Replace declaration with that of...
4959         (make_pass_cse_sincos): ...new function.
4960         (pass_optimize_bswap): Replace declaration with that of...
4961         (make_pass_optimize_bswap): ...new function.
4962         (pass_optimize_widening_mul): Replace declaration with that of...
4963         (make_pass_optimize_widening_mul): ...new function.
4964         (pass_warn_function_return): Replace declaration with that of...
4965         (make_pass_warn_function_return): ...new function.
4966         (pass_warn_function_noreturn): Replace declaration with that of...
4967         (make_pass_warn_function_noreturn): ...new function.
4968         (pass_cselim): Replace declaration with that of...
4969         (make_pass_cselim): ...new function.
4970         (pass_phiopt): Replace declaration with that of...
4971         (make_pass_phiopt): ...new function.
4972         (pass_forwprop): Replace declaration with that of...
4973         (make_pass_forwprop): ...new function.
4974         (pass_phiprop): Replace declaration with that of...
4975         (make_pass_phiprop): ...new function.
4976         (pass_tree_ifcombine): Replace declaration with that of...
4977         (make_pass_tree_ifcombine): ...new function.
4978         (pass_dse): Replace declaration with that of...
4979         (make_pass_dse): ...new function.
4980         (pass_nrv): Replace declaration with that of...
4981         (make_pass_nrv): ...new function.
4982         (pass_rename_ssa_copies): Replace declaration with that of...
4983         (make_pass_rename_ssa_copies): ...new function.
4984         (pass_sink_code): Replace declaration with that of...
4985         (make_pass_sink_code): ...new function.
4986         (pass_fre): Replace declaration with that of...
4987         (make_pass_fre): ...new function.
4988         (pass_check_data_deps): Replace declaration with that of...
4989         (make_pass_check_data_deps): ...new function.
4990         (pass_copy_prop): Replace declaration with that of...
4991         (make_pass_copy_prop): ...new function.
4992         (pass_vrp): Replace declaration with that of...
4993         (make_pass_vrp): ...new function.
4994         (pass_uncprop): Replace declaration with that of...
4995         (make_pass_uncprop): ...new function.
4996         (pass_return_slot): Replace declaration with that of...
4997         (make_pass_return_slot): ...new function.
4998         (pass_reassoc): Replace declaration with that of...
4999         (make_pass_reassoc): ...new function.
5000         (pass_rebuild_cgraph_edges): Replace declaration with that of...
5001         (make_pass_rebuild_cgraph_edges): ...new function.
5002         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
5003         (make_pass_remove_cgraph_callee_edges): ...new function.
5004         (pass_build_cgraph_edges): Replace declaration with that of...
5005         (make_pass_build_cgraph_edges): ...new function.
5006         (pass_local_pure_const): Replace declaration with that of...
5007         (make_pass_local_pure_const): ...new function.
5008         (pass_tracer): Replace declaration with that of...
5009         (make_pass_tracer): ...new function.
5010         (pass_warn_unused_result): Replace declaration with that of...
5011         (make_pass_warn_unused_result): ...new function.
5012         (pass_diagnose_tm_blocks): Replace declaration with that of...
5013         (make_pass_diagnose_tm_blocks): ...new function.
5014         (pass_lower_tm): Replace declaration with that of...
5015         (make_pass_lower_tm): ...new function.
5016         (pass_tm_init): Replace declaration with that of...
5017         (make_pass_tm_init): ...new function.
5018         (pass_tm_mark): Replace declaration with that of...
5019         (make_pass_tm_mark): ...new function.
5020         (pass_tm_memopt): Replace declaration with that of...
5021         (make_pass_tm_memopt): ...new function.
5022         (pass_tm_edges): Replace declaration with that of...
5023         (make_pass_tm_edges): ...new function.
5024         (pass_split_functions): Replace declaration with that of...
5025         (make_pass_split_functions): ...new function.
5026         (pass_feedback_split_functions): Replace declaration with that of...
5027         (make_pass_feedback_split_functions): ...new function.
5028         (pass_strength_reduction): Replace declaration with that of...
5029         (make_pass_strength_reduction): ...new function.
5030         (pass_ipa_lower_emutls): Replace declaration with that of...
5031         (make_pass_ipa_lower_emutls): ...new function.
5032         (pass_ipa_function_and_variable_visibility): Replace declaration with
5033         that of...
5034         (make_pass_ipa_function_and_variable_visibility): ...new function.
5035         (pass_ipa_tree_profile): Replace declaration with that of...
5036         (make_pass_ipa_tree_profile): ...new function.
5037         (pass_early_local_passes): Replace declaration with that of...
5038         (make_pass_early_local_passes): ...new function.
5039         (pass_ipa_whole_program_visibility): Replace declaration with that
5040         of...
5041         (make_pass_ipa_whole_program_visibility): ...new function.
5042         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5043         (make_pass_ipa_lto_gimple_out): ...new function.
5044         (pass_ipa_increase_alignment): Replace declaration with that of...
5045         (make_pass_ipa_increase_alignment): ...new function.
5046         (pass_ipa_inline): Replace declaration with that of...
5047         (make_pass_ipa_inline): ...new function.
5048         (pass_ipa_free_lang_data): Replace declaration with that of...
5049         (make_pass_ipa_free_lang_data): ...new function.
5050         (pass_ipa_free_inline_summary): Replace declaration with that of...
5051         (make_pass_ipa_free_inline_summary): ...new function.
5052         (pass_ipa_cp): Replace declaration with that of...
5053         (make_pass_ipa_cp): ...new function.
5054         (pass_ipa_reference): Replace declaration with that of...
5055         (make_pass_ipa_reference): ...new function.
5056         (pass_ipa_pure_const): Replace declaration with that of...
5057         (make_pass_ipa_pure_const): ...new function.
5058         (pass_ipa_pta): Replace declaration with that of...
5059         (make_pass_ipa_pta): ...new function.
5060         (pass_ipa_lto_finish_out): Replace declaration with that of...
5061         (make_pass_ipa_lto_finish_out): ...new function.
5062         (pass_ipa_tm): Replace declaration with that of...
5063         (make_pass_ipa_tm): ...new function.
5064         (pass_ipa_profile): Replace declaration with that of...
5065         (make_pass_ipa_profile): ...new function.
5066         (pass_ipa_cdtor_merge): Replace declaration with that of...
5067         (make_pass_ipa_cdtor_merge): ...new function.
5068         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5069         of...
5070         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5071         (pass_init_datastructures): Replace declaration with that of...
5072         (make_pass_init_datastructures): ...new function.
5073         (pass_fixup_cfg): Replace declaration with that of...
5074         (make_pass_fixup_cfg): ...new function.
5075         (pass_expand): Replace declaration with that of...
5076         (make_pass_expand): ...new function.
5077         (pass_instantiate_virtual_regs): Replace declaration with that of...
5078         (make_pass_instantiate_virtual_regs): ...new function.
5079         (pass_rtl_fwprop): Replace declaration with that of...
5080         (make_pass_rtl_fwprop): ...new function.
5081         (pass_rtl_fwprop_addr): Replace declaration with that of...
5082         (make_pass_rtl_fwprop_addr): ...new function.
5083         (pass_jump): Replace declaration with that of...
5084         (make_pass_jump): ...new function.
5085         (pass_jump2): Replace declaration with that of...
5086         (make_pass_jump2): ...new function.
5087         (pass_lower_subreg): Replace declaration with that of...
5088         (make_pass_lower_subreg): ...new function.
5089         (pass_cse): Replace declaration with that of...
5090         (make_pass_cse): ...new function.
5091         (pass_fast_rtl_dce): Replace declaration with that of...
5092         (make_pass_fast_rtl_dce): ...new function.
5093         (pass_ud_rtl_dce): Replace declaration with that of...
5094         (make_pass_ud_rtl_dce): ...new function.
5095         (pass_rtl_dce): Replace declaration with that of...
5096         (make_pass_rtl_dce): ...new function.
5097         (pass_rtl_dse1): Replace declaration with that of...
5098         (make_pass_rtl_dse1): ...new function.
5099         (pass_rtl_dse2): Replace declaration with that of...
5100         (make_pass_rtl_dse2): ...new function.
5101         (pass_rtl_dse3): Replace declaration with that of...
5102         (make_pass_rtl_dse3): ...new function.
5103         (pass_rtl_cprop): Replace declaration with that of...
5104         (make_pass_rtl_cprop): ...new function.
5105         (pass_rtl_pre): Replace declaration with that of...
5106         (make_pass_rtl_pre): ...new function.
5107         (pass_rtl_hoist): Replace declaration with that of...
5108         (make_pass_rtl_hoist): ...new function.
5109         (pass_rtl_store_motion): Replace declaration with that of...
5110         (make_pass_rtl_store_motion): ...new function.
5111         (pass_cse_after_global_opts): Replace declaration with that of...
5112         (make_pass_cse_after_global_opts): ...new function.
5113         (pass_rtl_ifcvt): Replace declaration with that of...
5114         (make_pass_rtl_ifcvt): ...new function.
5115         (pass_into_cfg_layout_mode): Replace declaration with that of...
5116         (make_pass_into_cfg_layout_mode): ...new function.
5117         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5118         (make_pass_outof_cfg_layout_mode): ...new function.
5119         (pass_loop2): Replace declaration with that of...
5120         (make_pass_loop2): ...new function.
5121         (pass_rtl_loop_init): Replace declaration with that of...
5122         (make_pass_rtl_loop_init): ...new function.
5123         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5124         (make_pass_rtl_move_loop_invariants): ...new function.
5125         (pass_rtl_unswitch): Replace declaration with that of...
5126         (make_pass_rtl_unswitch): ...new function.
5127         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5128         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5129         (pass_rtl_doloop): Replace declaration with that of...
5130         (make_pass_rtl_doloop): ...new function.
5131         (pass_rtl_loop_done): Replace declaration with that of...
5132         (make_pass_rtl_loop_done): ...new function.
5133         (pass_web): Replace declaration with that of...
5134         (make_pass_web): ...new function.
5135         (pass_cse2): Replace declaration with that of...
5136         (make_pass_cse2): ...new function.
5137         (pass_df_initialize_opt): Replace declaration with that of...
5138         (make_pass_df_initialize_opt): ...new function.
5139         (pass_df_initialize_no_opt): Replace declaration with that of...
5140         (make_pass_df_initialize_no_opt): ...new function.
5141         (pass_reginfo_init): Replace declaration with that of...
5142         (make_pass_reginfo_init): ...new function.
5143         (pass_inc_dec): Replace declaration with that of...
5144         (make_pass_inc_dec): ...new function.
5145         (pass_stack_ptr_mod): Replace declaration with that of...
5146         (make_pass_stack_ptr_mod): ...new function.
5147         (pass_initialize_regs): Replace declaration with that of...
5148         (make_pass_initialize_regs): ...new function.
5149         (pass_combine): Replace declaration with that of...
5150         (make_pass_combine): ...new function.
5151         (pass_if_after_combine): Replace declaration with that of...
5152         (make_pass_if_after_combine): ...new function.
5153         (pass_ree): Replace declaration with that of...
5154         (make_pass_ree): ...new function.
5155         (pass_partition_blocks): Replace declaration with that of...
5156         (make_pass_partition_blocks): ...new function.
5157         (pass_match_asm_constraints): Replace declaration with that of...
5158         (make_pass_match_asm_constraints): ...new function.
5159         (pass_regmove): Replace declaration with that of...
5160         (make_pass_regmove): ...new function.
5161         (pass_split_all_insns): Replace declaration with that of...
5162         (make_pass_split_all_insns): ...new function.
5163         (pass_fast_rtl_byte_dce): Replace declaration with that of...
5164         (make_pass_fast_rtl_byte_dce): ...new function.
5165         (pass_lower_subreg2): Replace declaration with that of...
5166         (make_pass_lower_subreg2): ...new function.
5167         (pass_mode_switching): Replace declaration with that of...
5168         (make_pass_mode_switching): ...new function.
5169         (pass_sms): Replace declaration with that of...
5170         (make_pass_sms): ...new function.
5171         (pass_sched): Replace declaration with that of...
5172         (make_pass_sched): ...new function.
5173         (pass_ira): Replace declaration with that of...
5174         (make_pass_ira): ...new function.
5175         (pass_reload): Replace declaration with that of...
5176         (make_pass_reload): ...new function.
5177         (pass_clean_state): Replace declaration with that of...
5178         (make_pass_clean_state): ...new function.
5179         (pass_branch_prob): Replace declaration with that of...
5180         (make_pass_branch_prob): ...new function.
5181         (pass_value_profile_transformations): Replace declaration with that
5182         of...
5183         (make_pass_value_profile_transformations): ...new function.
5184         (pass_postreload_cse): Replace declaration with that of...
5185         (make_pass_postreload_cse): ...new function.
5186         (pass_gcse2): Replace declaration with that of...
5187         (make_pass_gcse2): ...new function.
5188         (pass_split_after_reload): Replace declaration with that of...
5189         (make_pass_split_after_reload): ...new function.
5190         (pass_branch_target_load_optimize1): Replace declaration with that
5191         of...
5192         (make_pass_branch_target_load_optimize1): ...new function.
5193         (pass_thread_prologue_and_epilogue): Replace declaration with that
5194         of...
5195         (make_pass_thread_prologue_and_epilogue): ...new function.
5196         (pass_stack_adjustments): Replace declaration with that of...
5197         (make_pass_stack_adjustments): ...new function.
5198         (pass_peephole2): Replace declaration with that of...
5199         (make_pass_peephole2): ...new function.
5200         (pass_if_after_reload): Replace declaration with that of...
5201         (make_pass_if_after_reload): ...new function.
5202         (pass_regrename): Replace declaration with that of...
5203         (make_pass_regrename): ...new function.
5204         (pass_cprop_hardreg): Replace declaration with that of...
5205         (make_pass_cprop_hardreg): ...new function.
5206         (pass_reorder_blocks): Replace declaration with that of...
5207         (make_pass_reorder_blocks): ...new function.
5208         (pass_branch_target_load_optimize2): Replace declaration with that
5209         of...
5210         (make_pass_branch_target_load_optimize2): ...new function.
5211         (pass_leaf_regs): Replace declaration with that of...
5212         (make_pass_leaf_regs): ...new function.
5213         (pass_split_before_sched2): Replace declaration with that of...
5214         (make_pass_split_before_sched2): ...new function.
5215         (pass_compare_elim_after_reload): Replace declaration with that of...
5216         (make_pass_compare_elim_after_reload): ...new function.
5217         (pass_sched2): Replace declaration with that of...
5218         (make_pass_sched2): ...new function.
5219         (pass_stack_regs): Replace declaration with that of...
5220         (make_pass_stack_regs): ...new function.
5221         (pass_stack_regs_run): Replace declaration with that of...
5222         (make_pass_stack_regs_run): ...new function.
5223         (pass_df_finish): Replace declaration with that of...
5224         (make_pass_df_finish): ...new function.
5225         (pass_compute_alignments): Replace declaration with that of...
5226         (make_pass_compute_alignments): ...new function.
5227         (pass_duplicate_computed_gotos): Replace declaration with that of...
5228         (make_pass_duplicate_computed_gotos): ...new function.
5229         (pass_variable_tracking): Replace declaration with that of...
5230         (make_pass_variable_tracking): ...new function.
5231         (pass_free_cfg): Replace declaration with that of...
5232         (make_pass_free_cfg): ...new function.
5233         (pass_machine_reorg): Replace declaration with that of...
5234         (make_pass_machine_reorg): ...new function.
5235         (pass_cleanup_barriers): Replace declaration with that of...
5236         (make_pass_cleanup_barriers): ...new function.
5237         (pass_delay_slots): Replace declaration with that of...
5238         (make_pass_delay_slots): ...new function.
5239         (pass_split_for_shorten_branches): Replace declaration with that of...
5240         (make_pass_split_for_shorten_branches): ...new function.
5241         (pass_split_before_regstack): Replace declaration with that of...
5242         (make_pass_split_before_regstack): ...new function.
5243         (pass_convert_to_eh_region_ranges): Replace declaration with that
5244         of...
5245         (make_pass_convert_to_eh_region_ranges): ...new function.
5246         (pass_shorten_branches): Replace declaration with that of...
5247         (make_pass_shorten_branches): ...new function.
5248         (pass_set_nothrow_function_flags): Replace declaration with that of...
5249         (make_pass_set_nothrow_function_flags): ...new function.
5250         (pass_dwarf2_frame): Replace declaration with that of...
5251         (make_pass_dwarf2_frame): ...new function.
5252         (pass_final): Replace declaration with that of...
5253         (make_pass_final): ...new function.
5254         (pass_rtl_seqabstr): Replace declaration with that of...
5255         (make_pass_rtl_seqabstr): ...new function.
5256         (pass_release_ssa_names): Replace declaration with that of...
5257         (make_pass_release_ssa_names): ...new function.
5258         (pass_early_inline): Replace declaration with that of...
5259         (make_pass_early_inline): ...new function.
5260         (pass_inline_parameters): Replace declaration with that of...
5261         (make_pass_inline_parameters): ...new function.
5262         (pass_update_address_taken): Replace declaration with that of...
5263         (make_pass_update_address_taken): ...new function.
5264         (pass_convert_switch): Replace declaration with that of...
5265         (make_pass_convert_switch): ...new function.
5266         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
5267         to a subclass of simple_ipa_opt_pass along with...
5268         (pass_data_ipa_tree_profile): ...new pass_data instance and...
5269         (make_pass_ipa_tree_profile): ...new function.
5270         * tree-sra.c (pass_sra_early): Convert from a global struct to a
5271         subclass of gimple_opt_pass along with...
5272         (pass_data_sra_early): ...new pass_data instance and...
5273         (make_pass_sra_early): ...new function.
5274         (pass_sra): Convert from a global struct to a subclass of
5275         gimple_opt_pass along with...
5276         (pass_data_sra): ...new pass_data instance and...
5277         (make_pass_sra): ...new function.
5278         (pass_early_ipa_sra): Convert from a global struct to a subclass of
5279         gimple_opt_pass along with...
5280         (pass_data_early_ipa_sra): ...new pass_data instance and...
5281         (make_pass_early_ipa_sra): ...new function.
5282         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
5283         subclass of gimple_opt_pass along with...
5284         (pass_data_ccp): ...new pass_data instance and...
5285         (make_pass_ccp): ...new function.
5286         (pass_fold_builtins): Convert from a global struct to a subclass of
5287         gimple_opt_pass along with...
5288         (pass_data_fold_builtins): ...new pass_data instance and...
5289         (make_pass_fold_builtins): ...new function.
5290         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
5291         subclass of gimple_opt_pass along with...
5292         (pass_data_copy_prop): ...new pass_data instance and...
5293         (make_pass_copy_prop): ...new function.
5294         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
5295         global struct to a subclass of gimple_opt_pass along with...
5296         (pass_data_rename_ssa_copies): ...new pass_data instance and...
5297         (make_pass_rename_ssa_copies): ...new function.
5298         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
5299         subclass of gimple_opt_pass along with...
5300         (pass_data_dce): ...new pass_data instance and...
5301         (make_pass_dce): ...new function.
5302         (pass_dce_loop): Convert from a global struct to a subclass of
5303         gimple_opt_pass along with...
5304         (pass_data_dce_loop): ...new pass_data instance and...
5305         (make_pass_dce_loop): ...new function.
5306         (pass_cd_dce): Convert from a global struct to a subclass of
5307         gimple_opt_pass along with...
5308         (pass_data_cd_dce): ...new pass_data instance and...
5309         (make_pass_cd_dce): ...new function.
5310         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
5311         subclass of gimple_opt_pass along with...
5312         (pass_data_dominator): ...new pass_data instance and...
5313         (make_pass_dominator): ...new function.
5314         (pass_phi_only_cprop): Convert from a global struct to a subclass of
5315         gimple_opt_pass along with...
5316         (pass_data_phi_only_cprop): ...new pass_data instance and...
5317         (make_pass_phi_only_cprop): ...new function.
5318         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
5319         subclass of gimple_opt_pass along with...
5320         (pass_data_dse): ...new pass_data instance and...
5321         (make_pass_dse): ...new function.
5322         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
5323         a subclass of gimple_opt_pass along with...
5324         (pass_data_forwprop): ...new pass_data instance and...
5325         (make_pass_forwprop): ...new function.
5326         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
5327         struct to a subclass of gimple_opt_pass along with...
5328         (pass_data_tree_ifcombine): ...new pass_data instance and...
5329         (make_pass_tree_ifcombine): ...new function.
5330         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
5331         subclass of gimple_opt_pass along with...
5332         (pass_data_ch): ...new pass_data instance and...
5333         (make_pass_ch): ...new function.
5334         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
5335         subclass of gimple_opt_pass along with...
5336         (pass_data_tree_loop): ...new pass_data instance and...
5337         (make_pass_tree_loop): ...new function.
5338         (pass_tree_loop_init): Convert from a global struct to a subclass of
5339         gimple_opt_pass along with...
5340         (pass_data_tree_loop_init): ...new pass_data instance and...
5341         (make_pass_tree_loop_init): ...new function.
5342         (pass_lim): Convert from a global struct to a subclass of
5343         gimple_opt_pass along with...
5344         (pass_data_lim): ...new pass_data instance and...
5345         (make_pass_lim): ...new function.
5346         (pass_tree_unswitch): Convert from a global struct to a subclass of
5347         gimple_opt_pass along with...
5348         (pass_data_tree_unswitch): ...new pass_data instance and...
5349         (make_pass_tree_unswitch): ...new function.
5350         (pass_predcom): Convert from a global struct to a subclass of
5351         gimple_opt_pass along with...
5352         (pass_data_predcom): ...new pass_data instance and...
5353         (make_pass_predcom): ...new function.
5354         (pass_vectorize): Convert from a global struct to a subclass of
5355         gimple_opt_pass along with...
5356         (pass_data_vectorize): ...new pass_data instance and...
5357         (make_pass_vectorize): ...new function.
5358         (pass_graphite): Convert from a global struct to a subclass of
5359         gimple_opt_pass along with...
5360         (pass_data_graphite): ...new pass_data instance and...
5361         (make_pass_graphite): ...new function.
5362         (pass_graphite_transforms): Convert from a global struct to a subclass
5363         of gimple_opt_pass along with...
5364         (pass_data_graphite_transforms): ...new pass_data instance and...
5365         (make_pass_graphite_transforms): ...new function.
5366         (pass_check_data_deps): Convert from a global struct to a subclass of
5367         gimple_opt_pass along with...
5368         (pass_data_check_data_deps): ...new pass_data instance and...
5369         (make_pass_check_data_deps): ...new function.
5370         (pass_iv_canon): Convert from a global struct to a subclass of
5371         gimple_opt_pass along with...
5372         (pass_data_iv_canon): ...new pass_data instance and...
5373         (make_pass_iv_canon): ...new function.
5374         (pass_scev_cprop): Convert from a global struct to a subclass of
5375         gimple_opt_pass along with...
5376         (pass_data_scev_cprop): ...new pass_data instance and...
5377         (make_pass_scev_cprop): ...new function.
5378         (pass_record_bounds): Convert from a global struct to a subclass of
5379         gimple_opt_pass along with...
5380         (pass_data_record_bounds): ...new pass_data instance and...
5381         (make_pass_record_bounds): ...new function.
5382         (pass_complete_unroll): Convert from a global struct to a subclass of
5383         gimple_opt_pass along with...
5384         (pass_data_complete_unroll): ...new pass_data instance and...
5385         (make_pass_complete_unroll): ...new function.
5386         (pass_complete_unrolli): Convert from a global struct to a subclass of
5387         gimple_opt_pass along with...
5388         (pass_data_complete_unrolli): ...new pass_data instance and...
5389         (make_pass_complete_unrolli): ...new function.
5390         (pass_parallelize_loops): Convert from a global struct to a subclass
5391         of gimple_opt_pass along with...
5392         (pass_data_parallelize_loops): ...new pass_data instance and...
5393         (make_pass_parallelize_loops): ...new function.
5394         (pass_loop_prefetch): Convert from a global struct to a subclass of
5395         gimple_opt_pass along with...
5396         (pass_data_loop_prefetch): ...new pass_data instance and...
5397         (make_pass_loop_prefetch): ...new function.
5398         (pass_iv_optimize): Convert from a global struct to a subclass of
5399         gimple_opt_pass along with...
5400         (pass_data_iv_optimize): ...new pass_data instance and...
5401         (make_pass_iv_optimize): ...new function.
5402         (pass_tree_loop_done): Convert from a global struct to a subclass of
5403         gimple_opt_pass along with...
5404         (pass_data_tree_loop_done): ...new pass_data instance and...
5405         (make_pass_tree_loop_done): ...new function.
5406         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
5407         struct to a subclass of gimple_opt_pass along with...
5408         (pass_data_cse_reciprocals): ...new pass_data instance and...
5409         (make_pass_cse_reciprocals): ...new function.
5410         (pass_cse_sincos): Convert from a global struct to a subclass of
5411         gimple_opt_pass along with...
5412         (pass_data_cse_sincos): ...new pass_data instance and...
5413         (make_pass_cse_sincos): ...new function.
5414         (pass_optimize_bswap): Convert from a global struct to a subclass of
5415         gimple_opt_pass along with...
5416         (pass_data_optimize_bswap): ...new pass_data instance and...
5417         (make_pass_optimize_bswap): ...new function.
5418         (pass_optimize_widening_mul): Convert from a global struct to a
5419         subclass of gimple_opt_pass along with...
5420         (pass_data_optimize_widening_mul): ...new pass_data instance and...
5421         (make_pass_optimize_widening_mul): ...new function.
5422         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
5423         subclass of gimple_opt_pass along with...
5424         (pass_data_phiopt): ...new pass_data instance and...
5425         (make_pass_phiopt): ...new function.
5426         (pass_cselim): Convert from a global struct to a subclass of
5427         gimple_opt_pass along with...
5428         (pass_data_cselim): ...new pass_data instance and...
5429         (make_pass_cselim): ...new function.
5430         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
5431         subclass of gimple_opt_pass along with...
5432         (pass_data_phiprop): ...new pass_data instance and...
5433         (make_pass_phiprop): ...new function.
5434         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
5435         subclass of gimple_opt_pass along with...
5436         (pass_data_pre): ...new pass_data instance and...
5437         (make_pass_pre): ...new function.
5438         (pass_fre): Convert from a global struct to a subclass of
5439         gimple_opt_pass along with...
5440         (pass_data_fre): ...new pass_data instance and...
5441         (make_pass_fre): ...new function.
5442         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
5443         subclass of gimple_opt_pass along with...
5444         (pass_data_reassoc): ...new pass_data instance and...
5445         (make_pass_reassoc): ...new function.
5446         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
5447         subclass of gimple_opt_pass along with...
5448         (pass_data_sink_code): ...new pass_data instance and...
5449         (make_pass_sink_code): ...new function.
5450         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
5451         subclass of gimple_opt_pass along with...
5452         (pass_data_strlen): ...new pass_data instance and...
5453         (make_pass_strlen): ...new function.
5454         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
5455         struct to a subclass of gimple_opt_pass along with...
5456         (pass_data_build_alias): ...new pass_data instance and...
5457         (make_pass_build_alias): ...new function.
5458         (pass_build_ealias): Convert from a global struct to a subclass of
5459         gimple_opt_pass along with...
5460         (pass_data_build_ealias): ...new pass_data instance and...
5461         (make_pass_build_ealias): ...new function.
5462         (pass_ipa_pta): Convert from a global struct to a subclass of
5463         simple_ipa_opt_pass along with...
5464         (pass_data_ipa_pta): ...new pass_data instance and...
5465         (make_pass_ipa_pta): ...new function.
5466         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
5467         subclass of gimple_opt_pass along with...
5468         (pass_data_uncprop): ...new pass_data instance and...
5469         (make_pass_uncprop): ...new function.
5470         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
5471         global struct to a subclass of gimple_opt_pass along with...
5472         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
5473         (make_pass_late_warn_uninitialized): ...new function.
5474         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
5475         to a subclass of gimple_opt_pass along with...
5476         (pass_data_init_datastructures): ...new pass_data instance and...
5477         (make_pass_init_datastructures): ...new function.
5478         (pass_early_warn_uninitialized): Convert from a global struct to a
5479         subclass of gimple_opt_pass along with...
5480         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
5481         (make_pass_early_warn_uninitialized): ...new function.
5482         (pass_update_address_taken): Convert from a global struct to a
5483         subclass of gimple_opt_pass along with...
5484         (pass_data_update_address_taken): ...new pass_data instance and...
5485         (make_pass_update_address_taken): ...new function.
5486         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
5487         struct to a subclass of gimple_opt_pass along with...
5488         (pass_data_release_ssa_names): ...new pass_data instance and...
5489         (make_pass_release_ssa_names): ...new function.
5490         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
5491         subclass of gimple_opt_pass along with...
5492         (pass_data_stdarg): ...new pass_data instance and...
5493         (make_pass_stdarg): ...new function.
5494         * tree-switch-conversion.c (pass_convert_switch): Convert from a
5495         global struct to a subclass of gimple_opt_pass along with...
5496         (pass_data_convert_switch): ...new pass_data instance and...
5497         (make_pass_convert_switch): ...new function.
5498         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
5499         to a subclass of gimple_opt_pass along with...
5500         (pass_data_tail_recursion): ...new pass_data instance and...
5501         (make_pass_tail_recursion): ...new function.
5502         (pass_tail_calls): Convert from a global struct to a subclass of
5503         gimple_opt_pass along with...
5504         (pass_data_tail_calls): ...new pass_data instance and...
5505         (make_pass_tail_calls): ...new function.
5506         * tree-vect-generic.c (pass_lower_vector): Convert from a global
5507         struct to a subclass of gimple_opt_pass along with...
5508         (pass_data_lower_vector): ...new pass_data instance and...
5509         (make_pass_lower_vector): ...new function.
5510         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
5511         gimple_opt_pass along with...
5512         (pass_data_lower_vector_ssa): ...new pass_data instance and...
5513         (make_pass_lower_vector_ssa): ...new function.
5514         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
5515         to a subclass of gimple_opt_pass along with...
5516         (pass_data_slp_vectorize): ...new pass_data instance and...
5517         (make_pass_slp_vectorize): ...new function.
5518         (pass_ipa_increase_alignment): Convert from a global struct to a
5519         subclass of simple_ipa_opt_pass along with...
5520         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
5521         (make_pass_ipa_increase_alignment): ...new function.
5522         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
5523         gimple_opt_pass along with...
5524         (pass_data_vrp): ...new pass_data instance and...
5525         (make_pass_vrp): ...new function.
5526         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
5527         subclass of simple_ipa_opt_pass along with...
5528         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
5529         (make_pass_ipa_free_lang_data): ...new function.
5530         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
5531         gimple_opt_pass along with...
5532         (pass_data_tsan): ...new pass_data instance and...
5533         (make_pass_tsan): ...new function.
5534         (pass_tsan_O0): Convert from a global struct to a subclass of
5535         gimple_opt_pass along with...
5536         (pass_data_tsan_O0): ...new pass_data instance and...
5537         (make_pass_tsan_O0): ...new function.
5538         * var-tracking.c (pass_variable_tracking): Convert from a global
5539         struct to a subclass of rtl_opt_pass along with...
5540         (pass_data_variable_tracking): ...new pass_data instance and...
5541         (make_pass_variable_tracking): ...new function.
5542         * web.c (pass_web): Convert from a global struct to a subclass of
5543         rtl_opt_pass along with...
5544         (pass_data_web): ...new pass_data instance and...
5545         (make_pass_web): ...new function.
5546         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
5547         declaration with that of...
5548         (make_pass_mode_switch_use): ...new function.
5549         (pass_resolve_sw_modes): Replace declaration with that of...
5550         (make_pass_resolve_sw_modes): ...new function.
5551         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
5552         from a global struct to a subclass of rtl_opt_pass along with...
5553         (pass_data_mode_switch_use): ...new pass_data instance and...
5554         (make_pass_mode_switch_use): ...new function.
5555         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
5556         from a global struct to a subclass of rtl_opt_pass along with...
5557         (pass_data_resolve_sw_modes): ...new pass_data instance and...
5558         (make_pass_resolve_sw_modes): ...new function.
5559         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
5560         struct to a subclass of rtl_opt_pass along with...
5561         (pass_data_insert_vzeroupper): ...new pass_data instance and...
5562         (make_pass_insert_vzeroupper): ...new function.
5563         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
5564         global struct to a subclass of rtl_opt_pass along with...
5565         (pass_data_work_around_errata): ...new pass_data instance and...
5566         (make_pass_work_around_errata): ...new function.
5567         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
5568         struct to a subclass of rtl_opt_pass along with...
5569         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
5570         (make_pass_mips_machine_reorg2): ...new function.
5572 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5574         * passes.c (pass_manager::operator new): New.
5576 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5578         Handwritten part of conversion of passes to C++ classes.
5580         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
5581         (toplev.o): Add dep on PASS_MANAGER_H.
5582         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
5583         of early local pases to reflect this moving from a global to a
5584         member of gcc::pass_manager.
5585         (cgraph_add_new_function): Likewise.
5586         * lto-cgraph.c (lto_output_node): Update for conversion of
5587         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
5588         * passes.c (opt_pass::clone): New.
5589         (opt_pass::gate): New.
5590         (opt_pass::execute): New.
5591         (opt_pass::opt_pass): New.
5592         (pass_manager::execute_early_local_passes): New.
5593         (pass_manager::execute_pass_mode_switching): new.
5594         (finish_optimization_passes): Convert to...
5595         (pass_manager::finish_optimization_passes): ...this.
5596         (finish_optimization_passes): Update for conversion of passes to
5597         C++ classes.
5598         (register_dump_files_1): Use has_gate since we cannot portably
5599         check a vtable entry against NULL.
5600         (dump_one_pass): Likewise.
5601         (ipa_write_summaries_2): Likewise.
5602         (ipa_write_optimization_summaries_1): Likewise.
5603         (ipa_read_summaries_1): Likewise.
5604         (ipa_read_optimization_summaries_1): Likewise.
5605         (execute_ipa_stmt_fixups): Likewise.
5606         (pass_manager::pass_manager): Rewrite pass-creation, invoking
5607         pass-creation functions rather than wiring up globals, and
5608         storing the results in fields of pass_manager generated using
5609         pass-instances.def.
5610         (pass_manager::dump_profile_report): Update for conversion of
5611         passes to C++ classes.
5612         (pass_manager::execute_ipa_summary_passes): Likewise.
5613         (execute_one_ipa_transform_pass): Likewise.
5614         (execute_one_pass): Use has_gate and has_execute since we cannot
5615         portably check a vtable entry against NULL.
5616         * pass_manager.h (pass_manager::finish_optimization_passes): New.
5617         (pass_manager): Use pass-instances.def to add fields for the
5618         various pass instances.
5619         * toplev.c (finalize): Update for move of
5620         finish_optimization_passes to a method of gcc::pass_manager.
5621         * toplev.h (finish_optimization_passes): Move to method of class
5622         pass_manager.
5623         * tree-pass.h (struct pass_data): New.
5624         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
5625         (opt_pass::gate): Convert to virtual function.
5626         (opt_pass::~opt_pass): New.
5627         (opt_pass::clone): New.
5628         (opt_pass::execute): Convert to virtual function.
5629         (opt_pass::opt_pass): New.
5630         (opt_pass::ctxt_): new.
5631         (gimple_opt_pass): Convert to subclass of opt_pass.
5632         (gimple_opt_pass::gimple_opt_pass): New.
5633         (rtl_opt_pass): Convert to subclass of opt_pass.
5634         (rtl_opt_pass::rtl_opt_pass): New.
5635         (ipa_opt_pass_d): Convert to subclass of opt_pass.
5636         (ipa_opt_pass_d::ipa_opt_pass_d): New.
5637         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
5638         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
5639         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
5640         invocation of pass_mode_switching to reflect this moving from a
5641         global to a member of gcc::pass_manager.
5642         (ix86_option_override): Rework how pass_insert_vzeroupper is
5643         added to the pass_manager to reflect autogenerated changes.
5644         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
5646 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
5648         PR rtl-optimization/57708
5649         * recog.c (peep2_find_free_register): Validate all regs in a
5650         multi-reg mode.
5652 2013-08-05  Jan Hubicka  <jh@suse.cz>
5654         PR lto/57602
5655         * cgraph.c (verify_cgraph_node): Accept local flags from other
5656         partitions.
5657         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
5658         (function_and_variable_visibility): Likewise.
5659         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
5661 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5663         * graph.c (init_graph_slim_pretty_print): Remove.
5664         (print_graph_cfg): Do not call it.  Use local pretty printer.
5665         (start_graph_dump): Likewise.
5667 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5669         * gimple-pretty-print.c (buffer): Remove.
5670         (initialized): Likewise.
5671         (maybe_init_pretty_print): Likewise.
5672         (print_gimple_stmt): Do not call it.  Use non-static local
5673         pretty_printer variable.
5674         (print_gimple_expr): Likewise.
5675         (print_gimple_seq): Likewise.
5676         (gimple_dump_bb): Likewise.
5678 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5680         * asan.c (asan_pp): Remove.
5681         (asan_pp_initialized): Likewise.
5682         (asan_pp_initialize): Likewise.
5683         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
5684         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
5685         (asan_add_global): Likewise.
5687 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5689         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
5690         * diagnostic.h (diagnostic_flush_buffer): Adjust.
5691         * pretty-print.c (pp_formatted_text_data): Likewise.
5692         (pp_indent): Rename from pp_base_indent.
5693         (pp_format): Rename from pp_base_format.
5694         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
5695         (pp_format_verbatim): Rename from pp_base_format_verbatim.
5696         (pp_flush): Rename from pp_base_flush.
5697         (pp_set_line_maximum_length): Rename from
5698         pp_base_set_line_maximum_length.
5699         (pp_clear_output_area): Rename from pp_base_clear_output_area.
5700         (pp_set_prefix): Rename from pp_base_set_prefix.
5701         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
5702         (pp_emit_prefix): Rename from pp_base_emit_prefix.
5703         (pp_append_text): Rename from pp_base_append_text.
5704         (pp_formatted_text): Rename from pp_base_formatted_text.
5705         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
5706         (pp_remaining_character_count_for_line): Rename from
5707         pp_base_remaining_character_count_for_line.
5708         (pp_newline): Rename from pp_base_newline.
5709         (pp_character): Rename from pp_base_character.
5710         (pp_string): Rename from pp_base_string.
5711         (pp_maybe_space): Rename from pp_base_maybe_space.
5712         * asan.c (asan_pp_string): Adjust.
5713         (asan_emit_stack_protection): Likewise.
5714         (asan_add_global): Likewise.
5715         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
5716         * tree-mudflap.c (mf_varname_tree): Likewise.
5717         * tree-pretty-print.c (pp_tree_identifier): Rename from
5718         pp_base_tree_identifier.
5719         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
5720         Declare as function.
5722 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5724         * pretty-print.h (pp_bar_bar): New.
5725         (pp_ampersand_ampersand): Likewise.
5726         (pp_less_equal): Likewise.
5727         (pp_greater_equal): Likewise.
5728         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
5729         printer functions instead of pp_string or operators and punctuators.
5730         (dump_gimple_call): Likewise.
5731         (dump_gimple_omp_for): Likewise.
5732         (dump_gimple_transaction): Likewise.
5733         (dump_gimple_phi): Likewise.
5734         (pp_gimple_stmt_1): Likewise.
5735         * sched-vis.c (print_insn): Likewise.
5736         * tree-mudflap.c (mf_varname_tree): Likewise.
5737         * tree-pretty-print.c (dump_block_node): Likewise.
5738         (dump_generic_node): Likewise.
5740 2013-08-02  Jan Hubicka  <jh@suse.cz>
5742         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
5743         boundaries.
5744         * lto-streamer-out.c (tree_is_indexable): Results decls and
5745         parm decls are not indexable.
5746         (DFS_write_tree_body): Do not follow args and results.
5747         (hash_tree): Likewise.
5748         (output_functions): Rearrange so struct function is needed
5749         only when real body is output; be able to also ouptut abstract
5750         functions; output DECL_ARGUMENTS and DECL_RESULT.
5751         (lto_output): When not in WPA, ale store abstract functions.
5752         (write_symbol): Do not care about RESULT_DECL.
5753         (output_symbol_p): Handle correctly sbtract decls.
5754         * lto-streamer-in.c (input_function): Rearrange so struct
5755         function can be NULL at entry; allow streaming of
5756         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
5757         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
5758         sanity check during LTO.
5759         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
5760         RESULT_DECl and DECL_ARGUMENTS.
5761         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
5762         Likewise.
5764 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
5766         * pretty-print.h (pp_underscore): New.
5767         (pp_comma): Tidy.
5768         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
5769         printer functions instead of pp_character.
5770         (dump_binary_rhs): Likewise.
5771         (dump_ternary_rhs): Likewise.
5772         (dump_gimple_call_args): Likewise.
5773         (pp_points_to_solution): Likewise.
5774         (dump_gimple_call): Likewise.
5775         (dump_gimple_switch): Likewise.
5776         (dump_gimple_cond): Likewise.
5777         (dump_gimple_bind): Likewise.
5778         (dump_gimple_try): Likewise.
5779         (dump_gimple_omp_for): Likewise.
5780         (dump_gimple_omp_continue): Likewise.
5781         (dump_gimple_omp_single): Likewise.
5782         (dump_gimple_omp_sections): Likewise.
5783         (dump_gimple_omp_block): Likewise.
5784         (dump_gimple_omp_critical): Likewise.
5785         (dump_gimple_transaction): Likewise.
5786         (dump_gimple_asm): Likewise.
5787         (dump_gimple_phi): Likewise.
5788         (dump_gimple_omp_parallel): Likewise.
5789         (dump_gimple_omp_task): Likewise.
5790         (dump_gimple_omp_atomic_load): Likewise.
5791         (dump_gimple_omp_atomic_store): Likewise.
5792         (dump_gimple_mem_ops): Likewise.
5793         (pp_gimple_stmt_1): Likewise.
5794         (pp_cfg_jump): Likewise.
5795         (dump_implicit_edges): Likewise.
5796         (gimple_dump_bb_for_graph): Likewise.
5797         * graph.c (draw_cfg_node): Likewise.
5798         * langhooks.c (lhd_print_error_function): Likewise.
5799         * sched-vis.c (print_exp): Likewise.
5800         (print_value): Likewise.
5801         (print_pattern): Likewise.
5802         (print_insn): Likewise.
5803         (rtl_dump_bb_for_graph): Likewise.
5804         * tree-pretty-print.c (dump_function_declaration): Likewise.
5805         (dump_array_domain): Likewise.
5806         (dump_omp_clause): Likewise.
5807         (dump_location): Likewise.
5808         (dump_generic_node): Likewise.
5809         (print_struct_decl): Likewise.
5810         * diagnostic.c (diagnostic_show_locus): Use pp_space.
5812 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
5814         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
5815         candidate table when replacing a candidate statement.
5816         (replace_rhs_if_not_dup): Likewise.
5817         (replace_one_candidate): Likewise.
5819 2013-08-02  Jan Hubicka  <jh@suse.cz>
5820             Martin Liska  <marxin.liska@gmail.com>
5822         * cgraphunit.c (add_new_function): Fix logic when adding from
5823         late IPA pass.
5824         (assemble_thunk): Rename to ...
5825         (expand_thunk); .. this one; export; get it working with
5826         general functions; make produced gimple valid.
5827         * cgraph.h (expand_thunk): Declare.
5829 2013-08-02  Jan Hubicka  <jh@suse.cz>
5831         * ipa-cp.c (gather_context_independent_values): Use
5832         ipa_get_param_move_cost.
5833         (get_replacement_map): Remove PARAM; move parameter folding
5834         into tree-inline.c
5835         (create_specialized_node): Update.
5836         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
5837         assert that we have gimple body; update move_cost.
5838         (count_formal_params): Assert that we have gimple body.
5839         (ipa_dump_param): New function.
5840         (ipa_alloc_node_params): Break out from ...
5841         (ipa_initialize_node_params): ... here.
5842         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
5843         (ipa_write_node_info): Stream move costs.
5844         (ipa_read_node_info): Read move costs.
5845         (ipa_update_after_lto_read): Do not recompute node params.
5846         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
5847         (ipa_get_param): Check we are not in WPA.
5848         (ipa_get_param_move_cost): New.
5849         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
5850         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
5851         parm numbers to be present.
5853 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5855         PR rtl-optimization/58048
5856         * lra-constraints.c (process_alt_operands): Don't check asm
5857         operand on register.
5859 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
5861         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
5862         the implied StoreLoad barrier for atomic operations if before.
5864 2013-08-02  Jan Hubicka  <jh@suse.cz>
5865             Martin Liska  <marxin.liska@gmail.com>
5867         * cgraph.c (cgraph_function_body_availability): Do not check
5868         cgraph flags.
5869         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
5870         symtab_node_availability): Declare.
5871         * ipa.c (can_replace_by_local_alias): New.
5872         (function_and_variable_visibility): Use it.
5873         * symtab.c (symtab_for_node_and_aliases,
5874         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
5876 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
5878         PR rtl-optimization/57963
5879         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
5880         (lra_constraints): Use them.
5882 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
5884         * config/arm/types.md (define_attr "type"): Add "load_acq"
5885         and "store_rel".
5886         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
5887         changes.
5888         (cortex_a53_store1): Likewise.
5890 2013-08-01  Jan Hubicka  <jh@suse.cz>
5892         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
5893         partitions are not needed.
5895 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
5897         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
5898         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
5899         MAYBE_NON_Q_CLASS_P where appropriate.
5901 2013-08-01  Jan Hubicka  <jh@suse.cz>
5903         * cgraph.h (release_function_body): Declare.
5904         * tree.c (free_lang_data_in_decl): Free, parameters and return values
5905         of unused delcarations.
5907 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5909         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
5910         RTL form when subtracting a constant.
5912 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5914         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
5915         Generate canonical plus rtx with negated immediate instead of minus
5916         where appropriate.
5917         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
5919 2013-08-01  Jan Hubicka  <jh@suse.cz>
5921         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
5922         (cgraph_release_function_body): Likewise.
5923         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
5924         * cgraph.h (cgrpah_node): Rename abstract_and_needed
5925         to used_as_abstract_origin.
5926         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
5927         symbols used as abstract origins.
5928         * cgraphunit.c (analyze_functions): Update.
5929         * ipa.c (symtab_remove_unreachable_nodes): Recompute
5930         used_as_abstract_origin.
5931         * tree-inline.c (tree_function_versioning): Update
5932         used_as_abstract_origin; be ready for DECL_RESULT and
5933         DECL_ARGUMENTS to be NULL.
5935         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
5936         for abstract functions.
5937         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
5938         real symbols.
5940 2013-08-01  Jan Hubicka  <jh@suse.cz>
5942         * profile.c (compute_value_histograms): Fix thinko.
5944 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5946         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
5947         aarch-common-protos.h to extra_headers.
5948         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5949         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
5950         * config/aarch64/t-aarch64 (aarch-common.o): Define.
5952 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
5954         * config/aarch64/aarch64.md (define_attr "type"): Delete.
5955         Include "../arm/types.md".  Define "type" attribute for all patterns.
5956         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
5957         attribute changes.
5959 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
5961         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
5962         to support power8 load fusion.
5963         (fusion_gpr_mem_load): Likewise.
5965         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
5967         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
5968         declarations for power8 load fusion.
5969         (emit_fusion_gpr_load): Likewise.
5971         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
5972         tuning for power8, turn on fusion mode by default.  Turn on sign
5973         extending fusion mode if normal fusion mode is on, and we are at
5974         -O2 or -O3.
5975         (fusion_gpr_load_p): New function, return true if we can fuse an
5976         addis instruction with a dependent load to a GPR.
5977         (emit_fusion_gpr_load): Emit the instructions for power8 load
5978         fusion to GPRs.
5980         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
5981         (VSX load fusion peepholes): New peepholes to fuse together an
5982         addi instruction with a VSX load instruction.
5984         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
5985         peepholes to fuse an addis instruction with a load to a GPR base
5986         register.  If we are supporting sign extending fusions, convert
5987         sign extending loads to zero extending loads and add an explicit
5988         sign extension.
5990 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
5992         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
5993         aarch-common-protos.h to extra_headers.
5994         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
5995         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
5996         (arm_early_store_addr_dep): Likewise.
5997         (arm_no_early_alu_shift_dep): Likewise.
5998         (arm_no_early_alu_shift_value_dep): Likewise.
5999         (arm_no_early_mul_dep): Likewise.
6000         (arm_no_early_store_addr_dep): Likewise.
6001         (arm_mac_accumulator_is_mul_result): Likewise.
6002         (arm_mac_accumulator_is_result): Likewise.
6003         * config/arm/aarch-common.c: ... here.  New file.
6004         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
6005         here to ...
6006         (arm_early_store_addr_dep): Likewise.
6007         (arm_no_early_alu_shift_dep): Likewise.
6008         (arm_no_early_alu_shift_value_dep): Likewise.
6009         (arm_no_early_mul_dep): Likewise.
6010         (arm_no_early_store_addr_dep): Likewise.
6011         (arm_mac_accumulator_is_mul_result): Likewise.
6012         (arm_mac_accumulator_is_result): Likewise.
6013         * config/arm/aarch-common-protos.h: ... here.  New file.
6014         * config/arm/t-arm (aarch-common.o): Define.
6016 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6018         * config/arm/arm.md: Include new file "types.md".
6019         (define_attr "type"): Move from here to ...
6020         (define_attr "mul32"): Likewise.
6021         (define_attr "mul64"): Likewise.
6022         * config/arm/types.md: ... here.  New file.
6024 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6026         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6027         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6029 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6031         * gen-pass-instances.awk: Fix offset of substr().
6033 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
6035         * Makefile.in (pass-instances.def): New.
6036         (passes.o): Replace dependency on passes.def with one on
6037         pass-instances.def
6039         * gen-pass-instances.awk: New.
6041         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6042         rather than passes.def, updating local definition of NEXT_PASS
6043         macro to add an extra NUM parameter (currently unused).
6045 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6047         * Makefile.in (PASS_MANAGER_H): New.
6048         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6049         (passes.o): Likewise.
6050         (statistics.o): Likewise.
6051         (cgraphunit.o): Likewise.
6052         (context.o): Depend on PASS_MANAGER_H.
6054         * pass_manager.h: New.
6056         * cgraphunit.c (cgraph_add_new_function): Update for moves
6057         of globals to fields of pass_manager.
6058         (analyze_function): Likewise.
6059         (expand_function): Likewise.
6060         (ipa_passes): Likewise.
6061         (compile): Likewise.
6063         * context.c (context::context): New.
6064         * context.h  (context::context): New.
6065         (context::get_passes): New.
6066         (context::passes_): New.
6068         * lto-cgraph.c (input_node): Update for moves of globals to
6069         fields of pass_manager.
6071         * passes.c (all_passes): Remove, in favor of a field of the
6072         same name within the new class pass_manager.
6073         (all_small_ipa_passes): Likewise.
6074         (all_lowering_passes): Likewise.
6075         (all_regular_ipa_passes): Likewise.
6076         (all_late_ipa_passes): Likewise.
6077         (all_lto_gen_passes): Likewise.
6078         (passes_by_id): Likewise.
6079         (passes_by_id_size): Likewise.
6080         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6081         the new class pass_manager.
6082         (set_pass_for_id): Convert to...
6083         (pass_manager::set_pass_for_id): ...method.
6084         (get_pass_for_id): Convert to...
6085         (pass_manager::get_pass_for_id): ...method.
6086         (register_one_dump_file): Move body of implementation into...
6087         (pass_manager::register_one_dump_file): ...here.
6088         (register_dump_files_1): Convert to...
6089         (pass_manager::register_dump_files_1): ...method.
6090         (register_dump_files): Convert to...
6091         (pass_manager::register_dump_files): ...method.
6092         (create_pass_tab): Update for moves of globals to fields of
6093         pass_manager.
6094         (dump_passes): Move body of implementation into...
6095         (pass_manager::dump_passes): ...here.
6096         (register_pass): Move body of implementation into...
6097         (pass_manager::register_pass): ...here.
6098         (init_optimization_passes): Convert into...
6099         (pass_manager::pass_manager): ...constructor for new
6100         pass_manager class, and initialize the pass_lists array.
6101         (check_profile_consistency): Update for moves of globals to
6102         fields of pass_manager.
6103         (dump_profile_report): Move body of implementation into...
6104         (pass_manager::dump_profile_report): ...here.
6105         (ipa_write_summaries_1): Update for moves of pass lists from
6106         being globals to fields of pass_manager.
6107         (ipa_write_optimization_summaries): Likewise.
6108         (ipa_read_summaries):  Likewise.
6109         (ipa_read_optimization_summaries): Likewise.
6110         (execute_all_ipa_stmt_fixups): Likewise.
6112         * statistics.c (statistics_fini): Update for moves of globals to
6113         fields of pass_manager.
6115         * toplev.c (general_init): Replace call to
6116         init_optimization_passes with construction of the pass_manager
6117         instance.
6119         * tree-pass.h (all_passes): Remove, in favor of a field of the
6120         same name within the new class pass_manager.
6121         (all_small_ipa_passes): Likewise.
6122         (all_lowering_passes): Likewise.
6123         (all_regular_ipa_passes): Likewise.
6124         (all_lto_gen_passes): Likewise.
6125         (all_late_ipa_passes): Likewise.
6126         (passes_by_id): Likewise.
6127         (passes_by_id_size): Likewise.
6128         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6129         the new class pass_manager.
6130         (get_pass_for_id): Remove.
6132 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6134         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6135         configs.
6137 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6139         * arm.md (mulhi3): New expand pattern.
6141 2013-07-30  Jan Hubicka  <jh@suse.cz>
6142             Martin Liska  <marxin.liska@gmail.com>
6144         * profile.c (compute_value_histograms): Do not ICE when
6145         there is mismatch only on some counters.
6147 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6149         PR rtl-optimization/57637
6150         * function.c (move_insn_for_shrink_wrap): Also check the
6151         GEN set of the LIVE problem for the liveness analysis
6152         if it exists, otherwise give up.
6154 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6156         PR tree-optimization/57993
6157         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6158         replaced statement in the candidate table.
6159         (phi_add_costs): Return infinite cost when the hidden basis does
6160         not dominate all phis on which the candidate is dependent.
6161         (replace_one_candidate): Record replaced statement in the
6162         candidate table.
6164 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
6166         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6167         (ashlv2si3): New expander.
6168         (*ashlv2si3_i): New define_insn_and_split.
6169         * predicates.md (float_operation): Allow patterns with three
6170         basic sub-patterns.
6172         PR rtl-optimization/58021
6173         * mode-switching.c (create_pre_exit): Always split off preceding
6174         insns if we are not at the basic block head.
6176 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
6178         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6179         (UCLIBC_DYNAMIC_LINKER): New macro.
6180         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6181         `-mnan=2008'.
6182         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6183         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
6184         `-mnan=2008'.
6185         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6186         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6187         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6188         for SF and DF modes.  Use ieee_quad_format for TF mode.
6189         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6190         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6191         (mips_option_override): Handle `-mnan=legacy'.
6192         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6193         `-mabs=2008' and `-mnan=2008'.
6194         (OPTION_DEFAULT_SPECS): Add "nan" default.
6195         (ASM_SPEC): Handle `-mnan='.
6196         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6197         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6198         comment accordingly.
6199         (neg<mode>2): Likewise.
6200         * config/mips/mips.opt (mabs, mnan): New options.
6201         * doc/install.texi (Configuration): Document `--with-nan=' option.
6202         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6203         `-mnan=' options.
6204         (MIPS Options): Document them.
6205         * config.gcc <mips*-*-*>: Handle `--with-nan='.
6206         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6207         * configure: Regenerate.
6208         * config.in: Regenerate.
6210 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6212         * config/i386/i386.md (float post-reload splitters): Do not check
6213         for subregs of SSE registers.
6215 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6216             H.J. Lu  <hongjiu.lu@intel.com>
6218         PR target/57954
6219         PR target/57988
6220         * config/i386/i386.md (post-reload splitter
6221         to avoid partial SSE reg dependency stalls): New pattern.
6223 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6225         * config/s390/s390.md ("movcc"): Swap load and store instructions.
6227 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
6229         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6230         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
6232 2013-07-26  Cary Coutant  <ccoutant@google.com>
6234         * dwarf2out.c (die_checksum_ordered): Don't include template
6235         instantiations in signature.
6236         (is_template_parameter): New function.
6237         (is_template_instantiation): New function.
6238         (generate_skeleton_bottom_up): Don't include template instantiations
6239         in type unit DIE.
6240         (generate_skeleton): Likewise.
6241         (break_out_comdat_types): Move recursive call to break out nested
6242         types earlier.
6243         (prune_unused_types_mark_generic_parms_dies): Call
6244         is_template_parameter.
6246 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
6248         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
6249         uses vector registers.
6250         * config/aarch64/iterators.md: Add attributes rtn and vas.
6252 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6253             Richard Earnshaw  <richard.earnshaw@arm.com>
6255         * combine.c (simplify_comparison): Re-canonicalize operands
6256         where appropriate.
6257         * config/arm/arm.md (movcond_addsi): New splitter.
6259 2013-07-25  Sterling Augustine  <saugustine@google.com>
6261         * dwarf2out.c (size_of_pubnames): Move code to...
6262         (include_pubname_in_output): ...here.  New.
6263         (want_pubnames): Rearrange.
6264         (output_pubnames): Call include_pubname_in_output.  Move assertion.
6266 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6268         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
6270 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6272         PR target/38836
6273         * doc/extend.texi: Remove obsolete builtins. Fix
6274         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
6276 2013-07-25  Jan Hubicka  <jh@suse.cz>
6278         * cgraph.c (release_function_body): Break out from ...
6279         (cgraph_release_function_body): ... this one; also release DECL_RESULT
6280         and DECL_ARGUMENTS.
6281         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
6282         old_tree in the map.
6283         (create_specialized_node): Update.
6284         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
6285         into index.
6286         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
6287         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
6288         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
6289         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
6290         DECL_RESULT.
6292 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6294         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
6295         addsi3_carryin_alt2_<optab>): Correct output template.
6297 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6299         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
6300         Adjust for arm_restrict_it.
6301         Remove trailing whitespace.
6303 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
6305         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
6306         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
6308         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
6310 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
6312         PR rtl-optimization/57960
6313         * lra-constraints.c (process_alt_operands): Use the right mode
6314         when checking strict_low.
6316 2013-07-25  Jan Hubicka  <jh@suse.cz>
6318         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
6319         * cgraph.c (cgraph_remove_node): Do not release function body
6320         when in cgraph streaming.
6321         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
6322         in other partitions are not considered reachable; fix handling of
6323         clones.
6325 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6327         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
6329 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6331         PR target/19599
6332         PR target/57731
6333         PR target/57837
6334         * config/arm/arm.md ("*sibcall_insn): Replace use of
6335         Ss with US. Adjust output for v5 and v4t.
6336         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
6338         * config/arm/constraints.md ("Ss"): Rename to US.
6340 2013-07-25  Terry Guo  <terry.guo@arm.com>
6342         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
6343         shift_add/shift_sub0/shift_sub1 RTXs.
6345 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
6346             Anton Blanchard  <anton@au1.ibm.com>
6348         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
6349         (altivec_vpks<VI_char>ss): Likewise.
6350         (altivec_vpks<VI_char>us): Likewise.
6351         (altivec_vpku<VI_char>us): Likewise.
6352         (altivec_vpku<VI_char>um): Likewise.
6354 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
6356         Introduce context class.
6358         * Makefile.in (CONTEXT_H): New.
6359         (OBJS): Add context.o.
6360         (toplev.o): Add CONTEXT_H to dependencies.
6361         (context.o): New.
6363         * toplev.c (general_init): Create the singleton gcc::context instance.
6365         * context.c: New.
6367         * context.h: New.
6369 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
6371         PR rtl-optimization/57968
6372         * mode-switching.c (create_pre_exit): Allow instructions that
6373         don't set a return register to need a non-exit mode.
6375 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6376             Anton Blanchard  <anton@au1.ibm.com>
6378         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
6379         operands to vperm for little endian.
6380         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
6381         of lvsl to create the control mask for a vperm for little endian.
6383 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6384             Anton Blanchard  <anton@au1.ibm.com>
6386         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
6387         two operands for little-endian.
6389 2013-07-23  Steve Ellcey  <sellcey@mips.com>
6391         * config/mips/mips.c (mips_case_values_threshold): New.
6392         (TARGET_CASE_VALUES_THRESHOLD): Define.
6394 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6395             Anton Blanchard  <anton@au1.ibm.com>
6397         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
6398         selection of field for vector splat in little endian mode.
6400 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6402         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
6403         expanders to rs6000.md.
6404         (ior<mode>3): Likewise.
6405         (and<mode>3): Likewise.
6406         (one_cmpl<mode>2): Likewise.
6407         (nor<mode>3): Likewise.
6408         (andc<mode>3): Likewise.
6409         (eqv<mode>3): Likewise.
6410         (nand<mode>3): Likewise.
6411         (orc<mode>3): Likewise.
6413         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
6414         declaration.
6416         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
6417         to split multi-word logical operations.
6418         (rs6000_split_logical_di): Likewise.
6419         (rs6000_split_logical): Likewise.
6421         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
6422         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
6423         and allow TImode operations in 32-bit.
6424         (vsx_and<mode>3_64bit): Likewise.
6425         (vsx_ior<mode>3_32bit): Likewise.
6426         (vsx_ior<mode>3_64bit): Likewise.
6427         (vsx_xor<mode>3_32bit): Likewise.
6428         (vsx_xor<mode>3_64bit): Likewise.
6429         (vsx_one_cmpl<mode>2_32bit): Likewise.
6430         (vsx_one_cmpl<mode>2_64bit): Likewise.
6431         (vsx_nor<mode>3_32bit): Likewise.
6432         (vsx_nor<mode>3_64bit): Likewise.
6433         (vsx_andc<mode>3_32bit): Likewise.
6434         (vsx_andc<mode>3_64bit): Likewise.
6435         (vsx_eqv<mode>3_32bit): Likewise.
6436         (vsx_eqv<mode>3_64bit): Likewise.
6437         (vsx_nand<mode>3_32bit): Likewise.
6438         (vsx_nand<mode>3_64bit): Likewise.
6439         (vsx_orc<mode>3_32bit): Likewise.
6440         (vsx_orc<mode>3_64bit): Likewise.
6442         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
6443         logical types in GPRs.
6445         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
6446         logical insns to rs6000.md, and allow TImode operations in
6447         32-bit.
6448         (altivec_ior<mode>3): Likewise.
6449         (altivec_xor<mode>3): Likewise.
6450         (altivec_one_cmpl<mode>2): Likewise.
6451         (altivec_nor<mode>3): Likewise.
6452         (altivec_andc<mode>3): Likewise.
6454         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
6455         attributes for moving the 128-bit logical operations into
6456         rs6000.md.
6457         (BOOL_REGS_OUTPUT): Likewise.
6458         (BOOL_REGS_OP1): Likewise.
6459         (BOOL_REGS_OP2): Likewise.
6460         (BOOL_REGS_UNARY): Likewise.
6461         (BOOL_REGS_AND_CR0): Likewise.
6462         (one_cmpl<mode>2): Add support for DI logical operations on
6463         32-bit, splitting the operations to 32-bit.
6464         (anddi3): Likewise.
6465         (iordi3): Likewise.
6466         (xordi3): Likewise.
6467         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
6468         changes to combine the 32/64-bit code, allow logical operations on
6469         TI mode in 32-bit, and to use similar match_operator patterns like
6470         scalar mode uses.  Combine the Altivec and VSX code for logical
6471         operations, and move it here.
6472         (ior<mode>3, 128-bit types): Likewise.
6473         (xor<mode>3, 128-bit types): Likewise.
6474         (one_cmpl<mode>3, 128-bit types): Likewise.
6475         (nor<mode>3, 128-bit types): Likewise.
6476         (andc<mode>3, 128-bit types): Likewise.
6477         (eqv<mode>3, 128-bit types): Likewise.
6478         (nand<mode>3, 128-bit types): Likewise.
6479         (orc<mode>3, 128-bit types): Likewise.
6480         (and<mode>3_internal): Likewise.
6481         (bool<mode>3_internal): Likewise.
6482         (boolc<mode>3_internal1): Likewise.
6483         (boolc<mode>3_internal2): Likewise.
6484         (boolcc<mode>3_internal1): Likewise.
6485         (boolcc<mode>3_internal2): Likewise.
6486         (eqv<mode>3_internal1): Likewise.
6487         (eqv<mode>3_internal2): Likewise.
6488         (one_cmpl1<mode>3_internal): Likewise.
6490 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6492         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6493         Rename flag_stack_usage to flag_stack_usage_info.
6495 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6497         * config/microblaze/sync.md: New file.
6498         * config/microblaze/microblaze.md: Include sync.md
6499         * config/microblaze/microblaze.c: Add print_operand 'y'.
6500         * config/microblaze/constraints.md: Add memory_contraint
6501         'Q' which is a single register.
6503 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
6505         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
6507 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6509         * reload.c (find_reloads): Exit loop once we find this operand
6510         cannot be reloaded somehow for this alternative.
6512         * reload.c (find_reloads): Exit loop once we find a hard register.
6514         * rtlanal.c (computed_jump_p): Exit loop once we find label
6515         reference is used.
6517         * i386.c (ix86_pad_returns): Exit loop after setting replace.
6519         * cfgloopmanip.c (remove_path): Exit loop after setting
6520         irred_invalidated.
6522         * gensupport.c (subst_dup): Avoid loop if code is not
6523         MATCH_DUP nor MATCH_OP_DUP.
6525 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
6527         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
6529 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6531         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
6532         true for SP_REGNUM if mode == ptr_mode.
6533         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
6534         with value R0_REGNUM + 31.
6536 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6538         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
6539         pad pointer-typed argument downward.
6541 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6543         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
6544         and __ILP32__ when the ILP32 model is in use.
6546 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6548         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
6549         (aarch64_load_symref_appropriately): In the case of
6550         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
6551         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
6552         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
6553         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
6554         if 'mode' doesn't equal to 'ptr_mode'.
6555         (aarch64_output_mi_thunk): Add an assertion on the alignment of
6556         'vcall_offset'; change to call aarch64_emit_move differently depending
6557         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
6558         to calculate the upper bound of 'vcall_offset'.
6559         (aarch64_cannot_force_const_mem): Change to also return true if
6560         mode != ptr_mode.
6561         (aarch64_legitimize_reload_address): In the case of large
6562         displacements, add new local variable 'xmode' and an assertion
6563         based on it; change to use 'xmode' to generate the new rtx and
6564         reload.
6565         (aarch64_asm_trampoline_template): Change to generate the template
6566         differently depending on TARGET_ILP32 or not; change to use
6567         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
6568         (aarch64_trampoline_size): Removed.
6569         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
6570         and replace immediate literals with it.  Change to use 'ptr_mode'
6571         instead of 'DImode' and call convert_memory_address if the mode
6572         of 'fnaddr' doesn't equal to 'ptr_mode'.
6573         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
6574         to output symbol.
6575         (aarch64_elf_asm_destructor): Likewise.
6576         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
6577         on TARGET_ILP32 instead of aarch64_trampoline_size.
6578         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
6579         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
6580         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
6581         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
6582         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
6583         (storewb_pair<GPI:mode>_<P:mode>): ... this.
6584         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
6585         depending on the value of 'mode'.
6586         (add_losym_<mode>): New.
6587         (ldr_got_small_<mode>): New, based on ldr_got_small.
6588         (ldr_got_small): Remove.
6589         (ldr_got_small_sidi): New.
6590         * config/aarch64/iterators.md (P): New.
6591         (PTR): Change to 'ptr_mode' in the condition.
6593 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6595         * config.gcc (aarch64*-*-*): Support --with-abi.
6596         (aarch64*-*-elf): Support --with-multilib-list.
6597         (aarch64*-*-linux*): Likewise.
6598         (supported_defaults): Add abi to aarch64*-*-*.
6599         * configure.ac: Mention AArch64 for --with-multilib-list.
6600         * configure: Re-generated.
6601         * config/aarch64/biarchilp32.h: New file.
6602         * config/aarch64/biarchlp64.h: New file.
6603         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
6604         (ABI_SPEC): Ditto.
6605         (MULTILIB_DEFAULTS): Ditto.
6606         (DRIVER_SELF_SPECS): Ditto.
6607         (ASM_SPEC): Update to also substitute -mabi.
6608         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
6609         file whose name depends on -mabi= and -mbig-endian.
6610         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
6611         TARGET_ILP32.
6612         (POINTER_SIZE): New define.
6613         (POINTERS_EXTEND_UNSIGNED): Ditto.
6614         (enum aarch64_abi_type): New enumeration tag.
6615         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
6616         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
6617         (TARGET_ILP32): New define.
6618         * config/aarch64/aarch64.opt (mabi): New.
6619         (aarch64_abi): New.
6620         (ilp32, lp64): New values for -mabi.
6621         * config/aarch64/t-aarch64 (comma): New define.
6622         (MULTILIB_OPTIONS): Ditto.
6623         (MULTILIB_DIRNAMES): Ditto.
6624         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
6625         * doc/invoke.texi: Document -mabi for AArch64.
6627 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
6629         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
6631 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6632             Anton Blanchard  <anton@au1.ibm.com>
6634         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
6635         endianness when selecting field to splat.
6637 2013-07-22  Eric Christopher  <echristo@gmail.com>
6639         * dwarf2out.c (die_odr_checksum): New function to use
6640         CHECKSUM_ macros and ULEB128 for DIE tag.
6641         (generate_type_signature): Use.
6643 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
6645         * config.gcc (sparc*-*-*): Accept leon3 processor.
6646         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
6647         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
6648         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
6649         * config/sparc/sparc.opt (enum processor_type): Add leon3.
6650         (mfix-ut699): Adjust comment.
6651         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
6652         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
6653         (CPP_CPU_SPEC): Likewise.
6654         (ASM_CPU_SPEC): Likewise.
6655         * config/sparc/sparc.c (leon3_cost): New constant.
6656         (sparc_option_override): Add leon3 support.
6657         (mem_ref): New function.
6658         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
6659         (sparc_do_work_around_errata): Look into the instruction in the delay
6660         slot and adjust accordingly.  Add fix for the data cache nullify issues
6661         of the UT699.  Change insertion position for the NOP.
6662         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
6663         (leon3_load): New reservation.
6664         (leon_store): Bump latency to 2.
6665         (grfpu): New automaton.
6666         (grfpu_alu): New unit.
6667         (grfpu_ds): Likewise.
6668         (leon_fp_alu): Adjust.
6669         (leon_fp_mult): Delete.
6670         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
6671         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
6672         * config/sparc/sparc.md (cpu): Add leon3.
6673         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
6674         (swapsi): Likewise.
6675         (atomic_test_and_set): Likewise.
6676         (ldstub): Likewise.
6678 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
6680         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
6681         default for R5900 targets.
6682         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
6683         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
6684         * config/mips/mips.c (mips_option_override): Report an error for
6685         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
6686         for -march=r5900 -mhard-float.
6688 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6690         * df-problems.c (can_move_insns_across): Exit loop once we
6691         find a non-fixed, non-global register.
6693         * ipa-pure-const.c (propagate_nothrow): Exit loop after
6694         setting can_throw.
6696         * omega.c (omega_eliminate_red): Break after setting red_found.
6697         (omega_problem_has_red_equations): Similarly after setting found.
6698         (omega_query_variable): Similarly after setting coupled.
6700 2013-07-22  Marek Polacek  <polacek@redhat.com>
6702         * gimplify.c: Don't include gimple.h twice.
6704 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6706         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
6707         instead of TARGET_THUMB1.
6708         (Pz): New constraint.
6709         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
6710         encodings.
6711         (compare_negsi_si): Likewise.
6712         (compare_addsi2_op0): Likewise.
6713         (compare_addsi2_op1): Likewise.
6714         (addsi3_carryin_<optab>): Likewise.
6715         (addsi3_carryin_alt2_<optab>): Likewise.
6716         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
6717         for arm_restrict_it.
6718         (subsi3_carryin): Likewise.
6719         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
6720         (minmax_arithsi): Disable for arm_restrict_it.
6721         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
6722         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
6723         (satsi_<SAT:code>_shift): Likewise.
6724         (arm_shiftsi3): Add alternative for 16-bit encoding.
6725         (arm32_movhf): Disable for arm_restrict_it.
6726         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
6727         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
6729 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
6731         * config/arm/arm.md (attribute "insn"): Delete.
6732         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
6733         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
6734         (not_shiftsi): Update for attribute change.
6735         (not_shiftsi_compare0): Likewise.
6736         (not_shiftsi_compare0_scratch): Likewise.
6737         (arm_one_cmplsi2): Likewise.
6738         (thumb1_one_cmplsi2): Likewise.
6739         (notsi_compare0): Likewise.
6740         (notsi_compare0_scratch): Likewise.
6741         (thumb1_movdi_insn): Likewise.
6742         (arm_movsi_insn): Likewise.
6743         (movhi_insn_arch4): Likewise.
6744         (movhi_bytes): Likewise.
6745         (arm_movqi_insn): Likewise.
6746         (thumb1_movqi_insn): Likewise.
6747         (arm32_movhf): Likewise.
6748         (thumb1_movhf): Likewise.
6749         (arm_movsf_soft_insn): Likewise.
6750         (thumb1_movsf_insn): Likewise.
6751         (thumb_movdf_insn): Likewise.
6752         (movsicc_insn): Likewise.
6753         (movsfcc_soft_insn): Likewise.
6754         (and_scc): Likewise.
6755         (cond_move): Likewise.
6756         (if_move_not): Likewise.
6757         (if_not_move): Likewise.
6758         (if_shift_move): Likewise.
6759         (if_move_shift): Likewise.
6760         (if_shift_shift): Likewise.
6761         (if_not_arith): Likewise.
6762         (if_arith_not): Likewise.
6763         (cond_move_not): Likewise.
6764         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
6765         (neon_mov<mode>): Likewise.
6766         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
6767         (thumb2_movsi_vfp): Likewise.
6768         (movsf_vfp): Likewise.
6769         (thumb2_movsf_vfp): Likewise.
6770         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
6771         change.
6772         (cortexa7_older_only): Likewise.
6773         (cortexa7_younger): Likewise.
6774         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
6775         (1020alu_shift_op): Likewise.
6776         (1020alu_shift_reg_op): Likewise.
6777         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
6778         (alu_shift_op): Likewise.
6779         (alu_shift_reg_op): Likewise.
6780         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
6781         (11_alu_shift_op): Likewise.
6782         (11_alu_shift_reg_op): Likewise.
6783         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
6784         (9_alu_shift_reg_op): Likewise.
6785         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
6786         change.
6787         (cortex_a15_alu_shift): Likewise.
6788         (cortex_a15_alu_shift_reg): Likewise.
6789         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
6790         (cortex_a5_alu_shift): Likewise.
6791         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
6792         change.
6793         (cortex_a53_alu_shift): Likewise.
6794         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
6795         change.
6796         (cortex_a7_alu_reg): Likewise.
6797         (cortex_a7_alu_shift): Likewise.
6798         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6799         (cortex_a8_alu_shift): Likewise.
6800         (cortex_a8_alu_shift_reg): Likewise.
6801         (cortex_a8_mov): Likewise.
6802         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
6803         (cortex_a9_dp_shift): Likewise.
6804         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
6805         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
6806         (cortex_r4_mov): Likewise.
6807         (cortex_r4_alu_shift): Likewise.
6808         (cortex_r4_alu_shift_reg): Likewise.
6809         * config/arm/fa526.md (526_alu_op): Update for attribute change.
6810         (526_alu_shift_op): Likewise.
6811         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
6812         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
6813         (626te_alu_shift_op): Likewise.
6814         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
6815         (726te_alu_op): Likewise.
6816         (726te_alu_shift_op): Likewise.
6817         (726te_alu_shift_reg_op): Likewise.
6818         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
6819         (mp626_alu_shift_op): Likewise.
6820         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
6821         (pj4_alu_e1_conds): Likewise.
6822         (pj4_alu): Likewise.
6823         (pj4_alu_conds): Likewise.
6824         (pj4_shift): Likewise.
6825         (pj4_shift_conds): Likewise.
6826         (pj4_alu_shift): Likewise.
6827         (pj4_alu_shift_conds): Likewise.
6829 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6831         * config/arm/predicates.md (shiftable_operator_strict_it):
6832         New predicate.
6833         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
6834         Disable cond_exec version for arm_restrict_it.
6835         (thumb2_smaxsi3): Convert to generate cond_exec.
6836         (thumb2_sminsi3): Likewise.
6837         (thumb32_umaxsi3): Likewise.
6838         (thumb2_uminsi3): Likewise.
6839         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
6840         (thumb2_neg_abssi2): Likewise.
6841         (thumb2_mov_scc): Add alternative for 16-bit encoding.
6842         (thumb2_movsicc_insn): Adjust alternatives.
6843         (thumb2_mov_negscc): Disable for arm_restrict_it.
6844         (thumb2_mov_negscc_strict_it): New pattern.
6845         (thumb2_mov_notscc_strict_it): New pattern.
6846         (thumb2_mov_notscc): Disable for arm_restrict_it.
6847         (thumb2_ior_scc): Likewise.
6848         (thumb2_ior_scc_strict_it): New pattern.
6849         (thumb2_cond_move): Adjust for arm_restrict_it.
6850         (thumb2_cond_arith): Disable for arm_restrict_it.
6851         (thumb2_cond_arith_strict_it): New pattern.
6852         (thumb2_cond_sub): Adjust for arm_restrict_it.
6853         (thumb2_movcond): Likewise.
6854         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
6855         (thumb2_zero_extendhisi2_v6): Likewise.
6856         (thumb2_zero_extendqisi2_v6): Likewise.
6857         (orsi_notsi_si): Likewise.
6858         (orsi_not_shiftsi_si): Likewise.
6860 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
6862         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
6863         instruction sequence is 1 byte shorter.
6865 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
6867         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
6868         it is not needed after split.
6870 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
6872         PR target/51784
6873         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
6874         second label for nonlocal goto receivers. Don't output pic base labels
6875         unless we're producing PIC; mark that action unreachable().
6876         (ix86_save_reg): If the function contains a nonlocal label, save the
6877         PIC base reg.
6878         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
6879         * config/darwin.c (emitted_pic_label_num): New GTY.
6880         (update_pic_label_number_if_needed): New.
6881         (machopic_output_function_base_name): Adjust for nonlocal receiver
6882         case.
6883         (machopic_should_output_picbase_label): New.
6884         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
6885         (nonlocal_goto_receiver): New insn and split.
6887 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
6889         * config/aarch64/aarch64-builtins.c
6890         (aarch64_fold_builtin): Fold abs in all modes.
6891         * config/aarch64/aarch64-simd-builtins.def
6892         (abs): Enable for all modes.
6893         * config/aarch64/arm_neon.h
6894         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
6895         (vabs_f64): Add missing intrinsic.
6897 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
6899         * config/aarch64/arm_neon.h (vabs_s64): New function
6901 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
6903         PR target/57516
6904         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
6905         * config/avr/avr.md (adjust_len): Add `round'.
6906         * config/avr/avr-protos.h (avr_out_round): New prototype.
6907         (avr_out_plus): Add `out_label' argument.
6908         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
6909         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
6910         Handle the case where `insn' is just a pattern.
6911         (avr_out_bitop): Handle the case where `insn' is just a pattern.
6912         (avr_out_round): New function.
6913         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
6915 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
6917         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6918         Add check for flag_stack_usage to handle -fstack-usage support
6920 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
6922         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
6923         interaction for new Power8 flags and VSX.
6925 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
6927         PR middle-end/57698
6928         * tree-inline.c (expand_call_inline): Emit errors during
6929         early_inlining only if optimization is not turned on.
6931 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6933         * passes.def: New.
6935         * passes.c (init_optimization_passes): Move the construction of
6936         the pass hierarchy into a new passes.def file.
6938         * Makefile.in (passes.o): Add dependency on passes.def.
6940 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6942         * passes.c (init_optimization_passes): Introduce macros for
6943         constructing the tree of passes (INSERT_PASSES_AFTER,
6944         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
6945         TERMINATE_PASS_LIST).
6947 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
6948             Wei Mi  <wmi@google.com>
6950         PR rtl-optimization/57878
6951         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
6952         top.
6953         (reload_pseudo_compare_func): Check nregs first for reload
6954         pseudos.
6956 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
6958         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
6960 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
6962         * read-rtl.c (validate_const_int): Once an invalid character is
6963         seen, quit the loop.
6965         * gengtype.c (write_roots): Similarly once we find the "deletable"
6966         or "if_marked" option.
6968 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6970         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
6971         "xtab" and "sat".  Move value "clz" from here to ...
6972         (attriubte "type"): ... here.
6973         (satsi_<SAT:code>): Delete "insn" attribute.
6974         (satsi_<SAT:code>_shift): Likewise.
6975         (arm_zero_extendqisi2addsi): Likewise.
6976         (arm_extendqisi2addsi): Likewise.
6977         (clzsi2): Update for attribute changes.
6978         (rbitsi2): Likewise.
6979         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
6980         attribute.
6981         (arm_usatsihi): Likewise.
6982         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
6984 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
6986         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
6987         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
6988         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
6989         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
6990         in alphabetical order.
6991         (attribute "core_cycles"): Update for attribute changes.
6992         (arm_addsi3): Likewise.
6993         (addsi3_compare0): Likewise.
6994         (addsi3_compare0_scratch): Likewise.
6995         (addsi3_compare_op1): Likewise.
6996         (addsi3_compare_op2): Likewise.
6997         (compare_addsi2_op0): Likewise.
6998         (compare_addsi2_op1): Likewise.
6999         (addsi3_carryin_shift_<optab>): Likewise.
7000         (subsi3_carryin_shift): Likewise.
7001         (rsbsi3_carryin_shift): Likewise.
7002         (arm_subsi3_insn): Likewise.
7003         (subsi3_compare0): Likewise.
7004         (subsi3_compare): Likewise.
7005         (arm_andsi3_insn): Likewise.
7006         (thumb1_andsi3_insn): Likewise.
7007         (andsi3_compare0): Likewise.
7008         (andsi3_compare0_scratch): Likewise.
7009         (zeroextractsi_compare0_scratch
7010         (andsi_not_shiftsi_si): Likewise.
7011         (iorsi3_insn): Likewise.
7012         (iorsi3_compare0): Likewise.
7013         (iorsi3_compare0_scratch): Likewise.
7014         (arm_xorsi3): Likewise.
7015         (thumb1_xorsi3_insn): Likewise.
7016         (xorsi3_compare0): Likewise.
7017         (xorsi3_compare0_scratch): Likewise.
7018         (satsi_<SAT:code>_shift): Likewise.
7019         (rrx): Likewise.
7020         (arm_shiftsi3): Likewise.
7021         (shiftsi3_compare0): Likewise.
7022         (not_shiftsi): Likewise.
7023         (not_shiftsi_compare0): Likewise.
7024         (not_shiftsi_compare0_scratch): Likewise.
7025         (arm_one_cmplsi2): Likewise.
7026         (thumb_one_complsi2): Likewise.
7027         (notsi_compare0): Likewise.
7028         (notsi_compare0_scratch): Likewise.
7029         (thumb1_zero_extendhisi2): Likewise.
7030         (arm_zero_extendhisi2): Likewise.
7031         (arm_zero_extendhisi2_v6): Likewise.
7032         (arm_zero_extendhisi2addsi): Likewise.
7033         (thumb1_zero_extendqisi2): Likewise.
7034         (thumb1_zero_extendqisi2_v6): Likewise.
7035         (arm_zero_extendqisi2): Likewise.
7036         (arm_zero_extendqisi2_v6): Likewise.
7037         (arm_zero_extendqisi2addsi): Likewise.
7038         (thumb1_extendhisi2): Likewise.
7039         (arm_extendhisi2): Likewise.
7040         (arm_extendhisi2_v6): Likewise.
7041         (arm_extendqisi): Likewise.
7042         (arm_extendqisi_v6): Likewise.
7043         (arm_extendqisi2addsi): Likewise.
7044         (thumb1_extendqisi2): Likewise.
7045         (thumb1_movdi_insn): Likewise.
7046         (arm_movsi_insn): Likewise.
7047         (movsi_compare0): Likewise.
7048         (movhi_insn_arch4): Likewise.
7049         (movhi_bytes): Likewise.
7050         (arm_movqi_insn): Likewise.
7051         (thumb1_movqi_insn): Likewise.
7052         (arm32_movhf): Likewise.
7053         (thumb1_movhf): Likewise.
7054         (arm_movsf_soft_insn): Likewise.
7055         (thumb1_movsf_insn): Likewise.
7056         (movdf_soft_insn): Likewise.
7057         (thumb_movdf_insn): Likewise.
7058         (arm_cmpsi_insn): Likewise.
7059         (cmpsi_shiftsi): Likewise.
7060         (cmpsi_shiftsi_swp): Likewise.
7061         (arm_cmpsi_negshiftsi_si): Likewise.
7062         (movsicc_insn): Likewise.
7063         (movsfcc_soft_insn): Likewise.
7064         (arith_shiftsi): Likewise.
7065         (arith_shiftsi_compare0
7066         (arith_shiftsi_compare0_scratch
7067         (sub_shiftsi): Likewise.
7068         (sub_shiftsi_compare0
7069         (sub_shiftsi_compare0_scratch
7070         (and_scc): Likewise.
7071         (cond_move): Likewise.
7072         (if_plus_move): Likewise.
7073         (if_move_plus): Likewise.
7074         (if_move_not): Likewise.
7075         (if_not_move): Likewise.
7076         (if_shift_move): Likewise.
7077         (if_move_shift): Likewise.
7078         (if_shift_shift): Likewise.
7079         (if_not_arith): Likewise.
7080         (if_arith_not): Likewise.
7081         (cond_move_not): Likewise.
7082         (thumb1_ashlsi3): Set type attribute.
7083         (thumb1_ashrsi3): Likewise.
7084         (thumb1_lshrsi3): Likewise.
7085         (thumb1_rotrsi3): Likewise.
7086         (shiftsi3_compare0_scratch): Likewise.
7087         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7088         (neon_mov<mode>): Likewise.
7089         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7090         attribute changes.
7091         (thumb2_movsi_insn): Likewise.
7092         (thumb2_cmpsi_neg_shiftsi): Likewise.
7093         (thumb2_extendqisi_v6): Likewise.
7094         (thumb2_zero_extendhisi2_v6): Likewise.
7095         (thumb2_zero_extendqisi2_v6): Likewise.
7096         (thumb2_shiftsi3_short): Likewise.
7097         (thumb2_addsi3_compare0_scratch): Likewise.
7098         (orsi_not_shiftsi_si): Likewise.
7099         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7100         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7101         changes.
7102         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7103         (1020alu_shift_op): Likewise.
7104         (1020alu_shift_reg_op): Likewise.
7105         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7106         (alu_shift_op): Likewise.
7107         (alu_shift_reg_op): Likewise.
7108         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7109         (11_alu_shift_op): Likewise.
7110         (11_alu_shift_reg_op): Likewise.
7111         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7112         (9_alu_shift_reg_op): Likewise.
7113         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7114         changes.
7115         (cortex_a15_alu_shift): Likewise.
7116         (cortex_a15_alu_shift_reg): Likewise.
7117         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7118         changes.
7119         (cortex_a5_alu_shift): Likewise.
7120         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7121         changes.
7122         (cortex_a53_alu_shift): Likewise.
7123         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7124         changes.
7125         (cortex_a7_alu_reg): Likewise.
7126         (cortex_a7_alu_shift): Likewise.
7127         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7128         changes.
7129         (cortex_a8_alu_shift): Likewise.
7130         (cortex_a8_alu_shift_reg): Likewise.
7131         (cortex_a8_mov): Likewise.
7132         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7133         (cortex_a9_dp_shift): Likewise.
7134         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7135         changes.
7136         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7137         changes.
7138         (cortex_r4_mov): Likewise.
7139         (cortex_r4_alu_shift): Likewise.
7140         (cortex_r4_alu_shift_reg): Likewise.
7141         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7142         (526_alu_shift_op): Likewise.
7143         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7144         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7145         (626te_alu_shift_op): Likewise.
7146         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7147         (726te_alu_op): Likewise.
7148         (726te_alu_shift_op): Likewise.
7149         (726te_alu_shift_reg_op): Likewise.
7150         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7151         (mp626_alu_shift_op): Likewise.
7152         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7153         (pj4_alu_e1_conds): Likewise.
7154         (pj4_alu): Likewise.
7155         (pj4_alu_conds): Likewise.
7156         (pj4_shift): Likewise.
7157         (pj4_shift_conds): Likewise.
7158         (pj4_alu_shift): Likewise.
7159         (pj4_alu_shift_conds): Likewise.
7160         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7161         changes.
7162         (cortexa7_older_only): Likewise.
7163         (cortexa7_younger): Likewise.
7165 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7167         * ipa-pure-const.c (generate_summary): Rename to...
7168         (pure_const_generate_summary): ... this.
7170 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
7172         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7174 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
7176         PR target/57909
7177         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7178         usage in HI mode.
7180 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7182         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7183         enabled without -march=zEC12.
7184         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7185         flags to be set.
7187 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7189         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7190         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7191         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7192         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7193         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7194         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7195         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7196         ISA_HAS_FP4.
7197         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7198         and ISA_HAS_NMADD3_NMSUB3.
7199         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7200         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7201         (nmsub4<mode>, nmsub3<mode>): Likewise.
7202         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7204 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7206         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7207         TARGET_MIPS5400 checking.
7209 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
7210             Peter Bergner  <bergner@vnet.ibm.com>
7212         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7213         registers in the comment.
7214         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7215         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7216         rather than FIRST_PSEUDO_REGISTERS.
7218 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
7220         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7221         enable extra ISA flags with TARGET_HTM.
7223 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7225         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7226         Fix comment typos.
7228 2013-07-15  Cong Hou  <congh@google.com>
7230         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7231         in compare function for sorting.
7233 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
7235         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
7236         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
7237         * config/rs6000/rs6000.opt: Add -mhtm option.
7238         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
7239         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
7240         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7241         __HTM__ if the HTM instructions are available.
7242         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
7243         htm_spr_reg_operand): New define_predicates.
7244         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
7245         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
7246         Include htm.md.
7247         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
7248         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
7249         HTM builtin functions.
7250         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
7251         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
7252         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
7253         (rs6000_builtin_mask_calculate): Likewise.
7254         (rs6000_option_override_internal): Likewise.
7255         (bdesc_htm): Add new HTM builtin support.
7256         (htm_spr_num): New function.
7257         (htm_spr_regno): Likewise.
7258         (rs6000_htm_spr_icode): Likewise.
7259         (htm_expand_builtin): Likewise.
7260         (htm_init_builtins): Likewise.
7261         (rs6000_expand_builtin): Add support for HTM builtin functions.
7262         (rs6000_init_builtins): Likewise.
7263         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
7264         option.
7265         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
7266         (TARGET_HTM, MASK_HTM): Define macros.
7267         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
7268         (FIXED_REGISTERS): Likewise.
7269         (CALL_USED_REGISTERS): Likewise.
7270         (CALL_REALLY_USED_REGISTERS): Likewise.
7271         (REG_ALLOC_ORDER): Likewise.
7272         (enum reg_class): Likewise.
7273         (REG_CLASS_NAMES): Likewise.
7274         (REG_CLASS_CONTENTS): Likewise.
7275         (REGISTER_NAMES): Likewise.
7276         (ADDITIONAL_REGISTER_NAMES): Likewise.
7277         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
7278         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
7279         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
7280         * config/rs6000/htm.md: New file.
7281         * config/rs6000/htmintrin.h: New file.
7282         * config/rs6000/htmxlintrin.h: New file.
7284 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7286         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7287         Define SYMBOL_TINY_GOT, update comment.
7288         * config/aarch64/aarch64.c
7289         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
7290         (aarch64_expand_mov_immediate): Likewise.
7291         (aarch64_print_operand): Likewise.
7292         (aarch64_classify_symbol): Likewise.
7293         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
7294         (ldr_got_tiny): Define.
7296 2013-07-13  Tobias Grosser  <tobias@grosser.es>
7298         PR tree-optimization/54094
7299         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
7300           scheduling dimension for the parallelism check from the polyhedral
7301           information in the AST.
7302         * graphite-dependences.c (carries_deps): Do not assume the schedule is
7303           in 2D + 1 form.
7305 2013-07-13  Jason Merrill  <jason@redhat.com>
7307         * print-tree.c (debug_vec_tree): Use debug_raw.
7308         (debug_raw (vec<tree, va_gc> &)): New.
7309         (debug_raw (vec<tree, va_gc> *)): New.
7310         * tree.h: Declare them.
7312 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
7314         * ifcvt.c (ifcvt_after_combine): New static variable.
7315         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
7316         for size.
7317         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
7318         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
7319         rest_of_handle_if_after_reload): Pass new argument for if_convert.
7321 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
7323         * config/mips/mips.c (mips_expand_call): Remove empty statement.
7325 2013-07-12  Michael Matz  <matz@suse.de>
7327         PR middle-end/55771
7328         * convert.c (convert_to_real): Reject non-float inner types.
7330 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
7332         * config/aarch64/aarch64-protos.h
7333         (aarch64_simd_immediate_valid_for_move): Remove.
7334         * config/aarch64/aarch64.c (simd_immediate_info): New member.
7335         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
7336         cases.
7337         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
7339 2013-07-11  Steve Ellcey  <sellcey@mips.com>
7341         * config/mips/mips.c (mips_conditional_register_usage): Do not
7342         use t[0-7] registers in MIPS16 mode when optimizing for size.
7344 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
7346         * config/i386/i386.c (dispatch_function_versions): Fix array
7347         indexing of function_version_info to match actual_versions.
7349 2013-07-11  Teresa Johnson  <tejohnson@google.com>
7351         * vec.h (struct va_gc): Move release out-of-line.
7352         (va_gc::release): Call ggc_free on released vec.
7354 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7356         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
7357         Require GOT register as additional operand in UNSPEC.
7358         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
7359         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7360         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7361         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
7362         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
7363         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
7364         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
7366 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
7368         PR target/57631
7369         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
7370         name seen by assembler/linker rather if available.
7372 2013-07-11  Andreas Schwab  <schwab@suse.de>
7374         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
7376 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
7378         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
7380 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
7382         * doc/tm.texi.in: Move hook documentation to ....
7383         * target.def: ... here.
7385         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
7386         text on @hook line.
7387         * doc/tm.texi: Regenerate.
7389 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
7391         PR c++/57869
7392         * doc/invoke.texi: Document -Wconditionally-supported.
7394 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7396         PR target/57844
7397         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
7398         of my_fp.
7400 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7402         PR target/57506
7403         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
7404         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
7405         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
7406         Remove duplicate devices.
7407         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
7408         * config/avr/t-multilib: Regenerate.
7409         * config/avr/avr-tables.opt: Regenerate.
7410         * doc/avr-mmcu.texi: Regenerate.
7412 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7414         PR target/56987
7415         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
7417 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
7419         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
7420         the cost of MULT when optimizing for size.
7422 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7424         * config/cr16/cr16-protos.h: Don't include target.h.
7426 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
7428         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
7429         adjust register size for TDmode and TFmode for VSX registers.
7431 2013-07-08  Kai Tietz  <ktietz@redhat.com>
7433         PR target/56892
7434         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
7435         hook_bool_const_tree_true.
7437 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7439         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
7440         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
7441         * config/s390/s390.md: Define FPR*_REGNUM constants.
7442         Fix FPR2_REGNUM constant (18 -> 17).
7443         ("*trunc<BFP:mode><DFP_ALL:mode>2")
7444         ("*trunc<DFP_ALL:mode><BFP:mode>2")
7445         ("trunc<BFP:mode><DFP_ALL:mode>2")
7446         ("trunc<DFP_ALL:mode><BFP:mode>2")
7447         ("*extend<BFP:mode><DFP_ALL:mode>2")
7448         ("*extend<DFP_ALL:mode><BFP:mode>2")
7449         ("extend<BFP:mode><DFP_ALL:mode>2")
7450         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
7451         FPR4_REGNUM.
7453 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
7455         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
7457 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7459         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
7460         and cfun_fpr_bit_p to cfun_fpr_save_p.
7461         (s390_frame_area, s390_register_info, s390_frame_info)
7462         (s390_emit_prologue, s390_emit_epilogue)
7463         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
7464         register numbers.
7465         * config/s390/s390.h: Define *_REGNUM macros for floating point
7466         register numbers.
7468 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
7470         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
7472 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
7474         PR rtl-optimization/57786
7475         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
7476         and break out of the loop when it is set to false.
7478 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
7480         PR target/57819
7481         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
7482         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
7483         (const_int 63)) 0)).
7484         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
7485         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
7486         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
7488         PR rtl-optimization/57829
7489         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
7490         mask bits outside of mode are just sign-extension from mode to HWI.
7492 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
7494         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
7495         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
7496         adjust_address instead of change_address to keep info about alignment.
7497         (emit_strmov): Remove.
7498         (emit_memmov): New function.
7499         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
7500         (expand_movmem_epilogue): Likewise and return updated rtx for
7501         destination.
7502         (expand_constant_movmem_prologue): Likewise and return updated rtx for
7503         destination and source.
7504         (decide_alignment): Refactor, handle vector_loop.
7505         (ix86_expand_movmem): Likewise.
7506         (ix86_expand_setmem): Likewise.
7507         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
7509 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
7511         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
7512         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
7514 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7516         * config/i386/sse.md (sse_movlhps): Change alternative 3
7517         of operand 2 to "m".
7519 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7521         PR target/57807
7522         * config/i386/sse.md (iptr): New mode attribute.
7523         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
7524         (<sse>_vm<plusminus_insn><mode>3): Ditto.
7525         (<sse>_vmmul<mode>3): Ditto.
7526         (<sse>_vmdiv<mode>3): Ditto.
7527         (sse_vmrcpv4sf2): Ditto.
7528         (<sse>_vmsqrt<mode>2): Ditto.
7529         (sse_vmrsqrtv4sf2): Ditto.
7530         (<sse>_vm<code><mode>3): Ditto.
7531         (avx_vmcmp<mode>3): Ditto.
7532         (<sse>_vmmaskcmp<mode>3): Ditto.
7533         (<sse>_comi): Ditto.
7534         (<sse>_ucomi): Ditto.
7535         (*xop_vmfrcz_<mode>): Ditto.
7536         (*fmai_fmadd_<mode>): Ditto.
7537         (*fmai_fmsub_<mode>): Ditto.
7538         (*fmai_fnmadd_<mode>): Ditto.
7539         (*fmai_fnmsub_<mode>): Ditto.
7540         (*fma4i_vmfmadd_<mode>): Ditto.
7541         (*fma4i_vmfmsub_<mode>): Ditto.
7542         (*fma4i_vmfnmadd_<mode>): Ditto.
7543         (*fma4i_vmfnmsub_<mode>): Ditto.
7544         (*xop_vmfrcz_<mode>): Ditto.
7545         (sse_cvtps2pi): Ditto.
7546         (sse_cvttps2pi): Ditto.
7547         (sse_cvtss2si): Ditto.
7548         (sse_cvtss2si_2): Ditto.
7549         (sse_cvtss2siq_2): Ditto.
7550         (sse_cvttss2si): Ditto.
7551         (sse_cvttss2siq): Ditto.
7552         (sse_cvtsd2si): Ditto.
7553         (sse_cvtsd2si_2): Ditto.
7554         (sse_cvtsd2siq_2): Ditto.
7555         (sse_cvttsd2si): Ditto.
7556         (sse_cvttsd2siq): Ditto.
7557         (sse_cvtsd2ss): Ditto.
7558         (sse_cvtss2sd): Ditto.
7559         (avx2_pbroadcast<mode>): Ditto.
7560         (avx2_pbroadcast<mode>_1): Ditto.
7561         (*avx_vperm_broadcast_v4sf): Ditto.
7563         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
7564         (sse_movlhps): Ditto.
7565         (sse_storehps): Ditto.
7566         (sse_loadhps): Ditto.
7567         (sse_storelps): Ditto.
7568         (sse_loadlps): Ditto.
7569         (*vec_concatv4sf): Ditto.
7570         (*vec_interleave_highv2df): Ditto.
7571         (*vec_interleave_lowv2df): Ditto.
7572         (*vec_extractv2df_1_sse): Ditto.
7573         (*vec_extractv2df_0_sse): Ditto.
7574         (sse2_storelpd): Ditto.
7575         (sse2_loadlpd): Ditto.
7576         (sse2_movsd): Ditto.
7577         (*vec_concatv4si): Ditto.
7578         (vec_concatv2di): Ditto.
7580         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
7581         for Intel asm dialect.
7582         (mmx_punpcklwd): Ditto.
7583         (mmx_punpckldq): Ditto.
7585         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
7586         for intel assembler dialect.
7588 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
7590         PR target/29776
7591         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
7592         for BUILT_IN_C{LZ,LRSB}*.
7593         * tree.h (CASE_INT_FN): Add FN##IMAX case.
7594         * tree-vrp.c (extract_range_basic): Handle
7595         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
7596         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
7597         fall thru to code calling set_value*.
7598         * builtins.c (expand_builtin): Remove *IMAX cases.
7599         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
7600         if width is bigger than 2*HWI.
7602 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
7604         PR rtl-optimization/55342
7605         * lra-int.h (lra_subreg_reload_pseudos): New.
7606         * lra.c: Add undoing optional reloads to the block diagram.
7607         (lra_subreg_reload_pseudos): New.
7608         (lra_optional_reload_pseudos): Change comments.
7609         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
7610         lra_optional_reload_pseudos after undo transformations.
7611         * lra-assigns.c (pseudo_prefix_title): New.
7612         (lra_setup_reg_renumber): Use it.
7613         (spill_for): Ditto.  Check subreg reload pseudos too.
7614         (assign_by_spills): Consider subreg reload pseudos too.
7615         * lra-constraints.c (simplify_operand_subreg): Use
7616         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
7617         (curr_insn_transform): Recognize and do optional reloads.
7618         (undo_optional_reloads): New.
7619         (lra_undo_inheritance): Call undo_optional_reloads.
7621 2013-07-05  Thomas Quinot  <quinot@adacore.com>
7623         * tree-complex.c (expand_complex_operations_1): Fix typo.
7625 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
7627         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
7628         (tune_params): New member 'const vec_costs'.
7629         * config/aarch64/aarch64.c (generic_vector_cost): New.
7630         (generic_tunings): New member 'generic_vector_cost'.
7631         (aarch64_builtin_vectorization_cost): New.
7632         (aarch64_add_stmt_cost): New.
7633         (TARGET_VECTORIZE_ADD_STMT_COST): New.
7634         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
7636 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
7638         PR target/57777
7639         * config/i386/predicates.md (vsib_address_operand): Disallow
7640         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
7642 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
7644         PR middle-end/55030
7645         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
7646         expand_builtin_setjmp_receiver.
7647         (expand_label): Adjust, call expand_builtin_setjmp_receiver
7648         with NULL for the label parameter.
7649         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
7650         the frame-pointer.  Adjust comments.
7651         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
7652         only if LABEL is non-NULL.
7654 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
7656         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
7657         (ARM_ABI_AAPCS64): Ditto.
7658         (arm_abi): Ditto.
7659         (ARM_DEFAULT_ABI): Ditto.
7661 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
7663         * config/aarch64/aarch64-builtins.c
7664         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
7665         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
7666         (st1): Likewise.
7667         * config/aarch64/aarch64-simd.md
7668         (aarch64_ld1<VALL:mode>): New.
7669         (aarch64_st1<VALL:mode>): Likewise.
7670         * config/aarch64/arm_neon.h
7671         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
7673 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
7675         * config/i386/i386.c (gate_insert_vzeroupper): Check if
7676         target ISA is AVX.
7677         (ix86_option_override_internal):Turn on all -mavx target flags by
7678         default as they are dependent on AVX anyway.
7680 2013-07-02  Cary Coutant  <ccoutant@google.com>
7682         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
7683         deterministic hash.
7684         (loc_checksum_ordered): Likewise.
7685         (hash_loc_operands): Remove inline keyword.
7687 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
7689         PR tree-optimization/57741
7690         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
7691         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
7692         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
7693         Allow REAL_CST step_exprs if flag_associative_math.
7694         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
7696 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7698         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
7700 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
7702         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
7704 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7706         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
7707         encoding.
7708         (iorsi3_insn): Likewise.
7709         (arm_xorsi3): Likewise.
7711 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
7713         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
7714         to ...
7715         (attribute "type"): ... here, and prefix with "wmmx_".
7716         (attribute "core_cycles"): Update for attribute changes.
7717         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
7718         (tbcstv4hi): Likewise.
7719         (tbcstv2si): Likewise.
7720         (iwmmxt_iordi3): Likewise.
7721         (iwmmxt_xordi3): Likewise.
7722         (iwmmxt_anddi3): Likewise.
7723         (iwmmxt_nanddi3): Likewise.
7724         (iwmmxt_arm_movdi): Likewise.
7725         (iwmmxt_movsi_insn): Likewise.
7726         (mov<mode>_internal): Likewise.
7727         (and<mode>3_iwmmxt): Likewise.
7728         (ior<mode>3_iwmmxt): Likewise.
7729         (xor<mode>3_iwmmxt): Likewise.
7730         (add<mode>3_iwmmxt): Likewise.
7731         (ssaddv8qi3): Likewise.
7732         (ssaddv4hi3): Likewise.
7733         (ssaddv2si3): Likewise.
7734         (usaddv8qi3): Likewise.
7735         (usaddv4hi3): Likewise.
7736         (usaddv2si3): Likewise.
7737         (sub<mode>3_iwmmxt): Likewise.
7738         (sssubv8qi3): Likewise.
7739         (sssubv4hi3): Likewise.
7740         (sssubv2si3): Likewise.
7741         (ussubv8qi3): Likewise.
7742         (ussubv4hi3): Likewise.
7743         (ussubv2si3): Likewise.
7744         (mulv4hi3_iwmmxt): Likewise.
7745         (smulv4hi3_highpart): Likewise.
7746         (umulv4hi3_highpart): Likewise.
7747         (iwmmxt_wmacs): Likewise.
7748         (iwmmxt_wmacsz): Likewise.
7749         (iwmmxt_wmacu): Likewise.
7750         (iwmmxt_wmacuz): Likewise.
7751         (iwmmxt_clrdi): Likewise.
7752         (iwmmxt_clrv8qi): Likewise.
7753         (iwmmxt_clr4hi): Likewise.
7754         (iwmmxt_clr2si): Likewise.
7755         (iwmmxt_uavgrndv8qi3): Likewise.
7756         (iwmmxt_uavgrndv4hi3): Likewise.
7757         (iwmmxt_uavgv8qi3): Likewise.
7758         (iwmmxt_uavgv4hi3): Likewise.
7759         (iwmmxt_tinsrb): Likewise.
7760         (iwmmxt_tinsrh): Likewise.
7761         (iwmmxt_tinsrw): Likewise.
7762         (iwmmxt_textrmub): Likewise.
7763         (iwmmxt_textrmsb): Likewise.
7764         (iwmmxt_textrmuh): Likewise.
7765         (iwmmxt_textrmsh): Likewise.
7766         (iwmmxt_textrmw): Likewise.
7767         (iwmxxt_wshufh): Likewise.
7768         (eqv8qi3): Likewise.
7769         (eqv4hi3): Likewise.
7770         (eqv2si3): Likewise.
7771         (gtuv8qi3): Likewise.
7772         (gtuv4hi3): Likewise.
7773         (gtuv2si3): Likewise.
7774         (gtv8qi3): Likewise.
7775         (gtv4hi3): Likewise.
7776         (gtv2si3): Likewise.
7777         (smax<mode>3_iwmmxt): Likewise.
7778         (umax<mode>3_iwmmxt): Likewise.
7779         (smin<mode>3_iwmmxt): Likewise.
7780         (umin<mode>3_iwmmxt): Likewise.
7781         (iwmmxt_wpackhss): Likewise.
7782         (iwmmxt_wpackwss): Likewise.
7783         (iwmmxt_wpackdss): Likewise.
7784         (iwmmxt_wpackhus): Likewise.
7785         (iwmmxt_wpackwus): Likewise.
7786         (iwmmxt_wpackdus): Likewise.
7787         (iwmmxt_wunpckihb): Likewise.
7788         (iwmmxt_wunpckihh): Likewise.
7789         (iwmmxt_wunpckihw): Likewise.
7790         (iwmmxt_wunpckilb): Likewise.
7791         (iwmmxt_wunpckilh): Likewise.
7792         (iwmmxt_wunpckilw): Likewise.
7793         (iwmmxt_wunpckehub): Likewise.
7794         (iwmmxt_wunpckehuh): Likewise.
7795         (iwmmxt_wunpckehuw): Likewise.
7796         (iwmmxt_wunpckehsb): Likewise.
7797         (iwmmxt_wunpckehsh): Likewise.
7798         (iwmmxt_wunpckehsw): Likewise.
7799         (iwmmxt_wunpckelub): Likewise.
7800         (iwmmxt_wunpckeluh): Likewise.
7801         (iwmmxt_wunpckeluw): Likewise.
7802         (iwmmxt_wunpckelsb): Likewise.
7803         (iwmmxt_wunpckelsh): Likewise.
7804         (iwmmxt_wunpckelsw): Likewise.
7805         (ror<mode>3): Likewise.
7806         (ashr<mode>3_iwmmxt): Likewise.
7807         (lshr<mode>3_iwmmxt): Likewise.
7808         (ashl<mode>3_iwmmxt): Likewise.
7809         (ror<mode>3_di): Likewise.
7810         (ashr<mode>3_di): Likewise.
7811         (lshr<mode>3_di): Likewise.
7812         (ashl<mode>3_di): Likewise.
7813         (iwmmxt_wmadds): Likewise.
7814         (iwmmxt_wmaddu): Likewise.
7815         (iwmmxt_tmia): Likewise.
7816         (iwmmxt_tmiaph): Likewise.
7817         (iwmmxt_tmiabb): Likewise.
7818         (iwmmxt_tmiatb): Likewise.
7819         (iwmmxt_tmiabt): Likewise.
7820         (iwmmxt_tmiatt): Likewise.
7821         (iwmmxt_tmovmskb): Likewise.
7822         (iwmmxt_tmovmskh): Likewise.
7823         (iwmmxt_tmovmskw): Likewise.
7824         (iwmmxt_waccb): Likewise.
7825         (iwmmxt_wacch): Likewise.
7826         (iwmmxt_waccw): Likewise.
7827         (iwmmxt_waligni): Likewise.
7828         (iwmmxt_walignr): Likewise.
7829         (iwmmxt_walignr0): Likewise.
7830         (iwmmxt_walignr1): Likewise.
7831         (iwmmxt_walignr2): Likewise.
7832         (iwmmxt_walignr3): Likewise.
7833         (iwmmxt_wsadb): Likewise.
7834         (iwmmxt_wsadh): Likewise.
7835         (iwmmxt_wsadbz): Likewise.
7836         (iwmmxt_wsadhz): Likewise.
7837         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
7838         (iwmmxt_wabsdiffb): Likewise.
7839         (iwmmxt_wabsdiffh): Likewise.
7840         (iwmmxt_wabsdiffw): Likewise.
7841         (iwmmxt_waddsubhx): Likewise
7842         (iwmmxt_wsubaddhx): Likewise.
7843         (addc<mode>3): Likewise.
7844         (iwmmxt_avg4): Likewise.
7845         (iwmmxt_avg4r): Likewise.
7846         (iwmmxt_wmaddsx): Likewise.
7847         (iwmmxt_wmaddux): Likewise.
7848         (iwmmxt_wmaddsn): Likewise.
7849         (iwmmxt_wmaddun): Likewise.
7850         (iwmmxt_wmulwsm): Likewise.
7851         (iwmmxt_wmulwum): Likewise.
7852         (iwmmxt_wmulsmr): Likewise.
7853         (iwmmxt_wmulumr): Likewise.
7854         (iwmmxt_wmulwsmr): Likewise.
7855         (iwmmxt_wmulwumr): Likewise.
7856         (iwmmxt_wmulwl): Likewise.
7857         (iwmmxt_wqmulm): Likewise.
7858         (iwmmxt_wqmulwm): Likewise.
7859         (iwmmxt_wqmulmr): Likewise.
7860         (iwmmxt_wqmulwmr): Likewise.
7861         (iwmmxt_waddbhusm): Likewise.
7862         (iwmmxt_waddbhusl): Likewise.
7863         (iwmmxt_wqmiabb): Likewise.
7864         (iwmmxt_wqmiabt): Likewise.
7865         (iwmmxt_wqmiatb): Likewise.
7866         (iwmmxt_wqmiatt): Likewise.
7867         (iwmmxt_wqmiabbn): Likewise.
7868         (iwmmxt_wqmiabtn): Likewise.
7869         (iwmmxt_wqmiatbn): Likewise.
7870         (iwmmxt_wqmiattn): Likewise.
7871         (iwmmxt_wmiabb): Likewise.
7872         (iwmmxt_wmiabt): Likewise.
7873         (iwmmxt_wmiatb): Likewise.
7874         (iwmmxt_wmiatt): Likewise.
7875         (iwmmxt_wmiabbn): Likewise.
7876         (iwmmxt_wmiabtn): Likewise.
7877         (iwmmxt_wmiatbn): Likewise.
7878         (iwmmxt_wmiattn): Likewise.
7879         (iwmmxt_wmiawbb): Likewise.
7880         (iwmmxt_wmiawbt): Likewise.
7881         (iwmmxt_wmiawtb): Likewise.
7882         (iwmmxt_wmiawtt): Likewise.
7883         (iwmmxt_wmiawbbn): Likewise.
7884         (iwmmxt_wmiawbtn): Likewise.
7885         (iwmmxt_wmiawtbn): Likewise.
7886         (iwmmxt_wmiawttn): Likewise.
7887         (iwmmxt_wmerge): Likewise.
7888         (iwmmxt_tandc<mode>3): Likewise.
7889         (iwmmxt_torc<mode>3): Likewise.
7890         (iwmmxt_torvsc<mode>3): Likewise.
7891         (iwmmxt_textrc<mode>3): Likewise.
7892         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
7893         (wmmxt_pack): Likewise.
7894         (wmmxt_mult_c1): Likewise.
7895         (wmmxt_mult_c2): Likewise.
7896         (wmmxt_alu_c1): Likewise.
7897         (wmmxt_alu_c2): Likewise.
7898         (wmmxt_alu_c3): Likewise.
7899         (wmmxt_transfer_c1): Likewise.
7900         (wmmxt_transfer_c2): Likewise.
7901         (wmmxt_transfer_c3): Likewise.
7902         (marvell_f_iwmmxt_wstr): Likewise.
7903         (marvell_f_iwmmxt_wldr): Likewise.
7905 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
7907         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
7909 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7911         Revert:
7912         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7913         * lra-constraints.c (need_for_split_p): Check call used hard regs
7914         living through calls.
7916         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
7917         call used regs for call insn.
7919 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
7921         PR target/57736
7922         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
7923         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
7925 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7927         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
7929 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
7931         * lra-constraints.c (need_for_split_p): Check call used hard regs
7932         living through calls.
7934 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
7936         PR target/57744
7937         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
7938         to tie with any other modes.  Eliminate Altivec vector mode tests,
7939         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
7940         code, to return 0 if testing MODE2 for a condition, if we've
7941         already tested MODE1 for the same condition.
7943 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7945         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
7946         layout.
7948 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7950         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7951         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
7953 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7955         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
7956         Define.
7957         (aarch64_symbolic_constant_p): Remove.
7958         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
7959         static.  Fix line length and white space.
7960         (aarch64_symbolic_constant_p): Remove.
7961         * config/aarch64/predicates.md (aarch64_valid_symref):
7962         Use aarch64_classify_symbol_expression.
7964 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7966         * config/arm/constraints.md (Ts): New constraint.
7967         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
7968         16-bit encodings.
7969         (compare_scc): Use "Ts" constraint for operand 0.
7970         (ior_scc_scc): Likewise.
7971         (and_scc_scc): Likewise.
7972         (and_scc_scc_nodom): Likewise.
7973         (ior_scc_scc_cmp): Likewise for operand 7.
7974         (and_scc_scc_cmp): Likewise.
7975         * config/arm/thumb2.md (thumb2_movsi_insn):
7976         Add alternatives for 16-bit encodings.
7977         (thumb2_movhi_insn): Likewise.
7978         (thumb2_movsicc_insn): Likewise.
7979         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
7980         (thumb2_negscc): Use "Ts" constraint.
7981         Move mvn instruction outside cond_exec block.
7982         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
7983         for 16-bit encodings.
7985 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7987         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
7988         encoding.
7989         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
7990         (mulsi3subsi): Likewise.
7991         (mulsidi3adddi): Likewise.
7992         (mulsidi3_v6): Likewise.
7993         (umulsidi3_v6): Likewise.
7994         (umulsidi3adddi_v6): Likewise.
7995         (smulsi3_highpart_v6): Likewise.
7996         (umulsi3_highpart_v6): Likewise.
7997         (mulhisi3tb): Likewise.
7998         (mulhisi3bt): Likewise.
7999         (mulhisi3tt): Likewise.
8000         (maddhisi4): Likewise.
8001         (maddhisi4tb): Likewise.
8002         (maddhisi4tt): Likewise.
8003         (maddhidi4): Likewise.
8004         (maddhidi4tb): Likewise.
8005         (maddhidi4tt): Likewise.
8006         (zeroextractsi_compare0_scratch): Likewise.
8007         (insv_zero): Likewise.
8008         (insv_t2): Likewise.
8009         (anddi_notzesidi_di): Likewise.
8010         (anddi_notsesidi_di): Likewise.
8011         (andsi_notsi_si): Likewise.
8012         (iordi_zesidi_di): Likewise.
8013         (xordi_zesidi_di): Likewise.
8014         (andsi_iorsi3_notsi): Likewise.
8015         (smax_0): Likewise.
8016         (smax_m1): Likewise.
8017         (smin_0): Likewise.
8018         (not_shiftsi): Likewise.
8019         (unaligned_loadsi): Likewise.
8020         (unaligned_loadhis): Likewise.
8021         (unaligned_loadhiu): Likewise.
8022         (unaligned_storesi): Likewise.
8023         (unaligned_storehi): Likewise.
8024         (extv_reg): Likewise.
8025         (extzv_t2): Likewise.
8026         (divsi3): Likewise.
8027         (udivsi3): Likewise.
8028         (arm_zero_extendhisi2addsi): Likewise.
8029         (arm_zero_extendqisi2addsi): Likewise.
8030         (compareqi_eq0): Likewise.
8031         (arm_extendhisi2_v6): Likewise.
8032         (arm_extendqisi2addsi): Likewise.
8033         (arm_movt): Likewise.
8034         (thumb2_ldrd): Likewise.
8035         (thumb2_ldrd_base): Likewise.
8036         (thumb2_ldrd_base_neg): Likewise.
8037         (thumb2_strd): Likewise.
8038         (thumb2_strd_base): Likewise.
8039         (thumb2_strd_base_neg): Likewise.
8040         (arm_negsi2): Add alternative for 16-bit encoding.
8041         (arm_one_cmplsi2): Likewise.
8043 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8045         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8046         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8047         (movdfcc): Likewise.
8048         * config/arm/vfp.md (*thumb2_movsf_vfp):
8049         Disable predication for arm_restrict_it.
8050         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8051         (*thumb2_movdfcc_vfp): Likewise.
8052         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8053         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8054         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8055         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8056         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8057         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8058         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8059         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8060         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8061         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8062         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8063         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8064         Disable predication for arm_restrict_it.
8066 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8068         * config/i386/bmiintrin.h (_bextr_u32): New.
8069         (_bextr_u64): Ditto.
8071 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8073         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8074         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8075         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8076         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8077         * config/mips/n32-elf.h: ...this new file.
8079 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8081         PR target/57224
8082         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8083         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8085 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8087         * config/mips/mips-tables.opt: Regenerate.
8088         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8089         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8090         * doc/invoke.texi: Add -m14kc.
8092 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8094         PR target/57623
8095         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8096         constraints of operand 1 and 2.
8098         PR target/57623
8099         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8100         to match RTL canonicalization.  Swap predicates and
8101         constraints of operand 1 and 2.
8103 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8105         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8106         Process OP_INOUT regs for splitting too.
8108 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8110         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8111         decl before the loop, initialize to NULL.
8112         (vectorizable_load): Initialize ptr_incr to NULL.
8114 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8116         PR lto/57208
8117         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8118         * ipa-ref.c (ipa_maybe_record_reference): New function.
8119         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8120         * ipa-cp.c (create_specialized_node): Record potential references from
8121         aggvals.
8122         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8124 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8126         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8127         parameter 'mode' of type 'enum machine_mode mode'; change to pass
8128         'mode' to force_reg.
8129         (aarch64_add_offset): Update calls to aarch64_force_temporary.
8130         (aarch64_expand_mov_immediate): Likewise.
8132 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8134         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8135         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8137 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8139         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8140         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8141         (struct machine_function): Add tbegin_p.
8142         (s390_canonicalize_comparison): Fold CC mode compares to
8143         conditional jump if possible.
8144         (s390_emit_jump): Return the emitted jump.
8145         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8146         Handle CCRAWmode compares.
8147         (s390_option_override): Default to -mhtm if available.
8148         (s390_reg_clobbered_rtx): Handle floating point regs as well.
8149         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8150         FPRs instead of df_regs_ever_live_p.
8151         (s390_optimize_nonescaping_tx): New function.
8152         (s390_init_frame_layout): Extend clobbered_regs array to cover
8153         FPRs as well.
8154         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8155         (s390_expand_tbegin): New function.
8156         (enum s390_builtin): New enum definition.
8157         (code_for_builtin): New array definition.
8158         (s390_init_builtins): New function.
8159         (s390_expand_builtin): New function.
8160         (TARGET_INIT_BUILTINS): Define.
8161         (TARGET_EXPAND_BUILTIN): Define.
8162         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8163         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8164         (s390_alc_comparison): Likewise.
8165         * config/s390/s390-modes.def: Add CCRAWmode.
8166         * config/s390/s390.h (processor_flags): Add PF_TX.
8167         (TARGET_CPU_HTM): Define macro.
8168         (TARGET_HTM): Define macro.
8169         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8170         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8171         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8172         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8173         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8174         (TBEGIN_MASK, TBEGINC_MASK): New constants.
8175         ("*cc_to_int"): Move up.
8176         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8177         constants other than 0.
8178         ("*ccraw_to_int"): New insn and splitter definition.
8179         ("tbegin", "tbegin_nofloat", "tbegin_retry")
8180         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8181         ("tx_assist"): New expander.
8182         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8183         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8184         * config/s390/s390.opt: Add -mhtm option.
8185         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8186         * config/s390/htmxlintrin.h: New file.
8187         * config/s390/htmintrin.h: New file.
8188         * config/s390/s390intrin.h: New file.
8189         * doc/extend.texi: Document htm builtins.
8190         * config.gcc: Add the new header files to extra_headers.
8192 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
8194         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8195         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8197 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
8198             Pat Haugen  <pthaugen@us.ibm.com>
8199             Peter Bergner  <bergner@vnet.ibm.com>
8201         * config/rs6000/power8.md: New.
8202         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8203         setting for power8 entry.
8204         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8205         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8206         test for Power4/Power5 only.
8207         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8208         support.
8209         (force_new_group): Adjust comment.
8210         * config/rs6000/rs6000.md: Include power8.md.
8212 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
8214         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8215         * config/arm/arm-protos.h (arm_max_conditional_execute): New
8216         declaration.
8217         (tune_params): Update comment.
8218         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8219         (arm_max_conditional_execute): New function.
8220         (thumb2_final_prescan_insn): Use max_insn_skipped and
8221         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8223 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
8225         PR tree-optimization/57705
8226         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8227         SSA_NAME step, provided that it is not defined inside the loop.
8228         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8229         (get_initial_def_for_induction): Handle SSA_NAME IV step.
8231 2013-06-25  Martin Jambor  <mjambor@suse.cz>
8233         PR middle-end/57670
8234         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
8235         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
8236         calls in the dump.
8237         (ipa_note_param_call): Initialize member_ptr flag.
8238         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
8239         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
8240         (ipa_write_indirect_edge_info): Stream member_ptr flag.
8241         (ipa_read_indirect_edge_info): Likewise.
8243 2013-06-25  Richard Biener  <rguenther@suse.de>
8245         PR middle-end/56977
8246         * passes.c (init_optimization_passes): Move pass_fold_builtins
8247         and pass_dce earlier with -Og.
8249 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
8251         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
8252         <BIT_FIELD_REF>: Remove trailing TAB.
8253         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
8254         remove blank line.
8256 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8258         PR tree-optimization/57358
8259         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
8260         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
8261         (ipa_analyze_params_uses): Generate pessimistic info when true.
8263 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8265         PR tree-optimization/57539
8266         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
8267         global.inlined_to of the new node to it.  All callers changed.
8268         * ipa-inline-transform.c (clone_inlined_nodes): New variable
8269         inlining_into, pass it to cgraph_clone_node.
8270         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
8271         ipa_free_edge_args_substructures.
8272         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
8273         rdesc linked list.  Do not assert rdesc edges have inlined caller.
8274         Assert we have found an rdesc in the rdesc list.
8276 2013-06-24  Richard Biener  <rguenther@suse.de>
8278         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
8279         (pointer_set_lookup): Declare.
8280         (class pointer_map): New template class implementing a
8281         generic pointer to T map.
8282         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
8283         pointer_map<T>::contains, pointer_map<T>::insert,
8284         pointer_map<T>::traverse): New functions.
8285         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
8286         (pointer_set_lookup): New function.
8287         (pointer_set_contains): Use pointer_set_lookup.
8288         (pointer_set_insert): Likewise.
8289         (insert_aux): Remove.
8290         (struct pointer_map_t): Embed a pointer_set_t.
8291         (pointer_map_create): Adjust.
8292         (pointer_map_destroy): Likewise.
8293         (pointer_map_contains): Likewise.
8294         (pointer_map_insert): Likewise.
8295         (pointer_map_traverse): Likewise.
8296         * tree-streamer.h (struct streamer_tree_cache_d): Use a
8297         pointer_map<unsigned> instead of a pointer_map_t.
8298         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
8299         (streamer_tree_cache_lookup): Likewise.
8300         (streamer_tree_cache_create): Likewise.
8301         (streamer_tree_cache_delete): Likewise.
8302         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
8303         pointer_map<unsigned> instead of a pointer_map_t.
8304         (lto_init_tree_ref_encoder): Adjust.
8305         (lto_destroy_tree_ref_encoder): Likewise.
8306         * lto-section-out.c (lto_output_decl_index): Likewise.
8307         (lto_record_function_out_decl_state): Likewise.
8308         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
8310 2013-06-24  Richard Biener  <rguenther@suse.de>
8312         PR tree-optimization/57488
8313         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
8315 2013-06-24  Alan Modra  <amodra@gmail.com>
8317         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
8318         (gen_easy_altivec_constant): Likewise.
8319         * config/rs6000/predicates.md (easy_vector_constant_add_self,
8320         easy_vector_constant_msb): Likewise.
8322 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
8324         PR target/57688
8325         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
8326         add missing return true.
8328 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
8330         PR target/52483
8331         * config/sh/predicates.md (general_extend_operand): Invoke
8332         general_movsrc_operand for memory operands.
8333         (general_movsrc_operand): Allow reg+reg addressing, do not use
8334         general_operand for memory operands.
8336 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
8338         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
8339         when current target options does not apply.
8340         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
8341         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
8342         * config/i386/bmiintrin.h: Pass appropriate target
8343         attributes to header.
8344         * config/i386/mmintrin.h: Ditto.
8345         * config/i386/nmmintrin.h: Ditto.
8346         * config/i386/avx2intrin.h: Ditto.
8347         * config/i386/fxsrintrin.h: Ditto.
8348         * config/i386/tbmintrin.h: Ditto.
8349         * config/i386/xsaveintrin.h: Ditto.
8350         * config/i386/f16cintrin.h: Ditto.
8351         * config/i386/xtestintrin.h: Ditto.
8352         * config/i386/xsaveoptintrin.h: Ditto.
8353         * config/i386/bmi2intrin.h: Ditto.
8354         * config/i386/lzcntintrin.h: Ditto.
8355         * config/i386/smmintrin.h: Ditto.
8356         * config/i386/wmmintrin.h: Ditto.
8357         * config/i386/x86intrin.h: Remove all header include guards.
8358         * config/i386/prfchwintrin.h: Ditto.
8359         * config/i386/pmmintrin.h: Ditto.
8360         * config/i386/tmmintrin.h: Ditto.
8361         * config/i386/xmmintrin.h: Ditto.
8362         * config/i386/popcntintrin.h: Ditto.
8363         * config/i386/rdseedintrin.h: Ditto.
8364         * config/i386/ammintrin.h: Ditto.
8365         * config/i386/emmintrin.h: Ditto.
8366         * config/i386/immintrin.h: Remove all header include guards.
8367         * config/i386/fma4intrin.h: Ditto.
8368         * config/i386/lwpintrin.h: Ditto.
8369         * config/i386/xopintrin.h: Ditto.
8370         * config/i386/ia32intrin.h: Ditto.
8371         * config/i386/avxintrin.h: Ditto.
8372         * config/i386/rtmintrin.h: Ditto.
8373         * config/i386/fmaintrin.h: Ditto.
8374         * config/i386/mm3dnow.h: Ditto.
8376 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
8378         * common/config/i386/i386-common.c: Handle LZCNT.
8380 2013-06-22  Andi Kleen  <ak@linux.intel.com>
8382         * doc/extend.texi: Use __atomic_store_n instead of
8383         __atomic_store in HLE example.
8385 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
8387         * config/sh/sh.c: Remove <cstdlib> workaround.
8389 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8391         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
8393 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8395         * doc/extend.texi: Document that __atomic_clear and
8396           __atomic_test_and_set should only be used with bool.
8398 2013-06-20  Jan Hubicka  <jh@suse.cz>
8400         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
8401         types_same_for_odr.
8402         * tree.c (decls_same_for_odr): New function.
8403         (same_for_edr): New function.
8404         (types_same_for_odr): New function.
8405         (get_binfo_at_offset): Use it.
8406         * tree.h (types_same_for_odr): Declare.
8408 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
8409             Jason Merrill  <jason@redhat.com>
8411         * system.h: Include <cstdlib> as well as <stdlib.h>.
8413 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
8415         PR target/57655
8416         * config/i386/i386.c (construct_container): Report error if
8417         long double is used with disabled x87 float returns.
8419 2013-06-20  Jan Hubicka  <jh@suse.cz>
8421         * lto-cgraph.c (input_symtab): Do not set cgraph state.
8423 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
8425         PR rtl-optimization/57425
8426         PR rtl-optimization/57569
8427         * alias.c (write_dependence_p): Remove parameters mem_mode and
8428         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
8429         Changed all callers.
8430         (canon_anti_dependence): Get comments and semantics in sync.
8431         Add parameter mem_canonicalized.  Changed all callers.
8432         * rtl.h (canon_anti_dependence): Update prototype.
8434 2013-06-20  Richard Biener  <rguenther@suse.de>
8436         * data-streamer-in.c (streamer_read_uhwi): Optimize single
8437         byte case, inline streamer_read_uchar and defer section
8438         overrun check.
8440 2013-06-20  Richard Biener  <rguenther@suse.de>
8442         PR tree-optimization/57584
8443         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
8444         SSA names into the expanded expression that take part in
8445         abnormal coalescing.
8447 2013-06-19  Sharad Singhai  <singhai@google.com>
8449         * gcov.c (print_usage): Handle new option.
8450         (process_args): Ditto.
8451         (get_gcov_intermediate_filename): New function.
8452         (output_intermediate_file): New function.
8453         (output_gcov_file): New function
8454         (generate_results): Handle new option.
8455         (release_function): Relase demangled name.
8456         (read_graph_file): Handle demangled name.
8457         (output_lines): Ditto.
8458         * doc/gcov.texi: Document gcov intermediate format.
8460 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
8462         PR bootstrap/57604
8463         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
8464         (lra_emit_add): Use the functions.  Add comment about Y as an
8465         address segment.
8467 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
8469         PR driver/57652
8470         * collect2.c (collect_atexit): New.
8471         (collect_exit): Delete.
8472         (main): Register collect_atexit with atexit.
8473         (collect_wait): Change collect_exit to exit.
8474         (do_wait): Same.
8475         * collect2.h (collect_exit): Delete.
8476         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
8478 2013-06-19  Wei Mi  <wmi@google.com>
8480         PR rtl-optimization/57518
8481         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
8482         if regno is used in paradoxical subreg.
8483         (update_equiv_regs): Check pdx_subregs[regno] before
8484         set a reg to be equivalent with a mem.
8486 2013-06-19  Matthias Klose  <doko@ubuntu.com>
8488         PR driver/57651
8489         * file-find.h (find_a_file): Add a mode parameter.
8490         * file-find.c (find_a_file): Likewise.
8491         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
8492         with X_OK for the executables.
8493         * collect2.c (main): Call find_a_file with X_OK.
8495 2013-06-19  Steve Ellcey  <sellcey@mips.com>
8497         PR target/56942
8498         * config/mips/mips.md (casesi_internal_mips16_<mode>):
8499         Use NEXT_INSN instead of next_real_insn.
8501 2013-06-19  Jan Hubicka  <jh@suse.cz>
8503         * cgraph.h (const_value_known_p): Replace by ...
8504         (ctor_for_folding): .. this one.
8505         * cgraphunit.c (process_function_and_variable_attributes): Use it.
8506         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
8507         * expr.c (expand_expr_real_1): Likewise.
8508         (string_constant): Likewise.
8509         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
8510         * ipa.c (process_references): Likewise.
8511         (symtab_remove_unreachable_nodes): Likewise.
8512         * ipa-inline-analysis.c (param_change_prob): Likewise.
8513         * gimple-fold.c (canonicalize_constructor_val): Likewise.
8514         (get_base_constructor): Likwise.
8515         * varpool.c (varpool_remove_node): Likewise.
8516         (varpool_remove_initializer): LIkewise.
8517         (dump_varpool_node): LIkwise.
8518         (const_value_known_p): Rewrite to ...
8519         (ctor_for_folding): ... this one.
8521 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
8523         PR driver/57651
8524         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
8525         PERSONALITY in $PATH derived prefixes.
8527 2013-06-19  Jeff Law  <law@redhat.com>
8529         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
8530         in comment.
8532         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
8533         (simplify_bitwise_binary): Use it to simpify certain binary ops on
8534         booleans.
8536 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
8538         * config/arm/vfp.md: Move VFP instruction classification documentation
8539         to ...
8540         * config/arm/arm.md: ... here.  Update instruction classification
8541         documentation.
8543 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
8545         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
8546         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
8547         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
8549 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
8551         PR target/57609
8552         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8553         with NEXT_INSN.  Use tablejump_p to check for jump table data
8554         insns.
8556 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
8558         PR c++/56544
8559         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
8560         that now in C++ the value is correct per the C++ standards.
8562 2013-06-19  Richard Biener  <rguenther@suse.de>
8564         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
8565         for global context.
8567 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8569         Revert:
8570         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8572         PR target/57609
8573         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8574         with next_active_insn.
8576 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
8578         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
8579         functions are inlined during failures to flag an error.
8580         * tree-inline.c (expand_call_inline): Allow the error to be flagged
8581         in early inline pass.
8583 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
8585         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
8586         in comments.
8588 2013-06-18  Julian Brown  <julian@codesourcery.com>
8590         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
8591         Permit virtual register pre-reload if !strict.
8592         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
8593         change.
8594         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
8595         prototype.
8596         * config/arm/neon.md (movmisalign<mode>): Use
8597         neon_perm_struct_or_reg_operand instead of
8598         neon_struct_or_register_operand.
8599         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
8600         neon_permissive_struct_operand instead of neon_struct_operand.
8601         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
8602         neon_vector_mem_operand.
8603         * config/arm/predicates.md (neon_struct_operand): Adjust call to
8604         neon_vector_mem_operand.
8605         (neon_permissive_struct_operand): New.
8606         (neon_struct_or_register_operand): Rename to...
8607         (neon_perm_struct_or_reg_operand): This. Adjust call to
8608         neon_vector_mem_operand.
8610 2013-06-18  Richard Biener  <rguenther@suse.de>
8612         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
8613         * lto-streamer.h: Include pointer-set.h.
8614         (struct lto_decl_slot): Remove.
8615         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
8616         Remove next_index entry.
8617         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8618         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8619         (lto_init_tree_ref_encoder): Adjust.
8620         (lto_destroy_tree_ref_encoder): Likewise.
8621         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8622         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8623         (lto_output_decl_index): Adjust.
8624         (lto_new_out_decl_state): Likewise.
8625         (lto_record_function_out_decl_state): Likewise.
8626         * lto-streamer-out.c (copy_function): Likewise.
8628 2013-06-18  Richard Biener  <rguenther@suse.de>
8630         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
8631         * cgraphunit.c: Include cfgloop.h.
8632         (init_lowered_empty_function): Initialize the loop tree.
8633         (assemble_thunk): Insert new BBs into loops.
8635 2013-06-18  Richard Biener  <rguenther@suse.de>
8637         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
8638         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
8639         the map from cache entry to cache index optional.
8640         (streamer_tree_cache_replace_tree): Adjust accordingly.
8641         (streamer_tree_cache_append): Likewise.
8642         (streamer_tree_cache_delete): Likewise.
8643         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
8644         streamer cache map from cache entry to cache index.
8645         * lto-streamer-out.c (create_output_block): Adjust.
8647 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
8649         * config/arm/arm.md (attribute "insn"): Move multiplication and
8650         division attributes to...
8651         (attribute "type"): ... here.  Remove mult.
8652         (attribute "mul32"): New attribute.
8653         (attribute "mul64"): Add umaal.
8654         (*arm_mulsi3): Update attributes.
8655         (*arm_mulsi3_v6): Likewise.
8656         (*thumb_mulsi3): Likewise.
8657         (*thumb_mulsi3_v6): Likewise.
8658         (*mulsi3_compare0): Likewise.
8659         (*mulsi3_compare0_v6): Likewise.
8660         (*mulsi_compare0_scratch): Likewise.
8661         (*mulsi_compare0_scratch_v6): Likewise.
8662         (*mulsi3addsi): Likewise.
8663         (*mulsi3addsi_v6): Likewise.
8664         (*mulsi3addsi_compare0): Likewise.
8665         (*mulsi3addsi_compare0_v6): Likewise.
8666         (*mulsi3addsi_compare0_scratch): Likewise.
8667         (*mulsi3addsi_compare0_scratch_v6): Likewise.
8668         (*mulsi3subsi): Likewise.
8669         (*mulsidi3adddi): Likewise.
8670         (*mulsi3addsi_v6): Likewise.
8671         (*mulsidi3adddi_v6): Likewise.
8672         (*mulsidi3_nov6): Likewise.
8673         (*mulsidi3_v6): Likewise.
8674         (*umulsidi3_nov6): Likewise.
8675         (*umulsidi3_v6): Likewise.
8676         (*umulsidi3adddi): Likewise.
8677         (*umulsidi3adddi_v6): Likewise.
8678         (*smulsi3_highpart_nov6): Likewise.
8679         (*smulsi3_highpart_v6): Likewise.
8680         (*umulsi3_highpart_nov6): Likewise.
8681         (*umulsi3_highpart_v6): Likewise.
8682         (mulhisi3): Likewise.
8683         (*mulhisi3tb): Likewise.
8684         (*mulhisi3bt): Likewise.
8685         (*mulhisi3tt): Likewise.
8686         (maddhisi4): Likewise.
8687         (*maddhisi4tb): Likewise.
8688         (*maddhisi4tt): Likewise.
8689         (maddhidi4): Likewise.
8690         (*maddhidi4tb): Likewise.
8691         (*maddhidi4tt): Likewise.
8692         (divsi3): Likewise.
8693         (udivsi3): Likewise.
8694         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
8695         (thumb2_mulsi_short_compare0): Likewise.
8696         (thumb2_mulsi_short_compare0_scratch): Likewise.
8697         * config/arm/arm1020e.md (1020mult1): Update attribute change.
8698         (1020mult2): Likewise.
8699         (1020mult3): Likewise.
8700         (1020mult4): Likewise.
8701         (1020mult5): Likewise.
8702         (1020mult6): Likewise.
8703         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
8704         change.
8705         (cortex_a15_mult64): Likewise.
8706         (cortex_a15_sdiv): Likewise.
8707         (cortex_a15_udiv): Likewise.
8708         * config/arm/arm1026ejs.md (mult1): Update attribute change.
8709         (mult2): Likewise.
8710         (mult3): Likewise.
8711         (mult4): Likewise.
8712         (mult5): Likewise.
8713         (mult6): Likewise.
8714         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
8715         (pj4_ir_div): Likewise.
8716         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
8717         (11_mult2): Likewise.
8718         (11_mult3): Likewise.
8719         (11_mult4): Likewise.
8720         (11_mult5): Likewise.
8721         (11_mult6): Likewise.
8722         (11_mult7): Likewise.
8723         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
8724         (cortex_a8_mla): Likewise.
8725         (cortex_a8_mull): Likewise.
8726         (cortex_a8_smulwy): Likewise.
8727         (cortex_a8_smlald): Likewise.
8728         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
8729         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
8730         (cortex_r4_mul_3): Likewise.
8731         (cortex_r4_mla_4): Likewise.
8732         (cortex_r4_mla_3): Likewise.
8733         (cortex_r4_smlald): Likewise.
8734         (cortex_r4_mull): Likewise.
8735         (cortex_r4_sdiv): Likewise.
8736         (cortex_r4_udiv): Likewise.
8737         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
8738         (cortex_a7_idiv): Likewise.
8739         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
8740         (9_mult2): Likewise.
8741         (9_mult3): Likewise.
8742         (9_mult4): Likewise.
8743         (9_mult5): Likewise.
8744         (9_mult6): Likewise.
8745         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
8746         (cortex_a53_sdiv): Likewise.
8747         (cortex_a53_udiv): Likewise.
8748         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
8749         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
8750         (mp626_mult2): Likewise.
8751         (mp626_mult3): Likewise.
8752         (mp626_mult4): Likewise.
8753         * config/arm/fa526.md (526_mult1): Update attribute change.
8754         (526_mult2): Likewise.
8755         * config/arm/arm-generic.md (mult): Update attribute change.
8756         (mult_ldsched_strongarm): Likewise.
8757         (mult_ldsched): Likewise.
8758         (multi_cycle): Likewise.
8759         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
8760         * config/arm/fa606te.md (606te_mult1): Update attribute change.
8761         (606te_mult2): Likewise.
8762         (606te_mult3): Likewise.
8763         (606te_mult4): Likewise.
8764         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
8765         (cortex_a9_mac16): Likewise.
8766         (cortex_a9_multiply): Likewise.
8767         (cortex_a9_mac): Likewise.
8768         (cortex_a9_multiply_long): Likewise.
8769         * config/arm/fa626te.md (626te_mult1): Update attribute change.
8770         (626te_mult2): Likewise.
8771         (626te_mult3): Likewise.
8772         (626te_mult4): Likewise.
8774 2013-06-18  Richard Biener  <rguenther@suse.de>
8776         PR lto/57334
8777         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
8779 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8781         PR target/57609
8782         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8783         with next_active_insn.
8785 2013-06-18  Alan Modra  <amodra@gmail.com>
8787         * config/rs6000/rs6000.h (enum data_align): New.
8788         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
8789         (DATA_ABI_ALIGNMENT): Define.
8790         (CONSTANT_ALIGNMENT): Correct comment.
8791         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
8792         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
8794 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
8796         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
8797         ATTRIBUTE_UNUSED marking.
8799 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
8801         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
8802         alternative and update.
8803         (aarch64_dup_lanedi): Delete.
8804         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
8805         * config/aarch64/aarch64-simd-builtins.def: Update.
8807 2013-06-17  Richard Biener  <rguenther@suse.de>
8809         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
8810         (lto_input_scc): Declare.
8811         (lto_input_tree_1): Likewise.
8812         (struct lto_stats_d): Add num_tree_bodies_output and
8813         num_pickle_refs_output.
8814         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
8815         (lto_read_tree_1): Split out from ...
8816         (lto_read_tree): ... this.
8817         (lto_input_scc): New function.
8818         (lto_input_tree_1): Split out from ...
8819         (lto_input_tree): ... this.  Handle LTO_tree_scc.
8820         (lto_data_in_create): Create the streamer cache without hashes.
8821         * lto-streamer-out.c (create_output_block): Create the streamer
8822         cache with hashes when not doing WPA.
8823         (lto_write_tree_1): Split out from ...
8824         (lto_write_tree): ... this.
8825         (get_symbol_initial_value): New function.
8826         (lto_output_tree_1): Split out from ...
8827         (lto_output_tree): ... this.  Write trees as series of SCCs
8828         using a DFS walk via DFS_write_tree.
8829         (struct sccs, struct scc_entry): New types.
8830         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
8831         (DFS_write_tree_body): New function.
8832         (DFS_write_tree): Likewise.
8833         (hash_tree): Likewise.
8834         (scc_entry_compare): Likewise.
8835         (hash_scc): Likewise.
8836         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
8837         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
8838         TREE_CHAIN as regular reference.
8839         (streamer_read_integer_cst): Remove.
8840         (streamer_get_pickled_tree): Adjust.
8841         * tree-streamer-out.c (streamer_write_chain): Disable streaming
8842         of DECL_EXTERNALs in BLOCK_VARS for now.
8843         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
8844         reference.
8845         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
8846         Add hash value argument and record that if hashes are recorded
8847         in the cache.
8848         (streamer_tree_cache_insert_1): Adjust.
8849         (streamer_tree_cache_insert): Likewise.
8850         (streamer_tree_cache_insert_at): Rename to ...
8851         (streamer_tree_cache_replace_tree): ... this and adjust.
8852         (streamer_tree_cache_append): Adjust.
8853         (record_common_node): Likewise.
8854         (streamer_tree_cache_create): Add argument whether to
8855         record hash values together with trees.
8856         (streamer_tree_cache_delete): Adjust.
8857         * tree-streamer.h (struct streamer_tree_cache_d): Add
8858         vector of hashes.
8859         (streamer_read_integer_cst): Remove.
8860         (streamer_tree_cache_insert): Adjust.
8861         (streamer_tree_cache_append): Likewise.
8862         (streamer_tree_cache_insert_at): Rename to ...
8863         (streamer_tree_cache_replace_tree): ... this and adjust.
8864         (streamer_tree_cache_create): Add argument whether to record hashes.
8865         (streamer_tree_cache_get): Rename to ...
8866         (streamer_tree_cache_get_tree): ... this.
8867         (streamer_tree_cache_get_hash): New function.
8868         * tree.c (cache_integer_cst): New function.
8869         * tree.h (cache_integer_cst): Declare.
8870         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
8871         * lto-symtab.c (lto_varpool_replace_node): Only release
8872         DECL_INITIAL of non-prevailing decls.
8873         * varpool.c (varpool_remove_initializer): Do not release
8874         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
8876 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8878         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
8879         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
8880         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
8881         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
8882         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
8883         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
8884         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
8885         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
8886         instead of TARGET_64BIT.
8887         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
8888         Require ISA_HAS_<D>DIV.
8890 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
8892         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
8893         (mips*-*-linux*): Move default with_llsc setting to where other
8894         defaults are set.
8895         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
8896         with_arch block.
8897         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
8898         Likewise.  Remove default with_tune setting.  Move default float
8899         setting to its own block.  Handle with_llsc in the same block as above.
8901 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
8903         PR rtl-optimization/57425
8904         PR rtl-optimization/57569
8905         * alias.c (write_dependence_p): Add new parameters mem_mode,
8906         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
8907         Changed all callers.
8908         (canon_anti_dependence): New function.
8909         * cse.c (check_dependence): Use canon_anti_dependence.
8910         * cselib.c (cselib_invalidate_mem): Likewise.
8911         * rtl.h (canon_anti_dependence): Declare.
8913 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
8915         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
8916         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
8917         ".set mips3" for 64-bit targets.
8919 2013-06-15  Dehao Chen  <dehao@google.com>
8921         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
8922         * gimple-low.c (gimple_check_call_matching_types): Likewise.
8923         (gimple_check_call_args): Likewise.
8924         * value-prof.c (check_ic_target): Likewise.
8925         * ipa-inline.c (early_inliner): Likewise.
8926         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
8927         * cgraph.c (cgraph_create_edge_1): Likewise.
8928         (cgraph_make_edge_direct): Likewise.
8930 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
8932         PR target/57615
8933         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
8934         rs6000_output_move_128bit to handle emitting quad memory
8935         operations.  Set attribute length to 8 bytes.
8937 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
8939         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
8940         New pattern.
8941         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
8942         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
8943         (aarch64_<su>mlsl<mode>): Likewise.
8945 2013-06-14  Mike Stump  <mikestump@comcast.net>
8947         * Makefile.in (TARGET_H): Add insn-codes.h.
8949 2013-06-14  Alan Modra  <amodra@gmail.com>
8951         PR middle-end/57134
8952         PR middle-end/57586
8953         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
8954         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
8955         bitfield expansion when EXPAND_MEMORY.
8956         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
8958 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
8960         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
8961         test for clearing quad memory on 32-bit later.
8963 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8965         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
8966         (fold_negate_expr): Likewise.
8967         (fold_real_zero_addition_p): Handle vectors.
8968         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
8970 2013-06-14  Alan Modra  <amodra@gmail.com>
8972         * varasm.c (force_const_mem): Revert 2013-06-07 change.
8974 2013-06-13  Jan Hubicka  <jh@suse.cz>
8976         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
8977         Local comdats are not externally visible.
8978         * symtab.c (dump_symtab_base): Dump externally visible.
8979         (verify_symtab_base): Verify back links in the symtab hash.
8981 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8983         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
8984         CONVERT_EXPR as equal nodes.
8986 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
8988         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
8990 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
8992         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
8993         Generalize to complex and vector.
8994         * tree.c (build_all_ones_cst): New function.
8995         * tree.h (build_all_ones_cst): Declare it.
8997 2013-06-13  Alan Modra  <amodra@gmail.com>
8999         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
9000         * config/rs6000/rs6000.md (signbittf2): New insn.
9001         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
9002         (abstf2_internal, cmptf_internal2): Likewise.
9003         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
9005 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
9006             Pat Haugen  <pthaugen@us.ibm.com>
9007             Peter Bergner  <bergner@vnet.ibm.com>
9009         * config/rs6000/rs6000.c (emit_load_locked): Add support for
9010         power8 byte, half-word, and quad-word atomic instructions.
9011         (emit_store_conditional): Likewise.
9012         (rs6000_expand_atomic_compare_and_swap): Likewise.
9013         (rs6000_expand_atomic_op): Likewise.
9015         * config/rs6000/sync.md (larx): Add new modes for power8.
9016         (stcx): Likewise.
9017         (AINT): New mode iterator to include TImode as well as normal
9018         integer modes on power8.
9019         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9020         that VSX registers are not considered.  Use AINT mode iterator
9021         instead of INT1 to allow inclusion of quad word atomic operations
9022         on power8.
9023         (load_locked<mode>): Likewise.
9024         (store_conditional<mode>): Likewise.
9025         (atomic_compare_and_swap<mode>): Likewise.
9026         (atomic_exchange<mode>): Likewise.
9027         (atomic_nand<mode>): Likewise.
9028         (atomic_fetch_<fetchop_name><mode>): Likewise.
9029         (atomic_nand_fetch<mode>): Likewise.
9030         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9031         each type.
9032         (ATOMIC): On power8, add QImode, HImode modes.
9033         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9034         modes that promote to SImode.
9035         (load_lockedti): Convert TImode arguments to PTImode, so that we
9036         get a guaranteed even/odd register pair.
9037         (load_lockedpti): Likewise.
9038         (store_conditionalti): Likewise.
9039         (store_conditionalpti): Likewise.
9041         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9042         atomic load/store instructions.
9043         (HSI): Likewise.
9045 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9047         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9048         loads.
9049         (insn_count): New attribute, with most cases extracted from...
9050         (length): ...here.  Redefine most cases in terms of insn_count.
9051         (single_insn): Delete.
9052         (can_delay): Use insn_count to check for single instructions.
9053         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9054         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9055         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9056         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9057         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9058         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9059         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9060         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9061         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9062         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9063         rather than "length".
9064         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9065         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9066         Use "insn_count" rather than "length".
9067         * config/mips/mips-dsp.md
9068         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9069         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9070         length attributes.
9072 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9074         PR tree-optimization/57361
9075         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9077 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9079         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9080         to split.
9081         (aarch64_simd_combine<mode>): New instruction expansion.
9082         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9083         function prototype.
9084         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9085         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9087 2013-06-12  Jan Hubicka  <jh@suse.cz>
9089         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9090         decl has when in streaming stage.
9091         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9092         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9094 2013-06-12  Roland Stigge  <stigge@antcom.de>
9096         PR target/57578
9097         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9099 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9101         PR tree-optimization/57537
9102         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9103         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9105 2013-06-12  Richard Biener  <rguenther@suse.de>
9107         * data-streamer.h (streamer_write_char_stream): CSE
9108         obs->current_pointer.
9109         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9110         streamer_write_char_stream manually and optimize the resulting loop.
9111         (streamer_write_hwi_stream): Likewise.
9113 2013-06-12  Jan Hubicka  <jh@suse.cz>
9115         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9116         * cgraph.h (varpool_create_empty_node): Declare.
9117         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9118         duplicated nodes.
9119         * symtab.c (symtab_unregister_node): Be lax about missin entries
9120         in node hash.
9121         (symtab_get_node): Update comment.
9122         * varpool.c (varpool_create_empty_node): Break out from ...
9123         (varpool_node_for_decl): ... here.
9124         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9126 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
9128         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9129         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9130         part.  Use straight-line flow at the end.
9131         <COMPONENT_REF>: Remove superfluous else.
9132         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9134 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9136         PR target/56564
9137         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9138         target hook even for !TREE_PUBLIC decls.  If no resolution info
9139         is available, return false for common and external decls.
9141 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9143         * config/rl78/constraints.md (U): New constraint.
9144         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9145         valloc attribute.
9147 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9149         PR target/57589
9150         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9151         to allow returning address to AT_PLATFORM name.
9153 2013-06-11  Jan Hubicka  <jh@suse.cz>
9155         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9156         * cgraph.h (symtab_node_base): Add weakref flag.
9157         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9158         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9159         (output_weakrefs): Use weakref flag.
9160         * fold-const.c (simple_operand_p): Handle WEAK.
9161         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9162         * ipa.c (varpool_externally_visible_p): Drop weakref.
9163         (function_and_variable_visibility): Update comment; fix weakref
9164         sanity checks; do not clear DECL_WEAK on them.
9165         * lto-cgraph.c (lto_output_node): update.
9166         (lto_output_varpool_node): Update.
9167         (input_overwrite_node): Update.
9168         (input_node): Update.
9169         (input_varpool_node): Update.
9170         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9171         (lto_symtab_merge_symbols): Add sanity check.
9172         (lto_symtab_prevailing_decl): Do not special case weakrefs.
9173         * passes.c (rest_of_decl_compilation): Set static flag, too.
9174         * symtab.c (dump_symtab_base): Dump weakref.
9175         (verify_symtab_base): Sanity check weakrefs.
9176         (symtab_make_decl_local): Remove duplicated code.
9177         (symtab_alias_ultimate_target): Simplify.
9178         * varpool.c (varpool_create_variable_alias): Set weakref flag.
9180 2013-06-11  Tom de Vries  <tom@codesourcery.com>
9182         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
9183         sequence_vect == NULL.
9185 2013-06-11  DJ Delorie  <dj@redhat.com>
9187         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9188         (rl78_unwind_word_mode): New.
9190 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9192         * final.c (debug_prefix_maps): Make static.
9194 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9196         * function.c (initial_trampoline): Remove stray copy.
9198 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
9200         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9202 2013-06-11  Martin Jambor  <mjambor@suse.cz>
9204         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9205         within bounds at the beginning of the function.
9207 2013-06-11  Alan Modra  <amodra@gmail.com>
9209         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
9210         reporting.
9211         (get_named_section): Don't NULL !DECL_P decl.
9213 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
9215         * doc/invoke.texi (core-avx2): Document.
9216         (slm): Likewise.
9217         (atom): Updated with MOVBE.
9219 2013-06-11  Richard Biener  <rguenther@suse.de>
9221         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9223 2013-06-11  Anton Blanchard  <anton@samba.org>
9225         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9226         correct shift value in little-endian mode.
9228 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
9230         PR target/56564
9231         * varasm.c (get_variable_align): Move #endif to the right place.
9233 2013-06-10  Cary Coutant  <ccoutant@google.com>
9235         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
9236         for hash so that hash table traversal order is deterministic.
9238 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
9239             Pat Haugen  <pthaugen@us.ibm.com>
9240             Peter Bergner  <bergner@vnet.ibm.com>
9242         * config/rs6000/vector.md (GPR move splitter): Do not split moves
9243         of vectors in GPRS if they are direct moves or quad word load or
9244         store moves.
9246         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
9247         declaration.
9248         (direct_move_p): Likewise.
9249         (quad_load_store_p): Likewise.
9251         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
9252         classes into bins based on the physical register type.
9253         (reg_class_to_reg_type): Likewise.
9254         (IS_STD_REG_TYPE): Likewise.
9255         (IS_FP_VECT_REG_TYPE): Likewise.
9256         (reload_fpr_gpr): Arrays to determine what insn to use if we can
9257         use direct move instructions.
9258         (reload_gpr_vsx): Likewise.
9259         (reload_vsx_gpr): Likewise.
9260         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
9261         information that is a simplification of register classes.  Also
9262         precalculate direct move reload helpers.
9263         (direct_move_p): New function to return true if the operation can
9264         be done as a direct move instruciton.
9265         (quad_load_store_p): New function to return true if the operation
9266         is a quad memory operation.
9267         (rs6000_legitimize_address): If quad memory, only allow register
9268         indirect for TImode addresses.
9269         (rs6000_legitimate_address_p): Likewise.
9270         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
9271         (rs6000_reload_register_type): Likewise.
9272         (register_to_reg_type): Return register type.
9273         (rs6000_secondary_reload_simple_move): New helper function for
9274         secondary reload and secondary memory needed to identify anything
9275         that is a simple move, and does not need reloading.
9276         (rs6000_secondary_reload_direct_move): New helper function for
9277         secondary reload to identify cases that can be done with several
9278         instructions via the direct move instructions.
9279         (rs6000_secondary_reload_move): New helper function for secondary
9280         reload to identify moves between register types that can be done.
9281         (rs6000_secondary_reload): Add support for quad memory operations
9282         and for direct move.
9283         (rs6000_secondary_memory_needed): Likewise.
9284         (rs6000_debug_secondary_memory_needed): Change argument names.
9285         (rs6000_output_move_128bit): New function to return the move to
9286         use for 128-bit moves, including knowing about the various
9287         limitations of quad memory operations.
9289         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
9290         memory operations.  call rs6000_output_move_128bit for the actual
9291         instruciton(s) to generate.
9292         (vsx_movti_64bit): Likewise.
9294         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
9295         (UNSPEC_P8V_MTVSRWZ): Likewise.
9296         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
9297         (UNSPEC_P8V_MTVSRD): Likewise.
9298         (UNSPEC_P8V_XXPERMDI): Likewise.
9299         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
9300         (UNSPEC_FUSION_GPR): Likewise.
9301         (FMOVE128_GPR): New iterator for direct move.
9302         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
9303         (f32_sv): Likewise.
9304         (f32_dm): Likewise.
9305         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
9306         loads and direct move instructions.
9307         (zero_extendsidi2_lfiwzx): Likewise.
9308         (extendsidi2_lfiwax): Likewise.
9309         (extendsidi2_nocell): Likewise.
9310         (floatsi<mode>2_lfiwax): Likewise.
9311         (lfiwax): Likewise.
9312         (floatunssi<mode>2_lfiwzx): Likewise.
9313         (lfiwzx): Likewise.
9314         (fix_trunc<mode>_stfiwx): Likewise.
9315         (fixuns_trunc<mode>_stfiwx): Likewise.
9316         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
9317         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
9318         (parity<mode>2_cmpb): Set length/type attr.
9319         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
9320         for 'mr.' to fast_compare.
9321         (bpermd_<mode>): Change type attr to popcnt.
9322         (p8_fmrgow_<mode>): New insns for power8 direct move support.
9323         (p8_mtvsrwz_1): Likewise.
9324         (p8_mtvsrwz_2): Likewise.
9325         (reload_fpr_from_gpr<mode>): Likewise.
9326         (p8_mtvsrd_1): Likewise.
9327         (p8_mtvsrd_2): Likewise.
9328         (p8_xxpermdi_<mode>): Likewise.
9329         (reload_vsx_from_gpr<mode>): Likewise.
9330         (reload_vsx_from_gprsf): Likewise.
9331         (p8_mfvsrd_3_<mode>): LIkewise.
9332         (reload_gpr_from_vsx<mode>): Likewise.
9333         (reload_gpr_from_vsxsf): Likewise.
9334         (p8_mfvsrd_4_disf): Likewise.
9335         (multi-word GPR splits): Do not split direct moves or quad memory
9336         operations.
9338 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
9340         * tree-into-ssa.c (interesting_blocks): Make static.
9342 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
9344         PR target/56564
9345         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
9346         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
9347         Use DATA_ABI_ALIGNMENT for that case instead if defined.
9348         (get_variable_align): New function.
9349         (get_variable_section, emit_bss, emit_common,
9350         assemble_variable_contents, place_block_symbol): Use
9351         get_variable_align instead of DECL_ALIGN.
9352         (assemble_noswitch_variable): Add align argument, use it
9353         instead of DECL_ALIGN.
9354         (assemble_variable): Adjust caller.  Use get_variable_align
9355         instead of DECL_ALIGN.
9356         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
9357         caller.
9358         (DATA_ABI_ALIGNMENT): Define.
9359         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
9360         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
9361         opt is false, only return the psABI mandated alignment increase.
9362         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
9363         (DATA_ABI_ALIGNMENT): ... this.
9364         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
9365         (DATA_ABI_ALIGNMENT): ... this.
9366         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
9367         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
9368         (DATA_ABI_ALIGNMENT): ... this.
9369         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
9370         * doc/tm.texi: Regenerated.
9372 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
9374         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
9375         cmp_code to construct REG_EQUAL note.
9377 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
9379         PR target/57568
9380         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
9381         that operands[2] doesn't overlap with operands[0].
9383 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
9384             Jan Hubicka  <jh@suse.cz>
9386         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
9387         hack to mark symbols as used.
9389 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
9391         PR rtl-optimization/57559
9392         * lra-constraints.c (process_alt_operands): Don't discourage
9393         memory with known offset for offsetable memory constraint.
9394         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
9396 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
9398         * varasm.c (struct oc_local_state): Reorder fields.
9399         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
9400         and adjust accordingly.
9401         (output_constructor): Reorder initialization code and adjust call to
9402         output_constructor_bitfield.
9404 2013-06-07  Jan Hubicka  <jh@suse.cz>
9406         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
9408 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
9410         * tree-object-size.c (unknown): Make const.
9412 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9414         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
9415         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
9416         for last alternative in the cpu_facility attribute.
9418 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9420         PR target/56315
9421         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
9422         (xordi3): Change operand 2 constraint to arm_xordi_operand.
9423         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
9424         * config/arm/constraints.md (Dg): New constraint.
9425         * config/arm/neon.md (xordi3_neon): Remove.
9426         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
9427         * config/arm/predicates.md (arm_xordi_operand): New predicate.
9429 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9431         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
9432         Clean up alternatives.
9434 2013-06-07  Alan Modra  <amodra@gmail.com>
9436         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
9437         va_list_gpr_size.
9439 2013-06-07  Alan Modra  <amodra@gmail.com>
9441         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
9443 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9445         * config/arm/constraints.md (Df): New constraint.
9446         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
9447         Correct length attribute for last two alternatives.
9449 2013-06-07  Alan Modra  <amodra@gmail.com>
9451         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9452         override user -mfp-in-toc.
9453         (offsettable_ok_by_alignment): Consider just the current access
9454         rather than the whole object, unless BLKmode.  Handle
9455         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
9456         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
9457         for -mcmodel=medium.
9458         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
9459         override user -mfp-in-toc or -msum-in-toc.  Default to
9460         -mno-fp-in-toc for -mcmodel=medium.
9462 2013-06-06  DJ Delorie  <dj@redhat.com>
9464         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
9465         TARGET_VALID_POINTER_MODE.
9467 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
9468             Pat Haugen  <pthaugen@us.ibm.com>
9469             Peter Bergner  <bergner@vnet.ibm.com>
9471         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9472         Document new power8 builtins.
9474         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
9475         condition code register, to allow 128-bit logical operations to be
9476         done in the VSX or GPR registers.
9477         (nor<mode>3): Use the canonical form for nor.
9478         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
9479         vclz*, and vpopcnt* vector instructions.
9480         (nand<mode>3): Likewise.
9481         (orc<mode>3): Likewise.
9482         (clz<mode>2): LIkewise.
9483         (popcount<mode>2): Likewise.
9485         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
9486         that only the GPRs are recognized.
9488         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9489         support for new power8 builtins.
9491         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
9492         builtin functions.
9493         (xscvdpspn): Likewise.
9494         (vclz): Likewise.
9495         (vclzb): Likewise.
9496         (vclzh): Likewise.
9497         (vclzw): Likewise.
9498         (vclzd): Likewise.
9499         (vpopcnt): Likewise.
9500         (vpopcntb): Likewise.
9501         (vpopcnth): Likewise.
9502         (vpopcntw): Likewise.
9503         (vpopcntd): Likewise.
9504         (vgbbd): Likewise.
9505         (vmrgew): Likewise.
9506         (vmrgow): Likewise.
9507         (eqv): Likewise.
9508         (eqv_v16qi3): Likewise.
9509         (eqv_v8hi3): Likewise.
9510         (eqv_v4si3): Likewise.
9511         (eqv_v2di3): Likewise.
9512         (eqv_v4sf3): Likewise.
9513         (eqv_v2df3): Likewise.
9514         (nand): Likewise.
9515         (nand_v16qi3): Likewise.
9516         (nand_v8hi3): Likewise.
9517         (nand_v4si3): Likewise.
9518         (nand_v2di3): Likewise.
9519         (nand_v4sf3): Likewise.
9520         (nand_v2df3): Likewise.
9521         (orc): Likewise.
9522         (orc_v16qi3): Likewise.
9523         (orc_v8hi3): Likewise.
9524         (orc_v4si3): Likewise.
9525         (orc_v2di3): Likewise.
9526         (orc_v4sf3): Likewise.
9527         (orc_v2df3): Likewise.
9529         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
9530         allow power8 quad mode in 64-bit.
9531         (rs6000_builtin_vectorized_function): Add support to vectorize
9532         ISA 2.07 count leading zeros, population count builtins.
9533         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
9534         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
9535         (builtin_function_type): Add vgbbd builtin function which takes an
9536         unsigned argument.
9537         (altivec_expand_vec_perm_const): Add support for new power8 merge
9538         instructions.
9540         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
9541         that does not include TImdoe for use with 32-bit.
9542         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
9543         instructions.
9544         (UNSPEC_VSX_CVDPSPN): Likewise.
9545         (vsx_xscvdpspn): Likewise.
9546         (vsx_xscvspdpn): Likewise.
9547         (vsx_xscvdpspn_scalar): Likewise.
9548         (vsx_xscvspdpn_directmove): Likewise.
9549         (vsx_and<mode>3): Split logical operations into 32-bit and
9550         64-bit. Add support to do logical operations on TImode as well as
9551         VSX vector types.  Allow logical operations to be done in either
9552         VSX registers or in general purpose registers in 64-bit mode.  Add
9553         splitters if GPRs were used. For AND, add clobber of CCmode to
9554         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
9555         encoding.
9556         (vsx_and<mode>3_32bit): Likewise.
9557         (vsx_and<mode>3_64bit): Likewise.
9558         (vsx_ior<mode>3): Likewise.
9559         (vsx_ior<mode>3_32bit): Likewise.
9560         (vsx_ior<mode>3_64bit): Likewise.
9561         (vsx_xor<mode>3): Likewise.
9562         (vsx_xor<mode>3_32bit): Likewise.
9563         (vsx_xor<mode>3_64bit): Likewise.
9564         (vsx_one_cmpl<mode>2): Likewise.
9565         (vsx_one_cmpl<mode>2_32bit): Likewise.
9566         (vsx_one_cmpl<mode>2_64bit): Likewise.
9567         (vsx_nor<mode>3): Likewise.
9568         (vsx_nor<mode>3_32bit): Likewise.
9569         (vsx_nor<mode>3_64bit): Likewise.
9570         (vsx_andc<mode>3): Likewise.
9571         (vsx_andc<mode>3_32bit): Likewise.
9572         (vsx_andc<mode>3_64bit): Likewise.
9573         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
9574         and xxlorc instructions.
9575         (vsx_eqv<mode>3_64bit): Likewise.
9576         (vsx_nand<mode>3_32bit): Likewise.
9577         (vsx_nand<mode>3_64bit): Likewise.
9578         (vsx_orc<mode>3_32bit): Likewise.
9579         (vsx_orc<mode>3_64bit): Likewise.
9581         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
9583         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
9584         instruction.
9585         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
9586         (p8_vmrgow): Likewise.
9587         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
9588         GPRs to be split under VSX.
9589         (p8v_clz<mode>2): Add power8 count leading zero support.
9590         (p8v_popcount<mode>2): Add power8 population count support.
9591         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
9592         support.
9594         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
9595         instruction.
9597         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
9598         builtin functions.
9599         (vec_nand): Likewise.
9600         (vec_vclz): Likewise.
9601         (vec_vclzb): Likewise.
9602         (vec_vclzd): Likewise.
9603         (vec_vclzh): Likewise.
9604         (vec_vclzw): Likewise.
9605         (vec_vgbbd): Likewise.
9606         (vec_vmrgew): Likewise.
9607         (vec_vmrgow): Likewise.
9608         (vec_vpopcnt): Likewise.
9609         (vec_vpopcntb): Likewise.
9610         (vec_vpopcntd): Likewise.
9611         (vec_vpopcnth): Likewise.
9612         (vec_vpopcntw): Likewise.
9614 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9616         PR rtl-optimization/57468
9617         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
9618         spilled pseudos.
9620 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9622         PR rtl-optimization/57459
9623         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
9624         type when setting live regs.
9626 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
9628         * config/s390/s390.opt (mlra): New option.
9629         * config/s390/s390.c (s390_decompose_address): Check displacement
9630         for all registers for LRA.
9631         (s390_secondary_reload): Don't used secondary reloads for LRA.
9632         (s390_lra_p): New function.
9633         (TARGET_LRA_P): Define.
9634         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
9635         of attribute cpu_facility to zarch for the last alternative.
9636         (*cmpmem_short): Ditto.
9638 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
9640         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
9641         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
9642         (arm_expand_prologue): Likewise.
9644 2013-06-06  Teresa Johnson  <tejohnson@google.com>
9646         PR c++/53743
9647         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
9648         as this is now done by redirect_edge_and_branch_force.
9649         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
9650         barriers, and fix interaction with splitting.
9651         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
9652         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
9653         reflect changes made in the routine.
9654         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
9655         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
9656         since this is called in cfglayout mode, and replace partition fixup
9657         with assert as that is now done by force_nonfallthru_and_redirect.
9658         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
9659         already be marked with region crossing note.
9660         (insert_section_boundary_note): Make non-static, gate on flag
9661         has_bb_partition, rewrite to also check for multiple partitions.
9662         (rest_of_handle_reorder_blocks): Remove call to
9663         insert_section_boundary_note, now done later during free_cfg.
9664         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
9665         * bb-reorder.h (insert_section_boundary_note): Declare.
9666         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
9667         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
9668         invoke insert_section_boundary_note.
9669         (try_redirect_by_replacing_jump): Remove unnecessary
9670         check for region crossing note.
9671         (fixup_partition_crossing): New function.
9672         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
9673         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
9674         in non-cfglayout mode.
9675         (force_nonfallthru_and_redirect): Fixup partition boundaries,
9676         remove old code that tried to do this. Emit barrier correctly
9677         when we are in cfglayout mode.
9678         (last_bb_in_partition): New function.
9679         (rtl_split_edge): Correctly fixup partition boundaries.
9680         (commit_one_edge_insertion): Remove old code that tried to
9681         fixup region crossing edge since this is now handled in
9682         split_block, and set up insertion point correctly since
9683         block may now end in a jump.
9684         (verify_hot_cold_block_grouping): Guard against checking when not in
9685         linearized RTL mode.
9686         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
9687         notes.
9688         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
9689         rtl_verify_flow_info, so not called in cfglayout mode.
9690         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
9691         (fixup_reorder_chain): Remove old code that attempted to fixup region
9692         crossing note as this is now handled in force_nonfallthru_and_redirect.
9693         (duplicate_insn_chain): Don't duplicate switch section notes.
9694         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
9695         note.
9696         * basic-block.h (emit_barrier_after_bb): Declare.
9698 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9700         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
9701         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
9702         arm_usatsihi): Adjust alternatives for arm_restrict_it.
9704 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9706         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
9707         where appropriate.
9708         * config/arm/ldmstm.md: Regenerate.
9710 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9712         * config/arm/sync.md (atomic_loaddi_1):
9713         Disable predication for arm_restrict_it.
9714         (arm_load_exclusive<mode>): Likewise.
9715         (arm_load_exclusivesi): Likewise.
9716         (arm_load_exclusivedi): Likewise.
9717         (arm_load_acquire_exclusive<mode>): Likewise.
9718         (arm_load_acquire_exclusivesi): Likewise.
9719         (arm_load_acquire_exclusivedi): Likewise.
9720         (arm_store_exclusive<mode>): Likewise.
9721         (arm_store_exclusive<mode>): Likewise.
9722         (arm_store_release_exclusivedi): Likewise.
9723         (arm_store_release_exclusive<mode>): Likewise.
9725 2013-06-06  Richard Biener  <rguenther@suse.de>
9727         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
9728         after LTO_null.
9729         (lto_tag_is_tree_code_p): Adjust.
9730         (lto_tag_is_gimple_code_p): Likewise.
9731         (lto_gimple_code_to_tag): Likewise.
9732         (lto_tag_to_gimple_code): Likewise.
9733         (lto_tree_code_to_tag): Likewise.
9734         (lto_tag_to_tree_code): Likewise.
9735         * data-streamer.h (streamer_write_hwi_in_range): Use
9736         uhwi streaming to stream the normalized range.
9737         (streamer_read_hwi_in_range): Likewise.
9739 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9741         * config/arm/arm.md (enabled_for_depr_it): New attribute.
9742         (predicable_short_it): Likewise.
9743         (predicated): Likewise.
9744         (enabled): Handle above.
9745         (define_cond_exec): Set predicated attribute to yes.
9747 2013-06-05  Mike Stump  <mikestump@comcast.net>
9749         * gdbinit.in (__FUNCTION__): Add.
9751 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
9753         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
9754         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
9756 2013-06-05  Jan Hubicka  <jh@suse.cz>
9758         * varasm.c (mark_decl_referenced): Revert the removal until targets
9759         are fixed.
9761 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
9763         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
9764         instead of mark_decl_referenced.
9766 2013-06-05  Jan Hubicka  <jh@suse.cz>
9768         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
9769         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
9770         and symtab_used_from_object_file_p.
9771         (cgraph_make_node_local_1): Clear forced_by_abi.
9772         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
9773         * cgraph.h (symtab_node_base): Add forced_by_abi.
9774         (decide_is_variable_needed): Remove.
9775         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
9776         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
9777         (decide_is_symbol_needed): ... this one; handle symbols in general;
9778         always analyze virtuals; honnor forced_by_abi.
9779         (cgraph_finalize_function): Update.
9780         (varpool_finalize_decl): Update.
9781         (symbol_defined_and_needed): Remove.
9782         (analyze_functions): Update.
9783         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9784         output_refs, input_overwrite_node): Handle forced_by_abi.
9785         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
9786         (address_taken_from_non_vtable_p): ... this one.
9787         (comdat_can_be_unshared_p_1): New function.
9788         (cgraph_comdat_can_be_unshared_p): Rename to ...
9789         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
9790         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
9791         (function_and_variable_visibility): Clear forced_by_abi as needed.
9792         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
9793         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
9794         * symtab.c (dump_symtab_base): Dump forced_by_abi.
9795         * varpool.c (decide_is_variable_needed): Remove.
9797 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9799         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
9800         (arm_option_override): Override arm_restrict_it where appropriate.
9801         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
9802         * config/arm/arm.opt (mrestrict-it): New command-line option.
9803         * doc/invoke.texi: Document -mrestrict-it.
9805 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9807         * tsan.c (tsan_atomic_table): Make const.
9809 2013-06-05  Richard Biener  <rguenther@suse.de>
9811         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
9812         index associated with the tree we are supposed to replace.
9813         * tree-streamer-out.c (pack_ts_base_value_fields): Output
9814         TREE_ASM_WRITTEN as zero for everything but SSA names.
9816 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
9818         * tree-ssa-structalias.c (call_stmt_vars): Make static.
9820 2013-06-04  Jan Hubicka  <jh@suse.cz>
9822         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
9823         (input_node, input_varpool_node): Handle correctly external same
9824         body aliases.
9825         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
9826         nodes at ltrans stage.
9828 2013-06-04  Jan Hubicka  <jh@suse.cz>
9830         * ipa-inline.c (update_caller_keys): Fix availability test.
9831         (update_callee_keys): Likewise.
9832         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
9833         to follow ELF standard.
9835 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
9837         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
9838         (mips64r5900el-*-elf*): New configurations.
9839         * config/mips/mips-cpus.def (r5900): New processor.
9840         * config/mips/mips-tables.opt: Regenerate.
9841         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
9842         (mips_issue_rate): Handle PROCESSOR_R5900.
9843         (mips_reorg_process_insns): Force reorder mode for the R5900.
9844         * config/mips/mips.h (TARGET_MIPS5900): Define.
9845         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
9846         TARGET_MIPS5900.
9847         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
9848         TARGET_MIPS5900.
9849         * config/mips/mips.md (processor): Add r5900.
9850         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
9852 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9854         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
9855         into function to generate MOVI instruction.
9856         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
9857         (aarch64_preferred_simd_mode): Turn into wrapper.
9858         (aarch64_output_scalar_simd_mov_immediate): New function.
9859         * config/aarch64/aarch64-protos.h: Add prototype for above.
9861 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9863         * config/aarch64/aarch64.c (simd_immediate_info): Remove
9864         element_char member.
9865         (sizetochar): Return signed char.
9866         (aarch64_simd_valid_immediate): Remove elchar and other
9867         unnecessary variables.
9868         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
9869         Calculate element_char as required.
9870         * config/aarch64/aarch64-protos.h: Update and move prototype
9871         for aarch64_output_simd_mov_immediate.
9872         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9873         Update arguments.
9875 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9877         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
9878         information completed by aarch64_simd_valid_immediate.
9879         (aarch64_legitimate_constant_p): Update arguments.
9880         (aarch64_simd_valid_immediate): Work with struct rather than many
9881         pointers.
9882         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
9883         (aarch64_simd_make_constant): Update arguments.
9884         (aarch64_output_simd_mov_immediate): Work with struct rather than
9885         many pointers.  Output immediate directly rather than as operand.
9886         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9887         Update prototype.
9888         * config/aarch64/constraints.md (Dn): Update arguments.
9890 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9892         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
9893         longer static.
9894         (aarch64_simd_immediate_valid_for_move): Remove.
9895         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
9896         (aarch64_simd_make_constant): Update call.
9897         (aarch64_output_simd_mov_immediate): Update call.
9898         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
9899         Add prototype.
9900         * config/aarch64/constraints.md (Dn): Update call.
9902 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
9904         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
9905         return type to bool for prototype.
9906         (aarch64_legitimate_constant_p): Check for true instead of not -1.
9907         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
9908         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
9910 2013-06-04  Catherine Moore  <clm@codesourcery.com>
9912         * config/mips/mips.opt (meva): New.
9913         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
9914         (ASM_SPEC): Handle -meva.
9915         * doc/invoke.texi (meva):  Document.
9917 2013-06-04  Alan Modra  <amodra@gmail.com>
9919         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
9920         constant output.
9922 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9924         * rtl.def: Add extra fourth optional field to define_cond_exec.
9925         * gensupport.c (process_one_cond_exec): Process attributes from
9926         define_cond_exec.
9927         * doc/md.texi: Document fourth field in define_cond_exec.
9929 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
9931         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
9932         out the processing order as in store_bit_field_1.
9934 2013-06-04  Jan Hubicka  <jh@suse.cz>
9936         PR middle-end/57500
9937         * cgraphunit.c (cgraph_process_same_body_aliases): Create
9938         non-VAR_DECL node if it does not exist yet.
9940 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
9942         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
9943         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
9944         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
9945         target_cpu_default setting.
9947 2013-06-03  Teresa Johnson  <tejohnson@google.com>
9949         * dumpfile.c (opt_info_switch_p): Change -fopt-info
9950         default to -fopt-info=optimized instead of all.
9951         * doc/invoke.texi: Ditto.
9952         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
9953         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
9954         (execute_vect_slp): Emit BB vectorization success under
9955         MSG_OPTIMIZED_LOCATIONS.
9956         * tree-vect-slp.c (vect_slp_transform_bb): Change
9957         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
9958         * tree-vect-loop.c (vect_transform_loop): Ditto.
9960 2013-06-03  Jason Merrill  <jason@redhat.com>
9962         PR c++/57415
9963         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9964         Use TARGET_EXPR for C++.
9966 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
9968         PR rtl-optimization/57268
9969         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
9970         if DEBUG_INSN_P (insn).
9972         Reapply
9973         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
9975         PR rtl-optimization/57268
9976         * sched-deps.c (sched_analyze_2): Flush dependence lists if
9977         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
9979 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
9981         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
9982         (ix86_avoid_lea_for_addr): Likewise.
9983         (exact_dependency_1): Likewise.
9984         (ix86_adjust_cost): Likewise.
9985         (swap_top_of_ready_list): Fix formatting and !reload_completed check
9986         removed.
9987         (do_reorder_for_imul): Fix typo, formatting and
9988         !reload_completed check removed.
9989         (ix86_sched_reorder): Fix typo and formatting.
9990         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
9991         list.
9993 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
9995         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
9997 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
9999         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
10000         <STRING_CST>: Likewise.
10001         <VECTOR_CST>: Likewise.
10003 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
10004             Mikael Morin  <mikael@gcc.gnu.org>
10006         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
10007         * config.in: Regenerated.
10008         * configure: Regenerated.
10010 2013-06-01  Jan Hubicka  <jh@suse.cz>
10012         PR middle-end/57366
10013         * cgraphunit.c (compile): When weakref is not supported,
10014         set up transparent aliases before final output pass.
10015         * varasm.c (assemble_alias): Do not try to do it here.
10017 2013-06-01  Jan Hubicka  <jh@suse.cz>
10019         PR middle-end/57467
10020         * passes.c (for_per_function): Skip unanalyzed functions.
10022 2013-06-01  Jan Hubicka  <jh@suse.cz>
10024         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10025         (lto_symtab_merge_symbols_1): ... this one.
10026         (lto_symtab_merge_cgraph_nodes): Rename to ...
10027         (lto_symtab_merge_symbols): ... this one; simplify.
10028         * cgraph.c (same_body_aliases_done): Rename to ...
10029         (cpp_implicit_aliases_done): ... this one.
10030         (cgraph_create_function_alias): Update.
10031         (cgraph_same_body_alias): Update.
10032         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10033         (verify_edge_corresponds_to_fndecl): Simplify.
10034         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10035         (cgraph_node): Remove same_body_alias.
10036         (varpool_node): Remove alias_of and extra_name_alias.
10037         (same_body_aliases_done): Rename to ..
10038         (cpp_implicit_aliases_done): ... this one.
10039         (symtab_alias_ultimate_target): Add default parameter.
10040         (symtab_resolve_alias): New function.
10041         (fixup_same_cpp_alias_visibility): Declare.
10042         (cgraph_function_node): Add default parameter.
10043         (cgraph_node_asm_name): Likewise.
10044         (cgraph_function_or_thunk_node): Add default parameter; do
10045         not ICE when it is NULL.
10046         (varpool_variable_node): Likewise.
10047         * tree-emutls.c (create_emultls_var): Update.
10048         (ipa_lower_emutls): Update.
10049         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10050         (cgraph_reset_node): Reset alias info.
10051         (cgraph_finalize_function): Update.
10052         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10053         (analyze_function): Simplify.
10054         (cgraph_process_same_body_aliases): Simplify.
10055         (analyze_functions): Fixup same body aliases.
10056         (handle_alias_pairs): Simplify.
10057         (assemble_thunk): Update.
10058         (assemble_thunks_and_aliases): Update.
10059         (output_weakrefs): Rewrite.
10060         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10061         (lto_output_varpool_node): Likewise.
10062         (compute_ltrans_boundary): Remve assert.
10063         (get_alias_symbol): New functoin.
10064         (input_node): Rewrite alias handling.
10065         (input_varpool_node): Likewise.
10066         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10067         * ipa.c (process_references): Handle weakrefs correctly.
10068         (symtab_remove_unreachable_nodes): Likewise.
10069         * trans-mem.c (get_cg_data): Update.
10070         (ipa_tm_create_version_alias): Update.
10071         (ipa_tm_execute): Update.
10072         * symtab.c (dump_symtab_base): Dump aliases.
10073         (verify_symtab_base): Verify aliases.
10074         (symtab_node_availability): New function.
10075         (symtab_alias_ultimate_target): Simplify.
10076         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10077         handle all the fixup cases.
10078         (symtab_resolve_alias): New function.
10079         * passes.c (ipa_write_summaries): Handle weakrefs.
10080         * varpool.c (varpool_analyze_node): Simplify.
10081         (assemble_aliases): Update.
10082         (varpool_create_variable_alias): Simplify.
10083         (varpool_extra_name_alias): Simplify.
10084         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10085         (lto_symtab_merge_symbols): ... this one.
10087 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10089         Revert
10090         PR rtl-optimization/57268
10091         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10092         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10094 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10096         Partially reverted:
10097         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10099         PR middle-end/57073
10100         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10101         further up.
10103 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10105         PR rtl-optimization/57268
10106         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10107         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10109 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10111         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10112         unordered comparison operators when -fno-trapping-math is in effect
10113         on the e500.
10114         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10115         and implement unordered comparison operators properly on the e500.
10117 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10119         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10120         for constant scalar integers.
10121         (simplify_relational_operation_1): Likewise.
10123 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10125         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10126         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
10127         Fix comment.
10129 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10130             Igor Zamyatin  <igor.zamyatin@intel.com>
10132         Silvermont (SLM) architecture performance tuning.
10133         * config/i386/i386.h (enum ix86_tune_indices): Add
10134         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10135         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10137         * config/i386/i386.c (initial_ix86_tune_features)
10138         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10139         (ix86_lea_outperforms): Handle Silvermont tuning.
10140         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10141         call.
10142         (ix86_use_lea_for_mov): Likewise.
10143         (ix86_avoid_lea_for_addr): Likewise.
10144         (ix86_lea_for_add_ok): Likewise.
10145         (exact_dependency_1): New function.
10146         (exact_store_load_dependency): Likewise.
10147         (ix86_adjust_cost): Handle Silvermont tuning.
10148         (do_reoder_for_imul): Likewise.
10149         (swap_top_of_ready_list): New function.
10150         (ix86_sched_reorder): Changed to handle Silvermont tuning.
10152         * config/i386/i386.md (peepholes that split memory operand in fp
10153         converts): New.
10155 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10157         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10158         Remove un-necessary braces.
10160 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10162         * config/aarch64/aarch64.c (aarch64_classify_symbol):
10163         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10165 2013-05-31  Tobias Burnus  <burnus@net-b.de>
10167         PR middle-end/57073
10168         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10170 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10172         PR target/56315
10173         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10174         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10175         * config/arm/neon.md (iordi3_neon): Remove.
10176         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10177         * config/arm/predicates.md (imm_for_neon_logic_operand):
10178         Move to earlier in the file.
10179         (neon_logic_op2): Likewise.
10180         (arm_iordi_operand_neon): New predicate.
10182 2013-05-31  Richard Biener  <rguenther@suse.de>
10184         PR tree-optimization/57478
10185         PR tree-optimization/57453
10186         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10187         are life as well.
10189 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10191         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10192         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10194 2013-05-30  Tobias Burnus  <burnus@net-b.de>
10195             Thomas Koenig  <tkoenig@gcc.gnu.org>
10197         PR middle-end/57073
10198         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10199         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10201 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
10203         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10205 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
10207         * target.def (register_usage_leveling_p): New hook.
10208         * targhooks.c (default_register_usage_leveling_p): New.
10209         * targhooks.h (default_register_usage_leveling_p): New prototype.
10210         * lra-assigns.c (register_usage_leveling_p): Use the hook.
10211         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10212         * doc/tm.texi: Update.
10213         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10215 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
10217         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10218         (*insv_reg<mode>): New define_insn.
10220 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10222         PR rtl-optimization/57439
10223         * postreload.c (move2add_valid_value_p): Check that we have
10224         a zero subreg_regno_offset when accessing the register in
10225         the requested mode.
10227 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10228             Igor Zamyatin  <igor.zamyatin@intel.com>
10230         Silvermont (SLM) architecture pipeline model, tuning and
10231         insn selection.
10232         * config.gcc: Add slm config options and target.
10234         * config/i386/slm.md: New.
10236         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
10238         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
10239         PROCESSOR_SLM.
10240         (ix86_target_macros_internal): Likewise.
10242         * config/i386/i386.c (slm_cost): New cost.
10243         (m_SLM): New macro flag.
10244         (initial_ix86_tune_features): Set m_SLM.
10245         (x86_accumulate_outgoing_args): Likewise.
10246         (x86_arch_always_fancy_math_387): Likewise.
10247         (processor_target_table): Add slm cost.
10248         (cpu_names): Add slm cpu name.
10249         (x86_option_override_internal): Set SLM ISA.
10250         (ix86_issue_rate): New case PROCESSOR_SLM.
10251         (ia32_multipass_dfa_lookahead): Likewise.
10252         (fold_builtin_cpu): Add slm.
10254         * config/i386/i386.h (TARGET_SLM): New target macro.
10255         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
10256         (processor_type): Add PROCESSOR_SLM.
10258         * config/i386/i386.md (cpu): Add new value "slm".
10259         (slm.md): Include slm.md.
10261 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
10262             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10264         * config/arm/arm-protos.h: Add and update function protos.
10265         * config/arm/arm.c (use_simple_return_p): New added.
10266         (thumb2_expand_return): Check simple_return flag.
10267         * config/arm/arm.md: Add simple_return and conditional simple_return.
10268         * config/arm/iterators.md: Add iterator for return and simple_return.
10270 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10272         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
10273         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
10274         (arm_emit_vfp_multi_reg_pop): Likewise.
10275         (thumb2_emit_ldrd_pop): Likewise.
10276         (arm_expand_epilogue): Add misc REG_CFA notes.
10277         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
10279 2013-05-29  Lawrence Crowl  <crowl@google.com>
10281         * config/arm/t-arm: Update for below.
10283         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
10284         Change type to hash_table.  Update dependent calls and types.
10286         * config/i386/t-cygming: Update for below.
10288         * config/i386/t-interix: Update for below.
10290         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
10291         Change type to hash_table.  Update dependent calls and types.
10292         (i386_find_on_wrapper_list::wrappers): Likewise.
10294         * config/ia64/t-ia64: Update for below.
10296         * config/ia64/ia64.c (bundle_state_table):
10297         Change type to hash_table.  Update dependent calls and types.
10299         * config/mips/mips.c (mips_reorg_process_insns::htab):
10300         Change type to hash_table.  Update dependent calls and types.
10302         * config/sol2.c (solaris_comdat_htab):
10303         Change type to hash_table.  Update dependent calls and types.
10305         * config/t-sol2: Update for above.
10307 2013-05-29  Teresa Johnson  <tejohnson@google.com>
10309         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
10310         functions are not yet marked as defined.
10312 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
10313             Pat Haugen  <pthaugen@us.ibm.com>
10314             Peter Bergner  <bergner@vnet.ibm.com>
10316         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
10317         instructions.
10318         (VEC_A): Likewise.
10319         (VEC_C): Likewise.
10320         (vrotl<mode>3): Likewise.
10321         (vashl<mode>3): Likewise.
10322         (vlshr<mode>3): Likewise.
10323         (vashr<mode>3): Likewise.
10325         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10326         support for power8 V2DI builtins.
10328         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
10329         power8 V2DI builtins.
10330         (vupkhsw): Likewise.
10331         (vupklsw): Likewise.
10332         (vaddudm): Likewise.
10333         (vminsd): Likewise.
10334         (vmaxsd): Likewise.
10335         (vminud): Likewise.
10336         (vmaxud): Likewise.
10337         (vpkudum): Likewise.
10338         (vpksdss): Likewise.
10339         (vpkudus): Likewise.
10340         (vpksdus): Likewise.
10341         (vrld): Likewise.
10342         (vsld): Likewise.
10343         (vsrd): Likewise.
10344         (vsrad): Likewise.
10345         (vsubudm): Likewise.
10346         (vcmpequd): Likewise.
10347         (vcmpgtsd): Likewise.
10348         (vcmpgtud): Likewise.
10349         (vcmpequd_p): Likewise.
10350         (vcmpgtsd_p): Likewise.
10351         (vcmpgtud_p): Likewise.
10352         (vupkhsw): Likewise.
10353         (vupklsw): Likewise.
10354         (vaddudm): Likewise.
10355         (vmaxsd): Likewise.
10356         (vmaxud): Likewise.
10357         (vminsd): Likewise.
10358         (vminud): Likewise.
10359         (vpksdss): Likewise.
10360         (vpksdus): Likewise.
10361         (vpkudum): Likewise.
10362         (vpkudus): Likewise.
10363         (vrld): Likewise.
10364         (vsld): Likewise.
10365         (vsrad): Likewise.
10366         (vsrd): Likewise.
10367         (vsubudm): Likewise.
10369         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
10370         support for power8 V2DI instructions.
10372         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
10373         power8 V2DI instructions.  Combine pack and unpack insns to use an
10374         iterator for each mode.  Check whether a particular mode supports
10375         Altivec instructions instead of just checking TARGET_ALTIVEC.
10376         (UNSPEC_VPKUWUM): Likewise.
10377         (UNSPEC_VPKSHSS): Likewise.
10378         (UNSPEC_VPKSWSS): Likewise.
10379         (UNSPEC_VPKUHUS): Likewise.
10380         (UNSPEC_VPKSHUS): Likewise.
10381         (UNSPEC_VPKUWUS): Likewise.
10382         (UNSPEC_VPKSWUS): Likewise.
10383         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
10384         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
10385         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
10386         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
10387         (UNSPEC_VUPKHSB): Likewise.
10388         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
10389         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
10390         (UNSPEC_VUPKHSH): Likewise.
10391         (UNSPEC_VUPKLSB): Likewise.
10392         (UNSPEC_VUPKLSH): Likewise.
10393         (VI2): Likewise.
10394         (VI_char): Likewise.
10395         (VI_scalar): Likewise.
10396         (VI_unit): Likewise.
10397         (VP): Likewise.
10398         (VP_small): Likewise.
10399         (VP_small_lc): Likewise.
10400         (VU_char): Likewise.
10401         (add<mode>3): Likewise.
10402         (altivec_vaddcuw): Likewise.
10403         (altivec_vaddu<VI_char>s): Likewise.
10404         (altivec_vadds<VI_char>s): Likewise.
10405         (sub<mode>3): Likewise.
10406         (altivec_vsubcuw): Likewise.
10407         (altivec_vsubu<VI_char>s): Likewise.
10408         (altivec_vsubs<VI_char>s): Likewise.
10409         (altivec_vavgs<VI_char>): Likewise.
10410         (altivec_vcmpbfp): Likewise.
10411         (altivec_eq<mode>): Likewise.
10412         (altivec_gt<mode>): Likewise.
10413         (altivec_gtu<mode>): Likewise.
10414         (umax<mode>3): Likewise.
10415         (smax<mode>3): Likewise.
10416         (umin<mode>3): Likewise.
10417         (smin<mode>3): Likewise.
10418         (altivec_vpkuhum): Likewise.
10419         (altivec_vpkuwum): Likewise.
10420         (altivec_vpkshss): Likewise.
10421         (altivec_vpkswss): Likewise.
10422         (altivec_vpkuhus): Likewise.
10423         (altivec_vpkshus): Likewise.
10424         (altivec_vpkuwus): Likewise.
10425         (altivec_vpkswus): Likewise.
10426         (altivec_vpks<VI_char>ss): Likewise.
10427         (altivec_vpks<VI_char>us): Likewise.
10428         (altivec_vpku<VI_char>us): Likewise.
10429         (altivec_vpku<VI_char>um): Likewise.
10430         (altivec_vrl<VI_char>): Likewise.
10431         (altivec_vsl<VI_char>): Likewise.
10432         (altivec_vsr<VI_char>): Likewise.
10433         (altivec_vsra<VI_char>): Likewise.
10434         (altivec_vsldoi_<mode>): Likewise.
10435         (altivec_vupkhsb): Likewise.
10436         (altivec_vupkhs<VU_char>): Likewise.
10437         (altivec_vupkls<VU_char>): Likewise.
10438         (altivec_vupkhsh): Likewise.
10439         (altivec_vupklsb): Likewise.
10440         (altivec_vupklsh): Likewise.
10441         (altivec_vcmpequ<VI_char>_p): Likewise.
10442         (altivec_vcmpgts<VI_char>_p): Likewise.
10443         (altivec_vcmpgtu<VI_char>_p): Likewise.
10444         (abs<mode>2): Likewise.
10445         (vec_unpacks_hi_v16qi): Likewise.
10446         (vec_unpacks_hi_v8hi): Likewise.
10447         (vec_unpacks_lo_v16qi): Likewise.
10448         (vec_unpacks_hi_<VP_small_lc>): Likewise.
10449         (vec_unpacks_lo_v8hi): Likewise.
10450         (vec_unpacks_lo_<VP_small_lc>): Likewise.
10451         (vec_pack_trunc_v8h): Likewise.
10452         (vec_pack_trunc_v4si): Likewise.
10453         (vec_pack_trunc_<mode>): Likewise.
10455         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
10456         V2DI builtins.
10457         (vec_vmaxsd): Likewise.
10458         (vec_vmaxud): Likewise.
10459         (vec_vminsd): Likewise.
10460         (vec_vminud): Likewise.
10461         (vec_vpksdss): Likewise.
10462         (vec_vpksdus): Likewise.
10463         (vec_vpkudum): Likewise.
10464         (vec_vpkudus): Likewise.
10465         (vec_vrld): Likewise.
10466         (vec_vsld): Likewise.
10467         (vec_vsrad): Likewise.
10468         (vec_vsrd): Likewise.
10469         (vec_vsubudm): Likewise.
10470         (vec_vupkhsw): Likewise.
10471         (vec_vupklsw): Likewise.
10473 2013-05-29  Jan Hubicka  <jh@suse.cz>
10475         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
10476         flags; reorder rest of fields in more consistent way.
10477         (varpool_node): Remove analyzed, finalized and alias.
10478         (cgraph_ndoe): Likewise.
10479         (symtab_alias_ultimate_target): New function.
10480         (cgraph_function_node): Move offline.
10481         (cgraph_reset_node): Declare.
10482         (cgraph_comdat_can_be_unshared_p): Remove.
10483         (varpool_remove_initializer): Declare.
10484         (varpool_first_defined_variable, varpool_next_defined_variable
10485         cgraph_first_defined_function, cgraph_next_defined_function): Update.
10486         (cgraph_function_with_gimple_body_p): Update.
10487         (varpool_all_refs_explicit_p): Update.
10488         (symtab_alias_target): New function.
10489         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
10490         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
10491         (cgraph_function_or_thunk_node): Simplify using
10492         symtab_alias_ultimate_target.
10493         (varpool_variable_node): Likewise.
10494         * cgraph.c (cgraph_create_function_alias): Update.
10495         (cgraph_add_thunk): Update.
10496         (cgraph_remove_node): Update.
10497         (dump_cgraph_node): Do not dump removed flags.
10498         (cgraph_function_body_availability): Update.
10499         (cgraph_propagate_frequency): Update.
10500         (verify_cgraph_node): Check sanity of local flag.
10501         (cgraph_function_node): Move here from cgraph.h; revamp for
10502         cgraph_function_or_thunk_node.
10503         * lto-symtab.c (lto_varpool_replace_node): Update.
10504         (lto_symtab_resolve_can_prevail_p): Update.
10505         (lto_symtab_merge_cgraph_nodes): Update.
10506         * ipa-cp.c (determine_versionability, initialize_node_lattices,
10507         propagate_constants_accross_call, devirtualization_time_bonus,
10508         ipcp_propagate_stage): Update.
10509         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
10510         * ipa-inline-transform.c (clone_inlined_nodes,
10511         preserve_function_body_p): Update.
10512         * ipa-reference.c (propagate): Update.
10513         (write_node_summary_p): Update.
10514         * toplev.c (wrapup_global_declaration_2): Update.
10515         * cgraphunit.c (cgraph_analyze_function): Rename to ...
10516         (analyze_function) ... this one.
10517         (cgraph_process_new_functions): Update.
10518         (cgraph_reset_node): Export.
10519         (cgraph_finalize_function): Update.
10520         (cgraph_add_new_function): Update.
10521         (process_function_and_variable_attributes): Update.
10522         (varpool_finalize_decl): Update.
10523         (symbol_finalized): Remove.
10524         (symbol_finalized_and_needed): Rename to ...
10525         (symbol_defined_and_needed): ... update.
10526         (cgraph_analyze_functions): Update.
10527         (handle_alias_pairs): Update.
10528         (mark_functions_to_output): Update.
10529         (assemble_thunk): Update.
10530         (output_in_order): Update.
10531         (output_weakrefs): Update.
10532         (finalize_compilation_unit): Update.
10533         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
10534         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
10535         input_node, input_varpool_node): Update.
10536         * dbxout.c (dbxout_expand_expr): Update.
10537         * cgraphclones.c (cgraph_clone_node): Update.
10538         (cgraph_copy_node_for_versioning): Update.
10539         (cgraph_materialize_clone): Update.
10540         (cgraph_materialize_all_clones): Update.
10541         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
10542         propagate_pure_const, propagate_nothrow): Update.
10543         * lto-streamer-out.c (lto_output, write_symbol): Update.
10544         * ipa-utils.c (ipa_reverse_postorder): Update.
10545         * ipa-inline.c (can_inline_edge_p): Update.
10546         (update_caller_keys, ipa_inline): Update.
10547         * dwarf2out.c (reference_to_unused,
10548         premark_types_used_by_global_vars_helper): Update.
10549         * tree-eh.c (tree_could_trap_p): Update.
10550         * ipa-split.c (consider_split, execute_split_functions): Update.
10551         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
10552         has_addr_references_p): Update; move ahead in file for better
10553         readability.
10554         (process_references): Simplify.
10555         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
10556         bodies are removed.
10557         (cgraph_comdat_can_be_unshared_p): Make static.
10558         (cgraph_externally_visible_p): Update.
10559         (varpool_externally_visible_p): Update.
10560         (function_and_variable_visibility): Update.
10561         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
10562         ipa_tm_mark_force_output_node): Update.
10563         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
10564         estimate_edge_devirt_benefit, inline_generate_summary,
10565         inline_write_summary): Update.
10566         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
10567         * ipa-prop.c (ipa_compute_jump_functions): Update.
10568         (ipa_print_node_params, ipa_prop_read_section,
10569         ipa_update_after_lto_read, read_replacements_section): Update.
10570         * varasm.c (mark_decl_referenced): Update.
10571         (assemble_alias, dump_tm_clone_pairs): Update.
10572         * tree-inline.c (copy_bb): Update.
10573         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
10574         Update.
10575         * symtab.c (dump_symtab_base): Print new flags.
10576         (verify_symtab_base): Verify new flags.
10577         (symtab_alias_ultimate_target): New function.
10578         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
10579         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
10580         Update.
10581         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
10582         Update.
10583         * i386.c (ix86_get_function_versions_dispatcher,
10584         ix86_generate_version_dispatcher_body): Update.
10585         (fold_builtin_cpu): Use varpool_add_new_variable.
10586         * varpool.c (varpool_remove_initializer): Break out from ...
10587         (varpool_remove_node): ... this one.
10588         (dump_varpool_node, varpool_node_for_asm,
10589         cgraph_variable_initializer_availability, varpool_analyze_node,
10590         varpool_assemble_decl, varpool_remove_unreferenced_decls,
10591         varpool_finalize_named_section_flags, varpool_create_variable_alias):
10592         Update.
10594 2013-05-29  Jan Hubicka  <jh@suse.cz>
10596         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
10598 2013-05-29  Easwaran Raman  <eraman@google.com>
10600         PR tree-optimization/57442
10601         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
10602         when control exits the main loop.
10604 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
10606         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
10607         and RX600.
10608         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
10609         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
10610         * rx/t-rx: Add rx100 under multi library matches option for nofpu
10611         option.
10613 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10615         PR tree-optimization/57441
10616         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
10617         Don't limit size of incr_vec to number of candidates.
10619 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
10621         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
10622         and mips16 directories.
10623         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
10624         (MULTILIB_DIRNAMES): Ditto.
10625         (MULTILIB_EXCEPTIONS): Add new exceptions.
10626         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
10627         (MULTILIB_DIRNAMES): Ditto.
10628         (MULTILIB_EXCEPTIONS): Add new exceptions.
10630 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10631             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10633         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
10634         SYMBOL_TINY_ABSOLUTE.
10635         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
10636         SYMBOL_TINY_ABSOLUTE.
10637         (aarch64_expand_mov_immediate): Likewise.
10638         (aarch64_classify_symbol): Likewise.
10639         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
10640         Permit SYMBOL_TINY_ABSOLUTE.
10641         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
10643 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10644             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10646         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
10647         Refactor if/switch.  Replace gcc_assert with if.
10649 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10651         * config/i386/i386.c (initial_ix86_tune_features): Enable
10652         FP Reassociation for AMD bdver1 and bdver2.
10654 2013-05-29  Martin Jambor  <mjambor@suse.cz>
10656         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
10657         and IMAGPART_EXPR do not occur within other handled_components.
10659 2013-05-29  Richard Biener  <rguenther@suse.de>
10661         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
10662         access on whether the use is in the BB we currently try to
10663         vectorize.
10664         (vect_bb_vectorization_profitable_p): Pass the BB we currently
10665         vectorize to vect_bb_slp_scalar_cost.
10667 2013-05-29  Richard Biener  <rguenther@suse.de>
10669         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
10670         computing scalar cost offsetted by stmts that are kept live
10671         by scalar uses.
10672         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
10673         for computation of scalar cost.
10675 2013-05-28  Steve Ellcey  <sellcey@mips.com>
10677         * config/mips/mips-cpus.def (mips32r2): Change processor type.
10679 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
10681         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
10682         array notation built-in reduction functions.
10683         * doc/passes.texi (Passes): Added documentation about changes done
10684         for Cilk Plus.
10685         * doc/invoke.texi (C Dialect Options): Added documentation about
10686         the -fcilkplus flag.
10687         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
10688         (BUILTINS_DEF): Depend on cilkplus.def.
10689         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
10690         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
10691         * cilkplus.def: New file.
10693 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
10695         PR rtl-optimization/57439
10696         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
10698 2013-05-28  Easwaran Raman  <eraman@google.com>
10700         PR tree-optimization/57337
10701         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
10702         (find_insert_point): Correctly identify the insertion point
10703         when two statements with the same UID is compared.
10705 2013-05-28  Richard Biener  <rguenther@suse.de>
10707         PR tree-optimization/56787
10708         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
10709         from the list of data references.
10710         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
10711         clobbers.
10712         (vect_analyze_loop_operations): Likewise.
10713         (vect_transform_loop): Remove clobbers.
10715 2013-05-28  Martin Jambor  <mjambor@suse.cz>
10717         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
10718         and REALPART_EXPRs have scalar type.
10720 2013-05-28  Richard Biener  <rguenther@suse.de>
10722         PR tree-optimization/57411
10723         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
10724         virtual operands.
10725         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
10726         virtual operand propagation.
10728 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10730         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
10731         destination register for bmasksi_vis.
10732         (vector_init_bshuffle): Likewise.
10733         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
10735 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
10737         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
10738         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
10739         mode if the instruction isn't available in the original mode.
10740         * config/sparc/sparc.opt (mfix-ut699): New option.
10741         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
10742         (divdf3): Turn into expander.
10743         (divdf3_nofix): New insn.
10744         (divdf3_fix): Likewise.
10745         (divsf3): Disable if -mfix-ut699.
10746         (sqrtdf2): Turn into expander.
10747         (sqrtdf2_nofix): New insn.
10748         (sqrtdf2_fix): Likewise.
10749         (sqrtsf2): Disable if -mfix-ut699.
10751 2013-05-27  Richard Biener  <rguenther@suse.de>
10753         PR middle-end/57412
10754         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
10755         block for the new loop.
10757 2013-05-27  Richard Biener  <rguenther@suse.de>
10759         PR tree-optimization/57343
10760         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
10761         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
10762         (number_of_iterations_cond): Do not build the folded tree.
10764 2013-05-27  Richard Biener  <rguenther@suse.de>
10766         Revert
10767         PR middle-end/57381
10768         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10769         OEP_CONSTANT_ADDRESS_OF retained.
10771         PR tree-optimization/57417
10772         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
10773         for unchanged base.
10774         (set_ssa_val_to): Compare addresses using
10775         get_addr_base_and_unit_offset.
10777 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
10779         PR rtl-optimization/56833
10780         * postreload.c (move2add_record_mode): New function.
10781         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
10782         (move2add_use_add2_insn): Use move2add_record_sym_value.
10783         (move2add_use_add3_insn): Likewise.
10784         (reload_cse_move2add): Use move2add_valid_value_p and
10785         move2add_record_mode.  Invalidate call-clobbered and REG_INC
10786         affected regs by setting reg_mode to VOIDmode.
10787         (move2add_note_store): Don't pretend the inside of a SUBREG is
10788         the actual destination.  Invalidate single/leading registers by
10789         setting reg_mode to VOIDmode.
10790         Use move2add_record_sym_value, move2add_valid_value_p and
10791         move2add_record_mode.
10793 2013-05-27  Richard Biener  <rguenther@suse.de>
10795         PR tree-optimization/57396
10796         * tree-affine.c (double_int_constant_multiple_p): Properly
10797         return false for val == 0 and div != 0.
10799 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10801         * config/mips/mips.h: Use #elif in preprocessor conditions.
10803 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10805         PR target/53916
10806         * config/mips/constraints.md (kl): New constraint.
10807         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
10808         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
10809         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
10810         from LO for MIPS16.
10811         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
10813 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
10815         PR target/55777
10816         * config/mips/mips.c (mips_can_inline_p): New function.
10817         (TARGET_CAN_INLINE_P): Define.
10819 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
10821         * sched-int.h (ds_t, dw_t): Make unsigned int.
10822         Fix documentation that describes how all the ds_t bits are used.
10823         Reserve the last bit for delayed-branch scheduling.
10824         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
10825         (BITS_PER_DEP_WEAK): Fix definition and documentation.
10826         (gen_dep_weak_1): Remove prototype.
10827         * sched-deps.c (get_dep_weak_1): Make static.
10828         * target.def (speculate_insn, needs_block_p, gen_spec_check,
10829         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
10830         * doc/tm.texi: Regenerate.
10831         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
10833 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
10835         PR debug/56950
10836         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
10838 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
10839             Sandra Loosemore  <sandra@codesourcery.com>
10841         * config.gcc (powerpc-*): Allow native for with-cpu.
10843 2013-05-24  Jeff Law  <law@redhat.com>
10845         PR tree-optimization/57124
10846         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
10847         conversion feeding a condition if the range has an overflow
10848         if -fstrict-overflow.  Add warnings for when we do make the
10849         transformation.
10851 2013-05-24  Dehao Chen  <dehao@google.com>
10853         * tree-cfg.c (locus_discrim_map): Fix the typo.
10854         (locus_discrim_hasher): Likewise.
10855         (locus_discrim_hasher::hash): Likewise.
10856         (locus_discrim_hasher::equal): Likewise.
10858 2013-05-24  Martin Jambor  <mjambor@suse.cz>
10860         PR tree-optimization/57294
10861         * cgraph.h (ipa_record_stmt_references): Declare.
10862         * cgraphbuild.c (ipa_record_stmt_references): New function.
10863         (build_cgraph_edges): Use ipa_record_stmt_references.
10864         (rebuild_cgraph_edges): Likewise.
10865         (cgraph_rebuild_references): Likewise.
10866         * ipa-prop.c (ipa_modify_call_arguments): Discard references
10867         associated with the old statement and build references from the
10868         newly built statements.
10869         * ipa-ref.c (ipa_remove_stmt_references): New function.
10870         * ipa-ref.h (ipa_remove_stmt_references): Declare.
10872 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
10874         * lra-constraints.c (emit_spill_move): Use smaller mode for
10875         mem-mem moves.
10876         (check_and_process_move): Consider mem-reg moves for secondary
10877         too.
10878         (curr_insn_transform): Don't lose insns emitted before for
10879         secondary memory moves.
10880         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
10881         reg set up in the current insn.
10883 2013-05-24  Dehao Chen  <dehao@google.com>
10885         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
10886         hash function.
10887         (locus_descrim_hasher::equal): Likewise.
10888         (build_gimple_cfg): New discriminator assignment algorithm.
10889         (make_edges): Likewise.
10890         (next_discriminator_for_locus): Likewise.
10891         (same_line_p): Likewise.
10892         (assign_discriminators): Likewise.
10893         (make_cond_expr_edges): Likewise.
10894         (make_gimple_switch_edges): Likewise.
10895         (make_goto_expr_edges): Likewise.
10896         (make_gimple_asm_edges): Likewise.
10898 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
10900         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
10901         X format specifier to only display bottom 16 bits.
10902         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
10903         immediate to match for operand 2, since it will be masked.
10905 2013-05-24  Richard Biener  <rguenther@suse.de>
10907         PR tree-optimization/57287
10908         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
10909         all SSA names that occur in abnormal PHIs.
10911 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
10913         PR tree-ssa/57385
10914         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
10915         that index is not negative.
10917 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
10919         PR rtl-optimization/55177
10920         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
10921         (simplify_byte_swapping_operation): New.
10922         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
10923         (simplify_relational_operation_1): Deal with BSWAP.
10925 2013-05-23  Richard Henderson  <rth@redhat.com>
10927         PR target/56742
10928         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
10929         (ix86_reorg): Call it.
10931 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
10933         PR target/57379
10934         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
10935         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
10936         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
10938 2013-05-23  Christian Bruel  <christian.bruel@st.com>
10940         PR debug/57351
10941         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
10943 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10944             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10946         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
10947         * config/aarch64/constraints.md (Usa): Remove.
10948         * doc/md.texi (AArch64 Usa): Remove.
10950 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
10951             Marcus Shawcroft  <marcus.shawcroft@arm.com>
10953         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
10954         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
10955         * config/aarch64/predicates.md (aarch64_const_address): Remove.
10956         (aarch64_mov_operand): Use aarch64_mov_operand_p.
10958 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
10960         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
10961         instruction (AdvSIMD).
10962         * config/aarch64/aarch64-builtins.c
10963         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
10964         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
10966 2013-05-23  Martin Jambor  <mjambor@suse.cz>
10968         PR middle-end/57347
10969         * tree.h (contains_bitfld_component_ref_p): Declare.
10970         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
10971         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
10972         caller.
10973         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
10974         not access a bit-field.  Assert all final offsets are byte-aligned.
10976 2013-05-23  Richard Biener  <rguenther@suse.de>
10978         PR tree-optimization/57380
10979         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
10980         least one invariant or re-used load.
10981         * passes.c (init_optimization_passes): Move pass_phiprop before
10982         pass_forwprop.
10984 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
10986         * config/aarch64/aarch64-simd.md
10987         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
10989 2013-05-23  Richard Biener  <rguenther@suse.de>
10991         PR middle-end/57381
10992         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
10993         OEP_CONSTANT_ADDRESS_OF retained.
10995 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
10997         PR middle-end/57344
10998         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
10999         don't lower unit.  Handle unit not being always BITS_PER_WORD.
11001 2013-05-23  Richard Biener  <rguenther@suse.de>
11003         PR rtl-optimization/57341
11004         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
11005         instead of true_dependence.
11007 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
11009         * bb-reorder.c (branch_threshold): Make const.
11010         (exec_threshold): Ditto.
11012 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11013             Pat Haugen  <pthaugen@us.ibm.com>
11014             Peter Bergner  <bergner@vnet.ibm.com>
11016         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11017         documentation for the power8 crypto builtins.
11019         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11021         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11022         macros for defining power8 builtin functions.
11023         (BU_P8V_AV_2): Likewise.
11024         (BU_P8V_AV_P): Likewise.
11025         (BU_P8V_VSX_1): Likewise.
11026         (BU_P8V_OVERLOAD_1): Likewise.
11027         (BU_P8V_OVERLOAD_2): Likewise.
11028         (BU_CRYPTO_1): Likewise.
11029         (BU_CRYPTO_2): Likewise.
11030         (BU_CRYPTO_3): Likewise.
11031         (BU_CRYPTO_OVERLOAD_1): Likewise.
11032         (BU_CRYPTO_OVERLOAD_2): Likewise.
11033         (XSCVSPDP): Fix typo, point to the correct instruction.
11034         (VCIPHER): Add power8 crypto builtins.
11035         (VCIPHERLAST): Likewise.
11036         (VNCIPHER): Likewise.
11037         (VNCIPHERLAST): Likewise.
11038         (VPMSUMB): Likewise.
11039         (VPMSUMH): Likewise.
11040         (VPMSUMW): Likewise.
11041         (VPERMXOR_V2DI): Likewise.
11042         (VPERMXOR_V4SI: Likewise.
11043         (VPERMXOR_V8HI: Likewise.
11044         (VPERMXOR_V16QI: Likewise.
11045         (VSHASIGMAW): Likewise.
11046         (VSHASIGMAD): Likewise.
11047         (VPMSUM): Likewise.
11048         (VPERMXOR): Likewise.
11049         (VSHASIGMA): Likewise.
11051         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11052         __CRYPTO__ if the crypto instructions are available.
11053         (altivec_overloaded_builtins): Add support for overloaded power8
11054         builtins.
11056         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11057         support for power8 crypto builtins.
11058         (builtin_function_type): Likewise.
11059         (altivec_init_builtins): Add support for builtins that take vector
11060         long long (V2DI) arguments.
11062         * config/rs6000/crypto.md: New file, define power8 crypto
11063         instructions.
11065 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11066             Pat Haugen  <pthaugen@us.ibm.com>
11067             Peter Bergner  <bergner@vnet.ibm.com>
11069         * doc/invoke.texi (Option Summary): Add power8 options.
11070         (RS/6000 and PowerPC Options): Likewise.
11072         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11073         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11074         wm, wn, wr documentation.
11076         * config/rs6000/constraints.md (wm): New constraint for VSX
11077         registers if direct move instructions are enabled.
11078         (wn): New constraint for no registers.
11079         (wq): New constraint for quad word even GPR registers.
11080         (wr): New constraint if 64-bit instructions are enabled.
11081         (wv): New constraint if power8 vector instructions are enabled.
11082         (wQ): New constraint for quad word memory locations.
11084         * config/rs6000/predicates.md (const_0_to_15_operand): New
11085         constraint for 0..15 for crypto instructions.
11086         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11087         as GPR and floating point registers.
11088         (int_reg_operand): New predicate to match only GPR registers.
11089         (base_reg_operand): New predicate to match base registers.
11090         (quad_int_reg_operand): New predicate to match even GPR registers
11091         for quad memory operations.
11092         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11093         operations in both GPR and VSX registers.
11094         (quad_memory_operand): New predicate for quad memory operations.
11095         (reg_or_indexed_operand): New predicate for direct move support.
11097         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11098         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11099         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11100         (POWERPC_MASKS): Add power8 options.
11101         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11102         various options.
11104         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11105         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11107         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11108         (-mpower8-fusion): New power8 options.
11109         (-mpower8-fusion-sign): Likewise.
11110         (-mpower8-vector): Likewise.
11111         (-mcrypto): Likewise.
11112         (-mdirect-move): Likewise.
11113         (-mquad-memory): Likewise.
11115         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11116         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11117         registers.
11118         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11119         (rs6000_debug_vector_unit): Add p8_vector.
11120         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11121         definitions.  Also print fusion state.
11122         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11123         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11124         (rs6000_option_override_internal): Add support for power8.
11125         (rs6000_common_init_builtins): Add debugging for skipped builtins
11126         if -mdebug=builtin.
11127         (rs6000_adjust_cost): Add power8 support.
11128         (rs6000_issue_rate): Likewise.
11129         (insn_must_be_first_in_group): Likewise.
11130         (insn_must_be_last_in_group): Likewise.
11131         (force_new_group): Likewise.
11132         (rs6000_register_move_cost): Likewise.
11133         (rs6000_opt_masks): Likewise.
11135         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11136         power8 capable assembler, default to power7 options.
11137         (TARGET_DIRECT_MOVE): Likewise.
11138         (TARGET_CRYPTO): Likewise.
11139         (TARGET_P8_VECTOR): Likewise.
11140         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11141         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11142         (VECTOR_MEM_P8_VECTOR_P): Likewise.
11143         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11144         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11145         (TARGET_XSCVDPSPN): Likewise.
11146         (TARGET_XSCVSPDPN): Likewsie.
11147         (TARGET_SYNC_HI_QI): Likewise.
11148         (TARGET_SYNC_TI): Likewise.
11149         (MASK_CRYPTO): Likewise.
11150         (MASK_DIRECT_MOVE): Likewise.
11151         (MASK_P8_FUSION): Likewise.
11152         (MASK_P8_VECTOR): Likewise.
11153         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11154         temporary used by some of the direct move instructions to get two FP
11155         temporary registers does not force creation of a stack frame.
11156         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11157         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11158         that any VSX registers are tieable, even if they are also an
11159         Altivec vector mode.
11160         (r6000_reg_class_enum): Add wm, wr, wv constraints.
11161         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11162         (RS6000_BTM_CRYPTO): Likewise.
11163         (RS6000_BTM_COMMON): Likewise.
11165         * config/rs6000/rs6000.md (cpu attribute): Add power8.
11166         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11167         (enum rs6000_vector): Add power8 vector support.
11169 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11171         PR target/19599
11172         PR target/57340
11173         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11174         (any_sibcall_could_use_r3): this and handle indirect calls.
11175         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11177 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11179         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11181 2013-05-22  Richard Biener  <rguenther@suse.de>
11183         PR middle-end/57349
11184         * profile.c (branch_prob): Do not split blocks that are
11185         abnormally receiving from ECF_RETURNS_TWICE functions.
11187 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11189         * recog.c (offsettable_address_addr_space_p): Fix calculation of
11190         address mode.  Move pointer mode initialization to the same place.
11192 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11194         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11195         while it has any effect.
11197 2013-05-21  Easwaran Raman  <eraman@google.com>
11199         PR tree-optimization/57322
11200         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11201         UID of the statement added to the BB to be 1.
11203 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
11205         PR tree-optimization/57331
11206         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11207         of conversion from pointer type to integral type with integer.
11209 2013-05-21  Martin Jambor  <mjambor@suse.cz>
11211         PR lto/57289
11212         * ipa-prop.c (ipa_read_node_info): Process param_used and
11213         controlled_uses in the same order as when writing.
11215 2013-05-21  Magnus Granberg  <baldrick@free.fr>
11217         PR plugins/56754
11218         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11220 2013-05-21  Richard Biener  <rguenther@suse.de>
11222         PR tree-optimization/57318
11223         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11224         estimate stmts with side-effects as likely eliminated.
11226 2013-05-21  Richard Biener  <rguenther@suse.de>
11228         PR tree-optimization/57330
11229         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11230         preserve the call stmts fntype.
11232 2013-05-21  Richard Biener  <rguenther@suse.de>
11234         PR tree-optimization/57303
11235         * tree-ssa-sink.c (statement_sink_location): Improve killing
11236         stmt detection and properly handle self-assignments.
11238 2013-05-21  Christian Bruel  <christian.bruel@st.com>
11240         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
11241         spanning registers. LEAF_REG_REMAP is supported only for contiguous
11242         registers. Set register size out of the PARALLEL loop.
11244 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
11246         PR target/56547
11247         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
11248         (*fmasf4, *fmasf4_media): New insns.
11250 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11252         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
11253         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
11254         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
11255         (mips_idiv_insns): Update the comments to say that the returned
11256         instruction counts are in units of BASE_INSN_LENGTH.
11257         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
11258         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
11259         using 2 rather than 4 as the length of indirect MIPS16 and
11260         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
11261         length of a NOP.  Don't divide MIPS16 lengths by 2.
11262         (mips16_split_long_branches): Assume a branch is long if the
11263         length is greater than 4 rather than 8.
11264         * config/mips/mips.md (length): Give MIPS16 lengths directly,
11265         rather than multiplying them by 2.  Multiply instruction counts
11266         by BASE_INSN_LENGTH rather than 4.
11267         (*jump_mips16, tls_get_tp_mips16_<mode>)
11268         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
11270 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11272         * config/mips/mips.md (extended_mips16): Remove branch case.
11273         (length): Remove duplicated extended_mips16 test.
11275 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11277         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
11279 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
11281         * recog.h (Recog_data): Rename to...
11282         (recog_data_d): ...this.
11283         (recog_data): Update accordingly.
11284         * recog.c (recog_data): Likewise.
11285         * reload.c (save_recog_data): Likewise.
11286         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
11287         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
11289 2013-05-17  Julian Brown  <julian@codesourcery.com>
11291         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
11292         found in a REG_EQUAL note, invalidate it.
11294 2013-05-17  Easwaran Raman  <eraman@google.com>
11296         * tree-ssa-reassoc.c (find_insert_point): New function.
11297         (insert_stmt_after): Likewise.
11298         (get_def_stmt): Likewise.
11299         (ensure_ops_are_available): Likewise.
11300         (not_dominated_by): Likewise.
11301         (rewrite_expr_tree): Do not move statements beyond what is
11302         necessary. Remove call to swap_ops_for_binary_stmt...
11303         (reassociate_bb): ... and move it here.
11304         (build_and_add_sum): Assign UIDs for new statements.
11305         (linearize_expr): Likewise.
11306         (do_reassoc): Renumber gimple statement UIDs.
11308 2013-05-17  Jan Hubicka  <jh@suse.cz>
11310         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
11311         weakrefs.
11312         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
11313         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
11314         weakrefs.
11315         (output_weakrefs): Update.
11317 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
11318             Martin Jambor  <mjambor@suse.cz>
11320         PR middle-end/57276
11321         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
11322         value that corresponds to the given aggval is found in values vector.
11324 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
11326         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
11327         sse, sse2, sse3, ssse3 and sse4a flags to options.
11329 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
11331         * gengtype-state.c: (s_expr_writer): New class, to handle
11332         prettifying of output layout of s-expressions.
11333         (state_writer): New class, to write out gtype.state.
11334         (state_written_type_count): Move this variable into member data of
11335         state_writer.
11336         (s_expr_writer::s_expr_writer): New code: constructor for new class
11337         (state_writer::state_writer(): ditto
11338         (s_expr_writer::write_new_line): New function
11339         (s_expr_writer::write_any_indent): ditto
11340         (s_expr_writer::begin_s_expr): ditto
11341         (s_expr_writer::end_s_expr): ditto
11342         (write_state_fileloc): convert to method of state_writer...
11343         (state_writer:: write_state_fileloc): ...and use methods of
11344         s_expr_writer to write indentation into the gtype.state output file
11345         to visually represent the hierarchical structure of the list
11346         structures
11347         (write_state_fields): ditto, renaming to...
11348         (state_writer::write_state_fields)
11349         (write_state_a_string): ditto, renaming to...
11350         (state_writer::write_state_a_string)
11351         (write_state_string_option): ditto, renaming to...
11352         (state_writer::write_state_string_option)
11353         (write_state_type_option): ditto, renaming to...
11354         (state_writer::write_state_type_option)
11355         (write_state_nested_option): ditto, renaming to...
11356         (state_writer::write_state_nested_option)
11357         (write_state_option): ditto, renaming to...
11358         (state_writer::write_state_option)
11359         (write_state_options): ditto, renaming to...
11360         (state_writer::write_state_options)
11361         (write_state_lang_bitmap): ditto, renaming to...
11362         (state_writer::write_state_lang_bitmap)
11363         (write_state_version): ditto, renaming to...
11364         (state_writer::write_state_version)
11365         (write_state_scalar_type): ditto, renaming to...
11366         (state_writer::write_state_scalar_type)
11367         (write_state_string_type): ditto, renaming to...
11368         (state_writer::write_state_string_type)
11369         (write_state_undefined_type): ditto, renaming to...
11370         (state_writer::write_state_undefined_type)
11371         (write_state_struct_union_type): ditto, renaming to...
11372         (state_writer::write_state_struct_union_type)
11373         (write_state_struct_type): ditto, renaming to...
11374         (state_writer::write_state_struct_type)
11375         (write_state_user_struct_type): ditto, renaming to...
11376         (state_writer::write_state_user_struct_type)
11377         (write_state_lang_struct_type): ditto, renaming to...
11378         (state_writer::write_state_lang_struct_type)
11379         (write_state_param_struct_type): ditto, renaming to...
11380         (state_writer::write_state_param_struct_type)
11381         (write_state_pointer_type): ditto, renaming to...
11382         (state_writer::write_state_pointer_type)
11383         (write_state_array_type): ditto, renaming to...
11384         (state_writer::write_state_array_type)
11385         (write_state_gc_used): ditto, renaming to...
11386         (state_writer::write_state_gc_used)
11387         (write_state_common_type_content): ditto, renaming to...
11388         (state_writer::write_state_common_type_content)
11389         (write_state_type): ditto, renaming to...
11390         (state_writer::write_state_type)
11391         (write_state_pair_list): ditto, renaming to...
11392         (state_writer::write_state_pair_list)
11393         (write_state_pair): ditto, renaming to...
11394         (state_writer::write_state_pair)
11395         (write_state_typedefs): ditto, renaming to...
11396         (state_writer::write_state_typedefs)
11397         (write_state_structures): ditto, renaming to...
11398         (state_writer::write_state_structures)
11399         (write_state_param_structs): ditto, renaming to...
11400         (state_writer::write_state_param_structs)
11401         (write_state_variables): ditto, renaming to...
11402         (state_writer::write_state_variables)
11403         (write_state_srcdir): ditto, renaming to...
11404         (state_writer::write_state_srcdir)
11405         (write_state_files_list): ditto, renaming to...
11406         (state_writer::write_state_files_list)
11407         (write_state_languages): ditto, renaming to...
11408         (state_writer::write_state_languages)
11409         (write_state): create a state_writer instance and use it when
11410         writing out the state file
11412 2013-05-17  Mike Stump  <mikestump@comcast.net>
11414         PR rtl-optimization/57304
11415         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
11416         accessing DF_REF_REAL_LOC.
11418 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11420         PR rtl-optimization/57281
11421         PR rtl-optimization/57300
11422         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
11423         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
11424         what the other splitter did if the registers are dead.
11426 2013-05-17  Richard Biener  <rguenther@suse.de>
11428         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
11429         MEM_REF offsets.
11431 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11433         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
11434         linking.
11436 2013-05-17  Marek Polacek  <polacek@redhat.com>
11438         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
11439         length when doing non-zero store of storing '\0' to '\0'.
11441 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11443         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
11444         vect_external_def oprnd1 with loop_vinfo, try to emit
11445         optional cast, negation and and stmts on the loop preheader
11446         edge instead of into the pattern def seq.
11448         PR tree-optimization/57051
11449         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
11450         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
11452 2013-05-16  Nick Clifton  <nickc@redhat.com>
11454         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
11455         (rl78_is_naked_func): New function.
11456         (rl78_expand_prologue): Skip prologue generation for naked functions.
11457         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
11458         * doc/extend.texi (naked): Add RL78 to the list of processors
11459         that supports this attribute.
11461 2013-05-16  Jeff Law  <law@redhat.com>
11463         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
11465 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11467         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
11468         cache parameters using detect_caches_amd also for CYRIX,
11469         NSC and TM2 signatures.
11471 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11472             Dzianis Kahanovich  <mahatma@eu.by>
11474         PR target/45359
11475         PR target/46396
11476         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11477         VIA/Centaur processors and determine their cache parameters
11478         using detect_caches_amd.
11480 2013-05-16  Teresa Johnson  <tejohnson@google.com>
11482         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
11483         (rtl_verify_edges): New function.
11484         (rtl_verify_bb_insns): Ditto.
11485         (rtl_verify_bb_pointers): Ditto.
11486         (rtl_verify_bb_insn_chain): Ditto.
11487         (rtl_verify_fallthru): Ditto.
11488         (rtl_verify_bb_layout): Ditto.
11489         (rtl_verify_flow_info_1): Outline checks into new functions.
11490         (rtl_verify_flow_info): Ditto.
11492 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
11494         * cfghooks.c (copy_bbs): Add update_dominance argument.
11495         * cfghooks.h (copy_bbs): Update prototype.
11496         * tree-cfg.c (gimple_duplicate_sese_region):
11497         Add update_dominance argument.
11498         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
11499         * tree-ssa-loop-ch.c (copy_loop_headers): Update
11500         gimple_duplicate_sese_region call.
11501         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
11502         Update copy_bbs call.
11503         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
11504         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
11506 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11508         * tree-vectorizer.h (NUM_PATTERNS): Increment.
11509         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
11510         vect_recog_rotate_pattern.
11511         (vect_recog_rotate_pattern): New function.
11513 2013-05-16  Jason Merrill  <jason@redhat.com>
11515         * Makefile.in (LLINKER): New variable.
11516         (mostlyclean): Remove link mutex.
11517         * configure.ac: Handle --enable-link-mutex.
11518         * lock-and-run.sh: New script.
11520 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11522         PR target/19599
11523         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
11524         for NULL decl.
11526 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11528         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
11530 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11532         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
11533         * config/arm/arm.c (next_consecutive_mem): New function.
11534         (gen_movmem_ldrd_strd): Likewise.
11535         * config/arm/arm.md (movmemqi): Update condition and code.
11536         (unaligned_loaddi, unaligned_storedi): New patterns.
11538 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11540         * config.gcc: Obsolete *-*-solaris2.9*.
11541         * doc/install.texi (Specific, *-*-solaris2*): Document it.
11543 2013-05-16  Richard Biener  <rguenther@suse.de>
11545         * passes.c (init_optimization_passes): Move pass_parallelize_loops
11546         earlier, after GRAPHITE transforms and IV canonicalization.
11548 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11550         * omp-low.c (extract_omp_for_data): For collapsed loops,
11551         if at least one of the loops is known at compile time to
11552         iterate zero times, set count to 0.
11553         (expand_omp_regimplify_p): New function.
11554         (expand_omp_for_generic): For collapsed loops, if at least
11555         one of the loops isn't known to iterate at least once,
11556         add runtime check with setting count to 0.
11557         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
11558         For unsigned types if it isn't known at compile time that
11559         the loop will iterate at least once, add runtime check to bypass
11560         the whole loop if initial condition isn't true.
11562 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
11564         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
11566 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
11568         PR middle-end/57286
11569         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
11570         transformations to avoid an infinite loop.
11572 2013-05-16  Marek Polacek  <polacek@redhat.com>
11574         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
11576 2013-05-15  Leif Ekblad  <leif@rdos.net>
11578         * config/i386/i386.c (ix86_decompose_address): Use
11579         DEFAULT_TLS_SEG_REG to access TLS segment register.
11580         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
11581         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
11582         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
11584 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
11586         PR target/57260
11587         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
11588         sibling calls to functions that would normally be lazily bound,
11589         unless $gp is call-clobbered.
11591 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
11593         * config/i386/i386.c (ix86_option_override_internal): Update
11594         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
11595         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
11596         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
11597         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
11598         of TARGET_3DNOW.
11599         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
11601 2013-05-15  Andreas Schwab  <schwab@suse.de>
11603         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
11604         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
11605         third operand.
11607 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11609         * loop-unroll.c (report_unroll_peel): Check decision before
11610         emitting unroll/peel message.
11612 2013-05-15  Teresa Johnson  <tejohnson@google.com>
11614         * function.h (has_bb_partition): New rtl_data flag.
11615         (bb_reorder_complete): Ditto.
11616         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
11617         instead of flag_reorder_blocks_and_partition.
11618         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
11619         with some enhancements.
11620         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
11621         * bb-reorder.c (connect_traces): Check for has_bb_partition
11622         instead of flag_reorder_blocks_and_partition.
11623         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
11624         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
11625         verify_hot_cold_block_grouping.
11626         (partition_hot_cold_basic_blocks): Set has_bb_partition.
11628 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11630         PR target/19599
11631         * config/arm/predicates.md (call_insn_operand): New predicate.
11632         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
11633         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
11634         if insn is not a tail call.
11635         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
11636         registers.
11637         * config/arm/arm.h (enum reg_class): New caller save register class.
11638         (REG_CLASS_NAMES): Likewise.
11639         (REG_CLASS_CONTENTS): Likewise.
11640         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
11641         without decls.
11643 2013-05-15  Richard Biener  <rguenther@suse.de>
11645         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
11646         of MSG_OPTIMIZED_LOCATIONS.
11647         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
11648         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
11649         message.
11650         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
11651         of MSG_OPTIMIZED_LOCATIONS.
11652         (execute_vect_slp): Likewise.
11653         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
11654         (vect_create_cond_for_alias_checks): Likewise.
11655         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
11656         (vect_recog_widen_mult_pattern): Likewise.
11657         (vect_recog_widen_sum_pattern): Likewise.
11658         (vect_recog_over_widening_pattern): Likewise.
11659         (vect_recog_widen_shift_pattern): Likewise.
11660         (vect_recog_vector_vector_shift_pattern): Likewise.
11661         (vect_recog_divmod_pattern): Likewise.
11662         (vect_recog_mixed_size_cond_pattern): Likewise.
11663         (vect_recog_bool_pattern): Likewise.
11664         (vect_pattern_recog_1): Likewise.
11666 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11668         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
11669         non-functions to builtin_unreachable.
11670         * ipa-inline-transform.c (inline_call): Do not assert estimates were
11671         correct when new direct edges were discovered.
11673 2013-05-15  Martin Jambor  <mjambor@suse.cz>
11675         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
11676         header, print symbol order instead of node uid, print more information
11677         about indirect edge targets.
11678         (ipa_make_edge_direct_to_target): Print symbol order instead of node
11679         uids.
11680         (ipa_make_edge_direct_to_target): Likewise.
11681         (remove_described_reference): Likewise.
11682         (propagate_controlled_uses): Likewise.
11683         (ipa_print_node_params): Also print symbol order.
11684         (ipcp_transform_function): Print symbol order instead of node uids.
11685         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
11686         (cgraph_get_create_real_symbol_node): Likewise.
11687         * ipa-cp.c (print_lattice): Likewise.
11688         (print_all_lattices): Likewise.
11689         (determine_versionability): Likewise.
11690         (initialize_node_lattices): Likewise.
11691         (estimate_local_effects): Likewise.
11692         (update_profiling_info): Likewise.
11693         (create_specialized_node): Likewise.
11694         (perhaps_add_new_callers): Likewise.
11695         (decide_about_value): Likewise.
11696         (decide_whether_version_node): Likewise.
11697         (identify_dead_nodes): Likewise.
11698         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
11699         (dump_inline_summary): Likewise.
11700         (estimate_node_size_and_time): Likewise.
11701         (inline_analyze_function): Likewise.
11702         * ipa-inline.c (report_inline_failed_reason): Likewise.
11703         (want_early_inline_function_p): Likewise.
11704         (edge_badness): Likewise.
11705         (update_edge_key): Likewise.
11706         (inline_small_functions): Likewise.  Add dumping of order to two other
11707         dumps.
11708         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
11709         instead of node uids.
11710         (propagate_pure_const): Likewise.
11711         (propagate_pure_const): Likewise.
11712         * ipa-utils.c (dump_cgraph_node_set): Likewise.
11713         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
11714         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
11715         of node uids.
11716         * tree-pretty-print.c (dump_function_header): Likewise.
11717         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
11718         Print symbol order instead of node uids.
11720 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11722         * config/s390/s390.c (s390_register_move_cost): Don't impose the
11723         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
11725 2013-05-15  Richard Biener  <rguenther@suse.de>
11727         PR tree-optimization/57275
11728         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
11729         return value for fail to do runtime alias checks for gather loads.
11731 2013-05-15  Jan Hubicka  <jh@suse.cz>
11733         PR lto/57038
11734         PR lto/47375
11735         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
11736         weakrefs are not external.
11737         (lto_symtab_merge_decls): Fix thinko when dealing with
11738         non-lto_symtab decls.
11739         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
11740         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
11741         * varpool.c (dump_varpool_node): Dump more flags.
11743 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11745         * config/i386/i386.c (processor_alias_table): Add instruction
11746         FSGSBASE for AMD bdver3 architecture.
11748 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11750         * tree.c (warn_deprecated_use): Print file:line using locus color.
11751         * diagnostic.c (diagnostic_report_current_module): Print file:line
11752         and file:line:column using locus color.
11754 2013-05-14  Mike Stump  <mikestump@comcast.net>
11756         * gdbinit.in: Add __null.
11758 2013-05-14  Mike Stump  <mikestump@comcast.net>
11760         * recog.h: Rename struct recog_data to Recog_data.
11761         * recog.c: Likewise.
11762         * reload.c (can_reload_into): Likewise.
11763         * config/picochip/picochip.c: Likewise.
11765 2013-05-14  Mike Stump  <mikestump@comcast.net>
11767         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
11769 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
11771         * resource.h (struct resources): Remove unch_memory member.
11772         (CLEAR_RESOURCE): Don't clear unch_memory.
11773         * resource.c (mark_referenced_resources): Don't set it.
11774         (mark_set_resources): Likewise.
11775         (mark_target_live_regs): Don't clear it.
11776         (init_resource_info): Likewise.
11777         * reorg.c (resource_conflicts_p): Don't compare it.
11778         (redundant_insn): Don't set it.
11780         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
11781         Remove prototypes.
11782         * emit-rtl.c (next_label): Remove unused function.
11783         (skip_consecutive_labels, link_cc0_insns): Move to ...
11784         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
11785         only place where these functions are used, and make them static.
11787 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11789         * fold-const.c (fold_negate_expr): Handle vectors.
11790         (fold_truth_not_expr): Make it static.
11791         (fold_invert_truthvalue): New static function.
11792         (invert_truthvalue_loc): Handle vectors. Do not call
11793         fold_truth_not_expr directly.
11794         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
11795         <TRUTH_NOT_EXPR>: Do not cast to boolean.
11796         (fold_comparison): Handle vector constants.
11797         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
11798         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
11799         * tree.h (fold_truth_not_expr): Remove declaration.
11801 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
11803         * config/aarch64/aarch64-simd.md
11804         (aarch64_vcond_internal<mode>): Rename to...
11805         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
11806         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
11807         float modes. Clarify all iterator modes.
11808         (vcond<mode><mode>): Use new name for vcond expanders.
11809         (vcond<v_cmp_result><mode>): Likewise.
11810         (vcondu<mode><mode>: Likewise.
11811         * config/aarch64/iterators.md (VDQF_COND): New.
11813 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
11815         PR bootstrap/57266
11816         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
11817         variable for the shift amount. Check that we shift by non-negative
11818         amounts.
11820 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
11822         PR target/42017
11823         * config/arm/arm.h (EPILOGUE_USES): Only return true
11824         for LR_REGNUM after epilogue_completed.
11826 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
11828         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
11829         is error_mark_node.
11831 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11833         PR target/57261
11834         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
11835         and Solaris 11+/x86 with gld.
11836         * configure: Regenerate.
11838 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11840         * expmed.c (expand_shift_1): Canonicalize rotates by
11841         constant bitsize / 2 to bitsize - 1.
11842         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
11843         case ROTATERT>: Likewise.
11845         Revert:
11846         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
11848         * config/i386/i386.md (rotateinv): New code attr.
11849         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
11850         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
11851         roll $31, %eax, etc.
11853 2013-05-14  Richard Biener  <rguenther@suse.de>
11855         PR middle-end/57235
11856         * tree-eh.c (sink_clobbers): Give up for successors with
11857         multiple predecessors and no virtual uses.
11859 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
11861         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
11862         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
11864 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
11866         PR middle-end/57251
11867         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
11868         the case when both op0 and op1 have VOIDmode.
11870 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11872         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
11873         in multiply-accumulate mode.
11875 2013-05-13  Guozhi Wei  <carrot@google.com>
11877         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
11879 2013-05-13  Kai Tietz  <ktietz@redhat.com>
11881         PR target/56975
11882         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
11883         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
11884         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
11885         * config/i386/i386.c (ix86_option_override_internal): Likewise.
11886         (ix86_expand_prologue): Likewise.
11887         (ix86_expand_split_stack_prologue): Likewise.
11888         (legitimate_pic_address_disp_p): Likewise.
11889         (legitimize_pic_address): Likewise.
11890         (legitimize_tls_address): Likewise.
11891         (legitimize_pe_coff_symbol): Likewise.
11892         (output_pic_addr_const): Likewise.
11893         (construct_plt_address): Likewise.
11894         (ix86_expand_call): Likewise.
11895         (x86_output_mi_thunk): Likewise.
11896         (x86_function_profiler): Likewise.
11898 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
11900         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
11901         similar switch cases.
11902         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
11903         (aarch64_simd_mov_to_<mode>low): Delete.
11904         (aarch64_simd_mov_to_<mode>high): Delete.
11905         (move_lo_quad_<mode>): Add w<-r alternative.
11906         (aarch64_simd_move_hi_quad_<mode>): Likewise.
11907         (aarch64_simd_mov_from_*): Update type attribute.
11908         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
11909         statement.
11911 2013-05-13  Jan Hubicka  <jh@suse.cz>
11913         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
11914         * config/i386/i386.c (ix86_compute_frame_layout,
11915         ix86_expand_epilogue, emit_i387_cw_initialization,
11916         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
11917         ix86_local_alignment): Fix use of size/speed predicates.
11919 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11921         PR tree-optimization/45216
11922         PR tree-optimization/57157
11923         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
11924         the (-Y) & (B - 1) variant if OP is |.
11925         * expmed.c (expand_shift_1): For rotations by const0_rtx just
11926         return shifted.  Use (-op1) & (prec - 1) as other_amount
11927         instead of prec - op1.
11929 2013-05-13  Martin Jambor  <mjambor@suse.cz>
11931         PR middle-end/42371
11932         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
11933         (ipa_constant_data): New type.
11934         (ipa_jump_func): Use ipa_constant_data to hold information about
11935         constant jump functions.
11936         (ipa_get_jf_constant): Adjust to jump function type changes.
11937         (ipa_get_jf_constant_rdesc): New function.
11938         (ipa_param_descriptor): New field controlled_uses.
11939         (ipa_get_controlled_uses): New function.
11940         (ipa_set_controlled_uses): Likewise.
11941         * ipa-ref.h (ipa_find_reference): Declare.
11942         * ipa-prop.c (ipa_cst_ref_desc): New type.
11943         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
11944         changes.
11945         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
11946         New parameter cs.  Adjust all callers.
11947         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
11948         (remove_described_reference): New function.
11949         (jfunc_rdesc_usable): Likewise.
11950         (try_make_edge_direct_simple_call): Decrement controlled use count,
11951         attempt to remove reference if it hits zero.
11952         (combine_controlled_uses_counters): New function.
11953         (propagate_controlled_uses): Likewise.
11954         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
11955         (ipa_edge_duplication_hook): Duplicate reference descriptions.
11956         (ipa_print_node_params): Print described use counter.
11957         (ipa_write_jump_function): Adjust to jump function type changes.
11958         (ipa_read_jump_function): New parameter CS, pass it to
11959         ipa_set_jf_constant.  Adjust caller.
11960         (ipa_write_node_info): Stream controlled use count
11961         (ipa_read_node_info): Likewise.
11962         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
11963         asserting.
11964         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
11965         count.  Remove cloning-added reference if it reaches zero.
11966         * ipa-ref.c (ipa_find_reference): New function.
11968 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11970         * config/i386/i386.c (processor_target_table): Modified default
11971         alignment values for AMD BD and BT architectures.
11973 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
11975         * tree-vect-generic.c (uniform_vector_p): Move ...
11976         * tree.c (uniform_vector_p): ... here.
11977         * tree.h (uniform_vector_p): Declare it.
11978         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
11979         into a scalar.
11981 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
11983         PR tree-optimization/57230
11984         * tree-ssa-strlen.c (handle_char_store): Record length for
11985         array store from STRING_CST.
11987         PR tree-optimization/57230
11988         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
11989         check.
11991 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
11993         * config/epiphany/epiphany.c (epiphany_init): Check size of
11994         NUM_MODES_FOR_MODE_SWITCHING.
11995         (epiphany_expand_prologue):
11996         Remove CONFIG_REGNUM initial value handling code.
11997         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
11998         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
11999         (emit_set_fp_mode, epiphany_mode_after): Likewise.
12000         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
12001         Don't return 1 for FP_MODE_NONE.
12002         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
12003         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
12004         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
12005         * config/epiphany/epiphany.md (save_config): New pattern.
12007 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
12009         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
12011 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
12013         * config/i386/i386.md (memory): Handle sseishft1.
12014         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12015         (*vec_extractv2di_1): Ditto.
12017 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
12019         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12020         saved registers.
12022 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12024         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12025         Add mthumb/march=armv7-a multilib.
12026         Add mthumb/march=armv7-r multilib.
12027         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12029 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
12031         * config/v850/t-rtems: Add more multilibs.
12033 2013-05-10  Richard Biener  <rguenther@suse.de>
12035         PR tree-optimization/57214
12036         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12037         not propagate from SSA names that occur in abnormal PHI nodes.
12039 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12041         * stor-layout.c (element_precision): New function.
12042         * machmode.h (element_precision): Declare it.
12043         * tree.c (build_minus_one_cst): New function.
12044         (element_precision): Likewise.
12045         * tree.h (build_minus_one_cst): Declare new function.
12046         (element_precision): Likewise.
12047         * fold-const.c (operand_equal_p): Use element_precision.
12048         (fold_binary_loc): Handle vector types.
12049         * convert.c (convert_to_integer): Use element_precision.
12050         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12051         separately.
12053 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12055         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12056         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12057         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12058         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12059         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12060         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12061         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12062         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12063         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12064         (Uuw8): New constraints.
12065         (Usb4): Move into alphabetical order.
12066         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12067         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12068         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12069         previously unnamed patterns.
12070         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12071         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12072         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12073         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12074         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12075         of specific lengths.
12077 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12079         * config/i386/i386.md (rotateinv): New code attr.
12080         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12081         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12082         roll $31, %eax, etc.
12084         PR tree-optimization/45216
12085         PR tree-optimization/57157
12086         * tree-ssa-forwprop.c (simplify_rotate): New function.
12087         (ssa_forward_propagate_and_combine): Call it.
12089 2013-05-10  Richard Biener  <rguenther@suse.de>
12091         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12092         disable peeling when we version for aliasing.
12093         (vector_alignment_reachable_p): Honor explicit user alignment.
12094         (vect_supportable_dr_alignment): Likewise.
12095         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12096         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12097         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12098         then peeling to arrange for the cost-model check to come first.
12100 2013-05-10  Alan Modra  <amodra@gmail.com>
12102         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12103         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12104         * configure: Regenerate.
12106 2013-05-10  Alan Modra  <amodra@gmail.com>
12108         PR target/55033
12109         * varasm.c (default_elf_select_section): Move !DECL_P check..
12110         (get_named_section): ..to here before calling get_section_name.
12111         Adjust assertion.
12112         (default_section_type_flags): Add DECL_P check.
12113         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12114         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12116 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12118         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12119         When using gen_stack_adjust_str with a register offset, add a
12120         REG_FRAME_RELATED_EXPR note.
12122 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12124         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12125         (*vec_extractv4si_zext_mem): Ditto.
12126         (*vec_extractv2di): Add 0->x and x->x alternatives.
12127         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12128         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12130 2013-05-09  Jason Merrill  <jason@redhat.com>
12132         N3639 C++1y VLA support
12133         * gimplify.c (gimplify_vla_decl): Don't touch an existing
12134         DECL_VALUE_EXPR.
12136         * tree.c (build_constructor_va): New.
12137         * tree.h: Declare it.
12139 2013-05-09  Martin Jambor  <mjambor@suse.cz>
12141         PR lto/57084
12142         * gimple-fold.c (canonicalize_constructor_val): Call
12143         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12145 2013-05-09  Jan Hubicka  <jh@suse.cz>
12146             Richard Biener  <rguenther@suse.de>
12148         PR lto/54095
12149         * symtab.c (symtab_make_decl_local): Do not add private names.
12151 2013-05-09  Jan Hubicka  <jh@suse.cz>
12153         PR lto/54095
12154         * symtab.c (insert_to_assembler_name_hash): Handle clones.
12155         (unlink_from_assembler_name_hash): Likewise.
12156         (symtab_prevail_in_asm_name_hash, symtab_register_node,
12157         symtab_unregister_node, symtab_initialize_asm_name_hash,
12158         change_decl_assembler_name): Update.
12160 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
12162         * config/aarch64/aarch64.md: New movtf split.
12163         (*movtf_aarch64): Update.
12164         (aarch64_movdi_tilow): Handle TF modes and rename to
12165         aarch64_movdi_<mode>low.
12166         (aarch64_movdi_tihigh): Handle TF modes and rename to
12167         aarch64_movdi_<mode>high
12168         (aarch64_movtihigh_di): Handle TF modes and rename to
12169         aarch64_mov<mode>high_di
12170         (aarch64_movtilow_di): Handle TF modes and rename to
12171         aarch64_mov<mode>low_di
12172         (aarch64_movtilow_tilow): Remove spurious whitespace.
12173         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12174         splits.
12175         (aarch64_print_operand): Update.
12177 2013-05-09  Alan Modra  <amodra@gmail.com>
12179         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12180         powerpc64le.
12181         * configure: Regenerate.
12183 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12185         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12186         splitter preparation statements.
12187         * config/i386/sse.md (*vec_extract* splitters): Ditto.
12188         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12189         adjust_address_nv.
12191 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12193         * gimple-ssa-strength-reduction.c (count_candidates): Change
12194         return value to int.
12195         (analyze_candidates_and_replace): Change type of length to int.
12197 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12199         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12200         (*vec_extract<mode>): Use VI12_128 mode iterator.
12201         (*vec_extract<mode>_mem): Ditto.
12202         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12203         attribute.
12205 2013-05-08  Diego Novillo  <dnovillo@google.com>
12207         PR bootstrap/54659
12209         Revert:
12210         2012-08-17  Diego Novillo  <dnovillo@google.com>
12212         PR bootstrap/54281
12213         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12214         * config.in: Regenerate.
12215         * configure: Regenerate.
12216         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12218 2013-05-08  Jan Hubicka  <jh@suse.cz>
12220         PR lto/54095
12221         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12222         * cgraph.h (symtab_node_base): Add unique_name.
12223         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12224         input_overwrite_node, input_varpool_node): Stream unique_name.
12225         * cgraphclones.c (cgraph_create_virtual_clone,
12226         cgraph_function_versioning): Set unique_name.
12227         * ipa.c (function_and_variable_visibility): Set unique_name.
12229 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12231         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
12232         (alloc_cand_and_find_basis): Restrict conditional candidate
12233         processing to CAND_MULTs.
12235 2013-05-08  Jan Hubicka  <jh@suse.cz>
12237         PR lto/54095
12238         lto-symtab.c (lto_symtab_symbol_p): New function.
12239         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
12240         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
12241         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
12242         Skip static symbols.
12244 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
12246         PR tree-optimization/57200
12247         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
12248         Only call inform if the preceding warning_at returns true.
12250 2013-05-07  Han Shen  <shenhan@google.com>
12252         * cfgexpand.c (record_or_union_type_has_array_p): New function.
12253         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
12254         * common.opt (fstack-protector-strong): New option.
12255         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
12256         * doc/invoke.texi (Optimization Options): Document
12257         "-fstack-protector-strong".
12258         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
12260 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12262         * config/mips/mips.c (mips_machine_reorg2): Return 0.
12264 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
12266         * ira.c (update_equiv_regs): Add insn having equiv memory even if
12267         it is not lhs of the insn.
12268         (setup_reg_equiv): Remove insn having equiv memory which it is not
12269         lhs of the insn.
12270         * lra-constraints.c (process_address): Try to improve generation
12271         code for address base + disp.
12272         (lra_constraints): Make correct the code for checking insn setting
12273         up backward equivalence.  Remove insn only if it is in the init
12274         insn list.
12275         * lra-eliminations.c (update_reg_eliminate): Change return value.
12276         (lra_eliminate): Use the result.
12278 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
12280         * config/i386/sse.md (ssescalarnummask): New mode attribute.
12281         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
12282         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
12283         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
12284         register target operands.
12285         (*vec_extractv8hi_sse2): New pattern.
12286         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
12287         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
12288         (*vec_extract<mode>_mem): New insn and split pattern.
12290 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
12292         * config/arm/arm.c (arm_asan_shadow_offset): New function.
12293         (TARGET_ASAN_SHADOW_OFFSET): Define.
12294         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
12295         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
12297 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12299         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
12300         (incr_vec_index): Return -1 if increment not found.
12301         (create_add_on_incoming_edge): Assert if increment not found.
12302         (record_increment): Limit number of increments recorded.
12303         (all_phi_incrs_profitable): Return false if an increment not found.
12304         (replace_profitable_candidates): Don't process increments that were
12305         not recorded.
12306         (analyze_candidates_and_replace): Limit size of incr_vec.
12308 2013-05-07  Richard Biener  <rguenther@suse.de>
12310         * calls.c (special_function_p): setjmp-like functions are leaf.
12311         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
12312         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
12314 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
12316         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
12317         (aarch64_simd_mov<mode>): New expander.
12318         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
12319         (aarch64_simd_mov_to_<mode>high): Likewise.
12320         (aarch64_simd_mov_from_<mode>low): Likewise.
12321         (aarch64_simd_mov_from_<mode>high): Likewise.
12322         (aarch64_dup_lane<mode>): Update.
12323         (aarch64_dup_lanedi): New instruction pattern.
12324         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
12325         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
12327 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12329         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
12330         (replace_mult_candidate): Remove unnecessary argument; remove
12331         unnecessary parameter from call to introduce_cast_before_cand.
12332         (replace_unconditional_candidate): Remove unnecessary parameter
12333         from call to replace_mult_candidate.
12334         (replace_conditional_candidate): Likewise.
12335         (insert_initializers): Use make_temp_ssa_name.
12336         (introduce_cast_before_cand): Remove unnecessary argument; use
12337         make_temp_ssa_name.
12338         (replace_one_candidate): Remove unnecessary argument; remove
12339         unnecessary parameter from calls to introduce_cast_before_cand.
12340         (replace_profitable_candidates): Remove unnecessary parameters
12341         from calls to replace_one_candidate.
12343 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12345         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12346         phi def as possibly hiding a basis for a CAND_ADD whose operands
12347         have been commuted in the analysis.
12348         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
12350 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12352         * config/aarch64/aarch64.md
12353         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
12354         shift value between 0-4.
12356 2013-05-07  Richard Biener  <rguenther@suse.de>
12358         * double-int.h (rshift): New overload.
12359         * double-int.c (rshift): New function.
12360         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
12361         (create_reference_ops_from_ref): Remove.
12362         (vn_reference_insert): Use shared ops for constructing the
12363         reference and copy it.
12365 2013-05-07  Richard Biener  <rguenther@suse.de>
12367         PR middle-end/57190
12368         * tree-eh.c (sink_clobbers): Properly propagate
12369         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
12371 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
12373         PR tree-optimization/57149
12374         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
12375         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
12376         collect_phi_def_edges, execute_late_warn_uninitialized): Use
12377         uninit_undefined_value_p instead of ssa_undefined_value_p.
12379         PR debug/57184
12380         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
12381         for modifier == EXPAND_INITIALIZER.
12383 2013-05-07  Anton Blanchard  <anton@samba.org>
12385         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
12386         for powerpc64 little endian.
12387         * configure: Regenerate.
12389 2013-05-06  Graham Stott  <grahams@btinternet.com>
12391         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
12392         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
12393         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
12394         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
12396 2013-05-06  Graham Stott  <grahams@btinternet.com>
12398         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
12399         codes which allow non-lvalues.
12401 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
12403         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
12404         components are all 1s.
12405         (integer_minus_onep): New function.
12406         * tree.h (integer_minus_onep): Declare it.
12407         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
12408         integer_minus_onep instead of integer_all_onesp.
12410 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12412         PR target/52933
12413         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
12414         variations of these patterns.
12416 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12418         * config/i386/i386.md (isa): Add x64_sse4 member.
12419         (enabled): Handle x64_sse4.
12420         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
12421         instruction for 64bit SSE4_1 targets.  Update insn attributes.
12422         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
12423         instruction for SSE4_1 targets.  Update insn attributes.
12424         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12425         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
12426         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
12427         const_1 selector.
12428         (*vec_extractv4si): Rename from *sse4_1_pextrd.
12429         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
12430         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
12432 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12434         PR target/57108
12435         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
12437 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
12439         * final.c (do_assembler_dialects): Don't handle curly braces and
12440         vertical bar escaped by % as dialect delimiters.
12441         (output_asm_insn): Print curly braces and vertical bar if escaped
12442         by % and ASSEMBLER_DIALECT defined.
12443         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
12444         * doc/tm.texi: Regenerated.
12446 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12448         * config/mips/mips.c: Include tree-pass.h.
12449         (mips_reorg): Split in pre- and post-dbr_schedule parts.
12450         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
12451         (pass_mips_machine_reorg2): New machine specific pass.
12452         (insert_pass_mips_machine_reorg2): New pass plugin definition.
12453         (mips_option_override): Register the new pass.
12454         * rtl.h (cleanup_barriers): Remove prototype.
12455         (dbr_schedule): Likewise.
12456         * jump.c (cleanup_barriers): Make static.
12457         * reorg.c (dbr_schedule): Likewise.
12459 2013-05-06  Richard Biener  <rguenther@suse.de>
12461         PR tree-optimization/57185
12462         * tree-parloops.c (add_field_for_reduction): Handle anonymous
12463         SSA names properly.
12465 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12467         PR target/57106
12468         * config/i386/i386.c (add_parameter_dependencies): Add dependence
12469         between "first_arg" and "insn", not "last" and "insn".
12471 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12473         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
12474         (find_candidates_in_block): Re-enable slsr_process_phi.
12475         (create_phi_basis): Fix double counting of candidate adjustment.
12477 2013-05-06  Richard Biener  <rguenther@suse.de>
12479         PR middle-end/57147
12480         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
12481         the edge is also fallthru, preserve it and just clear the
12482         abnormal flag.
12483         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
12484         also complex, preserve that and just clear the fallthru flag.
12485         * tree-inline.c (update_ssa_across_abnormal_edges): Also
12486         update virtual operands.
12488 2013-05-06  Alan Modra  <amodra@gmail.com>
12490         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
12491         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
12492         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
12493         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
12494         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12495         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
12497 2013-05-06  Alan Modra  <amodra@gmail.com>
12499         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
12500         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
12501         (DEFAULT_ASM_ENDIAN): Define.
12502         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12503         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
12504         Update -K PIC clause from sysv4.h.
12505         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
12506         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
12508 2013-05-06  Alan Modra  <amodra@gmail.com>
12510         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
12511         twice for little-endian.
12512         (ashrdi3_no_power, ashrdi3): Support little-endian.
12514 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12516         PR target/55303
12517         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
12518         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
12519         related expanders.
12520         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
12521         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
12522         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
12523         New predicates.
12525 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
12526             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12528         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
12529         * config/pa/pa.opt: Make mbig-switch a no-op.
12530         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
12531         (CASE_VECTOR_MODE): Always return SImode.
12532         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
12533         for the !TARGET_BIG_SWITCH case.
12534         * config/pa/pa-linux.h: Likewise.
12535         * config/pa/pa-openbsd.h: Likewise.
12536         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
12537         * config/pa/pa.md (short_jump): Remove define_insn.
12538         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
12539         (casesi0): Remove define_insn.
12540         (type): Remove btable_branch.
12541         (pa_combine_type): Likewise.
12542         (in_nullified_branch_delay): Likewise.
12543         (in_call_delay): Likewise.
12544         (define_delay): Likewise.
12545         (define_insn_reservation "Z3"): Likewise.
12546         (define_insn_reservation "Z4"): Likewise.
12547         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
12548         (pa_adjust_insn_length): Remove adjustment for btable branches.
12549         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
12550         and mno-big-switch
12552 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
12554         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12555         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
12556         Add m->r,x alternatives.
12557         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
12558         splitters using SWI48x mode iterator.
12559         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
12560         TARGET_64BIT.  Add m->x alternative.
12561         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
12562         Add o->x alternative.  Enable for TARGET_SSE.
12563         (sse_storeq): Remove expander.
12564         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
12565         with memory input operand.
12566         (*vec_extractv2di_1 splitter): New.
12567         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
12568         * config/i386/i386.md (ssevecmodelower): New mode attribute.
12570 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12572         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
12573         (INT_LOWPART): Delete.
12574         (extract_MB): Adjust.
12575         (extract_ME): Adjust.
12576         (print_operand): Adjust.
12578 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
12580         * config/rs6000/predicates.md (reg_or_add_cint_operand,
12581         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12582         (reg_or_logical_cint_operand, easy_fp_constant,
12583         logical_const_operand): Delete "CONST_DOUBLE" case.
12584         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
12585         "HOST_BITS_PER_WIDE_INT == 64" test.
12586         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
12587         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
12588         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
12589         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
12590         test.
12591         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
12592         CONST_DOUBLE DImode/VOIDmode case.
12593         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
12594         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12595         CONST_DOUBLE VOIDmode case.
12596         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12597         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
12598         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
12599         Delete CONST_DOUBLE case.
12600         (splitters for mov FMOVE64 const_double): Delete
12601         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
12602         "HOST_BITS_PER_WIDE_INT >= 64" test.
12603         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
12604         case.
12605         (mov DI const_double): Delete.
12607 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
12609         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
12610         on op shows all bits zero in mode of a lowpart subreg, return zero.
12612 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
12614         PR target/57150
12615         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
12616         to save TFmode registers and DImode to save TImode registers for
12617         caller save operations.
12618         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
12619         mark being partially clobbered since they only use the first
12620         double word.
12622         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
12623         and TDmode only use the upper 64-bits of each VSX register.
12625 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12627         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
12628         (find_candidates_in_block): Disable slsr_process_phi.
12630 2013-05-03  Guozhi Wei  <carrot@google.com>
12632         * coverage.c (coverage_obj_init): Move the construction of gcov
12633         constructor to ...
12634         (build_init_ctor): ... here.
12636 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12638         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
12639         (slsr_cand_d): Redefine def_phi.
12640         (stride_status, phi_adjust_status, count_phis_status): New enums.
12641         (find_phi_def): New.
12642         (find_basis_for_base_expr): New.
12643         (find_basis_for_candidate): Handle hidden bases.
12644         (alloc_cand_and_find_basis): Handle phi candidates.
12645         (slsr_process_phi): New.
12646         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
12647         (create_mul_imm_cand): Likewise.
12648         (create_add_ssa_cand): Exclude phi base candidates.
12649         (create_add_imm_cand): Likewise.
12650         (slsr_process_cast): Likewise.
12651         (slsr_process_copy): Likewise.
12652         (find_candidates_in_block): Handle phi candidates.
12653         (dump_candidate): Likewise.
12654         (unconditional_cands): Delete.
12655         (unconditional_cands_with_known_stride_p): Delete.
12656         (phi_dependent_cand_p): New.
12657         (cand_increment): Handle phi-dependent candidates.
12658         (replace_dependent): Delete.
12659         (replace_mult_candidate): New.
12660         (replace_unconditional_candidate): New.
12661         (incr_vec_index): Move to avoid forward reference.
12662         (create_add_on_incoming_edge): New.
12663         (create_phi_basis): New.
12664         (replace_dependents): Delete.
12665         (replace_conditional_candidate): New.
12666         (phi_add_costs): New.
12667         (replace_uncond_cands_and_profitable_phis): New.
12668         (record_increment): Handle phi adjustments.
12669         (record_phi_increments): New.
12670         (record_increments): Handle phi adjustments.
12671         (phi_incr_cost): New.
12672         (lowest_cost_path): Handle phis.
12673         (total_savings): Likewise.
12674         (analyze_increments): Likewise.
12675         (ncd_with_phi): New.
12676         (ncd_of_cand_and_phis): New.
12677         (nearest_common_dominator_for_cands): Handle phi increments.
12678         (all_phi_incrs_profitable): New.
12679         (replace_profitable_candidates): Handle phi-dependent candidates.
12680         (analyze_candidates_and_replace): Likewise.
12682 2013-05-03  Teresa Johnson  <tejohnson@google.com>
12684         PR bootstrap/57154
12685         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
12686         do not exceed REG_BR_PROB_BASE.
12688 2013-05-03  Jeff Law  <law@redhat.com>
12690         PR tree-optimization/57144
12691         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
12692         operand of the condition will bit into the new type when eliminating
12693         a cast feeding a condition.
12695 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
12697         PR rtl-optimization/57130
12698         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
12699         of COMPARE as in_code to the recursive call if needed.
12701 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
12703         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
12704         (enabled): Handle new members.
12705         * config/i386/sse.md (*vec_concatv2si): Merge from
12706         *vec_concatv2si_sse2 and vec_concatv2si_sse.
12707         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
12709 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
12711         PR tree-optimization/57027
12712         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
12713         for fnms opportunity, check we got the prerequisite kind
12714         of tree / gimple before using accessor functions.
12716 2013-05-03  Richard Biener  <rguenther@suse.de>
12718         * double-int.h (lshift): New overload without precision
12719         and arith argument.
12720         (operator *=, operator +=, operator -=): Move ...
12721         * double-int.c (operator *=, operator +=, operator -=): ... here
12722         and implement more efficiently.
12723         (mul_double_with_sign): Remove.
12724         (lshift_double): Adjust to take unsinged shift argument, push
12725         dispatching code to callers.
12726         (mul_double_wide_with_sign): Add early out for callers that
12727         are not interested in high parts or overflow.
12728         (lshift): New function.
12729         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
12730         dispatch code here.
12731         (lrotate, rrotate): Use logical shifts.
12732         * expr.c (get_inner_reference): Use lshift.
12733         * fixed-value.c (do_fixed_divide): Likewise.
12734         * tree-dfa.c (get_ref_base_and_extent): Likewise.
12735         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
12736         (indirect_refs_may_alias_p): Likewise.
12737         (stmt_kills_ref_p_1): Likewise.
12739 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12741         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
12743 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
12745         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
12746         scalar form of FABD instruction.
12748 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
12750         * lra-constraints.c (process_alt_operands): Add checking alt
12751         number to choose the best alternative.
12753 2013-05-02  Richard Biener  <rguenther@suse.de>
12755         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
12756         bitmap and its handling.
12757         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
12759 2013-05-02  Richard Biener  <rguenther@suse.de>
12761         PR middle-end/57140
12762         * tree-inline.c (copy_loops): Properly handle removed loops.
12763         (copy_cfg_body): Mark destination loops for fixup if source
12764         loops needed fixup.
12766 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
12768         PR target/56732
12769         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
12770         generating simple_return for naked functions.
12772 2013-05-02  Martin Jambor  <mjambor@suse.cz>
12774         PR middle-end/56988
12775         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
12776         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
12777         flags match.
12778         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
12779         ipa_agg_replacement_value structures.
12780         (known_aggs_to_agg_replacement_list): Likewise.
12781         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
12782         (read_agg_replacement_chain): Likewise.
12783         (ipcp_transform_function): Also check that by_ref flags match.
12785 2013-05-02  Richard Biener  <rguenther@suse.de>
12787         * graphds.h (struct graph): Add obstack member.
12788         * graphds.c (new_graph): Initialize obstack and allocate
12789         vertices from it.
12790         (add_edge): Allocate edge from the obstack.
12791         (free_graph): Free the obstack instead of all edges and vertices.
12793 2013-05-02  Teresa Johnson  <tejohnson@google.com>
12795         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
12796         divides.
12797         * cfg.c (update_bb_profile_for_threading): Ditto.
12798         * tree-inline.c (copy_bb): Ditto.
12799         (copy_edges_for_bb): Ditto.
12800         (initialize_cfun): Ditto.
12801         (copy_cfg_body): Ditto.
12802         (expand_call_inline): Ditto.
12803         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
12804         (estimate_node_size_and_time): Ditto.
12805         (inline_merge_summary): Ditto.
12806         * cgraphclones.c (cgraph_clone_edge): Ditto.
12807         (cgraph_clone_node): Ditto.
12808         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12809         (compute_trg_info): Ditto.
12811 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12813         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
12814         S reg when fp attribute set.
12815         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
12817 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
12819         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
12820         New pattern.
12821         (*and_one_cmplsi3_compare0_uxtw): Likewise.
12822         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
12823         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
12825 2013-05-02  Richard Biener  <rguenther@suse.de>
12827         * tree-scalar-evolution.c (scev_info_hasher): Remove.
12828         (struct instantiate_cache_entry): New type.
12829         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
12830         (struct instantiate_cache_type): New type.
12831         (set_instantiated_value, get_instantiated_value): Remove.
12832         (get_instantiated_value_entry): New function.
12833         (instantiate_scev_name): Use the new cache and adjust.
12834         (instantiate_scev_poly): Adjust.
12835         (instantiate_scev_binary): Likewise.
12836         (instantiate_array_ref): Likewise.
12837         (instantiate_scev_convert): Likewise.
12838         (instantiate_scev_not): Likewise.
12839         (instantiate_scev_3): Likewise.
12840         (instantiate_scev_2): Likewise.
12841         (instantiate_scev_r): Likewise.
12842         (instantiate_scev): Likewise.
12843         (resolve_mixers): Likewise.
12845 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
12847         PR target/57091
12848         * lra-constraints.c (best_small_class_operands_num): Remove.
12849         (process_alt_operands): Remove small_class_operands_num.  Take
12850         small classes operands into losers and only if the operand is not
12851         matched.  Modify debugging output.
12852         (curr_insn_transform): Remove best_small_class_operands_num.
12853         Print insn name.
12855 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12857         * config/aarch64/aarch64-builtins.c
12858         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
12859         * config/aarch64/aarch64-simd-builtins.def
12860         (reduc_splus_): Add new modes.
12861         (reduc_uplus_): New.
12862         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
12863         (reduc_uplus_v4sf): Likewise.
12864         (reduc_splus_v4sf): Likewise.
12865         (aarch64_addv<mode>): Likewise.
12866         (reduc_uplus_<mode>): Likewise.
12867         (reduc_splus_<mode>): Likewise.
12868         (aarch64_addvv2di): Likewise.
12869         (reduc_uplus_v2di): Likewise.
12870         (reduc_splus_v2di): Likewise.
12871         (aarch64_addvv2si): Likewise.
12872         (reduc_uplus_v2si): Likewise.
12873         (reduc_splus_v2si): Likewise.
12874         (reduc_<sur>plus_<mode>): New.
12875         (reduc_<sur>plus_v2di): Likewise.
12876         (reduc_<sur>plus_v2si): Likewise.
12877         (reduc_<sur>plus_v4sf): Likewise.
12878         (aarch64_addpv4sf): Likewise.
12879         * config/aarch64/arm_neon.h
12880         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
12881         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
12882         add UNSPEC_SADDV, UNSPEC_UADDV.
12883         (SUADDV): New.
12884         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
12886 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12888         * config/aarch64/arm_neon.h
12889         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
12891 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12893         * config/aarch64/aarch64-builtins
12894         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
12896 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12898         * config/aarch64/aarch64-simd-builtins.def
12899         (reduc_smax_): New.
12900         (reduc_smin_): Likewise.
12901         (reduc_umax_): Likewise.
12902         (reduc_umin_): Likewise.
12903         (reduc_smax_nan_): Likewise.
12904         (reduc_smin_nan_): Likewise.
12905         (fmax): Remove.
12906         (fmin): Likewise.
12907         (smax): Update for V2SF, V4SF and V2DF modes.
12908         (smin): Likewise.
12909         (smax_nan): New.
12910         (smin_nan): Likewise.
12911         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
12912         (<su><maxmin><mode>3): ...This, refactor.
12913         (s<maxmin><mode>3): New.
12914         (<maxmin_uns><mode>3): Likewise.
12915         (reduc_<maxmin_uns>_<mode>): Refactor.
12916         (reduc_<maxmin_uns>_v4sf): Likewise.
12917         (reduc_<maxmin_uns>_v2si): Likewise.
12918         (aarch64_<fmaxmin><mode>: Remove.
12919         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
12920         new builtin names.
12921         (vmin<q>_f<32,64>): Likewise.
12922         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12923         (FMAXMIN): New.
12924         (su): Add mappings for smax, smin, umax, umin.
12925         (maxmin): New.
12926         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
12927         (FMAXMIN): Rename as...
12928         (FMAXMIN_UNS): ...This.
12929         (maxminv): Remove.
12930         (fmaxminv): Likewise.
12931         (fmaxmin): Likewise.
12932         (maxmin_uns): New.
12933         (maxmin_uns_op): Likewise.
12935 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12937         * config/aarch64/arm_neon.h
12938         (vac<ge, gt><sd>_f<32, 64>): Rename to...
12939         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
12940         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
12942 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12944         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
12945         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
12947 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12949         * config/aarch64/aarch64-simd.md
12950         (vcond<mode>_internal): Handle special cases for constant masks.
12951         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
12952         (vcondu<mode><mode>): Likewise.
12953         (vcond<v_cmp_result><mode>): New.
12955 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12957         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
12958         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
12959         * config/aarch64/aarch64-simd-builtins.def
12960         (cmeq): Update to BUILTIN_VALLDI.
12961         (cmgt): Likewise.
12962         (cmge): Likewise.
12963         (cmle): Likewise.
12964         (cmlt): Likewise.
12965         * config/aarch64/arm_neon.h
12966         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
12967         to builtins or C as appropriate.
12969 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
12971         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
12972         (cmgeu): ...This.
12973         (cmhi): Rename to...
12974         (cmgtu): ...This.
12975         * config/aarch64/aarch64-simd.md
12976         (simd_mode): Add SF.
12977         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
12978         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
12979         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
12980         (cstore<mode>_neg): ...This.
12981         * config/aarch64/iterators.md
12982         (VALLF): new.
12983         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
12984         (COMPARISONS): New.
12985         (UCOMPARISONS): Likewise.
12986         (optab): Add missing comparisons.
12987         (n_optab): New.
12988         (cmp_1): Likewise.
12989         (cmp_2): Likewise.
12990         (CMP): Likewise.
12991         (cmp): Remove.
12992         (VCMP_S): Likewise.
12993         (VCMP_U): Likewise.
12994         (V_cmp_result): Add DF, SF modes.
12995         (v_cmp_result): Likewise.
12996         (v): Likewise.
12997         (vmtype): Likewise.
12998         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
13000 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
13002         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
13003         define_insn to define_insn_and_split.
13004         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
13005         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
13006         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
13007         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
13008         (thumb2_negscc): Likewise.
13010 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13012         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13014 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13016         * config/arm/thumb2.md: Remove trailing whitespaces.
13018 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
13020         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13021         Use gen_int_mode rather than GEN_INT.
13023 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
13025         * value-prof.c (stream_in_histogram_value): Remove the strayed
13026         debug_gimple_stmt.
13028 2013-04-30  Richard Biener  <rguenther@suse.de>
13030         PR middle-end/57122
13031         * cfghooks.c (split_edge): Properly check for the loop latch edge.
13033 2013-04-30  Richard Biener  <rguenther@suse.de>
13035         PR middle-end/57107
13036         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13038 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13040         PR rtl-optimization/56957
13041         PR rtl-optimization/57105
13042         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13043         variable.  Use just INSN_UID for determining whether an insn
13044         should be only disconnected from the insn stream.
13045         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13047 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13049         PR tree-optimization/57104
13050         * tsan.c (instrument_expr): Don't instrument accesses to
13051         DECL_HARD_REGISTER VAR_DECLs.
13053 2013-04-30  Richard Biener  <rguenther@suse.de>
13055         * function.h (loops_for_fn): New inline function.
13056         (set_loops_for_fn): Likewise.
13057         * cfgloop.h (place_new_loop): Add struct function parameter.
13058         (get_loop): Likewise.
13059         (get_loops): Likewise.
13060         (number_of_loops): Likewise.
13061         (fel_next): Adjust.
13062         (fel_init): Likewise.
13063         * cfg.c (get_loop_copy): Adjust.
13064         * cfgloop.c (flow_loops_dump): Likewise.
13065         (record_loop_exits): Likewise.
13066         (verify_loop_structure): Likewise.
13067         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13068         (estimate_reg_pressure_cost): Likewise.
13069         (mark_loop_exit_edges): Likewise.
13070         * cfgloopmanip.c (place_new_loop): Likewise.
13071         (add_loop): Likewise.
13072         (duplicate_loop): Likewise.
13073         * graph.c (draw_cfg_nodes): Likewise.
13074         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13075         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13076         (extract_affine_chrec): Likewise.
13077         (build_scop_iteration_domain): Likewise.
13078         * graphite.c (graphite_initialize): Likewise.
13079         * ira-build.c (create_loop_tree_nodes): Likewise.
13080         (more_one_region_p): Likewise.
13081         (rebuild_regno_allocno_maps): Likewise.
13082         (mark_loops_for_removal): Likewise.
13083         (mark_all_loops_for_removal): Likewise.
13084         (remove_unnecessary_regions): Likewise.
13085         (ira_build): Likewise.
13086         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13087         * loop-init.c (fix_loop_structure): Likewise.
13088         (gate_rtl_move_loop_invariants): Likewise.
13089         (gate_rtl_unswitch): Likewise.
13090         (gate_rtl_unroll_and_peel_loops): Likewise.
13091         (rtl_doloop): Likewise.
13092         * lto-streamer-in.c (input_cfg): Likewise.
13093         * lto-streamer-out.c (output_cfg): Likewise.
13094         * modulo-sched.c (sms_schedule): Likewise.
13095         * predict.c (tree_estimate_probability): Likewise.
13096         (tree_estimate_probability_driver): Likewise.
13097         (estimate_loops): Likewise.
13098         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13099         (move_sese_region_to_fn): Likewise.
13100         (debug_loop_num): Likewise.
13101         * tree-chrec.c (chrec_evaluate): Likewise.
13102         (hide_evolution_in_other_loops_than_loop): Likewise.
13103         (chrec_component_in_loop_num): Likewise.
13104         (reset_evolution_in_loop): Likewise.
13105         (evolution_function_is_invariant_rec_p): Likewise.
13106         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13107         * tree-inline.c (copy_loops): Likewise.
13108         (copy_cfg_body): Likewise.
13109         (tree_function_versioning): Likewise.
13110         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13111         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13112         Likewise.
13113         (add_to_evolution_1): Likewise.
13114         (scev_const_prop): Likewise.
13115         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13116         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13117         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13118         (tree_ssa_lim_initialize): Likewise.
13119         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13120         (verify_loop_closed_ssa): Likewise.
13121         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13122         (tree_ssa_loop_im): Likewise.
13123         (tree_ssa_loop_unswitch): Likewise.
13124         (tree_vectorize): Likewise.
13125         (check_data_deps): Likewise.
13126         (tree_ssa_loop_ivcanon): Likewise.
13127         (tree_ssa_loop_bounds): Likewise.
13128         (tree_complete_unroll): Likewise.
13129         (tree_complete_unroll_inner): Likewise.
13130         (tree_parallelize_loops): Likewise.
13131         (tree_ssa_loop_prefetch): Likewise.
13132         (tree_ssa_loop_ivopts): Likewise.
13133         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13134         * tree-vectorizer.c (vectorize_loops): Likewise.
13136 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
13138         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13139         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13140         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13141         with EABI_LINK_SPEC.
13143 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13145         PR target/44578
13146         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13147         alternative.
13149 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
13151         PR target/57097
13152         * lra-constraints.c (process_alt_operands): Discourage a bit more
13153         using memory for pseudos.  Print cost dump for alternatives.
13154         Modify cost values for conflicts with early clobbers.
13155         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13157 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13159         PR target/57098
13160         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13162 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13164         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13165         from/to S register.
13166         (movdi_aarch64): Support LDR/STR from/to D register.
13168 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13170         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13171         or higher by default.
13173 2013-04-29  Richard Biener  <rguenther@suse.de>
13175         PR middle-end/57075
13176         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13177         even if not adding abnormal edges for calls that can make
13178         abnormal gotos.
13180 2013-04-29  Richard Biener  <rguenther@suse.de>
13182         PR middle-end/57103
13183         * tree-cfg.c (move_stmt_op): Fix condition under which to update
13184         TREE_BLOCK.
13185         (move_stmt_r): Remove redundant checking.
13187 2013-04-29  Teresa Johnson  <tejohnson@google.com>
13189         PR bootstrap/57077
13190         * basic-block.h (apply_scale): New function.
13191         (apply_probability): Use apply_scale.
13192         * gimple-streamer-in.c (input_bb): Ditto.
13193         * lto-streamer-in.c (input_cfg): Ditto.
13194         * lto-cgraph.c (merge_profile_summaries): Ditto.
13195         * tree-optimize.c (execute_fixup_cfg): Ditto.
13196         * tree-inline.c (copy_bb): Update comment to use apply_scale.
13197         (copy_edges_for_bb): Ditto.
13198         (copy_cfg_body): Ditto.
13200 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13202         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13203         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13204         (tail_merge_optimize): Handle current_loops == NULL.
13206 2013-04-26  Jeff Law  <law@redhat.com>
13208         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13209         (simplify_cond_using_ranges): Generalize code to simplify
13210         COND_EXPRs where one argument is a constant and the other
13211         is an SSA_NAME created by an integral type conversion.
13213 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13215         * config/arm/arm.md (store_minmaxsi): Use only when
13216         optimize_insn_for_size_p.
13218 2013-04-29  Christian Bruel  <christian.bruel@st.com>
13220         PR target/57108
13221         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13223 2013-04-29  Richard Biener  <rguenther@suse.de>
13225         PR middle-end/57089
13226         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13227         loop tree make sure to schedule a fixup for the child as well.
13228         (expand_omp_for_generic): Properly add loops.
13229         (expand_omp_for_static_nochunk): Likewise.
13230         (expand_omp_for_static_chunk): Likewise.
13231         (expand_omp_for): For the degenerate case fixup loops.
13232         (expand_omp_sections): Fix default bb placement in loops.
13233         (expand_omp_atomic_pipeline): Properly add loops.
13235 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13237         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
13239 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13241         * tree-ssa-tail-merge.c: Update header comment.
13243 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13245         * config/aarch64/arm_neon.h
13246         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
13247         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
13248         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
13249         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
13250         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
13251         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
13252         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
13253         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
13255 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13257         * config/aarch64/aarch64-simd.md
13258         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
13259         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
13260         fix_trunc, fixuns_trunc.
13261         (ftrunc<VDQF:mode>2): New.
13262         * config/aarch64/iterators.md (optab): Add fix, fixuns.
13263         (fix_trunc_optab): New.
13265 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13267         * config/aarch64/aarch64-builtins.c
13268         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
13269         iceilf, lround, iroundf.
13271 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13273         PR target/54349
13274         * config/i386/i386.h (enum ix86_tune_indices)
13275         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
13276         New, split from X86_TUNE_INTER_UNIT_MOVES.
13277         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
13278         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
13279         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
13280         (TARGET_INTER_UNIT_MOVES): Remove.
13281         * config/i386/i386.c (initial_ix86_tune_features): Update.
13282         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
13283         (ix86_expand_convert_uns_didf_sse): Use
13284         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13285         (ix86_expand_vector_init_one_nonzero): Ditto.
13286         (ix86_expand_vector_init_interleave): Ditto.
13287         (inline_secondary_memory_needed): Return true for moves from SSE class
13288         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
13289         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13290         * config/i386/constraints.md (Yi, Ym): Depend on
13291         TARGET_INTER_UNIT_MOVES_TO_VEC.
13292         (Yj, Yn): New constraints.
13293         * config/i386/i386.md (*movdi_internal): Change constraints of
13294         operand 1 from Yi to Yj and from Ym to Yn.
13295         (*movsi_internal): Ditto.
13296         (*movdf_internal): Ditto.
13297         (*movsf_internal): Ditto.
13298         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
13299         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13300         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
13301         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
13302         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
13303         * config/i386/sse.md (movdi_to_sse): Ditto.
13304         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
13305         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
13306         TARGET_INTER_UNIT_MOVES.
13307         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
13308         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
13309         instead of TARGET_INTER_UNIT_MOVES.
13310         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
13311         operand 1 from Yi to Yj and from Ym to Yn.
13313 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13315         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
13316         (float_truncate_hi_): Likewise.
13317         (float_extend_lo_): Likewise.
13318         (float_truncate_lo_): Likewise.
13319         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
13320         (aarch64_float_extend_lo_v2df): Likewise.
13321         (vec_unpacks_hi_v4sf): Likewise.
13322         (aarch64_float_truncate_lo_v2sf): Likewise.
13323         (aarch64_float_truncate_hi_v4sf): Likewise.
13324         (vec_pack_trunc_v2df): Likewise.
13325         (vec_pack_trunc_df): Likewise.
13327 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13329         * config/aarch64/aarch64-builtins.c
13330         (aarch64_fold_builtin): Fold float conversions.
13331         * config/aarch64/aarch64-simd-builtins.def
13332         (floatv2si, floatv4si, floatv2di): New.
13333         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
13334         * config/aarch64/aarch64-simd.md
13335         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
13336         * config/aarch64/iterators.md (FLOATUORS): New.
13337         (optab): Add float, floatuns.
13338         (su_optab): Likewise.
13340 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13342         * config/aarch64/aarch64-builtins.c
13343         (aarch64_builtin_vectorized_function): Use new names for
13344         fcvt builtins.
13345         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
13346         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
13347         (fcvtzu): Split as...
13348         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
13349         (fcvtas): Split as...
13350         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
13351         (fcvtau): Split as...
13352         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
13353         (fcvtps): Split as...
13354         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
13355         (fcvtpu): Split as...
13356         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
13357         (fcvtms): Split as...
13358         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
13359         (fcvtmu): Split as...
13360         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
13361         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
13362         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
13363         (lfrintnusf, lfrintnudf): Likewise.
13364         * config/aarch64/aarch64-simd.md
13365         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
13366         define_insn.
13367         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
13368         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
13369         (fcvt_pattern): Likewise.
13371 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13373         * config/aarch64/aarch64-simd.md
13374         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
13375         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
13377 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13379         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
13380         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
13381         (vrnd<a,m,n,p>_f32): Implement using builtins.
13382         (vrnd<i,x><q>_f<32, 64>): New.
13384 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13386         * config/aarch64/aarch64-builtins.c
13387         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
13388         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
13389         (frintz): Rename to...
13390         (btrunc): ...this.
13391         (frintp): Rename to...
13392         (ceil): ...this.
13393         (frintm): Rename to...
13394         (floor): ...this.
13395         (frinti): Rename to...
13396         (nearbyint): ...this.
13397         (frintx): Rename to...
13398         (rint): ...this.
13399         (frinta): Rename to...
13400         (round): ...this.
13401         * config/aarch64/aarch64-simd.md
13402         (aarch64_frint<frint_suffix><mode>): Delete.
13403         (<frint_pattern><mode>2): Convert to insn.
13404         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
13405         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
13406         (frint_pattern): Likewise.
13407         (frint_suffix): Likewise.
13409 2013-04-29  Richard Biener  <rguenther@suse.de>
13411         PR tree-optimization/57081
13412         * loop-init.c: Include tree-flow.h.
13413         (loop_optimizer_finalize): Free number of iteration estimates.
13414         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
13416 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
13418         PR tree-optimization/57083
13419         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
13420         non-singleton shift count range, zero extend low_bound for uns case.
13422         * config/i386/predicates.md (general_vector_operand): New predicate.
13423         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
13424         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
13425         if they aren't nonimmediate operands.  If their original values
13426         satisfy const_vector_equal_evenodd_p, don't shift them.
13427         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
13428         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
13429         if not nonimmediate_operand.
13430         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
13431         instead of register_operand.
13432         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
13434 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
13436         * stor-layout.c (finalize_size_functions): Allocate a structure and
13437         reset cfun before dumping the functions.
13439 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
13441         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
13443         PR target/56866
13444         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
13445         use xop_pmacsdqh if uns_p.
13446         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
13447         the immediate rotate count.
13449 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
13451         * rtl.h (struct rtx_def): Add comment for field jump.
13452         (LRA_SUBREG_P): New macro.
13453         * recog.c (register_operand): Check LRA_SUBREG_P.
13454         * lra.c (lra): Add note at the end of RTL code. Align non-empty
13455         stack frame.
13456         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13457         (lra_final_code_change): Skip subreg change for operators.
13458         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13459         if there are no operand changes.
13460         * lra-constraints.c (curr_insn_set): New.
13461         (match_reload): Set LRA_SUBREG_P.
13462         (emit_spill_move): Ditto.
13463         (check_and_process_move): Use curr_insn_set. Process only single
13464         set insns.  Don't initialize sec_mem_p and change_p.
13465         (simplify_operand_subreg): Use LRA_SUBREG_P.
13466         (reg_in_class_p): New function.
13467         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
13468         of #ifdef.  Add code to remove cycling.
13469         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
13470         non-null disp.  Reload inner instead of disp when base and index
13471         are null.  Try to put lo_sum into register.
13472         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
13473         (check_and_process_move): Move code for move cost check to
13474         simple_move_p.  Remove equiv_substitution.
13475         (simple_move_p): New function.
13476         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
13477         curr_insn_set.  Call check_and_process_move only for single set
13478         insns.  Use the new function.  Move call of check_and_process_move
13479         after operand equiv substitution and address process.
13481 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13483         PR go/57045
13484         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
13485         with nonlocal goto receivers or returns twice calls, ignore
13486         unininitialized values from abnormal edges to nl goto receiver
13487         or returns twice call.
13489 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13491         PR tree-optimization/57051
13492         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
13493         and VEC_RSHIFT_EXPR if shift count is a multiple of element
13494         bitsize.
13496 2013-04-26  Richard Biener  <rguenther@suse.de>
13498         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
13499         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
13500         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
13501         (fixup_loop_arrays_after_move): New function.
13502         (move_sese_region_to_fn): Properly outline the loop tree parts
13503         of the SESE region.
13505 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
13507         * config/i386/i386.md (type, unit): Fix long lines.
13509 2013-04-26  Richard Biener  <rguenther@suse.de>
13511         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
13512         (lto-streamer-out.o): Likewise.
13513         * cfgloop.c (init_loops_structure): Export, add struct function
13514         argument and adjust.
13515         (flow_loops_find): Adjust.
13516         * cfgloop.h (enum loop_estimation): Add EST_LAST.
13517         (init_loops_structure): Declare.
13518         * lto-streamer-in.c: Include cfgloop.h.
13519         (input_cfg): Input the loop tree.
13520         * lto-streamer-out.c: Include cfgloop.h.
13521         (output_cfg): Output the loop tree.
13522         (output_struct_function_base): Do not drop PROP_loops.
13524 2013-03-26  Richard Biener  <rguenther@suse.de>
13526         * tree-cfg.c (execute_build_cfg): Build the loop tree.
13527         (pass_build_cfg): Provide PROP_loops.
13528         (move_sese_region_to_fn): Remove loops that are outlined into fn
13529         for now.
13530         * tree-inline.c: Include cfgloop.h.
13531         (initialize_cfun): Do not drop PROP_loops.
13532         (copy_loops): New function.
13533         (copy_cfg_body): Copy loop structure.
13534         (tree_function_versioning): Initialize destination loop tree.
13535         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
13536         (pass_parallelize_loops): Do IL verification.
13537         * loop-init.c (loop_optimizer_init): Fixup loops if required.
13538         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
13539         the CFG make sure we fixup loops as well.
13540         * tree-ssa-tail-merge.c: Include cfgloop.h.
13541         (replace_block_by): When merging loop latches mark loops for fixup.
13542         * lto-streamer-out.c (output_struct_function_base): Drop
13543         PROP_loops for now.
13544         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
13545         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
13546         * ipa-split.c: Include cfgloop.h.
13547         (split_function): Add the new return block to the loop tree root.
13548         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
13549         whether we have removed the forwarder block.
13550         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
13551         * cfgloop.h (place_new_loop): Declare.
13552         * cfgloopmanip.c (place_new_loop): Export.
13553         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
13554         (tree-switch-conversion.o): Likewise.
13555         (tree-complex.o): Likewise.
13556         (tree-inline.o): Likewise.
13557         (tree-ssa-tailmerge.o): Likewise.
13558         (ipa-split.o): Likewise.
13559         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
13560         (tree-ssa-copy.o): Likewise.
13561         * tree-switch-conversion.c: Include cfgloop.h
13562         (process_switch): If we emit a bit-test cascade, schedule loops
13563         for fixup.
13564         * tree-complex.c: Include cfgloop.h.
13565         (expand_complex_div_wide): Properly add new basic-blocks to loops.
13566         * asan.c: Include cfgloop.h.
13567         (create_cond_insert_point): Properly add new basic-blocks to
13568         loops, schedule loop fixup.
13569         * cfgloop.c (verify_loop_structure): Check that looks are not
13570         marked for fixup.
13571         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
13572         to loops.
13573         (expand_omp_for_generic): Likewise.
13574         (expand_omp_sections): Likewise.
13575         (expand_omp_atomic_pipeline): Schedule loops for fixup.
13576         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
13577         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
13578         is initialized, not when loops are present.
13579         * tree-parloops.c (parallelize_loops): Remove checking here.
13580         * passes.c (init_optimization_passes): Schedule a copy-propagation
13581         pass before complete unrolling of inner loops.
13583 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13585         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
13586         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
13587         (colorize_init): Add argument to _WIN32 version.
13588         * toplev.c: Include diagnostic-color.h.
13589         (process_options): Default to -fdiagnostics-color=auto if
13590         GCC_COLORS env var is in the environment.
13591         * common.opt (fdiagnostics-color=): Add Var and Init.
13592         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
13593         env var is in the environment, the default is auto rather than never.
13595         * diagnostic.h (file_name_as_prefix): Add context argument.
13596         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
13597         the string as locus.
13598         * langhooks.c (lhd_print_error_function): Adjust caller.
13600 2013-04-25  Lawrence Crowl  <crowl@google.com>
13602         * var-tracking.c (shared_hash_def::htab):
13603         Change type to hash_table.  Update dependent calls and types.
13605 2013-04-25  Lawrence Crowl  <crowl@google.com>
13607         * Makefile.in: Update as needed below.
13609         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
13610         Move declaration to after the type's method definitons.
13612         * attribs.c (htab_t scoped_attributes::attribute_hash):
13613         Change type to hash_table.  Update dependent calls and types.
13615         * bitmap.c (htab_t bitmap_desc_hash):
13616         Change type to hash_table.  Update dependent calls and types.
13618         * cselib.c (htab_t cselib_hash_table):
13619         Change type to hash_table.  Update dependent calls and types.
13621         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
13622         (hash_string_slot_node): Move implementation into lto-streamer.h
13623         struct string_slot_hasher.
13624         (eq_string_slot_node): Likewise.
13626         * data-streamer-out.c: Update output_block::string_hash_table
13627         dependent calls and types.
13629         * dwarf2cfi.c (htab_t trace_index):
13630         Change type to hash_table.  Update dependent calls and types.
13632         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
13633         Change type to hash_table.  Update dependent calls and types.
13634         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
13635         (htab_t optimize_external_refs::map): Likewise.
13636         (htab_t output_comp_unit::extern_map): Likewise.
13637         (htab_t output_comdat_type_unit::extern_map): Likewise.
13638         (htab_t output_macinfo::macinfo_htab): Likewise.
13639         (htab_t optimize_location_lists::htab): Likewise.
13640         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
13642         * except.c (htab_t ehspec_hash_type):
13643         Change type to hash_table.  Update dependent calls and types.
13644         (assign_filter_values::ttypes): Likewise.
13645         (assign_filter_values::ehspec): Likewise.
13646         (sjlj_assign_call_site_values::ar_hash): Likewise.
13647         (convert_to_eh_region_ranges::ar_hash): Likewise.
13649         * gcse.c (htab_t pre_ldst_table):
13650         Change type to hash_table.  Update dependent calls and types.
13652         * ggc-common.c (htab_t saving_htab):
13653         Change type to hash_table.  Update dependent calls and types.
13654         (htab_t loc_hash): Likewise.
13655         (htab_t ptr_hash): Likewise.
13656         (call_count): Rename ggc_call_count.
13657         (call_alloc): Rename ggc_call_alloc.
13658         (loc_descriptor): Rename make_loc_descriptor.
13659         (add_statistics): Rename ggc_add_statistics.
13661         * ggc-common.c (saving_htab):
13662         Change type to hash_table.  Update dependent calls and types.
13664         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
13665         (push_gimplify_context): Likewise.
13666         (pop_gimplify_context): Likewise.
13667         (struct gimple_temp_hash_elt): Added.
13668         (struct gimplify_hasher): Likewise.
13669         (struct gimplify_ctx.temp_htab):
13670         Change type to hash_table.  Update dependent calls and types.
13672         * gimple-fold.c: Include gimplify-ctx.h.
13674         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
13675         Change type to hash_table.  Update dependent calls and types.
13676         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
13677         avoid potential global name collision.
13679         * gimplify.c: Include gimplify-ctx.h.
13680         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
13681         (htab_t gimplify_ctx::temp_htab):
13682         Update dependent calls and types for new type hash_table.
13683         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
13684         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
13686         * gimplify-ctx.h: New.
13687         (struct gimple_temp_hash_elt): Move from gimplify.c.
13688         (class gimplify_hasher): New.
13689         (struct gimplify_ctx): Move from gimple.h.
13690         (htab_t gimplify_ctx::temp_htab):
13691         Change type to hash_table.  Update dependent calls and types.
13693         * graphite-clast-to-gimple.c: Include graphite-htab.h.
13694         (htab_t ivs_params::newivs_index):
13695         Change type to hash_table.  Update dependent calls and types.
13696         (htab_t ivs_params::params_index): Likewise.
13697         (htab_t print_generated_program::params_index): Likewise.
13698         (htab_t gloog::newivs_index): Likewise.
13699         (htab_t gloog::params_index): Likewise.
13701         * graphite.c: Include graphite-htab.h.
13702         4htab_t graphite_transform_loops::bb_pbb_mapping):
13703         Change type to hash_table.  Update dependent calls and types.
13705         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
13706         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
13707         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
13709         * graphite-dependences.c: Include graphite-htab.h.
13710         (loop_is_parallel_p): Change hash table type of parameter.
13712         * graphite-htab.h: New.
13713         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
13714         (extern find_pbb_via_hash): Move from graphite-poly.h.
13715         (extern loop_is_parallel_p): Move from graphite-poly.h.
13716         (extern get_loop_body_pbbs): Move from graphite-poly.h.
13718         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
13719         (extern loop_is_parallel_p): Move to graphite-htab.h.
13720         (extern get_loop_body_pbbs): Move to graphite-htab.h.
13722         * haifa-sched.c (htab_t delay_htab):
13723         Change type to hash_table.  Update dependent calls and types.
13724         (htab_t delay_htab_i2): Likewise.
13726         * ira-color.c (htab_t allocno_hard_regs_htab):
13727         Change type to hash_table.  Update dependent calls and types.
13729         * ira-costs.c (htab_t cost_classes_htab):
13730         Change type to hash_table.  Update dependent calls and types.
13732         * loop-invariant.c (htab_t merge_identical_invariants::eq):
13733         Change type to hash_table.  Update dependent calls and types.
13735         * loop-iv.c (htab_t bivs):
13736         Change type to hash_table.  Update dependent calls and types.
13738         * loop-unroll.c (htab_t opt_info::insns_to_split):
13739         Change type to hash_table.  Update dependent calls and types.
13740         (htab_t opt_info::insns_with_var_to_expand): Likewise.
13742         * lto-streamer.h (struct string_slot): Move from data-streamer.h
13743         (struct string_slot_hasher): New.
13744         (htab_t output_block::string_hash_table):
13745         Change type to hash_table.  Update dependent calls and types.
13747         * lto-streamer-in.c (freeing_string_slot_hasher): New.
13748         (htab_t file_name_hash_table):
13749         Change type to hash_table.  Update dependent calls and types.
13751         * lto-streamer-out.c: Update output_block::string_hash_table dependent
13752         calls and types.
13754         * lto-streamer.c (htab_t tree_htab):
13755         Change type to hash_table.  Update dependent calls and types.
13757         * omp-low.c: Include gimplify-ctx.h.
13759         * passes.c (htab_t name_to_pass_map):
13760         Change type to hash_table.  Update dependent calls and types.
13761         (pass_traverse): Rename to passes_pass_traverse.
13763         * plugin.c (htab_t event_tab):
13764         Change type to hash_table.  Update dependent calls and types.
13766         * postreload-gcse.c (htab_t expr_table):
13767         Change type to hash_table.  Update dependent calls and types.
13768         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
13770         * sese.c (debug_rename_map_1): Make extern.
13771         (htab_t copy_bb_and_scalar_dependences::rename_map):
13772         Change type to hash_table.  Update dependent calls and types.
13774         * sese.h (extern debug_rename_map): Move to .c file.
13776         * store-motion.c (htab_t store_motion_mems_table):
13777         Change type to hash_table.  Update dependent calls and types.
13779         * trans-mem.c (htab_t tm_new_mem_hash):
13780         Change type to hash_table.  Update dependent calls and types.
13782         * tree-browser.c (htab_t TB_up_ht):
13783         Change type to hash_table.  Update dependent calls and types.
13785         * tree-cfg.c (htab_t discriminator_per_locus):
13786         Change type to hash_table.  Update dependent calls and types.
13788         * tree-complex.c: Include tree-hasher.h
13789         (htab_t complex_variable_components):
13790         Change type to hash_table.  Update dependent calls and types.
13792         * tree-eh.c (htab_t finally_tree):
13793         Change type to hash_table.  Update dependent calls and types.
13795         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
13796         struct int_tree_hasher.
13797         (extern int_tree_map_eq): Likewise.
13798         (uid_decl_map_hash): Removed.
13799         (extern decl_tree_map_eq): Likewise.
13801         * tree-hasher.h: New.
13802         (struct int_tree_hasher): New.
13803         (typedef int_tree_htab_type): New.
13805         * tree-inline.c: Include gimplify-ctx.h.
13807         * tree-mudflap.c: Include gimplify-ctx.h.
13809         * tree-parloops.c: Include tree-hasher.h.
13810         (htab_t eliminate_local_variables_stmt::decl_address):
13811         Change type to hash_table.  Update dependent calls and types.
13812         (htab_t separate_decls_in_region::decl_copies): Likewise.
13814         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
13815         Change type to hash_table.  Update dependent calls and types.
13817         * tree-sra.c (candidates):
13818         Change type to hash_table.  Update dependent calls and types.
13820         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
13821         in tree-flow.h.
13822         (int_tree_map_hash): Likewise.
13824         * tree-ssa-dom.c (htab_t avail_exprs):
13825         Change type to hash_table.  Update dependent calls and types.
13827         * tree-ssa-live.c (var_map_base_init::tree_to_index):
13828         Change type to hash_table.  Update dependent calls and types.
13830         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
13831         Change type to hash_table.  Update dependent calls and types.
13833         * tree-ssa-phiopt.c (seen_ssa_names):
13834         Change type to hash_table.  Update dependent calls and types.
13836         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
13837         Change type to hash_table.  Update dependent calls and types.
13839         * tree-ssa-uncprop.c (equiv):
13840         Change type to hash_table.  Update dependent calls and types.
13842 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
13844         PR rtl-optimization/57003
13845         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
13846         call note_stores with kill_clobbered_value callback again after
13847         killing regs_invalidated_by_call.
13849 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13851         * config/aarch64/aarch64-simd.md
13852         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
13853         (aarch64_simd_bsl<mode>): Likewise.
13854         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
13856 2013-04-25  Marek Polacek  <polacek@redhat.com>
13858         PR tree-optimization/57066
13859         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
13861 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
13863         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
13865 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13867         * config/aarch64/aarch64-builtins.c
13868         (aarch64_fold_builtin): New.
13869         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
13870         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
13871         * config/aarch64/aarch64-simd-builtins.def (abs): New.
13872         * config/aarch64/arm_neon.h
13873         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
13875 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13876             Tejas Belagod  <tejas.belagod@arm.com>
13878         * config/aarch64/aarch64-builtins.c
13879         (aarch64_gimple_fold_builtin): New.
13880         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
13881         * config/aarch64/aarch64-simd-builtins.def (addv): New.
13882         * config/aarch64/aarch64-simd.md (addpv4sf): New.
13883         (addvv4sf): Update.
13884         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
13886 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13888         * config/aarch64/aarch64.md
13889         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
13891 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13893         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
13894         (*ngcsi_uxtw): New pattern.
13896 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13897             Julian Brown  <julian@codesourcery.com>
13899         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
13900         (TB_DREG): Add T_V4HF.
13901         (v4hf_UP): New macro.
13902         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13903         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13904         Handle initialisation of V4HF. Adjust initialisation of reinterpret
13905         built-ins.
13906         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
13907         (arm_vector_mode_supported_p): Handle V4HF.
13908         (arm_mangle_map): Handle V4HFmode.
13909         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
13910         * config/arm/arm_neon_builtins.def: Add entries for
13911         vcvtv4hfv4sf, vcvtv4sfv4hf.
13912         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
13913         (neon_vcvtv4hfv4sf): Likewise.
13914         * config/arm/neon-gen.ml: Handle half-precision floating point
13915         features.
13916         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
13917         * config/arm/arm_neon.h: Regenerate.
13918         * config/arm/neon.ml (type elts): Add F16.
13919         (type vectype): Add T_float16x4, T_floatHF.
13920         (type vecmode): Add V4HF.
13921         (type features): Add Requires_FP_bit feature.
13922         (elt_width): Handle F16.
13923         (elt_class): Likewise.
13924         (elt_of_class_width): Likewise.
13925         (mode_of_elt): Refactor.
13926         (type_for_elt): Handle F16, fix error messages.
13927         (vectype_size): Handle T_float16x4.
13928         (vcvt_sh): New function.
13929         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
13930         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
13931         (string_of_mode): Handle V4HF.
13932         * doc/arm-neon-intrinsics.texi: Regenerate.
13934 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
13936         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
13937         format specifier in 'X' case.
13939 2013-04-25  Alan Modra  <amodra@gmail.com>
13941         PR target/57052
13942         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
13943         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
13944         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
13945         Repeat for many other rotate/shift and mask patterns using subregs.
13946         Name lshiftrt insns.
13947         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
13948         on WORDS_BIG_ENDIAN.
13950 2013-04-25  Alan Modra  <amodra@gmail.com>
13952         * config.gcc: Support little-endian powerpc-linux targets.
13953         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
13954         (LINK_OS_LINUX_SPEC): Define.
13955         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
13956         Preserve MASK_LITTLE_ENDIAN.
13957         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
13958         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
13959         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
13960         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
13961         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
13962         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
13963         Correct fp word order for little-endian.  Don't shift toc entries
13964         smaller than a word for little-endian.
13965         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
13966         (bswapdi2 splits): Correct low-part subreg for little-endian.
13967         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
13968         low/high where such is correct only for be.
13969         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
13970         little-endian for -mcall-aixdesc.
13972 2013-04-25  Alan Modra  <amodra@gmail.com>
13974         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
13975         replace_equiv_address_nv.
13977 2013-04-25  Alan Modra  <amodra@gmail.com>
13979         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
13981 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13983         Revert:
13984         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13985         * rtl.h (struct rtx_def): ...
13987 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
13989         PR rtl-optimizations/57046
13990         * lra-constraints (split_reg): Set up lra_risky_transformations_p
13991         for multi-reg splits.
13993 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
13995         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
13997 2013-04-24  Sterling Augustine  <saugustine@google.com>
13999         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
14000         (comp_dir_string, debug_str_dwo_section): New.
14001         (DEBUG_STR_DWO_SECTION): Rename to ...
14002         (DEBUG_DWO_STR_SECTION): ... this.
14003         (DEBUG_NORM_STR_SECTION): Delete.
14004         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
14005         (DEBUG_STR_DWO_SECTION_FLAGS): New.
14006         (find_AT_string): Move most logic to ...
14007         (find_AT_string_in_table): ... here.  New.
14008         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
14009         add_skeleton_AT_string.  Delete logic.
14010         (output_skeleton_debug_sections): Remove call to
14011         add_top_level_skeleton_die_attrs.
14012         (add_comp_dir_attribute): Move logic to comp_dir_string.
14013         (dwarf2out_init): Initialize debug_str_dwo_section.
14014         (output_indirect_string): Call find_string_form.
14015         (output_indirect_strings): Rewrite.
14016         (prune_unused_types): Empty skeleton_debug_str_hash.
14017         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14018         (dwarf2out_finish):  Call output_indirect_strings.
14020 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14022         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14024 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14026         * rtl.h (struct rtx_def): Add comment for field jump.
14027         (LRA_SUBREG_P): New macro.
14028         * recog.c (register_operand): Check LRA_SUBREG_P.
14029         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14030         stack frame.
14031         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14032         (lra_final_code_change): Skip subreg change for operators.
14033         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14034         if there are no operand changes.
14035         * lra-constraints.c (curr_insn_set): New.
14036         (match_reload): Set LRA_SUBREG_P.
14037         (emit_spill_move): Ditto.
14038         (check_and_process_move): Use curr_insn_set. Process only single
14039         set insns.  Don't initialize sec_mem_p and change_p.
14040         (simplify_operand_subreg): Use LRA_SUBREG_P.
14041         (reg_in_class_p): New function.
14042         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14043         of #ifdef.  Add code to remove cycling.
14044         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14045         non-null disp.  Reload inner instead of disp when base and index
14046         are null.  Try to put lo_sum into register.
14047         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14048         (check_and_process_move): Move code for move cost check to
14049         simple_move_p.  Remove equiv_substitution.
14050         (simple_move_p): New function.
14051         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14052         curr_insn_set.  Call check_and_process_move only for single set
14053         insns.  Use the new function.  Move call of check_and_process_move
14054         after operand equiv substitution and address process.
14056 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14058         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14059         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14060         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14062 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14064         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14066 2013-04-24  Marek Polacek  <polacek@redhat.com>
14068         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14069         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14070         (select_loops_exit_conditions): Likewise.
14071         (number_of_iterations_for_all_loops): Likewise.
14072         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14073         (scev_analysis): Likewise.
14075 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14076             Chao-ying Fu  <fu@mips.com>
14078         * config/mips/micromips.md (jraddiusp): New pattern.
14079         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14080         instruction if possible.
14082 2013-04-24  Alan Modra  <amodra@gmail.com>
14084         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14086 2013-04-24  Julian Brown  <julian@codesourcery.com>
14087             Chung-Lin Tang  <cltang@codesourcery.com>
14089         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14090         dependency behavior in enumeration type DIE generation. Add TODO note
14091         to comments about future DW_FORM_sdata/udata re-work of related code.
14093 2013-04-23  Lawrence Crowl  <crowl@google.com>
14095         * Makefile.in: Update as needed below.
14097         * hash-table.h (class hash_table):
14098         Correct many methods with parameter types compare_type to the correct
14099         value_type.  (Correct code was unlikely to notice the change.)
14100         (hash_table::elements_with_deleted) New.
14101         (class hashtable::iterator): New.
14102         (hashtable::begin()): New.
14103         (hashtable::end()): New.
14104         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14106         * statistics.c (statistics_hashes):
14107         Change type to hash_table.  Update dependent calls and types.
14109         * tree-into-ssa.c (var_infos):
14110         Change type to hash_table.  Update dependent calls and types.
14112         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14113         Change type to hash_table.  Update dependent calls and types.
14115         * tree-ssa-loop-im.c (struct mem_ref.refs):
14116         Change type to hash_table.  Update dependent calls and types.
14118         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14119         Change type to hash_table.  Update dependent calls and types.
14121         * tree-ssa-sccvn.c (vn_tables_s::nary):
14122         Change type to hash_table.  Update dependent calls and types.
14123         (vn_tables_s::phis): Likewise.
14124         (vn_tables_s::references): Likewise.
14126         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14127         (vn_reference_eq): Update parameter and return types.
14129         * tree-ssa-structalias.c (pointer_equiv_class_table):
14130         Change type to hash_table.  Update dependent calls and types.
14131         (location_equiv_class_table): Likewise.
14133         * tree-vect-data-refs.c: Consequential changes for making
14134         peeling a hash_table.
14136         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14137         (destroy_loop_vec_info): Dependent hash_table update.
14139         * tree-vectorizer.h (peeling_htab):
14140         Change type to hash_table.  Update dependent calls and types.
14142 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
14144         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14145         to check the register content is equal or not.
14146         * lra-constraints.c (match_reload): Use lra_assign_reg_val
14147         to assign register content record.
14148         * lra-eliminations.c (update_reg_eliminate): Use
14149         lra_update_reg_val_offset to update register content offset.
14150         * lra-int.h (struct lra_reg): Add offset member.
14151         (lra_reg_val_equal_p): New static inline function.
14152         (lra_update_reg_val_offset): New static inline function.
14153         (lra_assign_reg_val): New static inline function.
14154         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14155         to assign register content record.
14156         (initialize_lra_reg_info_element): Initial offset to zero.
14158 2013-04-23  Catherine Moore  <clm@codesourcery.com>
14160         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14161         operands.  Record compression.
14163 2013-04-23  Xinliang David Li  <davidxl@google.com>
14165         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14167 2013-04-23  Richard Biener  <rguenther@suse.de>
14169         PR middle-end/57036
14170         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14171         parameter, only add abnormal goto edges from the copied body
14172         if the call could perform abnormal gotos.
14173         (copy_cfg_body): Adjust.
14175 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
14177         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14179 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
14181         * coretypes.h (gimple_stmt_iterator): Add struct to make
14182         compatible with C.
14184 2013-04-23  Richard Biener  <rguenther@suse.de>
14186         PR tree-optimization/57026
14187         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14188         from SSA names occuring in abnormal PHI nodes.
14190 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14192         * lto/lto.c (print_lto_report_1): Fix LTO report names.
14194 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14196         * lto/lto.c (print_lto_report_1): Declare early.
14197         (read_cgraph_and_symbols): Call print_lto_report_1 early.
14199 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14201         * common.opt (-flto-report-wpa): Add.
14202         * doc/invoke.texi (-flto-report-wpa): Add.
14203         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14204         (lto_main): dito.
14206 2013-04-22  Xinliang David Li  <davidxl@google.com>
14208         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14209         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14210         * Makefile.in: New dependency
14212             David Daney <ddaney.cavm@gmail.com>
14214         * configure.ac (gcc_cv_as_micromips_support):  Use the
14215         --fatal-warnings option.
14216         * configure: Regenerate.
14218 2013-04-22  Marek Polacek  <polacek@redhat.com>
14220         PR sanitizer/56990
14221         * tsan.c (instrument_expr): Don't instrument expression
14222         in case its size is zero.
14224 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
14226         PR target/57032
14227         Revert:
14228         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14230         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14232 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14234         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
14235         (gimple_stmt_iterator): New typedef.
14236         * gimple.h (gimple_stmt_iterator): Rename to...
14237         (gimple_stmt_iterator_d): ... This.
14238         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
14239         trees be valid for GIMPLE and GENERIC.
14240         (TARGET_GIMPLE_FOLD_BUILTIN): New.
14241         * gimple-fold.c (gimple_fold_call): Call target hook
14242         gimple_fold_builtin.
14243         * hooks.c (hook_bool_gsiptr_false): New.
14244         * hooks.h (hook_bool_gsiptr_false): New.
14245         * target.def (fold_stmt): New.
14246         * doc/tm.texi: Regenerate.
14248 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
14250         PR target/57018
14251         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
14252         a set sp if no stack realignment.
14254 2013-04-22  Nick Clifton  <nickc@redhat.com>
14256         * config.gcc (tilegx-linux): Extend extra_objs rather than
14257         overwriting it.
14258         (tilepro-linux): Likewise.
14260 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14262         * config/aarch64/aarch64-builtins.c
14263         (CF): Remove.
14264         (CF0, CF1, CF2, CF3, CF4, CF10): New.
14265         (VAR<1-12>): Add MAP parameter.
14266         (BUILTIN_*): Likewise.
14267         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
14268         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
14269         (aarch64_ushl_n<mode>): Likewise.
14270         (aarch64_sshr_n<mode>): Likewise.
14271         (aarch64_ushr_n<mode>): Likewise.
14272         (aarch64_<maxmin><mode>): Likewise.
14273         (aarch64_sqrt<mode>): Likewise.
14274         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
14275         (vshr<q>_n_*): Likewise.
14277 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14279         * config/aarch64/aarch64-builtins.c
14280         (aarch64_simd_builtin_type_mode): Handle SF types.
14281         (sf_UP): Define.
14282         (BUILTIN_GPF): Define.
14283         (aarch64_init_simd_builtins): Handle SF types.
14284         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
14285         (frecps): Likewise.
14286         (frecpx): Likewise.
14287         * config/aarch64/aarch64-simd.md
14288         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
14289         (aarch64_frecpe<mode>): New.
14290         (aarch64_frecps<mode>): Likewise.
14291         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
14292         (v8type): Add frecp<esx>.
14293         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
14294         (aarch64_frecps<mode>): Likewise.
14295         * config/aarch64/iterators.md (FRECP): New.
14296         (frecp_suffix): Likewise.
14297         * config/aarch64/arm_neon.h
14298         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
14300 2013-04-22  Christian Bruel  <christian.bruel@st.com>
14302         PR target/56995
14303         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
14304         (REG_CLASS_NAMES): Idem.
14305         (REG_CLASS_CONTENTS): Idem.
14306         (REGCLASS_HAS_FP_REG): Idem.
14307         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
14308         (sh_conditional_register_usage): Idem.
14310 2013-04-21  Jeff Law  <law@redhat.com>
14312         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
14313         (ssa_forward_propagate_and_combine): Use it.
14315 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14317         * lra.c: Update the flow chart diagram.
14319 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14321         PR rtl-optimization/56847
14322         * lra-constraints.c (process_alt_operands): Discourage alternative
14323         with non-matche doffsettable memory constraint fro memory with
14324         known offset.
14326 2013-04-19  Richard Biener  <rguenther@suse.de>
14328         PR tree-optimization/56982
14329         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
14330         function.
14331         * gimplify.c (gimplify_call_expr): Notice special calls.
14332         (gimplify_modify_expr): Likewise.
14333         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
14334         abnormal control flow receivers.
14335         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
14336         in the same way as cfun->has_nonlocal_labels.
14337         (gimple_purge_dead_abnormal_call_edges): Likewise.
14338         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
14339         receivers start a basic-block.
14341 2013-04-19  Richard Biener  <rguenther@suse.de>
14343         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
14344         member ...
14345         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
14346         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
14347         (SLP_TREE_LOAD_PERMUTATION): Add.
14348         (vect_transform_slp_perm_load): Adjust prototype.
14349         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
14350         (vect_free_slp_instance): Likewise.
14351         (vect_create_new_slp_node): Likewise.
14352         (vect_supported_slp_permutation_p): Remove.
14353         (vect_slp_rearrange_stmts): Adjust.
14354         (vect_supported_load_permutation_p): Likewise.  Inline
14355         vect_supported_slp_permutation_p here.
14356         (vect_analyze_slp_instance): Compute load permutations per
14357         slp node instead of per instance.
14358         (vect_get_slp_defs): Adjust.
14359         (vect_transform_slp_perm_load): Likewise.
14360         (vect_schedule_slp_instance): Remove redundant code.
14361         (vect_schedule_slp): Remove hack for PR56270, add it ...
14362         * tree-vect-stmts.c (vectorizable_load): ... here, do not
14363         CSE loads for SLP.  Adjust.
14365 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14367         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
14368         spelling in two comments.
14370 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14372         PR target/56797
14373         * config/arm/arm.c (load_multiple_sequence): Require SP
14374         as base register for loads if SP is in the register list.
14376 2013-04-19  Martin Jambor  <mjambor@suse.cz>
14378         PR tree-optimization/56718
14379         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
14380         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
14381         and made public.  Adjusted all callers.
14382         (ipa_intraprocedural_devirtualization): New function.
14383         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
14384         (ipa_intraprocedural_devirtualization): Likewise.
14385         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
14387 2013-04-19  Richard Biener  <rguenther@suse.de>
14389         PR tree-optimization/57000
14390         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
14392 2013-04-19  Terry Guo  <terry.guo@arm.com>
14394         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
14395         Replace with ...
14396         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
14397         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
14398         (cortex_m4_fmacs): Use new reservations.
14399         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
14401 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
14403         PR rtl-optimization/56999
14404         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
14405         related code.
14406         (lra_coalesce): Remove split_origin_bitmap and related code.
14407         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
14408         ranges if necessary.
14410 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
14412         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
14413         New array.
14414         (ix86_expand_call): Remove clobbered_registers array and use
14415         x86_64_ms_sysv_extra_clobbered_registers instead.
14416         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
14417         Declare here.
14418         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
14419         predicate.
14420         * config/i386/i386.md (*call_rex64_ms_sysv): Use
14421         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
14422         (*call_value_rex64_ms_sysv): Ditto.
14424 2013-04-18  Cary Coutant  <ccoutant@google.com>
14426         * dwarf2out.c (output_pubnames): Check die_perennial_p of
14427         parent instead of die_mark.
14429 2013-04-18  Diego Novillo  <dnovillo@google.com>
14431         * gimple.c (create_gimple_tmp): New.
14432         (get_expr_type): New.
14433         (build_assign): New.
14434         (build_type_cast): New.
14435         * gimple.h (enum ssa_mode): Define.
14436         (gimple_seq_set_location): New.
14437         * asan.c (build_check_stmt): Change some gimple_build_* calls
14438         to use build_assign and build_type_cast.
14440 2013-04-18  Richard Biener  <rguenther@suse.de>
14442         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
14443         handle negative step.  Remove redundant checks.
14444         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
14445         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
14446         for negative step and grouped loads fail to vectorize.
14448 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14450         * emit-rtl.c (reset_insn_used_flags): New function.
14451         (reset_all_used_flags): Use it.
14452         (verify_insn_sharing): New function.
14453         (verify_rtl_sharing): Fix verification for SEQUENCEs.
14455 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
14457         PR tree-optimization/56984
14458         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
14459         and (x >> M) >= N don't register any assertion if N << M is the
14460         minimum value.
14462 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14464         * lower-subreg.c (resolve_simple_move): If called self-recursive,
14465         do not delete_insn insns that have not yet been emitted, only
14466         unlink them with remove_insn.
14467         * df-scan.c (df_insn_delete): Revert r197492.
14469 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
14471         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
14472         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
14474 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14476         * config/arm/arm.md (movsicc_insn): Convert define_insn into
14477         define_insn_and_split.
14478         (and_scc,ior_scc,negscc): Likewise.
14479         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
14481 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14483         * config/arm/arm.c (use_return_insn): Return 0 for targets that
14484         can benefit from using a sequence of LDRD instructions in epilogue
14485         instead of a single LDM instruction.
14487 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14489         PR 45688
14490         * doc/extend.texi: Fix typo.
14492 2013-04-17  Richard Biener  <rguenther@suse.de>
14494         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
14495         (vect_build_slp_tree): ... here.
14496         (vect_build_slp_tree_1): Compute which stmts of the SLP group
14497         match.  Remove special-casing of mismatched complex loads.
14498         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
14499         re-try the match with swapped commutative operands.
14500         (vect_supported_load_permutation_p): Remove special-casing of
14501         mismatched complex loads.
14502         (vect_analyze_slp_instance): Adjust.
14504 2013-04-17  Richard Biener  <rguenther@suse.de>
14506         PR rtl-optimization/56921
14507         * cfgloop.h (struct loop): Add simple_loop_desc member.
14508         (struct niter_desc): Mark with GTY(()).
14509         (simple_loop_desc): Do not use aux field but simple_loop_desc.
14510         * loop-iv.c (get_simple_loop_desc): Likewise.
14511         (free_simple_loop_desc): Likewise.
14513         Revert
14514         2013-04-16  Richard Biener  <rguenther@suse.de>
14516         PR rtl-optimization/56921
14517         * loop-init.c (pass_rtl_move_loop_invariants): Add
14518         TODO_do_not_ggc_collect to todo_flags_finish.
14519         (pass_rtl_unswitch): Same.
14520         (pass_rtl_unroll_and_peel_loops): Same.
14521         (pass_rtl_doloop): Same.
14523 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
14525         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
14526         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
14527         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
14528         references.
14529         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
14530         * tree-streamer.c (record_common_node): Adjust reference in comment.
14532 2013-04-17  Terry Guo  <terry.guo@arm.com>
14534         * config/arm/cortex-m4.md: Add a new bypass.
14536 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14538         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
14539         New pattern.
14540         (*subs_<optab><mode>_multp2): New pattern.
14541         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14542         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14544 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14546         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
14547         (*subs_mul_imm_<mode>): New pattern.
14549 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
14551         PR target/56948
14552         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
14553         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
14554         (vsx_movti_32bit): Change j->wa to O->wa.
14556 2013-04-16  Richard Biener  <rguenther@suse.de>
14558         PR rtl-optimization/56921
14559         * loop-init.c (pass_rtl_move_loop_invariants): Add
14560         TODO_do_not_ggc_collect to todo_flags_finish.
14561         (pass_rtl_unswitch): Same.
14562         (pass_rtl_unroll_and_peel_loops): Same.
14563         (pass_rtl_doloop): Same.
14565 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
14567         * config/arm/arm.c (emit_multi_reg_push): New declaration
14568         for an existing function.
14569         (arm_emit_strd_push): New function.
14570         (arm_expand_prologue): Used here.
14571         (arm_emit_ldrd_pop): New function.
14572         (arm_expand_epilogue): Used here.
14573         (arm_get_frame_offsets): Update condition.
14574         (arm_emit_multi_reg_pop): Add a special case for load of a single
14575         register with writeback.
14577 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
14579         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
14580         description.
14582 2013-04-16  Richard Biener  <rguenther@suse.de>
14584         PR tree-optimization/56756
14585         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
14586         (first_mem_ref_loc): New.
14587         (execute_sm): Place the load temporarily before a previous
14588         access instead of in the latch edge to ensure its SSA dependencies
14589         are defined at points dominating the load.
14591 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
14593         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
14594         correct fix by moving header and footer insn to the footer of
14595         the merged basic block.  Clear BB_END of the merged-away block.
14597         PR middle-end/43631
14598         * emit-rtl.c (make_note_raw): New function.
14599         (link_insn_into_chain): New static inline function.
14600         (add_insn): Use it.
14601         (add_insn_before, add_insn_after): Factor insn chain linking code...
14602         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
14603         using link_insn_into_chain.
14604         (note_outside_basic_block_p): New helper function for emit_note_after
14605         and emit_note_before.
14606         (emit_note_after): Use nobb variant of add_insn_after if the note
14607         should not be contained in a basic block.
14608         (emit_note_before): Use nobb variant of add_insn_before if the note
14609         should not be contained in a basic block.
14610         (emit_note_copy): Use make_note_raw.
14611         (emit_note): Likewise.
14612         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
14613         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
14614         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
14615         the moved barrier the tail of the basic block it follows.
14616         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
14618 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
14620         PR tree-optimization/56962
14621         * gimple-ssa-strength-reduction.c (record_increment): Only set
14622         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
14623         either rhs1 or rhs2 is equal to c->base_expr.
14625 2013-04-15  Richard Biener  <rguenther@suse.de>
14627         PR tree-optimization/56933
14628         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
14629         member.
14630         (GROUP_READ_WRITE_DEPENDENCE): Remove.
14631         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
14632         * tree-vect-data-refs.c (vect_analyze_group_access): Move
14633         dependence check ...
14634         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
14635         ... here.
14636         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
14637         GROUP_READ_WRITE_DEPENDENCE.
14639 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14641         * emit-rtl.c (reset_all_used_flags): New function.
14642         (verify_rtl_sharing): Call reset_all_used_flags before and after
14643         performing the checks.
14645 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14647         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
14648         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
14649         * config/arm/constraints.md (De): New constraint.
14650         * config/arm/neon.md (anddi3_neon): Delete.
14651         (neon_vand<mode>): Expand to standard anddi3 pattern.
14652         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
14653         Move earlier in the file.
14654         (neon_inv_logic_op2): Likewise.
14655         (arm_anddi_operand_neon): New predicate.
14657 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14659         * configure.ac (gcc_cv_ld_as_needed): Set
14660         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
14661         Use -z ignore, -z record on *-*-solaris2*.
14662         (HAVE_LD_AS_NEEDED): Update comment.
14663         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
14664         * configure: Regenerate.
14665         * config.in: Regenerate.
14666         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
14667         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
14668         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
14669         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
14670         equivalents.  Fix markup.
14671         * doc/tm.texi: Regenerate.
14673 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
14675         * config/i386/i386.opt: New option mstack-protector-guard=.
14676         * config/i386/i386-opts.h: Add enum stack_protector_guard.
14677         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
14678         TARGET_SSP_TLS_GUARD.
14679         * config/i386/i386.c (ix86_option_override_internal): Set
14680         ix86_stack_protector_guard.
14681         * config/i386/i386.md (stack_protect_set): Enable for
14682         TARGET_SSP_TLS_GUARD only.
14683         (stack_protect_set_<mode>): Ditto.
14684         (stack_protect_test): Ditto.
14685         (stack_protect_test_<mode>): Ditto.
14686         * doc/invoke.texi (i386 Option): Document.
14688 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
14690         PR target/56890
14691         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
14692         (S_MODES): Set H_MODE bit.
14693         (SF_MODES): Set only S_MODE and SF_MODE bits.
14694         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
14695         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
14696         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
14697         <MODE_FLOAT>: Likewise.
14699 2013-04-15  Joey Ye  <joey.ye@arm.com>
14701         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
14703 2013-04-15  Joey Ye  <joey.ye@arm.com>
14705         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
14706         for real far jump.
14707         (thumb_far_jump_used_p): Count instruction size and set
14708         far_jump_used.
14710 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
14712         * reorg.c (fill_simple_delay_slots): Reindent block of code.
14713         * resource.c (mark_target_live_regs): Reformat conditional block.
14715 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14717         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
14718         notes, they are emitted only just before final.
14719         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
14721 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
14723         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
14724         * cfgrtl.c (delete_insn): Call it here instead.
14725         * lra-spills.c (lra_final_code_change): Use delete_insn.
14726         * haifa-sched.c (sched_remove_insn): Likewise.
14727         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
14728         returning to the nop pool.
14729         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
14730         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
14732 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14734         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
14735         * doc/tm.texi: Regenerated.
14737 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
14739         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
14740         QImode checks.
14742 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
14744         * df-core.c (df_find_def): Compare register numbers.
14745         (df_find_use): Likewise.
14747 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
14749         PR target/56903
14750         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
14751         lra_in_progress for return.
14753 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14755         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
14756         define_insn into define_insn_and_split and emit movsicc patterns.
14758 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
14760         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
14762 2013-04-12  Richard Biener  <rguenther@suse.de>
14764         * tree-pass.h (TODO_do_not_ggc_collect): New.
14765         * passes.c (execute_one_ipa_transform_pass): Honor
14766         TODO_do_not_ggc_collect.
14767         (execute_one_pass): Likewise.
14769         Revert
14770         2013-04-10  Richard Biener  <rguenther@suse.de>
14772         * passes.c (init_optimization_passes): Remove reload pass.
14773         * ira.c (do_reload): Merge into ...
14774         (ira): ... this.
14775         (rest_of_handle_reload): Remove.
14776         (pass_reload): Likewise.
14777         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14778         of reload for vzeroupper pass placement.
14780 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
14782         PR tree-optimization/56918
14783         PR tree-optimization/56920
14784         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
14785         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
14786         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
14787         use wide_mul_with_sign method.
14789 2013-04-12  Richard Biener  <rguenther@suse.de>
14791         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
14792         not be considered a gimple constant.
14794 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
14796         * fold-const.c (const_binop): Handle vector shifts by a scalar.
14797         (fold_binary_loc): Call const_binop also for mixed vector-scalar
14798         operations.
14800 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14801             Jakub Jelinek  <jakub@redhat.com>
14803         * opts.c: Include diagnostic-color.h.
14804         (common_handle_option): Handle OPT_fdiagnostics_color_.
14805         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
14806         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
14807         (diagnostic-color.o): New.
14808         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
14809         (diagnostic_color_rule): New enum.
14810         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
14811         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
14812         the location string.
14813         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
14814         either NULL, or color kind.
14815         * diagnostic-color.c: New file.
14816         * diagnostic-color.h: New file.
14817         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
14818         arguments.
14819         * doc/invoke.texi (-fdiagnostics-color): Document.
14820         * pretty-print.h (pp_show_color): Define.
14821         (struct pretty_print_info): Add show_color field.
14822         * diagnostic.c: Include diagnostic-color.h.
14823         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
14824         macros.  Colorize error:, warning: etc. strings and also the location
14825         string.
14826         (diagnostic_show_locus): Colorize the caret line.
14827         * pretty-print.c: Include diagnostic-color.h.
14828         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
14829         inside of %< %> quotes or quoted through q format modifier.
14831 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14833         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
14835 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14837         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
14838         code in CC_NZ mode.
14839         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
14840         pattern.
14842 2013-04-11  Marek Polacek  <polacek@redhat.com>
14844         PR tree-optimization/48184
14845         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
14847 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
14849         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
14850         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
14851         (skip_simple_arithmetic): Tidy up.
14852         * tree.h (skip_simple_constant_arithmetic): Declare.
14854 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14856         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
14858 2013-04-11  Richard Biener  <rguenther@suse.de>
14860         * tree-vect-loop.c (get_initial_def_for_induction): Properly
14861         generate vector constants.
14863 2013-04-11  Richard Biener  <rguenther@suse.de>
14865         PR tree-optimization/56878
14866         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
14867         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
14868         New function.
14869         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14870         Prefer to align the DR with the most invariant base address.
14872 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14874         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
14875         comment.
14877 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
14879         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
14880         floating-point vector comparisons against 0.
14882 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
14884         PR tree-optimization/56899
14885         * fold-const.c (extract_muldiv_1): Apply distributive law
14886         only if TYPE_OVERFLOW_WRAPS (ctype).
14888 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
14890         PR target/56124
14891         * ira-costs.c (scan_one_insn): Check whether the source rtx of
14892         loading has side effect.
14894 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
14896         * config/sparc/sparc.c: Include tree-pass.h.
14897         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
14898         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
14899         head of file.  Change return type.  Split off gate function.
14900         (sparc_gate_work_around_errata): New function.
14901         (pass_work_around_errata): New pass definition.
14902         (insert_pass_work_around_errata) New pass insert definition to
14903         insert pass_work_around_errata just after delayed-branch scheduling.
14904         (sparc_option_override): Insert the pass.
14905         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
14907 2013-04-10  David S. Miller  <davem@davemloft.net>
14909         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
14910         or -mcpu=hypersparc.
14912         * target.def (cstore_mode): New hook.
14913         * target.h: Include insn-codes.h
14914         * targhooks.c: Likewise.
14915         (default_cstore_mode): New function.
14916         * targhooks.h: Declare it.
14917         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
14918         * doc/tm.texi: Rebuild.
14919         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
14920         target hook, rather than inspecting the insn_data.
14921         * config/sparc/sparc.c (sparc_cstore_mode): New function.
14922         (TARGET_CSTORE_MODE): Redefine.
14923         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
14924         result patterns.
14925         * config/sparc/predicates.md (cstore_result_operand): New special
14926         predicate.
14927         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
14928         Use it for operand 0.
14929         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
14930         (*snesi_special): Likewise.
14931         (*snesi_zero): Likewise.
14932         (*seqsi_zero): Likewise.
14933         (*sltu_insn): Likewise.
14934         (*sgeu_insn): Likewise.
14935         (*seqdi_special): Make operand 0 and comparison operation be of
14936         DImode.
14937         (*snedi_special): Likewise.
14938         (*snedi_special_vis3): Likewise.
14939         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
14940         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
14941         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
14942         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
14943         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
14944         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
14945         (*sltu_extend_sp64): Likewise.
14946         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
14947         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
14948         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
14949         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
14950         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
14952 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
14954         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
14955         (aarch64_start_file): Use the new function.
14957 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
14958             Jason Merrill  <jason@redhat.com>
14960         * common.opt: Add -gdwarf.
14961         * opts.c (common_handle_option): Handle it.
14962         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
14964 2013-04-10  Richard Biener  <rguenther@suse.de>
14966         * passes.c (execute_todo): Do not call ggc_collect conditional here.
14967         (execute_one_ipa_transform_pass): But unconditionally here.
14968         (execute_one_pass): And here.
14969         (init_optimization_passes): Remove reload pass.
14970         * tree-pass.h (TODO_ggc_collect): Remove.
14971         (pass_reload): Likewise.
14972         * ira.c (do_reload): Merge into ...
14973         (ira): ... this.
14974         (rest_of_handle_reload): Remove.
14975         (pass_reload): Likewise.
14976         * config/i386/i386.c (ix86_option_override): Refer to ira instead
14977         of reload for vzeroupper pass placement.
14978         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
14979         and todo_flags_finish of all passes.
14981 2013-04-10  Richard Biener  <rguenther@suse.de>
14983         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
14984         first_const_oprnd field, rename first_def_type to first_op_type.
14985         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
14986         (vect_get_and_check_slp_defs): Always use the type of the
14987         operand.  Allow mixed vect_external_def, vect_constant_def types.
14988         (vect_get_constant_vectors): Handle mixed vect_external_def,
14989         vect_constant_def types.
14991 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
14993         PR tree-optimization/55524
14994         * tree-ssa-math-opts.c
14995         (convert_mult_to_fma): Don't use an fms construct
14996         when we don't have an fms operation, but fnma, and it looks
14997         likely that we'll be able to use the latter.
14999 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
15001         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
15002         function.
15003         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
15004         inline fail caused by overwritable functions.
15006 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
15008         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
15009         unnecessary bits in the constant power of two case.
15011 2013-04-10  Richard Biener  <rguenther@suse.de>
15013         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15014         broken code swapping operands.
15015         (vect_build_slp_tree): Do not compute load permutations here.
15016         (vect_analyze_slp_instance): Compute load permutations here,
15017         after building the SLP tree.
15019 2013-04-09  Christian Bruel  <christian.bruel@st.com>
15021         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15022         of next/prev_real_insn.
15024 2013-04-09  Jan Hubicka  <jh@suse.cz>
15026         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15027         Drop aliased parameter.
15028         (function_and_variable_visibility): Do not handle alias pairs.
15029         * cgraph.c (varpool_externally_visible_p): Update prototype.
15030         * varpool.c (varpool_add_new_variable): Update.
15032 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15034         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15036 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
15038         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15040         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15042 2013-04-09  Marek Polacek  <polacek@redhat.com>
15044         PR tree-optimization/48762
15045         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15047 2013-04-09  Richard Biener  <rguenther@suse.de>
15049         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15050         dealing with cost.
15051         (vect_build_slp_tree): Likewise.
15052         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15053         calculating the cost of a SLP instance.
15054         (vect_analyze_slp_instance): Use it from here, after building
15055         the SLP tree.
15057 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15059         PR middle-end/56883
15060         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15061         expand_omp_for_static_chunk): Use simple_p = true in
15062         force_gimple_operand_gsi calls when assigning to addressable decls.
15064 2013-04-09  Jeff Law  <law@redhat.com>
15066         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15067         when the boolean was created by converting a wider object which
15068         had a boolean range.
15070 2013-04-09  Richard Biener  <rguenther@suse.de>
15072         * tree-vectorizer.h (slp_void_p): Remove.
15073         (slp_tree): Typedef before _slp_tree declaration.
15074         (struct _slp_tree): Use a vector of slp_tree as children.
15075         (vect_get_place_in_interleaving_chain): Remove.
15076         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15077         Move ...
15078         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15079         and make static.
15080         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15081         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15082         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15083         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15084         Use slp_node instead of slp_void_p and adjust.
15086 2013-04-09  Richard Biener  <rguenther@suse.de>
15088         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15089         work that is not necessary.
15091 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15093         PR tree-optimization/56854
15094         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15095         forward into clobber stmts if it would change MEM_REF lhs into
15096         non-MEM_REF.
15098 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15100         * tree.c (type_hash_lookup, type_hash_add): Make static.
15101         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15103 2013-04-09  Richard Biener  <rguenther@suse.de>
15105         * tree.h (unsave_expr_now): Remove.
15106         * tree-inline.c (mark_local_for_remap_r): Remove.
15107         (unsave_expr_1): Likewise.
15108         (unsave_r): Likewise.
15109         (unsave_expr_now): Likewise.
15110         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15111         (propagate_tree_value): Likewise.
15113 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15115         * doc/rtl.texi (sequence): Rewrite documentation to match the
15116         current use of SEQUENCE rtl objects.
15117         * rtl.def (SEQUENCE): Likewise.
15119         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15120         Update documentation.
15121         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15122         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15124         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15126 2013-04-08  Teresa Johnson  <tejohnson@google.com>
15128         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15129         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15130         methods.
15131         (estimate_edge_size_and_time): Add comment to suggest using rounding
15132         methods.
15133         (estimate_node_size_and_time): Ditto.
15134         (remap_edge_change_prob): Use helper rounding divide methods.
15135         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15136         (gimple_mod_pow2_value_transform): Ditto.
15137         (gimple_mod_subtract_transform): Ditto.
15138         (gimple_ic_transform): Ditto.
15139         (gimple_stringops_transform): Ditto.
15140         * stmt.c (conditional_probability): Ditto.
15141         (emit_case_dispatch_table): Ditto.
15142         * lto-cgraph.c (merge_profile_summaries): Ditto.
15143         * tree-optimize.c (execute_fixup_cfg): Ditto.
15144         * cfgcleanup.c (try_forward_edges): Ditto.
15145         * cfgloopmanip.c (scale_loop_profile): Ditto.
15146         (loopify): Ditto.
15147         (duplicate_loop_to_header_edge): Ditto.
15148         (lv_adjust_loop_entry_edge): Ditto.
15149         * tree-vect-loop.c (vect_transform_loop): Ditto.
15150         * profile.c (compute_branch_probabilities): Ditto.
15151         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15152         * lto-streamer-in.c (input_cfg): Ditto.
15153         * gimple-streamer-in.c (input_bb): Ditto.
15154         * ipa-cp.c (update_profiling_info): Ditto.
15155         (update_specialized_profile): Ditto.
15156         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15157         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15158         rounding methods.
15159         * sched-rgn.c (compute_dom_prob_ps): Ditto.
15160         (compute_trg_info): Ditto.
15161         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15162         (purge_dead_edges): Ditto.
15163         * loop-unswitch.c (unswitch_loop): Ditto.
15164         * cgraphclones.c (cgraph_clone_edge): Ditto.
15165         (cgraph_clone_node): Ditto.
15166         * tree-inline.c (copy_bb): Ditto.
15167         (copy_edges_for_bb): Ditto.
15168         (initialize_cfun): Ditto.
15169         (copy_cfg_body): Ditto.
15170         (expand_call_inline): Ditto.
15172 2013-04-08  Kai Tietz  <ktietz@redhat.com>
15174         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15175         TARGET_CYGWIN64 by TARGET_64BIT.
15177 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15179         * config/epiphany/epiphany.md (GPR_1): New constant.
15180         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15181         * config/epiphany/epiphany.c (gen_compare_reg):
15182         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15183         is already in place.
15184         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15185         Don't require being called during rtl expansion; If y operlaps r0,
15186         return 0.
15187         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15188         (epiphany_expand_epilogue): Likewise.
15190         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15191         Don't use CC_FPmode for ORDERED / UNORDERED.
15192         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15194         * config/epiphany/constraints.md (CnL): New constraint.
15195         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15196         * config/epiphany/predicates.md (add_operand): Allow 1024.
15198         * config/epiphany/epiphany.md (logical_op): New code iterator.
15199         (op_mnc): New code attribute.
15200         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15201         (mov_f+1, mov_f+2): New peephole2 patterns.
15203         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15204         (cstoresi4): Also allow re-use of zero result when doing a NE
15205         comparison to a non-zero operand.
15206         Use (clobber (scratch)) for first insn if the gpr output is not needed.
15208         * config/epiphany/epiphany.md (<insn_opname>v2si3):
15209         Use gen_addsi3_i / gen_subsi3_i.
15211 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15213         PR c++/34949
15214         PR c++/50243
15215         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15216         contain anything but clobbers, at most one __builtin_stack_restore,
15217         optionally debug stmts and final resx, and if it has at least one
15218         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
15219         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15220         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15221         which isn't defaut definition, remove them.
15222         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15223         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15224         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15225         with MEM_REF LHS with SSA_NAME address.
15227 2013-04-08  Jeff Law  <law@redhat.com>
15229         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15231 2013-04-08  Richard Biener  <rguenther@suse.de>
15233         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
15234         extra newline.
15235         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
15236         determined vector type.
15237         (vect_analyze_data_refs): Likewise.
15238         (vect_get_new_vect_var): Adjust.
15239         (vect_create_destination_var): Preserve SSA name versions.
15240         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
15241         not dump anything here.
15243 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15245         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
15246         Add member lr_slot_known.
15247         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
15248         if necessary.
15249         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
15250         Remove code that sets lr_slot_offset according to what a previous
15251         version of epiphany_emit_save_restore used to do.
15252         (epiphany_emit_save_restore): When doing an lr save or restore,
15253         set/verify lr_slot_known and lr_slot_offset.
15255 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
15257         PR target/54338
15258         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
15259         in ALL_REGS.
15261 2013-04-08  Richard Biener  <rguenther@suse.de>
15263         * alias.c (find_base_term): Fix thinko in previous change.
15265 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15267         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
15268         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
15269         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
15270         if possible to compute val.
15271         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
15272         For QImode integers don't require anything about precision.  Use
15273         const_with_all_bytes_same to find out if the constant doesn't have
15274         repeated bytes in it.
15276 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15278         * config/s390/s390.c (s390_expand_insv): Only accept insertions
15279         within mode size.
15281 2013-04-08  Marek Polacek  <polacek@redhat.com>
15283         PR rtl-optimization/48182
15284         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
15285         value to 1.
15287 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15289         PR target/55487
15290         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
15291         nuses, make sure we have a label.
15293 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15295         PR target/56843
15296         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
15297         (rs6000_emit_swdiv_low_precision): Remove.
15298         (rs6000_emit_swdiv): Rewrite to handle between one and four
15299         iterations of Newton-Raphson generally; modify required number of
15300         iterations for some cases.
15301         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
15303 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
15305         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
15306         set-but-unused variable.
15308         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
15309         basic blocks of released function bodies garbage-collectable.
15311         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
15312         (struct rtl_opt_pass): Add TODO_df_finish.
15314         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
15316 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15318         * config/arm/constraints.md (q): New constraint.
15319         * config/arm/ldrdstrd.md: New file.
15320         * config/arm/arm.md (ldrdstrd.md) New include.
15321         (arm_movdi): Use "q" instead of "r" constraint
15322         for double-word memory access.
15323         (movdf_soft_insn): Likewise.
15324         * config/arm/vfp.md (movdi_vfp): Likewise.
15325         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
15326         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
15327         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
15328         (mem_ok_for_ldrd_strd): Likewise.
15329         (output_move_double): Update assertion.
15331 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15333         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
15335 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15337         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
15338         define_insn_and_split.
15339         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
15341 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15343         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
15344         define_insn_and_split.
15345         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
15346         (shiftsi3_compare): New pattern.
15347         (rrx): New pattern.
15348         * config/arm/unspecs.md (UNSPEC_RRX): New.
15350 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15352         * config/arm/arm.md (negdi_extendsidi): New pattern.
15353         (negdi_zero_extendsidi): Likewise.
15355 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15357         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
15358         define_insn_and_split.
15359         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
15360         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
15362 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15364         * config/arm/arm.md (arm_subdi3): Convert define_insn into
15365         define_insn_and_split.
15366         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
15367         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
15369 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15371         * config/arm/arm.md (subsi3_carryin): New pattern.
15372         (subsi3_carryin_const): Likewise.
15373         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
15374         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
15376 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15378         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
15380 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15382         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
15383         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
15385 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15387         * config/arm/arm.c (arm_expand_builtin): Change fcode
15388         type to unsigned int.
15390 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15392         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
15394 2013-04-04  Ian Lance Taylor  <iant@google.com>
15396         * doc/standards.texi (Standards): The Go frontend supports the Go 1
15397         language standard.
15399 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
15401         PR middle-end/56729
15402         * df-scan.c (df_insn_delete): Disable failing assert.
15404 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15406         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
15407         New function prototype.
15408         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
15409         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
15410         (arm_builtin_vectorized_function): New function.
15412 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15414         * config/arm/arm_neon_builtins.def: New file.
15415         * config/arm/arm.c (neon_builtin_data): Move contents to
15416         arm_neon_builtins.def.
15417         (enum arm_builtins): Include neon builtin definitions.
15418         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
15419         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
15421 2013-04-04  Marek Polacek  <polacek@redhat.com>
15423         PR tree-optimization/48186
15424         * predict.c (maybe_hot_frequency_p): Return false if
15425         HOT_BB_FREQUENCY_FRACTION is 0.
15426         (cgraph_maybe_hot_edge_p): Likewise.
15428 2013-04-04  Richard Biener  <rguenther@suse.de>
15430         PR tree-optimization/56826
15431         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
15432         more accurately.
15434 2013-04-04  Richard Biener  <rguenther@suse.de>
15436         PR tree-optimization/56213
15437         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
15438         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
15440 2013-04-04  Richard Biener  <rguenther@suse.de>
15442         PR tree-optimization/56837
15443         * tree-loop-distribution.c (classify_partition): For non-zero
15444         values require that the value has the same precision as its
15445         mode to be useful as memset value.
15447 2013-04-03  Nick Clifton  <nickc@redhat.com>
15449         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
15450         (fmssf4): Use fmsf.s on E3V5 architectures.
15451         (fnmasf4): Use fnmaf.s on E3V5 architectures.
15452         (fnmssf4): Use fnmsf.s on E3V5 architectures.
15454 2013-04-03  Jeff Law  <law@redhat.com>
15456         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
15457         (lra-eliminations.o): Likewise.
15459 2013-04-03  Teresa Johnson  <tejohnson@google.com>
15461         * gcov-io.c (compute_working_sets): Moved most of body of old
15462         compute_working_sets here from profile.c.
15463         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
15464         (gcov_working_set_t): Moved typedef here from basic-block.h
15465         (compute_working_set): Declare.
15466         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
15467         (get_working_sets): Renamed from compute_working_set,
15468         replace most of body with call to new compute_working_sets.
15469         (get_exec_counts): Replace call to compute_working_sets
15470         to get_working_sets.
15471         * profile.h (get_working_sets): Renamed from compute_working_set.
15472         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
15473         to get_working_sets.
15474         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
15475         * gcov-dump.c (dump_working_sets): New function.
15477 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
15479         * hwint.c (sext_hwi, zext_hwi): New functions.
15480         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
15481         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
15482         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
15483         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
15484         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
15485         (sext_hwi, zext_hwi): New functions.
15487 2013-04-03  Jeff Law  <law@redhat.com>
15489         PR tree-optimization/56799
15490         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
15491         back test for widening conversion erroneously dropped in prior change.
15493 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15495         PR target/56809
15496         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
15497         instead of next_real_insn.
15499 2013-04-03  Marek Polacek  <polacek@redhat.com>
15501         PR sanitizer/55702
15502         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
15504 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15506         PR target/56809
15507         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
15508         next_real_insn.
15509         (thumb1_output_casesi): Likewise.
15510         (thumb2_output_casesi): Likewise.
15512 2013-04-03  Richard Biener  <rguenther@suse.de>
15514         PR tree-optimization/56817
15515         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
15516         Split out ...
15517         (tree_unroll_loops_completely_1): ... new function to manually
15518         walk the loop tree, properly defering outer loops of unrolled
15519         loops to later iterations.
15521 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15523         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
15524         (vectorizable_load): Likewise.
15525         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
15526         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
15528 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15530         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
15531         BIT_FIELD_REF.
15533 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15535         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
15537 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
15539         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
15541 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15543         PR tree-optimization/56790
15544         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
15545         folding.
15547 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15549         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
15550         Handle VEC_MERGE.
15551         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
15552         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
15553         equal arguments.
15555 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
15557         PR c/19449
15558         * tree.h (force_folding_builtin_constant_p): New decl.
15559         * builtins.c (force_folding_builtin_constant_p): New variable.
15560         (fold_builtin_constant_p): Fold immediately also if
15561         force_folding_builtin_constant_p.
15563 2013-04-03  Richard Biener  <rguenther@suse.de>
15565         PR tree-optimization/56812
15566         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
15567         DRs of the same interleaving chain are independent.
15569 2013-04-02  Jason Merrill  <jason@redhat.com>
15571         * gdbinit.in (pbb): Use debug fn.
15573 2013-04-02  Lawrence Crowl  <crowl@google.com>
15575         * sese.h (struct ivtype_map_elt_s): Remove unused.
15576         (extern debug_ivtype_map): Remove unused.
15577         (extern eq_ivtype_map_elts): Remove unused.
15578         * sese.c (debug_ivtype_map): Removed unused.
15579         (debug_ivtype_map_1): Removed unused.
15580         (debug_ivtype_elt): Remove unused.
15581         (eq_ivtype_map_elts): Remove unused.
15584 2013-04-02  Kai Tietz  <ktietz@redhat.com>
15586         PR target/52790
15587         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
15588         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
15589         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
15590         function.
15591         (legitimize_pe_coff_symbol): Likewise.
15592         (is_imported_p): New helper-function.
15593         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
15594         for Windows x64 targets.
15595         (ix86_expand_prologue): Optimize for pe-coff targets.
15596         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
15597         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
15598         medium/large code-model.
15599         (legitimize_pic_address): Likewise.
15600         (legitimize_tls_address): Likewise.
15601         (ix86_expand_call): Likewise.
15602         (x86_output_mi_thunk): Likewise.
15603         (get_dllimport_decl): Add new beimport argument.
15604         (construct_plt_address): Don't assert for x64 pe-coff targets.
15605         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
15606         targets.
15607         (SYMBOL_FLAG_STUBVAR): New macro.
15608         (SYMBOL_REF_STUBVAR_P): Likewise.
15609         * config/i386/winnt.c (stub_list): New structure.
15610         (stub_head): New local variable.
15611         (i386_pe_record_stub): New function.
15612         (i386_pe_file_end): Emit refptr-stubs.
15614 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
15616         PR rtl-optimization/56745
15617         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
15618         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
15620         PR c++/34949
15621         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
15622         and both of them are MEM_REFs, just compare first argument for
15623         equality and attempt to deal even with differing offsets.
15625         PR c++/34949
15626         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
15627         of gimple_clobber_p to be MEM_REF.
15628         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
15629         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
15630         after gimplification.
15631         * asan.c (get_mem_ref_of_assignment): Don't instrument
15632         gimple_clobber_p stmts.
15633         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
15634         gimple_clobber_p stmt if they have MEM_REF lhs and
15635         are dead because of another gimple_clobber_p stmt.
15636         * tree-ssa-live.c (clear_unused_block_pointer): Treat
15637         gimple_clobber_p stmts like debug stmts.
15638         (remove_unused_locals): Remove clobbers with MEM_REF lhs
15639         that refer to unused VAR_DECLs or uninitialized values.
15640         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
15641         gimple_clobber_p stmts if they refer to removed parameters.
15642         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
15643         formatting.
15645 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
15647         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
15648         using SWI48 mode attribute.
15650 2013-04-02  Wei Mi  <wmi@google.com>
15652         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
15653         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
15654         *<rotate_insn><mode>3_mask in i386.md.
15656 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
15658         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
15660 2013-04-02  Richard Biener  <rguenther@suse.de>
15662         PR tree-optimization/56778
15663         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15664         Runtime alias tests are not supported for gather loads.
15665         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
15666         stmts referenced from SSA operands before updating SSA form.
15668 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
15669             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15671         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
15672         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
15673         * config/arm/cortex-a53.md: New file.
15674         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
15675         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
15676         * config/arm/arm.c (arm_issue_rate): Likewise.
15677         * config/arm/arm-tune.md: Regenerate
15678         * config/arm/arm-tables.opt: Regenerate.
15679         * config/arm/arm-cores.def: Add cortex-a53.
15681 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
15683         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
15684         non-static link.
15686 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15688         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
15689         scalar load/store operations using B/H registers.
15690         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
15692 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
15694         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
15695         scalar move.
15696         * config/aarch64/aarch64.c
15697         (aarch64_simd_scalar_immediate_valid_for_move): New.
15698         * config/aarch64/aarch64-protos.h
15699         (aarch64_simd_scalar_immediate_valid_for_move): New.
15700         * config/aarch64/constraints.md (Dh, Dq): New.
15701         * config/aarch64/iterators.md (hq): New.
15703 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
15705         * reorg.c (get_branch_condition): Deal with conditional returns.
15706         (fill_simple_delay_slots): Remove dead code dealing with jumps.
15708 2013-04-01  Wei Mi  <wmi@google.com>
15710         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
15711         Truncate operand 2 using %b asm operand modifier.
15712         (*<shift_insn><mode>3_mask): Ditto.
15713         (*<rotate_insn><mode>3_mask): Ditto.
15715 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
15717         PR middle-end/56798
15718         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
15720 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
15722         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
15723         of next_real_insn.
15724         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
15726 2013-03-30  Lawrence Crowl  <crowl@google.com>
15728         * dse.c (clear_alias_sets): Remove never set.
15729         (disqualified_clear_alias_sets): Remove never set.
15730         (clear_alias_mode_pool): Remove never set.
15731         (dse_step0): Remove condition that is never true.
15732         (canon_address): Remove condition that is never true.
15733         (dse_step7): Remove condition that is never true.
15734         (rest_of_handle_dse): Remove condition that is never true.
15735         (rest_of_handle_dse::did_global): Remove never read from above.
15736         (dse_step2_spill): Remove never called from above.
15737         (dse_step5_spill): Remove never called from above.
15739 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
15741         * doc/md.texi (Standard Names) <casesi>: Update documentation for
15742         JUMP_TABLE_DATA changes.
15743         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
15744         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
15745         (Insns) <jump_table_data>: New entry.
15746         * doc/tm.texi: Regenerate.
15748         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
15750         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
15751         for table jump at the end of a basic block using tablejump_p.
15752         * targhooks.c (default_invalid_within_doloop): Likewise.
15753         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
15754         target hook implementation that is identical to the default hook.
15755         (rs6000_invalid_within_doloop): Remove.
15757         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
15758         unused variable from tablejump_p call.
15760         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
15761         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
15762         (INSN_DELETED_P): Likewise.
15763         (emit_jump_table_data): New prototype.
15764         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
15765         after 4th as unused.
15766         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
15767         * sched-vis.c (print_insn): Likewise.
15768         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
15769         insn for compatibility with back ends that use next_active_insn to
15770         identify jump table data.
15771         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
15772         (remove_insn): Likewise.
15773         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
15774         to be emitted.
15775         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
15776         (emit_jump_table_data): New function.
15778         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
15779         basic block, a JUMP_TABLE_DATA never is.
15780         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
15781         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
15782         off from code handling real insns.
15783         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
15784         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
15785         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
15786         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
15787         is not a NONDEBUG_INSN_P.
15788         * ira-costs.c (scan_one_insn): Likewise.
15789         * jump.c (mark_all_labels): Likewise.
15790         (mark_jump_label_1): Likewise.
15791         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15792         * lra.c (get_insn_freq): Expect all insns reaching here to be in
15793         a basic block.
15794         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
15795         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
15796         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
15797         JUMP_TABLE_DATA_P insns.
15798         (calculate_elim_costs_all_insns): Likewise.
15799         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
15800         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
15801         (delete_output_reload): Code style fixups.
15802         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
15803         insn flags on this non-insn.
15804         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
15805         as scheduling barriers, for pre-change compatibility.
15806         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
15807         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
15809         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
15810         redundant JUMP_TABLE_DATA_P test.
15811         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
15812         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15813         (frv_for_each_packet): Likewise.
15814         * config/i386/i386.c (min_insn_size): Likewise.
15815         (ix86_avoid_jump_mispredicts): Likewise.
15816         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15817         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
15818         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15819         (mips16_insn_length): Robustify.
15820         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
15821         (mips16_split_long_branches): Likewise.
15822         * config/pa/pa.c (pa_combine_instructions): Likewise.
15823         * config/rs6000/rs6000.c (get_next_active_insn): Treat
15824         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
15825         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
15826         as contributing to pool range lengths.
15827         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
15828         Remove redundant JUMP_TABLE_DATA_P test.
15829         (sh_loop_align): Likewise.
15830         (split_branches): Likewise.
15831         (sh_insn_length_adjustment): Likewise.
15832         * config/spu/spu.c (get_branch_target): Likewise.
15834 2013-03-29  Jan Hubicka  <jh@suse.cz>
15836         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
15837         gcov streaming; stream hot bb threshold to ltrans.
15838         * predict.c (get_hot_bb_threshold): Break out from ....
15839         (maybe_hot_count_p): ... here.
15840         (set_hot_bb_threshold): New function.
15841         * lto-section-in.c (lto_section_name): Add profile.
15842         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
15843         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
15844         and data-streamer.h
15845         (histogram_entry): New structure.
15846         (histogram, histogram_pool): New global vars.
15847         (histogram_hash): New structure.
15848         (histogram_hash::hash): New method.
15849         (histogram_hash::equal): Likewise.
15850         (account_time_size): New function.
15851         (cmp_counts): New function.
15852         (dump_histogram): New function.
15853         (ipa_profile_generate_summary): New function.
15854         (ipa_profile_write_summary): New function.
15855         (ipa_profile_read_summary): New function.
15856         (ipa_profile): Decide on threshold.
15857         (pass_ipa_profile): Add ipa_profile_write_summary and
15858         ipa_profile_read_summary.
15859         * Makefile.in (ipa.o): Update dependencies.
15860         * lto-streamer.h (LTO_section_ipa_profile): New section.
15862 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
15864         * tree.h (VAR_P): New.
15866 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
15868         PR lto/56777
15869         * doc/invoke.texi ([-fwhole-program]): Fix typo.
15871 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
15873         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
15874         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15875         (control_flow_insn_p): Likewise.
15876         * cfgrtl.c (duplicate_insn_chain): Likewise.
15877         * final.c (get_attr_length_1): Likewise.
15878         (shorten_branches): Likewise.
15879         (final_scan_insn): Likewise.
15880         * function.c (instantiate_virtual_regs): Likewise.
15881         * gcse.c (insert_insn_end_basic_block): Likewise.
15882         * ira-costs.c (scan_one_insn): Likewise.
15883         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
15884         * lra.c (check_rtl): Likewise.
15885         * reload1.c (elimination_costs_in_insn): Likewise.
15886         * reorg.c (follow_jumps): Likewise.
15888         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
15889         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
15890         (thumb_far_jump_used_p): Likewise.
15891         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
15892         (workaround_speculation): Likewise.
15893         (add_sched_insns_for_speculation): Likewise.
15894         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
15895         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
15896         (frv_for_each_packet): Likewise.
15897         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
15898         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
15899         (final_emit_insn_group_barriers): Likewise.
15900         * config/m32r/m32r.c (m32r_is_insn): Likewise.
15901         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
15902         (mips16_insn_length): Likewise.
15903         * config/pa/pa.c (pa_reorg): Likewise.
15904         (pa_combine_instructions): Likewise.
15905         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
15906         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
15907         (sh_reorg): Likewise.
15908         (split_branches): Likewise.
15909         * config/spu/spu.c (get_branch_target): Likewise.
15911         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
15912         JUMP_TABLE_DATA_P.
15914 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
15916         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
15917         Fix declaration name.
15919 2013-03-28  Lawrence Crowl  <crowl@google.com>
15921         * graphds.h (struct graph.indicies): Remove unused.
15922         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
15923         (SCOP_ORIGINAL_PDDRS): Remove unused.
15924         * sese.h (extern insert_loop_close_phis): Removed unused.
15925         (extern insert_guard_phis): Removed unused.
15926         (extern ivtype_map_elt_info): Removed unused.
15927         (new_ivtype_map_elt): Removed unused.
15928         * sese.c (ivtype_map_elt_info): Removed unused.
15930 2013-03-28  Lawrence Crowl  <crowl@google.com>
15932         * Makefile.in: Add several missing include dependences.
15933         (DUMPFILE_H): New.
15934         (test-dump.o): New.  This object is not added to any executable,
15935         but is present for ad-hoc testing.
15936         * bitmap.c
15937         (debug (const bitmap_head_def &)): New.
15938         (debug (const bitmap_head_def *)): New.
15939         * bitmap.h
15940         (extern debug (const bitmap_head_def &)): New.
15941         (extern debug (const bitmap_head_def *)): New.
15942         * cfg.c
15943         (debug (edge_def &)): New.
15944         (debug (edge_def *)): New.
15945         * cfghooks.c
15946         (debug (basic_block_def &)): New.
15947         (debug (basic_block_def *)): New.
15948         * dumpfile.h
15949         (dump_node (const_tree, int, FILE *)): Correct source file.
15950         * dwarf2out.c
15951         (debug (die_struct &)): New.
15952         (debug (die_struct *)): New.
15953         * dwarf2out.h
15954         (extern debug (die_struct &)): New.
15955         (extern debug (die_struct *)): New.
15956         * gimple-pretty-print.c
15957         (debug (gimple_statement_d &)): New.
15958         (debug (gimple_statement_d *)): New.
15959         * gimple-pretty-print.h
15960         (extern debug (gimple_statement_d &)): New.
15961         (extern debug (gimple_statement_d *)): New.
15962         * ira-build.c
15963         (debug (ira_allocno_copy &)): New.
15964         (debug (ira_allocno_copy *)): New.
15965         (debug (ira_allocno &)): New.
15966         (debug (ira_allocno *)): New.
15967         * ira-int.h
15968         (extern debug (ira_allocno_copy &)): New.
15969         (extern debug (ira_allocno_copy *)): New.
15970         (extern debug (ira_allocno &)): New.
15971         (extern debug (ira_allocno *)): New.
15972         * ira-lives.c
15973         (debug (live_range &)): New.
15974         (debug (live_range *)): New.
15975         * lra-int.h
15976         (debug (lra_live_range &)): New.
15977         (debug (lra_live_range *)): New.
15978         * lra-lives.c
15979         (debug (lra_live_range &)): New.
15980         (debug (lra_live_range *)): New.
15981         * omega.c
15982         (debug (omega_pb_d &)): New.
15983         (debug (omega_pb_d *)): New.
15984         * omega.h
15985         (extern debug (omega_pb_d &)): New.
15986         (extern debug (omega_pb_d *)): New.
15987         * print-rtl.c
15988         (debug (const rtx_def &)): New.
15989         (debug (const rtx_def *)): New.
15990         * print-tree.c
15991         (debug_tree (tree): Move within file.
15992         (debug_raw (const tree_node &)): New.
15993         (debug_raw (const tree_node *)): New.
15994         (dump_tree_via_hooks (const tree_node *, int)): New.
15995         (debug (const tree_node &)): New.
15996         (debug (const tree_node *)): New.
15997         (debug_verbose (const tree_node &)): New.
15998         (debug_verbose (const tree_node *)): New.
15999         (debug_head (const tree_node &)): New.
16000         (debug_head (const tree_node *)): New.
16001         (debug_body (const tree_node &)): New.
16002         (debug_body (const tree_node *)): New.
16003         (debug_vec_tree (tree): Move and reimplement in terms of dump.
16004         (debug (vec<tree, va_gc> &)): New.
16005         (debug (vec<tree, va_gc> *)): New.
16006         * rtl.h
16007         (extern debug (const rtx_def &)): New.
16008         (extern debug (const rtx_def *)): New.
16009         * sbitmap.c
16010         (debug_raw (simple_bitmap_def &)): New.
16011         (debug_raw (simple_bitmap_def *)): New.
16012         (debug (simple_bitmap_def &)): New.
16013         (debug (simple_bitmap_def *)): New.
16014         * sbitmap.h
16015         (extern debug (simple_bitmap_def &)): New.
16016         (extern debug (simple_bitmap_def *)): New.
16017         (extern debug_raw (simple_bitmap_def &)): New.
16018         (extern debug_raw (simple_bitmap_def *)): New.
16019         * sel-sched-dump.c
16020         (debug (vinsn_def &)): New.
16021         (debug (vinsn_def *)): New.
16022         (debug_verbose (vinsn_def &)): New.
16023         (debug_verbose (vinsn_def *)): New.
16024         (debug (expr_def &)): New.
16025         (debug (expr_def *)): New.
16026         (debug_verbose (expr_def &)): New.
16027         (debug_verbose (expr_def *)): New.
16028         (debug (vec<rtx> &)): New.
16029         (debug (vec<rtx> *)): New.
16030         * sel-sched-dump.h
16031         (extern debug (vinsn_def &)): New.
16032         (extern debug (vinsn_def *)): New.
16033         (extern debug_verbose (vinsn_def &)): New.
16034         (extern debug_verbose (vinsn_def *)): New.
16035         (extern debug (expr_def &)): New.
16036         (extern debug (expr_def *)): New.
16037         (extern debug_verbose (expr_def &)): New.
16038         (extern debug_verbose (expr_def *)): New.
16039         (extern debug (vec<rtx> &)): New.
16040         (extern debug (vec<rtx> *)): New.
16041         * sel-sched-ir.h
16042         (_list_iter_cond_expr): Make inline instead of static.
16043         * sreal.c
16044         (debug (sreal &)): New.
16045         (debug (sreal *)): New.
16046         * sreal.h
16047         (extern debug (sreal &)): New.
16048         (extern debug (sreal *)): New.
16049         * tree.h
16050         (extern debug_raw (const tree_node &)): New.
16051         (extern debug_raw (const tree_node *)): New.
16052         (extern debug (const tree_node &)): New.
16053         (extern debug (const tree_node *)): New.
16054         (extern debug_verbose (const tree_node &)): New.
16055         (extern debug_verbose (const tree_node *)): New.
16056         (extern debug_head (const tree_node &)): New.
16057         (extern debug_head (const tree_node *)): New.
16058         (extern debug_body (const tree_node &)): New.
16059         (extern debug_body (const tree_node *)): New.
16060         (extern debug (vec<tree, va_gc> &)): New.
16061         (extern debug (vec<tree, va_gc> *)): New.
16062         * tree-cfg.c
16063         (debug (struct loop &)): New.
16064         (debug (struct loop *)): New.
16065         (debug_verbose (struct loop &)): New.
16066         (debug_verbose (struct loop *)): New.
16067         * tree-dump.c: Add header dependence.
16068         * tree-flow.h
16069         (extern debug (struct loop &)): New.
16070         (extern debug (struct loop *)): New.
16071         (extern debug_verbose (struct loop &)): New.
16072         (extern debug_verbose (struct loop *)): New.
16073         * tree-data-ref.c
16074         (debug (data_reference &)): New.
16075         (debug (data_reference *)): New.
16076         (debug (vec<data_reference_p> &)): New.
16077         (debug (vec<data_reference_p> *)): New.
16078         (debug (vec<ddr_p> &)): New.
16079         (debug (vec<ddr_p> *)): New.
16080         * tree-data-ref.h
16081         (extern debug (data_reference &)): New.
16082         (extern debug (data_reference *)): New.
16083         (extern debug (vec<data_reference_p> &)): New.
16084         (extern debug (vec<data_reference_p> *)): New.
16085         (extern debug (vec<ddr_p> &)): New.
16086         (extern debug (vec<ddr_p> *)): New.
16087         * tree-ssa-alias.c
16088         (debug (pt_solution &)): New.
16089         (debug (pt_solution *)): New.
16090         * tree-ssa-alias.h
16091         (extern debug (pt_solution &)): New.
16092         (extern debug (pt_solution *)): New.
16093         * tree-ssa-alias.c
16094         (debug (_var_map &)): New.
16095         (debug (_var_map *)): New.
16096         (debug (tree_live_info_d &)): New.
16097         (debug (tree_live_info_d *)): New.
16098         * tree-ssa-alias.h
16099         (extern debug (_var_map &)): New.
16100         (extern debug (_var_map *)): New.
16101         (extern debug (tree_live_info_d &)): New.
16102         (extern debug (tree_live_info_d *)): New.
16104 2013-03-28  Jan Hubicka  <jh@suse.cz>
16106         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16108 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16110         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16111         record only when desired or required.
16113 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16115         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16116         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16118 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16120         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16121         (*andsi3_compare0_uxtw): New pattern.
16122         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16123         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16125 2013-03-28  Jan Hubicka  <jh@suse.cz>
16127         * data-streamer-in.c (streamer_read_gcov_count): New function.
16128         * gimple-streamer-out.c: Include value-prof.h.
16129         (output_gimple_stmt): Output histogram.
16130         (output_bb): Use streamer_write_gcov_count.
16131         * value-prof.c: Include data-streamer.h
16132         (dump_histogram_value): Add HIST_TYPE_MAX.
16133         (stream_out_histogram_value): New function.
16134         (stream_in_histogram_value): New function.
16135         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16136         (stream_out_histogram_value, stream_in_histogram_value): Declare.
16137         * data-streamer-out.c (streamer_write_gcov_count): New function.
16138         (streamer_write_gcov_count_stream): New function.
16139         * lto-cgraph.c (lto_output_edge): Update counter streaming.
16140         (lto_output_node): Likewise.
16141         (input_node, input_edge): Likewise.
16142         * lto-streamer-out.c (output_cfg): Update streaming.
16143         * lto-streamer-in.c (input_cfg): Likewise.
16144         * data-streamer.h (streamer_write_gcov_count,
16145         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16146         * gimple-streamer-in.c: Include value-prof.h
16147         (input_gimple_stmt): Input histograms.
16148         (input_bb): Update profile streaming.
16150 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16152         * genmodes.c (emit_max_int): New function.
16153         (emit_insn_modes_h): Added call to emit_max_function.
16154         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16155         Added doc.
16156         * machmode.def: Fixed comment.
16158 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16160         * combine.c (try_combine): Removed useless assert.
16161         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16163 2013-03-28  Marek Polacek  <polacek@redhat.com>
16164             Richard Biener  <rguenther@suse.de>
16166         PR tree-optimization/56695
16167         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16168         build signed result of a vector comparison.
16169         * tree-cfg.c (verify_gimple_comparison): Check that a result
16170         of a vector comparison has signed type.
16172 2013-03-28  Richard Biener  <rguenther@suse.de>
16174         PR tree-optimization/37021
16175         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16176         do not restrict gaps between groups.
16177         * tree-vect-stmts.c (vectorizable_load): Properly account for
16178         a gap between groups.
16180 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
16182         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16183         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16184         is not enabled.
16186 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
16188         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16189         * doc/extend.texi (Named Address Spaces): Ditto.
16190         (Variable Attributes): Ditto.
16192 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16194         * config.build: Add support for cygwin x64 target.
16195         * config.gcc: Likewise.
16196         * config.host: Likewise.
16197         * configure.ac: Likewise
16198         * configure: Regenerated.
16200 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16202         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16203         * config/i386/t-cygwin-w64: New file.
16204         * config/i386/cygwin-w64.h: New file.
16205         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16206         and add support for x64-cygwin target.
16207         (CPP_SPEC): Likewise.
16208         (CXX_WRAP_SPEC_LIST): Undefine before define.
16209         (LIBGCJ_SONAME): Use 15 as version.
16211 2013-03-27  Richard Biener  <rguenther@suse.de>
16213         PR tree-optimization/56716
16214         * tree-ssa-structalias.c (perform_var_substitution): Adjust
16215         dumping for ref nodes.
16217 2013-03-27  Martin Jambor  <mjambor@suse.cz>
16219         PR tree-optimization/55334
16220         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16221         restricted pointers to arrays.
16223 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16225         * Makefile.in (.SUFFIXES): Add .cc.
16226         (.c.o): Apply same recipe for implicit rule .cc.o.
16228 2013-03-27  Richard Biener  <rguenther@suse.de>
16230         PR tree-optimization/37021
16231         * tree-vect-data-refs.c (vect_check_strided_load): Allow
16232         REALPART/IMAGPART_EXPRs around the supported refs.
16233         * tree-ssa-structalias.c (find_func_aliases): Assume that
16234         floating-point values are not used to transfer pointers.
16236 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16238         * target.def (TARGET_HAS_IFUNC_P): New target hook.
16239         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
16240         * doc/tm.texi: Regenerate.
16241         * targhooks.h (default_has_ifunc_p): New.
16242         * targhooks.c (default_has_ifunc_p): Ditto.
16243         * config/linux-protos.h: New file.
16244         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
16245         hook for linux which disables support of indirect functions in android.
16246         * config/linux-android.c: New file.
16247         * config/t-linux-android.c: Ditto.
16248         * config.gcc: Added new object file linux-android.o.
16249         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
16250         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
16251         * varasm.c (do_assemble_alias): Likewise.
16252         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
16253         doesn't support indirect functions.
16254         * configure: Regenerate.
16256 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
16258         PR target/56102
16259         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
16260         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
16261         mult-word mode.
16263 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16265         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
16267 2013-03-27  Terry Guo  <terry.guo@arm.com>
16269         * config/arm/arm-cores.def: Added core cortex-r7.
16270         * config/arm/arm-tune.md: Regenerated.
16271         * config/arm/arm-tables.opt: Regenerated.
16272         * doc/invoke.texi: Added entry for core cortex-r7.
16274 2013-03-27  Walter Lee  <walt@tilera.com>
16276         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
16277         double-decrement of next_scratch_regno.
16279 2013-03-27  Walter Lee  <walt@tilera.com>
16281         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
16282         input operands.
16283         (insn_v1mulus): Ditto.
16284         (insn_v2muls): Ditto.
16286 2013-03-27  Walter Lee  <walt@tilera.com>
16288         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
16289         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
16291 2013-03-27  Walter Lee  <walt@tilera.com>
16293         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
16294         (*sibcall_value): Ditto.
16296 2013-03-27  Walter Lee  <walt@tilera.com>
16298         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
16299         (insn_mnz_v8qi): ... this ...
16300         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
16301         vector equivalent.
16302         (insn_v<n>mnz): Replaced by ...
16303         (insn_v1mnz): ... this ...
16304         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
16305         equivalent.
16306         (insn_mz_<mode>): Replaced by ...
16307         (insn_mz_v8qi): ... this ...
16308         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
16309         vector equivalent.
16310         (insn_v<n>mz): Replaced by ...
16311         (insn_v1mz): ... this ...
16312         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
16313         equivalent.
16315 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
16317         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
16319 2013-03-26  Roland McGrath  <mcgrathr@google.com>
16321         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
16322         than fprintf with a non-constant, non-format string.
16324 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16326         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
16327         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
16328         operand 0 predicate.
16329         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
16330         attribute.  Use general_x64nomem_operand as operand 1 predicate.
16331         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
16332         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16333         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
16334         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16335         (mov<mode>_insv_1): Remove expander.  Merge insn with
16336         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
16337         Use general_x64nomem_operand as operand 1 predicate.
16338         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
16339         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
16340         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16341         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
16342         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
16343         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
16344         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16345         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
16346         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
16347         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
16348         (general_x64nomem_operand): Ditto.
16350 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
16352         * config/rtems.opt: Add -pthread option.
16354 2013-03-26  Richard Biener  <rguenther@suse.de>
16356         * alias.c (find_base_term): Avoid redundant and not used recursion.
16357         (base_alias_check): Get the initial base term from the caller.
16358         (true_dependence_1): Compute and pass base terms to base_alias_check.
16359         (write_dependence_p): Likewise.
16360         (may_alias_p): Likewise.
16362 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
16364         * config/aarch64/aarch64.c (aarch64_classify_address): Support
16365         PC-relative load in SI modes and above only.
16367 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
16369         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
16370         * config/arm/iwmmxt.md (WCGR0): Update.
16371         (WCGR1, WCGR2, WCGR3): Likewise.
16373 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16375         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
16376         Use x64 and nox64 isa attributes.
16378 2013-03-26  Richard Biener  <rguenther@suse.de>
16380         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
16381         alignment computations and rely on get_object_alignment_1
16382         for the !TYPE_P case.
16383         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
16385 2013-03-26  Walter Lee  <walt@tilera.com>
16387         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
16388         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
16390 2013-03-25  Jeff Law  <law@redhat.com>
16392         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
16393         check for INTEGRAL_TYPE_P that was missing due to checking in
16394         wrong version of prior patch.
16396 2013-03-25  Walter Lee  <walt@tilera.com>
16398         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
16399         TILEGX_INSN_SHUFFLEBYTES1.
16400         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
16401         shufflebytes1.
16402         (tilegx_builtins): Ditto.
16403         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
16405 2013-03-25  Walter Lee  <walt@tilera.com>
16407         * config/tilegx/tilegx.md (floatsisf2): New pattern.
16408         (floatunssisf2): New pattern.
16409         (floatsidf2): New pattern.
16410         (floatunssidf2): New pattern.
16412 2013-03-25  Walter Lee  <walt@tilera.com>
16414         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
16415         tests for constraint J, K, N, P.
16417 2013-03-25  Walter Lee  <walt@tilera.com>
16419         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
16420         Use indirect/pcrel encoding.
16421         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
16422         Ditto.
16424 2013-03-25  Steve Ellcey  <sellcey@mips.com>
16426         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
16427         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
16428         * config/mips/mips.c (mips_option_override): Set IMADD default.
16429         * config/mips/mips.h (PTF_AVOID_IMADD): New.
16430         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
16431         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
16432         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
16433         * doc/invoke.texi (-mimadd/-mno-imadd): New.
16435 2013-03-25  Jeff Law  <law@redhat.com>
16437         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
16438         slightly to avoid creating and folding useless trees.  Simplify
16439         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
16441 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
16443         * config/i386/i386.md (*zero_extendsidi2): Merge with
16444         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
16445         * config/i386/predicates.md (x86_64_zext_operand): Rename from
16446         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
16447         targets.  Clarify comment.
16449 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16451         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
16452         pass-through jump functions differently.
16453         (ipa_read_jump_function): Likewise.  Also use setter functions to set
16454         up jump functions.
16456 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16458         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
16459         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
16460         process it.
16461         (ipa_get_indirect_edge_target): New function.
16462         (devirtualization_time_bonus): New parameter known_aggs, pass it to
16463         ipa_get_indirect_edge_target.  Update all callers.
16464         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
16465         ipa_get_indirect_edge_target_1 instead of calling
16466         ipa_get_indirect_edge_target.
16467         (create_specialized_node): Pass aggvlas to
16468         ipcp_discover_new_direct_edges.
16470 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16472         * config/arm/arm.md (f_sels, f_seld): New types.
16473         (*cmov<mode>): New pattern.
16474         * config/arm/predicates.md (arm_vsel_comparison_operator): New
16475         predicate.
16477 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16479         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
16480         POSIX-printf for mingw-hosted  builds.
16482 2013-03-25  Richard Biener  <rguenther@suse.de>
16484         PR middle-end/56694
16485         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
16486         must-not-throw stmt location.
16488 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16490         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
16491         Emit load-acquire versions when acq is true.
16492         (arm_emit_store_exclusive): Add rel parameter.
16493         Emit store-release versions when rel is true.
16494         (arm_split_compare_and_swap): Use acquire-release instructions
16495         instead.
16496         of barriers when appropriate.
16497         (arm_split_atomic_op): Likewise.
16498         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
16499         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
16500         (VUNSPEC_SLX): Likewise.
16501         (VUNSPEC_LDA): Likewise.
16502         (VUNSPEC_STL): Likewise.
16503         * config/arm/sync.md (atomic_load<mode>): New pattern.
16504         (atomic_store<mode>): Likewise.
16505         (arm_load_acquire_exclusive<mode>): Likewise.
16506         (arm_load_acquire_exclusivesi): Likewise.
16507         (arm_load_acquire_exclusivedi): Likewise.
16508         (arm_store_release_exclusive<mode>): Likewise.
16510 2013-03-25  Catherine Moore  <clm@codesourcery.com>
16512         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
16513         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
16514         * config/mip/predicates.md (lwsp_swsp_operand,
16515         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
16516         sb16_operand, db4_operand, db7_operand, ib3_operand,
16517         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
16518         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
16519         andi16_operand): New predicates.
16520         * config/mips/mips.md (compression): New attribute.
16521         (enabled): New attribute.
16522         (length): Consider compression in computing length.
16523         (shift_compression): New code attribute.
16524         (*add<mode>3): New operands. Record compression.
16525         (sub<mode>3): Likewise.
16526         (one_cmpl<mode>2): Likewise.
16527         (*and<mode>3): Likewise.
16528         (*ior<mode>3): Likewise.
16529         (unnamed pattern for xor): Likewise.
16530         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
16531         (*<optab><mode>3): Likewise.
16532         (*mov<mode>_internal: Likewise.
16533         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
16534         (mips_unsigned_immediate_p): New.
16535         (umips_lwsp_swsp_address_p): New.
16536         (m16_based_address_p): New.
16537         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
16538         (mips_unsigned_immediate_p): New prototype.
16539         (lwsp_swsp_address_p): New prototype.
16540         (m16_based_address_p): New prototype.
16541         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
16542         (mips_signed_immediate_p): New function.
16543         (m16_based_address_p): New function.
16544         (lwsp_swsp_address_p): New function.
16545         (mips_print_operand_punctuation): Recognize short delay slot insns
16546         for microMIPS.add<mode>3"
16548 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16550         PR target/56720
16551         * config/arm/iterators.md (v_cmp_result): New mode attribute.
16552         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
16554 2013-03-25  Richard Biener  <rguenther@suse.de>
16556         PR tree-optimization/56689
16557         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
16558         any edge.
16560 2013-03-25  Richard Biener  <rguenther@suse.de>
16562         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
16563         of bitmap.
16564         (memory_references): Likewise.
16565         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
16566         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
16567         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
16568         (gather_mem_refs_in_loops): Fold into ...
16569         (analyze_memory_references): ... this.  Move initialization
16570         to tree_ssa_lim_initialize.
16571         (fill_always_executed_in): Rename to ...
16572         (fill_always_executed_in_1): ... this.
16573         (fill_always_executed_in): Move contains_call computation to
16574         this new function from ...
16575         (tree_ssa_lim_initialize): ... here.
16576         (tree_ssa_lim): Call fill_always_executed_in.
16578 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16580         * postreload.c (reload_combine): Fix code detecting returns.
16582 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
16584         * function.c (emit_use_return_register_into_block): On cc0 targets,
16585         do not emit the sequence between cc0 setter and user.
16587 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16589         * config/i386/predicates.md (local_symbolic_operand): Interpret
16590         dll-imported symbols as none-local.
16592 2013-03-25  Richard Biener  <rguenther@suse.de>
16594         * tree-ssa-loop-im.c (struct depend): Remove.
16595         (struct lim_aux_data): Make depends a vec of gimples.
16596         (free_lim_aux_data): Adjust.
16597         (add_dependency): Likewise.
16598         (set_level): Likewise.
16600 2013-03-25  Richard Biener  <rguenther@suse.de>
16602         PR middle-end/56434
16603         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
16604         the pointer returned by calls with ECF_MALLOC set.
16606 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16608         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
16610 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
16612         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
16613         using MMXMODE mode iterator.
16614         (*move<mode>_internal): Merge with *movv2sf_internal and
16615         *movv2sf_internal_rex64 using MMXMODE mode iterator.
16617 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16619         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
16620         (record_last_mem_set_info): Likewise.
16622         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
16623         of XNEWVEC followed by memset.
16624         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
16626 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
16628         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
16629         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
16630         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
16631         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
16632         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
16633         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
16634         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
16635         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
16636         BARRIER_P instead of GET_CODE.
16638 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16640         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
16641         inaccuracy in the probing code.
16643         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
16644         (ctrapdi4): Likewise.
16646 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16648         * calls.c (expand_call): Add missing guard to code handling return
16649         of non-BLKmode structures in MSB.
16650         * function.c (expand_function_end): Likewise.
16652 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
16654         * combine.c (try_combine): Adjust comment.  Do not add the set of
16655         insn #0 if the destination indirectly is set or dies in insn #2.
16656         Tidy up code to distribute a new note.
16658 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16660         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
16661         also for alternatives 16 and 17.
16663 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
16665         * config/i386/sse.md (*mov<mode>_internal): Merge with
16666         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
16667         Emit insn template depending on type attribute.  Use
16668         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16669         movd instead of movq mnemonic for interunit moves.  Rewrite mode
16670         attribute calculation.  Remove unit attribute calculation.
16671         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
16672         Set prefix_data16 attribute for DImode ssemov types.
16673         Use Ym instead of y for SSE-MMX conversion alternatives.
16674         Reorder operand constraints.
16676 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
16678         * df.h (df_insn_delete): Adjust prototype.
16679         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
16680         and let it decide whether mark the basic block dirty.
16681         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
16682         * df-scan.c (df_insn_info_delete): New helper function, split
16683         off from df_insn_delete.
16684         (df_scan_free_bb_info): Use it.
16685         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
16686         Likewise.
16687         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
16688         that the insn is actually an insn and it has a non-NULL basic block.
16689         Do not mark basic block dirty if only deleting a DEBUG_INSN.
16691 2013-03-22  Richard Biener  <rguenther@suse.de>
16693         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
16694         dep_ref members.
16695         (mem_ref_alloc): Do not allocate them.
16696         (refs_independent_p): Do not query or maintain a cache.
16698 2013-03-22  Richard Biener  <rguenther@suse.de>
16700         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
16701         (gather_mem_refs_in_loops): Do not compute it.
16702         (analyze_memory_references): Do not allocate it.
16703         (tree_ssa_lim_finalize): Do not free it.
16704         (for_all_locs_in_loop): Do not query all_refs_in_loop.
16706 2013-03-22  Richard Biener  <rguenther@suse.de>
16708         * is-a.h (as_a): Use gcc_checking_assert.
16710 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
16712         * config/aarch64/aarch64.c (aarch64_print_operand): New
16713         format specifier for printing a constant in hex.
16714         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
16715         format specifier for printing second operand.
16717 2013-03-22  Richard Biener  <rguenther@suse.de>
16719         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
16720         bitmaps.
16721         (gather_mem_refs_in_loops): Perform store accumulation here.
16722         (create_vop_ref_mapping_loop): Remove.
16723         (create_vop_ref_mapping): Likewise.
16724         (analyze_memory_references): Initialize refs_stored_in_loop.
16725         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
16726         (record_indep_loop): Remove.
16727         (record_dep_loop): New function.
16728         (ref_indep_loop_p_1): Adjust to only walk over references
16729         in the loop, not its subloops.
16730         (ref_indep_loop_p): Rename to ...
16731         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
16732         maintaining a more fine-grained cache.
16733         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
16734         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
16736 2013-03-22  Richard Biener  <rguenther@suse.de>
16738         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
16739         (struct mem_ref): Make accesses_in_loop a vec of a vec of
16740         aggregate mem_ref_loc.
16741         (free_mem_ref_locs): Inline into ...
16742         (memref_free): ... this and adjust.
16743         (mem_ref_alloc): Adjust.
16744         (mem_ref_locs_alloc): Remove.
16745         (record_mem_ref_loc): Adjust.
16746         (get_all_locs_in_loop): Rewrite into ...
16747         (for_all_locs_in_loop): ... this iterator.
16748         (rewrite_mem_ref_loc): New functor.
16749         (rewrite_mem_refs): Use for_all_locs_in_loop.
16750         (sm_set_flag_if_changed): New functor.
16751         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
16752         (ref_always_accessed): New functor.
16753         (ref_always_accessed_p): Use for_all_locs_in_loop.
16755 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16757         * tree-pass.h (PROP_gimple_lvec): New.
16758         * passes.c (dump_properties): Handle PROP_gimple_lvec.
16759         (init_optimization_passes): Move pass_lower_vector.
16760         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
16761         PROP_gimple_lvec.
16762         (pass_lower_vector): Provide PROP_gimple_lvec.
16763         (pass_lower_vector_ssa): Likewise.
16764         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
16766 2013-03-21  Mark Wielaard  <mjw@redhat.com>
16768         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
16770 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16772         * config/i386/i386.md (*movdi_internal): Disparage slightly
16773         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
16774         conversion alternatives.
16776 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
16778         PR middle-end/48087
16779         * diagnostic.def (DK_WERROR): New kind.
16780         * diagnostic.h (werrorcount): Define.
16781         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
16782         promoted to DK_ERROR, increment DK_WERROR counter instead of
16783         DK_ERROR counter.
16784         * toplev.c (toplev_main): Call print_ignored_options even if
16785         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
16786         even if just werrorcount is non-zero.
16788         PR debug/55608
16789         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
16790         on failure.
16791         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
16792         (string_cst_pool_decl): New function.
16793         (optimize_one_addr_into_implicit_ptr): New function.
16794         (resolve_addr_in_expr): Optimize DWARF location expression
16795         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
16796         which doesn't live in memory, but has DW_AT_location or
16797         DW_AT_const_value, or refers to a string literal, into
16798         DW_OP_GNU_implicit_pointer.
16799         (optimize_location_into_implicit_ptr): New function.
16800         (resolve_addr): If removing DW_AT_location of a variable because
16801         it was DW_OP_addr of address of the variable, but the variable doesn't
16802         live in memory, try to emit const value attribute for the initializer.
16804 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16806         * tree.h (VECTOR_TYPE_P): New macro.
16807         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
16808         TYPE_MODE): Use it.
16809         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
16810         VEC_COND_EXPR cannot be lvalues.
16811         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
16813 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
16815         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
16816         Restrict the transformation to equal modes.
16818 2013-03-21  Richard Biener  <rguenther@suse.de>
16820         PR tree-optimization/39326
16821         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
16822         (MEM_ANALYZABLE): Adjust.
16823         (record_mem_ref_loc): Move bitmap ops ...
16824         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
16825         unanalyzable refs, do not record locations for it.
16826         (analyze_memory_references): Allocate ref zero as shared
16827         unanalyzable ref.
16828         (refs_independent_p): Do not test for unanalyzed mems here.
16829         (ref_indep_loop_p_1): Special-case disambiguation against
16830         the unanalyzed ref.
16831         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
16833 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
16835         * config/arm/arm-protos.h (tune_params): Add
16836         prefer_neon_for_64bits field.
16837         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
16838         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
16839         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
16840         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
16841         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
16842         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
16843         (arm_option_override): Handle -mneon-for-64bits new option.
16844         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
16845         (prefer_neon_for_64bits): Declare new variable.
16846         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
16847         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
16848         (arch_enabled): Handle new arch types. Remove support for onlya8
16849         and nota8.
16850         (one_cmpldi2): Use new arch names.
16851         (zero_extend<mode>di2, extend<mode>di2): Ditto.
16852         * config/arm/arm.opt (mneon-for-64bits): Add option.
16853         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
16854         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
16855         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
16856         of onlya8.
16857         * doc/invoke.texi (-mneon-for-64bits): Document.
16859 2013-03-21  Richard Biener  <rguenther@suse.de>
16861         PR tree-optimization/39326
16862         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
16863         (sort_bbs_in_loop_postorder_cmp): New function.
16864         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
16866 2013-03-21  Richard Biener  <rguenther@suse.de>
16868         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
16869         (vect_insert_into_interleaving_chain): Likewise.
16870         (vect_drs_dependent_in_basic_block): Inline ...
16871         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
16872         split out from ...
16873         (vect_analyze_data_ref_dependence): ... here.  Simplify.
16874         (vect_check_interleaving): Simplify.
16875         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
16876         (vect_slp_analyze_data_ref_dependences): ... this new function.
16877         (dr_group_sort_cmp): New function.
16878         (vect_analyze_data_ref_accesses): Compute data-reference groups
16879         here instead of in vect_analyze_data_ref_dependence.  Use
16880         a more efficient algorithm.
16881         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
16882         vect_slp_analyze_data_ref_dependences.  Call
16883         vect_analyze_data_ref_accesses earlier.
16884         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
16885         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
16886         (vect_slp_analyze_data_ref_dependences): New prototype.
16888 2013-03-21  Richard Biener  <rguenther@suse.de>
16890         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
16891         ref is stored in the loop.
16892         (find_refs_for_sm): Walk only over all stores.
16893         (store_motion_loop): Allocate from lim_bitmap_obstack.
16894         (store_motion): Likewise.
16896 2013-03-21  Richard Biener  <rguenther@suse.de>
16898         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
16899         Update virtual SSA form.
16901 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16903         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
16904         * configure: Regenerate.
16905         * config.in: Regenerate.
16906         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
16907         if !HAVE_LD_EH_FRAME_CIEV3.
16909 2013-03-21  Richard Biener  <rguenther@suse.de>
16911         * tree-cfg.c (verify_expr_no_block): New function.
16912         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
16913         nor DECL_VALUE_EXPR have locations with associated blocks.
16914         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
16915         (clear_unused_block_pointer): Remove code dealing with
16916         blocks in DECL_DEBUG_EXPR locations.
16918 2013-03-21  Richard Biener  <rguenther@suse.de>
16920         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
16921         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
16922         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
16923         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
16924         instead of DECL_DEBUG_EXPR_IS_FROM.
16925         * gimplify.c (gimplify_modify_expr): Likewise.
16926         * tree-cfg.c (verify_expr_location_1): Likewise.
16927         * tree-complex.c (create_one_component_var): Likewise.
16928         * tree-sra.c (create_access_replacement): Likewise.
16929         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
16930         (clear_unused_block_pointer): Likewise.
16931         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
16932         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
16933         * var-tracking.c (var_debug_decl): Likewise.
16934         (track_expr_p): Likewise.
16935         * tree-inline.c (add_local_variables): Likewise.  Set
16936         DECL_HAS_DEBUG_EXPR_P after copying it.
16937         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
16938         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
16940 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
16942         PR bootstrap/56656
16943         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
16944         * configure: Regenerate.
16945         * config.in: Regenerate.
16946         * config/i386/i386.md (*movdf_internal): Use
16947         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
16948         movd instead of movq mnemonic for interunit moves.
16949         (*movdi_internal): Ditto.
16951 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16953         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
16954         (abd<mode>_3): New pattern.
16955         (aba<mode>_3): New pattern.
16956         (fabd<mode>_3): New pattern.
16958 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16960         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
16961         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
16962         occurrence of REGISTER_PREFIX as its empty string.
16964 2013-03-20  Jeff Law  <law@redhat.com>
16966         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
16967         addititional equivalences for equality comparisons between an SSA_NAME
16968         and a constant where the SSA_NAME was set from a widening conversion.
16970 2013-03-20  Walter Lee  <walt@tilera.com>
16972         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
16974 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16976         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
16977         depending on type attribute.
16978         (*movti_internal): Ditto.
16979         (*movtf_internal): Ditto.
16980         (*movxf_internal): Ditto.
16981         (*movdf_internal): Ditto.
16982         (*movsf_internal): Ditto.
16984 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16986         * config/i386/i386.md (*movti_internal): Set prefix attribute to
16987         maybe_vex for sselog1 and ssemov types.
16988         (*movdi_internal): Reorder operand constraints.
16989         (*movsi_internal): Ditto.  Set prefix attribute to
16990         maybe_vex for sselog1 and ssemov types.
16991         (*movtf_internal): Set prefix attribute to maybe_vex
16992         for sselog1 and ssemov types.
16993         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
16994         DImode ssemov types.  Reorder operand constraints.
16995         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
16996         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
16997         attribute for SImode ssemov types.  Reorder operand constraints.
16999 2013-03-20  Martin Jambor  <mjambor@suse.cz>
17001         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
17002         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
17004 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
17006         * config/rs6000/predicates.md (indexed_address, update_address_mem
17007         update_indexed_address_mem): New predicates.
17008         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
17009         attribute for load/store instructions.
17010         * config/rs6000/dfp.md (movsd_store): Likewise.
17011         (movsd_load): Likewise.
17012         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
17013         (unnamed HI->DI extend define_insn): Likewise.
17014         (unnamed SI->DI extend define_insn): Likewise.
17015         (unnamed QI->SI extend define_insn): Likewise.
17016         (unnamed QI->HI extend define_insn): Likewise.
17017         (unnamed HI->SI extend define_insn): Likewise.
17018         (unnamed HI->SI extend define_insn): Likewise.
17019         (extendsfdf2_fpr): Likewise.
17020         (movsi_internal1): Likewise.
17021         (movsi_internal1_single): Likewise.
17022         (movhi_internal): Likewise.
17023         (movqi_internal): Likewise.
17024         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17025         attribute for load/store instructions.
17026         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17027         instructions.
17028         (mov<mode>_softfloat): Likewise.
17029         (mov<mode>_hardfloat32): Likewise.
17030         (mov<mode>_hardfloat64): Likewise.
17031         (mov<mode>_softfloat64): Likewise.
17032         (movdi_internal32): Likewise.
17033         (movdi_internal64): Likewise.
17034         (probe_stack_<mode>): Likewise.
17036 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
17038         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17039         floating point, and decimal floating point to reload iterator.
17041         * config/rs6000/constraints.md (wl constraint): New constraints to
17042         return FLOAT_REGS if certain options are used to reduce the number
17043         of separate patterns that exist in the file.
17044         (wx constraint): Likewise.
17045         (wz constraint): Likewise.
17047         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17048         -mdebug=reg, print wg, wl, wx, and wz constraints.
17049         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17050         Initialize the reload functions for 64-bit binary/decimal floating
17051         point types.
17052         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17053         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17054         create the buffer on the stack to overcome not having a 32-bit
17055         load and store.
17056         (rs6000_emit_move): Likewise.
17057         (rs6000_secondary_memory_needed_rtx): Likewise.
17058         (rs6000_alloc_sdmode_stack_slot): Likewise.
17059         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17060         via xxlxor, just like DFmode 0.0.
17062         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17063         define as 1 if we are running on a power7 or newer.
17064         (enum r6000_reg_class_enum): Add new constraints.
17066         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17067         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17068         with other moves by using conditional constraits (wg).  Use LFIWZX
17069         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17070         (movsd splitter): Likewise.
17071         (movsd_hardfloat): Likewise.
17072         (movsd_softfloat): Likewise.
17074         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17075         binary and decimal floating point moves.
17076         (fmove_ok): New attributes to combine binary and decimal floating
17077         point moves, and to combine power6x (mfpgpr) moves along normal
17078         floating moves.
17079         (real_value_to_target): Likewise.
17080         (f32_lr): Likewise.
17081         (f32_lm): Likewise.
17082         (f32_li): Likewise.
17083         (f32_sr): Likewise.
17084         (f32_sm): Likewise.
17085         (f32_si): Likewise.
17086         (movsf): Combine binary and decimal floating point moves.  Combine
17087         power6x (mfpgpr) moves with other moves by using conditional
17088         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17089         (mov<mode> for SFmode/SDmode); Likewise.
17090         (SFmode/SDmode splitters): Likewise.
17091         (movsf_hardfloat): Likewise.
17092         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17093         (movsf_softfloat): Likewise.
17094         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17096         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17097         wx and wz constraints.
17099         * config/rs6000/constraints.md (wg constraint): New constraint to
17100         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17102         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17103         constraint.
17105         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17106         -mdebug=reg, print wg, wl, wx, and wz constraints.
17107         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17108         Initialize the reload functions for 64-bit binary/decimal floating
17109         point types.
17110         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17111         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17112         create the buffer on the stack to overcome not having a 32-bit
17113         load and store.
17114         (rs6000_emit_move): Likewise.
17115         (rs6000_secondary_memory_needed_rtx): Likewise.
17116         (rs6000_alloc_sdmode_stack_slot): Likewise.
17117         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17118         via xxlxor, just like DFmode 0.0.
17120         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17121         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17122         with other moves by using conditional constraits (wg).  Use LFIWZX
17123         and STFIWX for loading SDmode on power7.
17124         (movdd splitters): Likewise.
17125         (movdd_hardfloat32): Likewise.
17126         (movdd_softfloat32): Likewise.
17127         (movdd_hardfloat64_mfpgpr): Likewise.
17128         (movdd_hardfloat64): Likewise.
17129         (movdd_softfloat64): Likewise.
17131         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17132         64-bit binary and decimal floating point moves.
17133         (FMOVE64X): Likewise.
17134         (movdf): Combine 64-bit binary and decimal floating point moves.
17135         Combine power6x (mfpgpr) moves with other moves by using
17136         conditional constraits (wg).
17137         (mov<mode> for DFmode/DDmode): Likewise.
17138         (DFmode/DDmode splitters): Likewise.
17139         (movdf_hardfloat32): Likewise.
17140         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17141         (movdf_softfloat32): Likewise.
17142         (movdf_hardfloat64_mfpgpr): Likewise.
17143         (movdf_hardfloat64): Likewise.
17144         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17145         (movdf_softfloat64): Likewise.
17146         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17147         (reload_<mode>_load): Move to later in the file so they aren't in
17148         the middle of the floating point move insns.
17149         (reload_<mode>_store): Likewise.
17151         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17152         constraint.
17154         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17155         constraint if -mdebug=reg.
17156         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17157         Enable using dd reload support if needed.
17159         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17160         binary and decimal floating point moves in rs6000.md.
17161         (movtd_internal): Likewise.
17163         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17164         decimal floating point moves.
17165         (movtf): Likewise.
17166         (movtf_internal): Likewise.
17167         (mov<mode>_internal, TDmode/TFmode): Likewise.
17168         (movtf_softfloat): Likewise.
17169         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17171         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17172         movdi_internal64, using wg constraint for move direct operations.
17173         (movdi_internal64): Likewise.
17175         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17176         MODES_TIEABLE_P for selected modes.  Print the numerical value of
17177         the various virtual registers. Use GPR/FPR first/last values,
17178         instead of hard coding the register numbers.  Print which modes
17179         have reload functions registered.
17180         (rs6000_option_override_internal): If -mdebug=reg, trace the options
17181         settings before/after setting cpu, target and subtarget settings.
17182         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17183         and for secondary reload failures in rs6000_secondary_reload_inner.
17184         (rs6000_secondary_reload_fail): Likewise.
17185         (rs6000_secondary_reload_inner): Likewise.
17187         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17188         macros for first/last GPR and FPR registers.
17189         (LAST_GPR_REGNO): Likewise.
17190         (FIRST_FPR_REGNO): Likewise.
17191         (LAST_FPR_REGNO): Likewise.
17193         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17194         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17195         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17196         (vcond<mode><mode>): Likewise.
17197         (vcondu<mode><mode>): Likewise.
17198         (vector_gtu<mode>): Likewise.
17199         (vector_gte<mode>): Likewise.
17200         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17201         to prevent the compiler from converting DImode operations to TImode.
17202         (ior<mode>3): Likewise.
17203         (and<mode>3): Likewise.
17204         (one_cmpl<mode>2): Likewise.
17205         (nor<mode>3): Likewise.
17206         (andc<mode>3): Likewise.
17208         * config/rs6000/constraints.md (wt constraint): New constraint
17209         that returns VSX_REGS if TImode is allowed in VSX registers.
17211         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17212         constant under VSX.
17214         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17215         similar to TImode, but it is restricted to being in the GPRs.
17217         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17218         TImode to occupy a single VSX register.
17220         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17221         -mvsx-timode for power7/power8.
17222         (power7 cpu): Likewise.
17223         (power8 cpu): Likewise.
17225         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17226         sure that TFmode/TDmode take up two registers if they are ever
17227         allowed in the upper VSX registers.
17228         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17229         registers.
17230         (rs6000_init_hard_regno_mode_ok): Likewise.
17231         (rs6000_debug_reg_global): Add debugging for PTImode and wt
17232         constraint.  Print if LRA is turned on.
17233         (rs6000_option_override_internal): Give an error if -mvsx-timode
17234         and VSX is not enabled.
17235         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
17236         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
17237         to reg+offset addressing.  Use PTImode when checking offset
17238         addresses for validity.
17239         (reg_offset_addressing_ok_p): Likewise.
17240         (rs6000_legitimate_offset_address_p): Likewise.
17241         (rs6000_legitimize_address): Likewise.
17242         (rs6000_legitimize_reload_address): Likewise.
17243         (rs6000_legitimate_address_p): Likewise.
17244         (rs6000_eliminate_indexed_memrefs): Likewise.
17245         (rs6000_emit_move): Likewise.
17246         (rs6000_secondary_reload): Likewise.
17247         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
17248         reloads to fpr registers to continue to use reg+offset addressing,
17249         but 64-bit reloads to altivec registers need reg+reg addressing.
17250         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
17251         it.  Treat LO_SUM like a PLUS operation.
17252         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
17253         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
17254         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
17255         registers to share a register with a smaller sized type, since VSX
17256         puts scalars in the upper 64-bits.
17257         (print_operand): Add support for PTImode.
17258         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
17259         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
17260         registers, but don't have arithmetic support.
17261         (rs6000_memory_move_cost): Add test for VSX.
17262         (rs6000_opt_masks): Add -mvsx-timode.
17264         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
17265         for TImode.
17266         (VSs): Likewise.
17267         (VSr): Use wt constraint for TImode.
17268         (VSv): Drop TImode support.
17269         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
17270         (vsx_movti_64bit): Likewise.
17271         (vsx_movti_32bit): Likewise.
17272         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
17273         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
17274         one '?' on the appropriate output constraint.  Do not allow TImode
17275         logical operations on 32-bit systems.
17276         (vsx_ior<mode>3): Likewise.
17277         (vsx_xor<mode>3): Likewise.
17278         (vsx_one_cmpl<mode>2): Likewise.
17279         (vsx_nor<mode>3): Likewise.
17280         (vsx_andc<mode>3): Likewise.
17281         (vsx_concat_<mode>): Likewise.
17282         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
17284         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
17285         OPTION_MASK_VSX_TIMODE.
17286         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
17287         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
17289         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
17290         (TI2 iterator): New iterator for TImode, PTImode.
17291         (wd mode attribute): Add values for vector types.
17292         (movti_string): Replace TI move operations with operations for TImode
17293         and PTImode.  Add support for TImode being allowed in VSX registers.
17294         (mov<mode>_string, TImode/PTImode): Likewise.
17295         (movti_ppc64): Likewise.
17296         (mov<mode>_ppc64, TImode/PTImode): Likewise.
17297         (TI mode splitters): Likewise.
17299         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
17300         constraint.
17302 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
17304         PR tree-optimization/56355
17305         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
17306         Also handle integers with undefined overflow.
17308 2013-03-20  Catherine Moore  <clm@codesourcery.com>
17309             Maciej W. Rozycki  <macro@codesourcery.com>
17310             Tom de Vries  <tom@codesourcery.com>
17311             Nathan Sidwell  <nathan@codesourcery.com>
17312             Iain Sandoe  <iain@codesourcery.com>
17313             Nathan Froyd  <froydnj@codesourcery.com>
17314             Chao-ying Fu  <fu@mips.com>
17316         * doc/extend.texi: (micromips, nomicromips, nocompression):
17317         Document new function attributes.
17318         * doc/invoke.texi (minterlink-compressed, mmicromips,
17319         m14k, m14ke, m14kec): Document new options.
17320         (minterlink-mips16): Update documentation.
17321         * doc/md.texi (ZC, ZD): Document new constraints.
17322         * configure.ac (gcc_cv_as_micromips): Check if linker
17323         supports the .set micromips directive.
17324         * configure: Regenerate.
17325         * config.in: Regenerate.
17326         * config/mips/mips-tables.opt: Regenerate.
17327         * config/mips/micromips.md: New file.
17328         * constraints.md (ZC, ZD): New constraints.
17329         * config/mips/predicates.md (movep_src_register): New predicate.
17330         (movep_src_operand): New predicate.
17331         (non_volatile_mem_operand): New predicate.
17332         * config/mips/mips.md (multimem): New type.
17333         (length): Differentiate between 17-bit and 18-bit branch offsets.
17334         (MOVEP1, MOVEP2): New mode iterator.
17335         (mov_<load>l): Use ZC constraint.
17336         (mov_<load>r): Likewise.
17337         (mov_<store>l): Likewise.
17338         (mov_<store>r): Likewise.
17339         (*branch_equality<mode>_inverted): Add microMIPS support.
17340         (*branch_equality<mode>): Likewise.
17341         (*jump_absolute): Likewise.
17342         (indirect_jump_<mode>): Likewise.
17343         (tablejump_<mode>): Likewise.
17344         (<optab>_internal): Likewise.
17345         (sibcall_internal): Likewise.
17346         (sibcall_value_internal): Likewise.
17347         (prefetch): Use constraint ZD.
17348         * config/mips/mips.opt (minterlink-compressed): New option.
17349         (minterlink-mips16): Now an alias for minterlink-compressed.
17350         (mmicromips): New option.
17351         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17352         (compare_and_swap_12): Likewise.
17353         (sync_add<mode>): Likewise.
17354         (sync_<optab>_12): Likewise.
17355         (sync_old_<optab>_12): Likewise.
17356         (sync_new_<optab>_12): Likewise.
17357         (sync_nand_12): Likewise.
17358         (sync_old_nand_12): Likewise.
17359         (sync_new_nand_12): Likewise.
17360         (sync_sub<mode>): Likewise.
17361         (sync_old_add<mode>): Likewise.
17362         (sync_old_sub<mode>): Likewise.
17363         (sync_new_add<mode>): Likewise.
17364         (sync_new_sub<mode>): Likewise.
17365         (sync_<optab><mode>): Likewise.
17366         (sync_old_<optab><mode>): Likewise.
17367         (sync_new_<optab><mode>): Likewise.
17368         (sync_nand<mode>): Likewise.
17369         (sync_old_nand<mode>): Likewise.
17370         (sync_new_nand<mode>): Likewise.
17371         (sync_lock_test_and_set<mode>): Likewise.
17372         (test_and_set_12): Likewise.
17373         (atomic_compare_and_swap<mode>): Likewise.
17374         (atomic_exchange<mode>_llsc): Likewise.
17375         (atomic_fetch_add<mode>_llsc): Likewise.
17376         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17377         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17378         (umips_save_restore_pattern_p): Likewise.
17379         (umips_load_store_pair_p): Likewise.
17380         (umips_output_load_store_pair): Likewise.
17381         (umips_movep_target_p): Likewise.
17382         (umips_12bit_offset_address_p): Likewise.
17383         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17384         (mips_base_mips16): Rename this...
17385         (mips_base_compression_flags): ...to this. Update all uses.
17386         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17387         (mips_mips16_decl_p): Delete.
17388         (mips_nomips16_decl_p): Delete.
17389         (mips_get_compress_on_flags): New function.
17390         (mips_get_compress_off_flags): New function.
17391         (mips_get_compress_mode): New function.
17392         (mips_get_compress_on_name): New function.
17393         (mips_get_compress_off_name): New function.
17394         (mips_insert_attributes): Support multiple compression types.
17395         (mips_merge_decl_attributes): Likewise.
17396         (umips_12bit_offset_address_p): New function.
17397         (mips_start_function_definition): Emit .set micromips directive.
17398         (mips_call_may_need_jalx_p): New function.
17399         (mips_function_ok_for_sibcall): Add microMIPS support.
17400         (mips_print_operand_punctuation): Support short delay slots and
17401         compact jumps.
17402         (umips_swm_mask, umips_swm_encoding): New.
17403         (umips_build_save_restore): New function.
17404         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17405         (was_mips16_p): Remove.
17406         (old_compression_mode): New.
17407         (mips_set_compression_mode): New function.
17408         (mips_set_current_function): Add microMIPS support.
17409         (mips_option_override): Likewise.
17410         (umips_save_restore_pattern_p): New function.
17411         (umips_output_save_restore): New function.
17412         (umips_load_store_pair_p_1): New function.
17413         (umips_load_store_pair_p): New function.
17414         (umips_output_load_store_pair_1): New function.
17415         (umips_output_load_store_pair): New function.
17416         (umips_movep_target_p) New function.
17417         (mips_prepare_pch_save): Add microMIPS support.
17418         * config/mips/mips.h (TARGET_COMPRESSION): New.
17419         (TARGET_CPU_CPP_BUILTINS): Update macro
17420         to use new compression flags and to support microMIPS.
17421         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17422         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17423         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17424         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17425         (ASM_SPEC): Support mmicromips and mno-micromips.
17426         (M16STORE_REG_P): New macro.
17427         (MIPS_CALL): Support TARGET_MICROMIPS.
17428         (MICROMIPS_J): New macro.
17429         (mips_base_mips16): Rename this...
17430         (mips_base_compression_flags): ...to this.
17431         (UMIPS_12BIT_OFFSET_P): New macro.
17432         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17433         (MULTILIB_DIRNAMES): Likewise.
17434 2013-03-20  Richard Biener  <rguenther@suse.de>
17436         PR tree-optimization/56661
17437         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
17438         the result does not have to be distinct.
17440 2013-03-20  Richard Biener  <rguenther@suse.de>
17442         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
17443         remap_gimple_op_r.
17445 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17446             Steven Bosscher  <steven@gcc.gnu.org>
17448         PR rtl-optimization/56605
17449         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
17451 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17453         PR bootstrap/56656
17454         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
17455         that require movd instead of movq.
17457 2013-03-20  Richard Biener  <rguenther@suse.de>
17459         * tree-ssa-structalias.c (struct variable_info): Add pointer
17460         to the first field of an aggregate with sub-vars.  Make
17461         this and the pointer to the next subfield its ID.
17462         (vi_next): New function.
17463         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
17464         storedanything_id, integer_id): Increment by one.
17465         (new_var_info, get_call_vi, lookup_call_clobber_vi,
17466         get_call_clobber_vi): Adjust.
17467         (solution_set_expand): Simplify and speedup.
17468         (solution_set_add): Inline into ...
17469         (set_union_with_increment): ... this.  Adjust accordingly.
17470         (do_sd_constraint): Likewise.
17471         (do_ds_constraint): Likewise.
17472         (do_complex_constraint): Simplify.
17473         (build_pred_graph): Adjust.
17474         (solve_graph): Likewise.  Simplify and speedup.
17475         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
17476         get_constraint_for_component_ref, get_constraint_for_1,
17477         first_vi_for_offset, first_or_preceding_vi_for_offset,
17478         create_function_info_for, create_variable_info_for_1,
17479         create_variable_info_for, intra_create_variable_infos): Adjust.
17480         (init_base_vars): Push NULL for ID zero.
17481         (compute_points_to_sets): Adjust.
17483 2013-03-20  Richard Biener  <rguenther@suse.de>
17485         * cfgloop.c (verify_loop_structure): Streamline and avoid
17486         ICEing on corrupt loop tree.
17487         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
17488         loop tree.
17490 2013-03-20  Richard Biener  <rguenther@suse.de>
17492         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
17493         check whether an SSA update is needed.
17495 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
17497         * config/mips/constraints.md (T): Rename to...
17498         (Yf): ...this.
17499         (U): Rename to...
17500         (Yd): ...this.
17501         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
17502         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
17504 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17506         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
17507         (*subsi3_carryin_uxtw): Likewise.
17509 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17511         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
17512         (*rorsi3_insn_uxtw): Likewise.
17514 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17516         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
17517         (*extrsi5_insn_uxtw): Likewise.
17519 2013-03-19  Richard Biener  <rguenther@suse.de>
17521         PR tree-optimization/56273
17522         * passes.c (init_optimization_passes): Move second VRP after DOM.
17524 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
17526         * config/i386/i386.md (*movti_internal): Merge from
17527         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
17528         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
17529         nox64 isa attributes.
17531 2013-03-18  Richard Biener  <rguenther@suse.de>
17533         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
17534         (unite): Likewise.
17535         (merge_node_constraints): Likewise.
17536         (build_succ_graph): Likewise.
17537         (valid_graph_edge): Inline into single caller.
17538         (unify_nodes): Likewise.  Use bitmap_set_bit return value
17539         and cache varinfo.
17540         (scc_visit): Fix formatting and variable use.
17541         (do_sd_constraint): Use gcc_checking_assert.
17542         (do_ds_constraint): Likewise.
17543         (do_complex_constraint): Likewise.
17544         (condense_visit): Likewise.  Cleanup.
17545         (dump_pred_graph): New function.
17546         (perform_var_substitution): Dump the pred-graph before
17547         variable substitution.
17548         (find_equivalent_node): Use gcc_checking_assert.
17549         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
17551 2013-03-18  Richard Biener  <rguenther@suse.de>
17553         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
17554         Remove cond_expr_stmt_list argument and do not gimplify the
17555         built expression.
17556         (vect_loop_versioning): Adjust.
17557         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
17558         Cleanup to use less temporaries.
17559         (vect_create_data_ref_ptr): Cleanup.
17561 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17563         PR tree-optimization/56635
17564         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
17565         require types_compatible_p types.
17567 2013-03-18  Nick Clifton  <nickc@redhat.com>
17569         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
17570         spurious backslash.
17572         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
17573         Add missing line to comment describing function.
17575 2013-03-18  Richard Biener  <rguenther@suse.de>
17577         PR tree-optimization/56210
17578         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17579         Handle string / character search functions.
17580         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
17582 2013-03-18  Richard Biener  <rguenther@suse.de>
17584         PR middle-end/56483
17585         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
17586         and implement properly.
17587         * gimple.h (gimple_cond_single_var_p): Remove.
17589 2013-03-18  Richard Biener  <rguenther@suse.de>
17591         * tree-data-ref.h (find_data_references_in_loop): Declare.
17592         * tree-data-ref.c (get_references_in_stmt): Use a stack
17593         vector pre-allocated in the callers.
17594         (find_data_references_in_stmt): Adjust.
17595         (graphite_find_data_references_in_stmt): Likewise.
17596         (create_rdg_vertices): Likewise.
17597         (find_data_references_in_loop): Export.
17598         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17599         Compute dependences here...
17600         (vect_analyze_data_refs): ...not here.  When we encounter
17601         a non-vectorizable data reference in basic-block vectorization
17602         truncate the data reference vector.  Do not bother to
17603         fixup data-dependence information for gather loads.
17604         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
17605         of data references, as reported.
17607 2013-03-18  Richard Biener  <rguenther@suse.de>
17609         PR tree-optimization/3713
17610         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
17611         has_constants and expr.
17612         (stmt_has_constants): Properly valueize SSA names when deciding
17613         whether the stmt has constants.
17615 2013-03-18  Richard Biener  <rguenther@suse.de>
17617         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
17618         whole function when there is nothing to do.
17619         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
17620         * tree-vectorizer.c (vectorize_loops): Update virtual and
17621         loop-closed SSA once.
17622         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
17624 2013-03-18  Richard Biener  <rguenther@suse.de>
17626         PR middle-end/56113
17627         * domwalk.c (bb_postorder): New global static.
17628         (cmp_bb_postorder): New function.
17629         (walk_dominator_tree): Replace scheme imposing an order for
17630         visiting dominator sons by one sorting them at the time they
17631         are pushed on the stack.
17633 2013-03-18  Richard Biener  <rguenther@suse.de>
17635         PR tree-optimization/39326
17636         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
17637         (struct mem_ref): Replace mem member with ao_ref typed member.
17638         (MEM_ANALYZABLE): Adjust.
17639         (memref_eq): Likewise.
17640         (mem_ref_alloc): Likewise.
17641         (gather_mem_refs_stmt): Likewise.
17642         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
17643         (execute_sm_if_changed_flag_set): Adjust.
17644         (execute_sm): Likewise.
17645         (ref_always_accessed_p): Likewise.
17646         (refs_independent_p): Likewise.
17647         (can_sm_ref_p): Likewise.
17649 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17651         PR c/56566
17652         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
17653         return 1 even for !unsignedp.
17655 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17657         * config/i386/i386.md (isa): Add x64 and nox64.
17658         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
17659         (*pushtf): Enable *roF alternative for x64 isa only.
17660         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
17661         mode attribute of integer alternatives to DImode for TARGET_64BIT.
17662         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
17663         (*movtf_internal): Merge from *movtf_internal_rex64 and
17664         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
17665         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
17666         nox64 isa attributes.
17667         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
17668         nox64 isa attributes.
17669         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
17671 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
17673         * config/alpha/alpha.c (TARGET_LRA_P): New define.
17675 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
17677         PR target/56640
17678         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
17679         class names.  Remove trailing comma after "ALL_REGS".
17681 2013-03-16  Jan Hubicka  <jh@suse.cz>
17683         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
17684         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
17685         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
17686         of cgraph_get_create_node.
17687         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17689 2013-03-16  Jason Merrill  <jason@redhat.com>
17691         PR debug/49090
17692         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
17693         with DW_AT_default_value.
17695 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
17697         * BASE-VER: Set to 4.9.0.
17699 2013-03-14  Andi Kleen  <ak@linux.intel.com>
17701         PR target/56619
17702         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
17703         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
17704         Document _x* TSX intrinsics.
17706 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
17707             David Holsgrove  <david.holsgrove@xilinx.com>
17709         * configure.ac: Add MicroBlaze TLS support detection.
17710         * configure: Regenerate.
17711         * config/microblaze/microblaze-protos.h
17712         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
17713         symbol_mentioned_p, label_mentioned_p): Add prototypes.
17714         * config/microblaze/microblaze.c (microblaze_address_type): Add
17715         ADDRESS_TLS and tls_reloc address types.
17716         (microblaze_address_info): Add tls_reloc.
17717         (TARGET_HAVE_TLS): Define.
17718         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
17719          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
17720          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
17721         load_tls_operand,  microblaze_call_tls_get_addr,
17722         microblaze_legitimize_tls_address): New functions.
17723         (microblaze_classify_unspec): Handle UNSPEC_TLS.
17724         (get_base_reg): Use microblaze_tls_symbol_p.
17725         (microblaze_classify_address): Handle TLS.
17726         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
17727         label_mentioned_p and microblaze_tls_referenced_p.
17728         (microblaze_legitimize_address): Handle TLS.
17729         (microblaze_address_insns): Handle ADDRESS_TLS.
17730         (pic_address_needs_scratch): Handle TLS.
17731         (print_operand_address): Handle TLS.
17732         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
17733         (microblaze_expand_move): Handle TLS.
17734         (microblaze_legitimate_constant_p): Check
17735         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
17736         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
17737         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
17738         (PIC_OFFSET_TABLE_REGNUM): Set.
17739         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
17740         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
17741         (addsi3, movsi_internal2, movdf_internal): Update constraints
17742         * config/microblaze/predicates.md (arith_plus_operand): Define
17743         (move_operand): Redefine as move_src_operand,
17744         check microblaze_tls_referenced_p.
17746 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17748         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
17749         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
17751 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
17753         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
17754         CC mode for AND.
17756 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
17758         PR tree-optimization/53265
17759         * common.opt (Waggressive-loop-optimizations): New option.
17760         * tree-ssa-loop-niter.c: Include tree-pass.h.
17761         (do_warn_aggressive_loop_optimizations): New function.
17762         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
17763         if number_of_latch_executions returned constant.
17764         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
17765         early.  If number_of_latch_executions returned constant, set
17766         nb_iterations_upper_bound back to it.
17767         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
17768         field.
17769         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
17770         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
17772         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
17773         (MULTILIB_OSDIRNAMES): Set.
17774         * genmultilib: If defaultosdirname doesn't start with :: , set
17775         defaultosdirname2 instead, clear it and emit two . multilib_raw
17776         entries instead of just one.
17778 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
17780         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17781         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
17782         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
17783         (SUBTARGET_OVERRIDE_OPTIONS): New.
17785 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
17787         PR target/49880
17788         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
17789         (musermode): Convert to Var(TARGET_USERMODE).
17790         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
17791         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
17792         * config/sh/sh.c (sh_option_override): Use
17793         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
17794         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
17795         condition.
17796         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
17797         TARGET_SH4.
17798         (udivsi3_i4_single, divsi3_i4_single): Use
17799         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
17801 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
17803         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
17804         default setting.
17806 2013-03-13  Richard Biener  <rguenther@suse.de>
17808         PR tree-optimization/56608
17809         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
17810         calls when vectorizing basic-blocks.
17812 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
17814         PR plugins/45078
17815         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
17816         tm_file.
17818 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
17820         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
17822 2013-03-11  Jan Hubicka  <jh@suse.cz>
17824         PR lto/56557
17825         * lto-streamer-out.c (output_symbol_p): Skip references from
17826         constructors of external variables.
17828 2013-03-11  Jan Hubicka  <jh@suse.cz>
17830         PR middle-end/56571
17831         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
17832         from pseudos.
17833         * emit-rtl.c (verify_rtx_sharing): Likewise.
17834         (copy_insn_1): Likewise.
17835         * rtl.c (copy_rtx): Likewise.
17837 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
17839         PR target/56591
17840         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
17841         output_operand_lossage message.
17843 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
17845         PR target/56470
17846         * arm.c (shift_op): Validate RTL pattern on the fly.
17847         (arm_print_operand, case 'S'): Don't use shift_operator to validate
17848         the RTL.
17850 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17852         PR target/56347
17853         * config/pa/pa.md (call_value): Check for calls to powf and direct to
17854         new call patterns that clobber %fr12.
17855         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
17856         split and postreload patterns.
17857         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
17858         registers %fr12 and %fr12R as call used.
17860 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
17862         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
17863         (canon_address, record_store, replace_read, check_mem_read_rtx,
17864         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
17865         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
17866         rest_of_handle_dse): Likewise.
17868 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
17870         PR middle-end/56524
17871         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
17872         Add base_optabs.
17873         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
17874         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
17875         (save_optabs_if_changed): Replace with...
17876         (init_tree_optimization_optabs): ...this.
17877         * optabs.c (save_optabs_if_changed): Rename to...
17878         (init_tree_optimization_optabs): ...this.  Take the optimization node
17879         as argument.  Do nothing if the base optabs are already correct.
17880         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
17881         to recompute optabs.
17882         * function.h (function): Remove optabs field.
17883         * function.c (invoke_set_current_function_hook): Call
17884         init_tree_optimization_optabs.  Use the result to initialize
17885         this_fn_optabs.
17887 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
17889         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
17890         if GTMA_HAS_NO_INSTRUMENTATION.
17891         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
17892         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
17893         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
17894         * gimple-pretty-print.c (dump_gimple_transaction): Handle
17895         GTMA_HAS_NO_INSTRUMENTATION.
17897 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
17899         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
17900         libasan_preinit.o.
17902 2013-03-08  Marek Polacek  <polacek@redhat.com>
17903             Jakub Jelinek  <jakub@redhat.com>
17905         PR tree-optimization/56478
17906         * predict.c (is_comparison_with_loop_invariant_p): Change the
17907         type of loop_step to tree.
17908         (predict_loops): Adjust.
17909         (predict_iv_comparison): Perform the computations on double_ints.
17911 2013-03-08  Richard Biener  <rguenther@suse.de>
17913         PR tree-optimization/56570
17914         * tree-cfg.c (verify_expr_location_1): Verify locations for
17915         DECL_DEBUG_EXPR.
17916         * tree-sra.c (create_access_replacement): Strip locations
17917         from DECL_DEBUG_EXPRs.
17919 2013-03-08  Richard Biener  <rguenther@suse.de>
17921         * tree-inline.c (expand_call_inline): Do not associate
17922         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
17923         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
17925 2013-03-08  Richard Biener  <rguenther@suse.de>
17927         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
17928         or block changes with -Og.  Fix for location / block encoding
17929         changes and PHI arguments with locations.
17931 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
17933         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
17934         for all counters.
17935         (struct output_info): Likewise.
17936         (register_overhead): Remove bad gcc_assert.
17937         (bitmap_find_bit): If there is only a single bitmap element, do not
17938         count a miss as a search.
17939         (print_statistics): Update for counter type changes.
17940         (dump_bitmap_statistics): Likewise.  Print headers such that they
17941         are properly lined up with the printed counters.
17943 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
17945         PR tree-optimization/56559
17946         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
17947         check that it has only a single use.
17949 2013-03-07  Richard Biener  <rguenther@suse.de>
17951         * doc/invoke.texi (fwhole-program): Discourage use in combination
17952         with -flto.
17954 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
17956         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
17958         PR tree-optimization/56539
17959         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
17960         instead of GSI_CONTINUE_LINKING as last argument to
17961         force_gimple_operand_gsi.  Adjust function comment.
17963         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
17964         aarch64-cores.def.
17966         PR middle-end/56548
17967         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
17968         promoted mode, convert the result back to the original mode.
17970 2013-03-06  Richard Biener  <rguenther@suse.de>
17972         PR middle-end/56294
17973         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
17974         (insert_updated_phi_nodes_compare_uids): New function.
17975         (update_ssa): Sort symbols_to_rename after UID before
17976         traversing it to insert PHI nodes.
17978 2013-03-06  Richard Biener  <rguenther@suse.de>
17980         PR middle-end/50494
17981         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
17982         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
17984         Revert
17985         2013-02-13  Richard Biener  <rguenther@suse.de>
17987         PR lto/50494
17988         * varasm.c (output_constant_def_1): Get the decl representing
17989         the constant as argument.
17990         (output_constant_def): Wrap output_constant_def_1.
17991         (make_decl_rtl): Use output_constant_def_1 with the decl
17992         representing the constant.
17993         (build_constant_desc): Optionally re-use a decl already
17994         representing the constant.
17995         (tree_output_constant_def): Adjust.
17997 2013-03-06  Joey Ye  <joey.ye@arm.com>
17999         PR lto/50293
18000         * gcc.c (convert_white_space): New function.
18001         (main): Handles white space in function name.
18003 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
18005         PR target/56529
18006         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
18007         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
18008         to SH_DIV_CALL_TABLE for TARGET_SH2.
18009         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
18010         list.
18011         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
18012         call-table options.
18014 2013-03-05  Sterling Augustine  <saugustine@google.com>
18015             Cary Coutant  <ccoutant@google.com>
18017         PR debug/55364
18018         * dwarf2out.c (resolve_addr): Don't call
18019         remove_loc_list_addr_table_entries a second time for the same
18020         expression.
18022 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18024         PR debug/56510
18025         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18026         (avoid_complex_debug_insns): New function.
18027         (expand_debug_locations): Call it.
18029         PR rtl-optimization/56484
18030         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18031         lifetimes of hard registers on small register class machines.
18033 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18035         * config/microblaze/microblaze-protos.h: Rename
18036         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18037         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18038         fast_interrupt.
18039         (microblaze_fast_interrupt_function_p): New function.
18040         (microblaze_is_interrupt_handler): Rename to
18041         microblaze_is_interrupt_variant and add fast_interrupt check.
18042         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18043         (save_restore_insns): Likewise.
18044         (compute_frame_size): Likewise.
18045         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18046         (microblaze_globalize_label): Likewise.
18047         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18048         * config/microblaze/microblaze.md: Use wrapper
18049         microblaze_is_interrupt_variant.
18051 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18053         * sdbout.c (sdbout_one_type): Switch to current function's section
18054         supporting cold/hot.
18056 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18058         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18059         -mxl-reorder.
18061 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18063         PR middle-end/56461
18064         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18065         if VALGRIND_GET_VBITS is defined, temporarily make object
18066         memory all defined, and restore previous valgrind addressability
18067         and definability afterwards.  Free this_object at the end.
18069         PR middle-end/56461
18070         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18071         right before calling lra_create_live_ranges, also call it
18072         when clearing live_p.  Only call lra_clear_live_ranges
18073         at the end if live_p.
18075         PR middle-end/56461
18076         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18078 2013-03-05  Richard Biener  <rguenther@suse.de>
18080         PR tree-optimization/56521
18081         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18082         value-id.
18084 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18086         PR c++/55135
18087         * except.h (remove_unreachable_eh_regions): New prototype.
18088         * except.c (remove_eh_handler_splicer): New function, split out
18089         of remove_eh_handler.
18090         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18091         warning about running it on many EH regions one at a time.
18092         (remove_unreachable_eh_regions_worker): New function, walk the
18093         EH tree in depth-first order and remove non-marked regions.
18094         (remove_unreachable_eh_regions): New function.
18095         * tree-eh.c (mark_reachable_handlers): New function, split out
18096         from remove_unreachable_handlers.
18097         (remove_unreachable_handlers): Use mark_reachable_handlers and
18098         remove_unreachable_eh_regions.
18099         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18100         and remove_unreachable_eh_regions.
18102 2013-03-05  Richard Biener  <rguenther@suse.de>
18104         PR middle-end/56525
18105         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18106         not freeing them until the end.
18108 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18110         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18112 2013-03-05  Richard Biener  <rguenther@suse.de>
18114         PR tree-optimization/56270
18115         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18116         of loads after scheduling an SLP instance.
18118 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18120         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18121         tic6x.exp.
18122         (check_gcc_parallelize): Run guality.exp as a separate job from
18123         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18124         struct-layout-1.exp with stackalign.exp.
18126         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18128         PR middle-end/56461
18129         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18130         load_index sbitmap even if some bit in it isn't set.
18132         PR middle-end/56461
18133         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18134         (discover_iteration_bound_by_body_walk): Change queues to
18135         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
18136         spelling in comment.  Call safe_push on queues[bound_index] directly.
18137         Release queues[queue_index] in every iteration unconditionally.
18138         Release bounds vector.
18140         PR middle-end/56461
18141         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18142         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18143         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18144         inner_phis vector.
18146 2013-03-05  Richard Biener  <rguenther@suse.de>
18148         PR lto/56515
18149         * tree-inline.c (remap_blocks_to_null): New function.
18150         (expand_call_inline): When expanding a call stmt without
18151         an associated block inline remap all callee blocks to NULL.
18153 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18155         PR rtl-optimization/56494
18156         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18157         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18158         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18160         PR middle-end/56461
18161         * sel-sched-ir.c (free_sched_pools): Release
18162         succs_info_pool.stack[succs_info_pool.max_top] vectors too
18163         if succs_info_pool.max_top isn't -1.
18165         PR bootstrap/56509
18166         * opts.c (opts_obstack, opts_concat): Moved to...
18167         * opts-common.c (opts_obstack, opts_concat): ... here.
18169 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18171         PR middle-end/56461
18172         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18174 2013-03-04  Martin Jambor  <mjambor@suse.cz>
18176         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18177         all appropriate places.
18179 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
18181         PR tree-optimization/56424
18182         * ipa-split.c (split_function): Do not set the RSO flag if result is
18183         not by reference and its type is a register type.
18185 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
18187         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18188         (microblaze_legitimate_pic_operand): Likewise
18189         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18190         new function microblaze_legitimate_pic_operand
18191         * config/microblaze/microblaze-protos.h
18192         (microblaze_legitimate_pic_operand): Declare.
18194 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18196         * config/microblaze/predicates.md (call_insn_simple_operand):
18197         New predicate for supported rtx code types.
18198         * config/microblaze/microblaze.md (call_internal1): Use
18199         call_insn_simple_operand predicate.
18201 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18203         PR middle-end/56461
18204         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18205         partitions.ordered_remove.
18207         PR middle-end/56461
18208         * tree-vect-stmts.c (vectorizable_conversion): Don't call
18209         vec_oprnds0.create (1) for modifier == NONE.
18211         PR middle-end/56461
18212         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18213         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18214         vec_oprnds1 right before pushing anything to it for
18215         scalar_shift_arg.
18217         PR middle-end/56461
18218         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18219         set nbbs to 0 instead of having separate code path.
18220         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18221         instead of false as last argument if returning NULL.
18223 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
18225         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18226         the attribute is now called "target" instead of "option".
18227         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18228         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
18229         attribute/pragma name for TARGET_OPTION_VALID_P and
18230         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
18231         * doc/tm.texi: Regenerated.
18233 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
18235         * config/microblaze/microblaze.c:
18236         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
18237         * config/microblaze/microblaze.h: Add -mxl-reorder to
18238         DRIVER_SELF_SPECS.
18239         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
18240         instructions emitted if TARGET_REORDER.
18241         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
18242         or 0 for -m/-mno case, but initialises as 2 to detect default use case
18243         separately.
18245 2013-03-01  Xinliang David Li  <davidxl@google.com>
18247         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
18248         walk length.
18250 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
18252         PR middle-end/56461
18253         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
18254         vector even when returning true.  Fix up function comment formatting.
18256         PR middle-end/56461
18257         * ira-build.c (ira_loop_nodes_count): New variable.
18258         (create_loop_tree_nodes): Initialize it.
18259         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
18261         PR middle-end/56461
18262         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
18263         method on dr_chain and result_chain.
18264         * tree-vect-stmts.c (vectorizable_store): Only call
18265         result_chain.create if j == 0.
18267         PR middle-end/56461
18268         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
18269         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
18270         before overwriting it.
18272 2013-03-01  Tobias Burnus  <burnus@net-b.de>
18274         * doc/extended.texi (C Extensions): Change order in @menu
18275         to match @node.
18276         (Other MIPS Built-in Functions): Move last MIPS entry before
18277         "picoChip Built-in Functions".
18278         (SH Built-in Functions): Move after RX Built-in Functions.
18279         * doc/gcc.texi (Introduction): Change order in @menu
18280         to match @node.
18281         * doc/md.texi (Constraints): Ditto.
18282         * gty.texi (Type Information): Ditto.
18283         (User-provided marking routines for template types): Make
18284         subsection.
18285         * doc/invoke.texi (AArch64 Options): Move before
18286         "Adapteva Epiphany Options".
18288 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
18289             Jakub Jelinek  <jakub@redhat.com>
18291         PR sanitizer/56454
18292         * asan.c (gate_asan): Lookup no_sanitize_address instead of
18293         no_address_safety_analysis attribute.
18294         * doc/extend.texi (no_address_safety_attribute): Rename to
18295         no_sanitize_address attribute, mention no_address_safety_analysis
18296         attribute as deprecated alias.
18298 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
18300         PR middle-end/56461
18301         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
18302         type to vec<vec<tree> > *.
18303         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
18304         to be vec<tree> instead of vec<tree> *, set vec_defs
18305         to vNULL and call vec_defs.create (number_of_vects), adjust other
18306         uses of vec_defs.
18307         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
18308         vectorizable_condition): Adjust vect_get_slp_defs callers.
18310 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18312         * config/aarch64/aarch64.c
18313         (aarch64_float_const_representable): Remove unused variable.
18315 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18317         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
18319 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18321         * config/aarch64/aarch64-builtins.c
18322         (aarch64_init_simd_builtins): Make static.
18324 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18326         * config/aarch64/aarch64.c
18327         (aarch64_simd_make_constant): Make static.
18329 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18331         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
18332         with no initialization to the RHS of debug statements.
18334 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18336         PR tree-optimization/56294
18337         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
18338         Adjust dumping.
18339         (get_access_replacement): Do not call create_access_replacement.
18340         Assert a replacement exists.
18341         (get_repl_default_def_ssa_name): Create the replacement declaration
18342         itself.
18344 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18346         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
18347         final_end_function.
18349 2013-02-28  Marek Polacek  <polacek@redhat.com>
18351         PR rtl-optimization/56466
18352         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
18353         if we're changing a loop.
18354         (peel_loops_completely): Likewise.
18356 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
18358         PR c++/55813
18359         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
18361 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18363         PR target/56445
18364         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
18365         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
18366         INTX_FTYPE_FX, FX_FTYPE_INTX.
18367         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
18369 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18371         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
18372         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
18373         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
18374         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
18375         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
18376         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
18377         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
18378         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
18379         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
18380         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
18381         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
18382         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
18383         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
18384         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
18385         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18386         (avrxmega6): Increase max flash segments from 5 to 6.
18387         * config/avr/t-multilib: Regenerate.
18388         * config/avr/avr-tables.opt: Regenerate.
18389         * doc/avr-mmcu.texi: Regenerate.
18391 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18393         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
18394         (avr_device_to_arch): Rename to avr_device_to_ld.
18395         (avr_device_to_as): New prototype.
18396         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
18397         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
18398         * config/avr/driver-avr.c (avr_device_to_as): New.
18399         (avr_device_to_arch): Rename to avr_device_to_ld.
18401 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18403         PR middle-end/56461
18404         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
18405         method on dr_chain and result_chain.
18407         PR middle-end/56461
18408         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
18409         pointer_set_destroy on not_executed_last_iteration.
18411         PR middle-end/56461
18412         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
18414         PR middle-end/56461
18415         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
18416         FOR_EACH_DEFINED_FUNCTION when freeing state.
18418         PR middle-end/56461
18419         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
18420         pool_free.
18421         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
18422         overwriting it.
18424         PR middle-end/56461
18425         * ipa-cp.c (decide_whether_version_node): Call vec_free on
18426         known_aggs[i].items and release known_aggs vector.
18428         PR middle-end/56461
18429         * ipa-reference.c (propagate): Free node_info even for alias nodes.
18431 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18433         * config/microblaze/microblaze.c (microblaze_emit_compare):
18434         Use xor for EQ/NE comparisions.
18435         * config/microblaze/microblaze.md (cstoresf4): Add constraints
18436         (cbranchsf4): Adjust operator to comparison_operator.
18438 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18440         PR middle-end/56461
18441         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
18442         vector.
18443         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
18444         vec_safe_push, always update *slot.
18445         (redirect_edge_var_map_clear): Use vec_free.
18446         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
18447         (free_var_map_entry): Use vec_free.
18448         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
18449         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
18451 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
18453         PR middle-end/45472
18454         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
18455         when the may_trap_p bit of the exprs being merged differs.
18456         Reorder tests for speculativeness in the logical and operator.
18458 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18460         * incpath.c (add_standard_paths): Use reconcat instead of concat
18461         where appropriate and avoid leaking memory.
18463         * opts.h: Include obstack.h.
18464         (opts_concat): New prototype.
18465         (opts_obstack): New declaration.
18466         * opts.c (opts_concat): New function.
18467         (opts_obstack): New variable.
18468         (init_options_struct): Call gcc_init_obstack on opts_obstack.
18469         (finish_options): Use opts_concat instead of concat
18470         and XOBNEWVEC instead of XNEWVEC.
18471         * opts-common.c (generate_canonical_option, decode_cmdline_option,
18472         generate_option): Likewise.
18473         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
18474         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
18476         PR target/56455
18477         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18478         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
18480 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18482         PR middle-end/56461
18483         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
18485 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18487         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
18488         (arm_block_move_unaligned_straight): Likewise.
18489         (arm_adjust_block_mem): Likewise.
18491 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18493         PR target/48901
18494         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
18495         temp, cond and label.
18496         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
18498         PR target/52500
18499         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
18500         * config/c6x/c6x.h (dbx_register_map): Update declaration.
18502         PR target/52501
18503         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
18504         of prologue/epilogue functions.
18506         PR target/52550
18507         * config/tilegx/tilegx.c (tilegx_expand_prologue):
18508         Remove unused variable cfa_offset.
18509         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
18511         PR target/54639
18512         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
18513         type promotion to unsigned.
18515         PR target/54640
18516         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
18517         for HOST_WIDE_INT of 32 bit / same size as int.
18518         (arm_block_move_unaligned_straight): Likewise.
18519         (arm_adjust_block_mem): Likewise.
18521         PR target/54662
18522         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
18523         ALL_CFLAGS.
18525 2013-02-26  Marek Polacek  <polacek@redhat.com>
18527         PR tree-optimization/56426
18528         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
18530 2013-02-26  Richard Biener  <rguenther@suse.de>
18532         PR target/56444
18533         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
18534         unused variable loops.
18536 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18538         PR tree-optimization/56448
18539         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
18540         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
18541         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
18542         later operands of the references, or even first operand for
18543         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
18545         PR tree-optimization/56443
18546         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
18547         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
18548         to type_for_mode langhook.
18550 2013-02-25  Matt Turner  <mattst88@gmail.com>
18552         * doc/invoke.texi: Document r4700.
18554 2013-02-25  Richard Biener  <rguenther@suse.de>
18556         PR tree-optimization/56175
18557         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
18558         split out from ...
18559         (simplify_bitwise_binary): ... here.  Also guard the conversion
18560         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
18562 2013-02-25  Catherine Moore  <clm@codesourcery.com>
18564         Revert:
18565         2013-02-24  Catherine Moore  <clm@codesourcery.com>
18566             Maciej W. Rozycki  <macro@codesourcery.com>
18567             Tom de Vries  <tom@codesourcery.com>
18568             Nathan Sidwell  <nathan@codesourcery.com>
18569             Iain Sandoe  <iain@codesourcery.com>
18570             Nathan Froyd  <froydnj@codesourcery.com>
18571             Chao-ying Fu  <fu@mips.com>
18573         * doc/extend.texi: (micromips, nomicromips, nocompression):
18574         Document new function attributes.
18575         * doc/invoke.texi (minterlink-compressed, mmicromips,
18576         m14k, m14ke, m14kec): Document new options.
18577         (minterlink-mips16): Update documentation.
18578         * doc/md.texi (ZC, ZD): Document new constraints.
18579         * configure.ac (gcc_cv_as_micromips): Check if linker
18580         supports the .set micromips directive.
18581         * configure: Regenerate.
18582         * config.in: Regenerate.
18583         * config/mips/mips-tables.opt: Regenerate.
18584         * config/mips/micromips.md: New file.
18585         * constraints.md (ZC, AD): New constraints.
18586         * config/mips/predicates.md (movep_src_register): New predicate.
18587         (movep_src_operand): New predicate.
18588         (non_volatile_mem_operand): New predicate.
18589         * config/mips/mips.md (multimem): New type.
18590         (length): Differentiate between 17-bit and 18-bit branch offsets.
18591         (MOVEP1, MOVEP2): New mode iterator.
18592         (mov_<load>l): Use ZC constraint.
18593         (mov_<load>r): Likewise.
18594         (mov_<store>l): Likewise.
18595         (mov_<store>r): Likewise.
18596         (*branch_equality<mode>_inverted): Add microMIPS support.
18597         (*branch_equality<mode>): Likewise.
18598         (*jump_absolute): Likewise.
18599         (indirect_jump_<mode>): Likewise.
18600         (tablejump_<mode>): Likewise.
18601         (<optab>_internal): Likewise.
18602         (sibcall_internal): Likewise.
18603         (sibcall_value_internal): Likewise.
18604         (prefetch): Use constraint ZD.
18605         * config/mips/mips.opt (minterlink-compressed): New option.
18606         (minterlink-mips16): Now an alias for minterlink-compressed.
18607         (mmicromips): New option.
18608         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18609         (compare_and_swap_12): Likewise.
18610         (sync_add<mode>): Likewise.
18611         (sync_<optab>_12): Likewise.
18612         (sync_old_<optab>_12): Likewise.
18613         (sync_new_<optab>_12): Likewise.
18614         (sync_nand_12): Likewise.
18615         (sync_old_nand_12): Likewise.
18616         (sync_new_nand_12): Likewise.
18617         (sync_sub<mode>): Likewise.
18618         (sync_old_add<mode>): Likewise.
18619         (sync_old_sub<mode>): Likewise.
18620         (sync_new_add<mode>): Likewise.
18621         (sync_new_sub<mode>): Likewise.
18622         (sync_<optab><mode>): Likewise.
18623         (sync_old_<optab><mode>): Likewise.
18624         (sync_new_<optab><mode>): Likewise.
18625         (sync_nand<mode>): Likewise.
18626         (sync_old_nand<mode>): Likewise.
18627         (sync_new_nand<mode>): Likewise.
18628         (sync_lock_test_and_set<mode>): Likewise.
18629         (test_and_set_12): Likewise.
18630         (atomic_compare_and_swap<mode>): Likewise.
18631         (atomic_exchange<mode>_llsc): Likewise.
18632         (atomic_fetch_add<mode>_llsc): Likewise.
18633         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18634         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18635         (umips_save_restore_pattern_p): Likewise.
18636         (umips_load_store_pair_p): Likewise.
18637         (umips_output_load_store_pair): Likewise.
18638         (umips_movep_target_p): Likewise.
18639         (umips_12bit_offset_address_p): Likewise.
18640         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18641         (mips_base_mips16): Rename this...
18642         (mips_base_compression_flags): ...to this. Update all uses.
18643         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18644         (mips_mips16_decl_p): Delete.
18645         (mips_nomips16_decl_p): Delete.
18646         (mips_get_compress_on_flags): New function.
18647         (mips_get_compress_off_flags): New function.
18648         (mips_get_compress_mode): New function.
18649         (mips_get_compress_on_name): New function.
18650         (mips_get_compress_off_name): New function.
18651         (mips_insert_attributes): Support multiple compression types.
18652         (mips_merge_decl_attributes): Likewise.
18653         (umips_12bit_offset_address_p): New function.
18654         (mips_start_function_definition): Emit .set micromips directive.
18655         (mips_call_may_need_jalx_p): New function.
18656         (mips_function_ok_for_sibcall): Add microMIPS support.
18657         (mips_print_operand_punctuation): Support short delay slots and
18658         compact jumps.
18659         (umips_swm_mask, umips_swm_encoding): New.
18660         (umips_build_save_restore): New function.
18661         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18662         (was_mips16_p): Remove.
18663         (old_compression_mode): New.
18664         (mips_set_compression_mode): New function.
18665         (mips_set_current_function): Add microMIPS support.
18666         (mips_option_override): Likewise.
18667         (umips_save_restore_pattern_p): New function.
18668         (umips_output_save_restore): New function.
18669         (umips_load_store_pair_p_1): New function.
18670         (umips_load_store_pair_p): New function.
18671         (umips_output_load_store_pair_1): New function.
18672         (umips_output_load_store_pair): New function.
18673         (umips_movep_target_p) New function.
18674         (mips_prepare_pch_save): Add microMIPS support.
18675         * config/mips/mips.h (TARGET_COMPRESSION): New.
18676         (TARGET_CPU_CPP_BUILTINS): Update macro
18677         to use new compression flags and to support microMIPS.
18678         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18679         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18680         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18681         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18682         (ASM_SPEC): Support mmicromips and mno-micromips.
18683         (M16STORE_REG_P): New macro.
18684         (MIPS_CALL): Support TARGET_MICROMIPS.
18685         (MICROMIPS_J): New macro.
18686         (mips_base_mips16): Rename this...
18687         (mips_base_compression_flags): ...to this.
18688         (UMIPS_12BIT_OFFSET_P): New macro.
18689         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18690         (MULTILIB_DIRNAMES): Likewise.
18692 2013-02-25  Tom de Vries  <tom@codesourcery.com>
18694         PR rtl-optimization/56131
18695         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
18696         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
18697         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
18699 2013-02-25  Tobias Burnus  <burnus@net-b.de>
18701         * doc/invoke.texi (-fsanitize=): Move from optimization
18702         to debugging options.
18704 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18706         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
18708 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
18709             Alexander Monakov  <amonakov@ispras.ru>
18711         PR middle-end/56077
18712         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
18713         flush pending lists also on non-jumps.  Adjust comment.
18715 2013-02-24  Catherine Moore  <clm@codesourcery.com>
18716             Maciej W. Rozycki  <macro@codesourcery.com>
18717             Tom de Vries  <tom@codesourcery.com>
18718             Nathan Sidwell  <nathan@codesourcery.com>
18719             Iain Sandoe  <iain@codesourcery.com>
18720             Nathan Froyd  <froydnj@codesourcery.com>
18721             Chao-ying Fu  <fu@mips.com>
18723         * doc/extend.texi: (micromips, nomicromips, nocompression):
18724         Document new function attributes.
18725         * doc/invoke.texi (minterlink-compressed, mmicromips,
18726         m14k, m14ke, m14kec): Document new options.
18727         (minterlink-mips16): Update documentation.
18728         * doc/md.texi (ZC, ZD): Document new constraints.
18729         * configure.ac (gcc_cv_as_micromips): Check if linker
18730         supports the .set micromips directive.
18731         * configure: Regenerate.
18732         * config.in: Regenerate.
18733         * config/mips/mips-tables.opt: Regenerate.
18734         * config/mips/micromips.md: New file.
18735         * constraints.md (ZC, AD): New constraints.
18736         * config/mips/predicates.md (movep_src_register): New predicate.
18737         (movep_src_operand): New predicate.
18738         (non_volatile_mem_operand): New predicate.
18739         * config/mips/mips.md (multimem): New type.
18740         (length): Differentiate between 17-bit and 18-bit branch offsets.
18741         (MOVEP1, MOVEP2): New mode iterator.
18742         (mov_<load>l): Use ZC constraint.
18743         (mov_<load>r): Likewise.
18744         (mov_<store>l): Likewise.
18745         (mov_<store>r): Likewise.
18746         (*branch_equality<mode>_inverted): Add microMIPS support.
18747         (*branch_equality<mode>): Likewise.
18748         (*jump_absolute): Likewise.
18749         (indirect_jump_<mode>): Likewise.
18750         (tablejump_<mode>): Likewise.
18751         (<optab>_internal): Likewise.
18752         (sibcall_internal): Likewise.
18753         (sibcall_value_internal): Likewise.
18754         (prefetch): Use constraint ZD.
18755         * config/mips/mips.opt (minterlink-compressed): New option.
18756         (minterlink-mips16): Now an alias for minterlink-compressed.
18757         (mmicromips): New option.
18758         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18759         (compare_and_swap_12): Likewise.
18760         (sync_add<mode>): Likewise.
18761         (sync_<optab>_12): Likewise.
18762         (sync_old_<optab>_12): Likewise.
18763         (sync_new_<optab>_12): Likewise.
18764         (sync_nand_12): Likewise.
18765         (sync_old_nand_12): Likewise.
18766         (sync_new_nand_12): Likewise.
18767         (sync_sub<mode>): Likewise.
18768         (sync_old_add<mode>): Likewise.
18769         (sync_old_sub<mode>): Likewise.
18770         (sync_new_add<mode>): Likewise.
18771         (sync_new_sub<mode>): Likewise.
18772         (sync_<optab><mode>): Likewise.
18773         (sync_old_<optab><mode>): Likewise.
18774         (sync_new_<optab><mode>): Likewise.
18775         (sync_nand<mode>): Likewise.
18776         (sync_old_nand<mode>): Likewise.
18777         (sync_new_nand<mode>): Likewise.
18778         (sync_lock_test_and_set<mode>): Likewise.
18779         (test_and_set_12): Likewise.
18780         (atomic_compare_and_swap<mode>): Likewise.
18781         (atomic_exchange<mode>_llsc): Likewise.
18782         (atomic_fetch_add<mode>_llsc): Likewise.
18783         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18784         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18785         (umips_save_restore_pattern_p): Likewise.
18786         (umips_load_store_pair_p): Likewise.
18787         (umips_output_load_store_pair): Likewise.
18788         (umips_movep_target_p): Likewise.
18789         (umips_12bit_offset_address_p): Likewise.
18790         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18791         (mips_base_mips16): Rename this...
18792         (mips_base_compression_flags): ...to this. Update all uses.
18793         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18794         (mips_mips16_decl_p): Delete.
18795         (mips_nomips16_decl_p): Delete.
18796         (mips_get_compress_on_flags): New function.
18797         (mips_get_compress_off_flags): New function.
18798         (mips_get_compress_mode): New function.
18799         (mips_get_compress_on_name): New function.
18800         (mips_get_compress_off_name): New function.
18801         (mips_insert_attributes): Support multiple compression types.
18802         (mips_merge_decl_attributes): Likewise.
18803         (umips_12bit_offset_address_p): New function.
18804         (mips_start_function_definition): Emit .set micromips directive.
18805         (mips_call_may_need_jalx_p): New function.
18806         (mips_function_ok_for_sibcall): Add microMIPS support.
18807         (mips_print_operand_punctuation): Support short delay slots and
18808         compact jumps.
18809         (umips_swm_mask, umips_swm_encoding): New.
18810         (umips_build_save_restore): New function.
18811         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18812         (was_mips16_p): Remove.
18813         (old_compression_mode): New.
18814         (mips_set_compression_mode): New function.
18815         (mips_set_current_function): Add microMIPS support.
18816         (mips_option_override): Likewise.
18817         (umips_save_restore_pattern_p): New function.
18818         (umips_output_save_restore): New function.
18819         (umips_load_store_pair_p_1): New function.
18820         (umips_load_store_pair_p): New function.
18821         (umips_output_load_store_pair_1): New function.
18822         (umips_output_load_store_pair): New function.
18823         (umips_movep_target_p) New function.
18824         (mips_prepare_pch_save): Add microMIPS support.
18825         * config/mips/mips.h (TARGET_COMPRESSION): New.
18826         (TARGET_CPU_CPP_BUILTINS): Update macro
18827         to use new compression flags and to support microMIPS.
18828         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18829         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18830         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18831         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18832         (ASM_SPEC): Support mmicromips and mno-micromips.
18833         (M16STORE_REG_P): New macro.
18834         (MIPS_CALL): Support TARGET_MICROMIPS.
18835         (MICROMIPS_J): New macro.
18836         (mips_base_mips16): Rename this...
18837         (mips_base_compression_flags): ...to this.
18838         (UMIPS_12BIT_OFFSET_P): New macro.
18839         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18840         (MULTILIB_DIRNAMES): Likewise.
18842 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
18844         PR target/52555
18845         * target-globals.c (save_target_globals): For init_reg_sets and
18846         target_reinit remporarily set this_fn_optabs to this_target_optabs.
18848 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
18850         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
18851         * config/aarch64/t-aarch64
18852         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
18854 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
18856         PR inline-asm/56148
18857         * lra-constraints.c (process_alt_operands): Reload operand
18858         conflicting with earlier clobber only if no more other conflicting
18859         operands.
18861 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
18863         PR sanitizer/56393
18864         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
18865         if not linking a shared library.
18867 2013-02-22  Seth LaForge  <sethml@google.com>
18869         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
18871 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
18873         * config/arm/arm.md (split for extendsidi): Update condition.
18874         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
18875         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
18876         (qhs_zextenddi_cstr): Likewise.
18878 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18880         PR middle-end/56420
18881         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
18882         avoid signed wrapping.
18883         (expand_mult): Handle properly multiplication by
18884         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
18885         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
18886         in the compiler if coeff is HOST_WIDE_INT_MIN.
18887         (expand_divmod): Don't make ext_op1 static, change it's type to
18888         uhwi.  Avoid undefined behavior in -INTVAL (op1).
18890         PR rtl-optimization/50339
18891         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
18892         field.
18893         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
18894         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
18895         into splitting_ashiftrt field.
18896         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
18897         ASHIFTRT.
18898         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
18899         choices.
18901 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
18903         PR middle-end/56108
18904         * trans-mem.c (execute_tm_mark): Do not expand transactions that
18905         are sure to go irrevocable.
18907 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
18909         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
18910         scalars are valid operands.
18912 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18914         PR tree-optimization/56310
18915         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
18916         only matching indices and non-negative final offsets.
18917         (intersect_aggregates_with_edge): Pass src_idx to
18918         agg_replacements_to_vector.  Pass src_idx insstead of index to
18919         intersect_with_agg_replacements.
18921 2013-02-21  Martin Jambor  <mjambor@suse.cz>
18923         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
18924         instead of hard-wired defaults.
18926 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
18928         * doc/invoke.texi (MIPS Options): Update documentation of the
18929         floating-point multiply-accumulate instruction restrictions.
18931 2013-02-21  Kostya Serebryany  <kcc@google.com>
18933         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
18934         asan_shadow_offset on x86_64 linux.
18936 2013-02-21  Richard Biener  <rguenther@suse.de>
18938         PR tree-optimization/56415
18939         Revert
18940         2013-02-11  Richard Biener  <rguenther@suse.de>
18942         PR tree-optimization/56273
18943         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
18944         first VRP run.
18946 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
18948         PR bootstrap/56258
18949         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
18950         instead of @itemx.
18952         PR inline-asm/56405
18953         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
18954         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
18956 2013-02-20  Jan Hubicka  <jh@suse.cz>
18958         PR tree-optimization/56265
18959         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
18960         when target is referenced for first time.
18962 2013-02-20  Richard Biener  <rguenther@suse.de>
18964         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
18965         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
18966         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
18967         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
18968         not return anything.
18969         (rename_ssa_copies): Do not remove unused locals.
18970         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
18971         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
18972         * passes.c (execute_function_todo): Do not schedule unused locals
18973         removal if cleanup_tree_cfg did something.
18974         * tree-ssa-live.c (remove_unused_locals): Dump statistics
18975         about the number of removed locals.
18977 2013-02-20  Richard Biener  <rguenther@suse.de>
18979         PR tree-optimization/56398
18980         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
18982 2013-02-20  Martin Jambor  <mjambor@suse.cz>
18984         PR tree-optimization/55334
18985         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
18986         restricted pointers to arrays.
18988 2013-02-20  Richard Biener  <rguenther@suse.de>
18989             Jakub Jelinek  <jakub@redhat.com>
18991         PR tree-optimization/56396
18992         * tree-ssa-ccp.c (n_const_val): New static variable.
18993         (get_value): Return NULL for SSA names we don't have a lattice
18994         entry for.
18995         (ccp_initialize): Initialize n_const_val.
18996         * tree-ssa-copy.c (n_copy_of): New static variable.
18997         (init_copy_prop): Initialize n_copy_of.
18998         (get_value): Return NULL_TREE for SSA names we don't have a
18999         lattice entry for.
19001 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19003         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
19005 2013-02-20  Richard Biener  <rguenther@suse.de>
19007         * genpreds.c (write_lookup_constraint): Do not compare first
19008         letter of the constraint again.
19010 2013-02-20  Richard Biener  <rguenther@suse.de>
19012         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
19013         and ceil_log2.
19014         (get_use_iv_cost): Terminate hashtable walk when coming across
19015         an empty entry.
19017 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
19019         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19020         reassociation for avx2 targets.
19022 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19024         * config/microblaze/microblaze.c: microblaze_has_clz = 0
19025         Add version check for v8.10.a to enable microblaze_has_clz
19026         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19027         version and TARGET_PATTERN_COMPARE check
19028         * config/microblaze/microblaze.md: New clzsi2 instruction
19030 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19032         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19033         function before branching.
19035 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19037         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19038         DUMP_INSN_RTX_UID.
19039         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19041 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19043         PR middle-end/55889
19044         * sel-sched.c: Include ira.h.
19045         (implicit_clobber_conflict_p): New function.
19046         (moveup_expr): Use it.
19047         * Makefile.in (sel-sched.o): Depend on ira.h.
19049 2013-02-19  Richard Biener  <rguenther@suse.de>
19051         PR tree-optimization/56384
19052         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19053         (vn_hash_type): Split out from ...
19054         (vn_hash_constant_with_type): ... here.
19055         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19056         (vn_phi_eq): Compare types from vn_phi_s structure.
19057         (vn_phi_lookup): Populate vn_phi_s type.
19058         (vn_phi_insert): Likewise.
19060 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19062         PR tree-optimization/56350
19063         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19064         if haven't found reduction or nested cycle operand, rather than
19065         asserting we must find it.
19067         PR tree-optimization/56381
19068         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19069         to fold_build3.
19071 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19072             Jakub Jelinek  <jakub@redhat.com>
19074         PR target/52555
19075         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19076         (swap_optab_enable): Same.
19077         (init_all_optabs): Use argument instead of global.
19078         * tree.h (struct tree_optimization_option): New field target_optabs.
19079         * expr.h (init_all_optabs): Add argument to prototype.
19080         (TREE_OPTIMIZATION_OPTABS): New.
19081         (save_optabs_if_changed): Protoize.
19082         * optabs.h: Declare this_fn_optabs.
19083         * optabs.c (save_optabs_if_changed): New.
19084         Declare this_fn_optabs.
19085         (init_optabs): Add argument to init_all_optabs() call.
19086         * function.c (invoke_set_current_function_hook): Handle per
19087         function optabs.
19088         * function.h (struct function): New field optabs.
19089         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19090         optimization_current_node has changed.
19091         * target-globals.h (save_target_globals_default_opts): Protoize.
19092         * target-globals.c (save_target_globals_default_opts): New.
19094 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19096         PR target/56347
19097         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19098         registers %fr12 and %fr12R as call used.
19100         PR target/56214
19101         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19102         and HImode, require all displacements to be an integer multiple of
19103         their mode size.
19104         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19105         only allow QImode and HImode when reload is in progress and strict is
19106         true.  Likewise for symbolic addresses.  Use base14_operand to check
19107         displacements in REG+BASE addresses.
19109 2013-02-18  Richard Biener  <rguenther@suse.de>
19111         PR tree-optimization/56366
19112         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19113         handle sign-conversion of outer-loop initial induction value.
19115 2013-02-18  Richard Biener  <rguenther@suse.de>
19117         PR middle-end/56349
19118         * cfghooks.c (merge_blocks): If we merge a latch into another
19119         block adjust references to it.
19120         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19121         (verify_loop_structure): Verify that a recorded latch is in fact
19122         a latch.
19124 2013-02-18  Richard Biener  <rguenther@suse.de>
19126         PR tree-optimization/56321
19127         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19128         order SSA name release and virtual operand unlinking.
19130 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19132         * config/microblaze/microblaze.md (save_stack_block): Define.
19133         (restore_stack_block): Likewise.
19135 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19137         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19138         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19139         * config/microblaze/microblaze.c (microblaze_option_override):
19140         Bail out early for PIC modes when target does not support PIC.
19142 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19144         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19145         Replace with a microblaze version.
19146         (microblaze_trampoline_init): Adapt for microblaze.
19147         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19148         microblaze.
19150 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
19151             Dodji Seketeli  <dodji@redhat.com>
19153         PR asan/56330
19154         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19155         (instrument_mem_region_access): Do not forget to always put
19156         instrumentation of the of 'base' and 'base + len' in a "if (len !=
19157         0) statement, even for cases where either 'base' or 'base + len'
19158         are not instrumented -- because they have been previously
19159         instrumented.  Simplify the logic by putting all the statements
19160         instrument 'base + len' inside a sequence, and then insert that
19161         sequence right before the current insertion point.  Then, to
19162         instrument 'base + len', just get an iterator on that statement.
19163         And do not forget to update the pointer to iterator the function
19164         received as argument.
19166 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
19168         PR rtl-optimization/56348
19169         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19171 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
19173         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19174         (clean_graph_dump_file): Pass base to start_graph_dump.
19176 2013-02-14  Richard Henderson  <rth@redhat.com>
19178         PR target/55941
19179         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19181 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
19183         * collect2-aix.h: Define F_LOADONLY.
19185 2013-02-14  Richard Biener  <rguenther@suse.de>
19187         PR lto/50494
19188         * varasm.c (output_constant_def_1): Get the decl representing
19189         the constant as argument.
19190         (output_constant_def): Wrap output_constant_def_1.
19191         (make_decl_rtl): Use output_constant_def_1 with the decl
19192         representing the constant.
19193         (build_constant_desc): Optionally re-use a decl already
19194         representing the constant.
19195         (tree_output_constant_def): Adjust.
19197 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
19199         Fix an asan crash
19200         * asan.c (instrument_builtin_call):  Really put the length of the
19201         second source argument into src1_len.
19203 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19205         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19206         argument.  If it is false, don't create edge from then_bb to
19207         fallthru_bb.
19208         (insert_if_then_before_iter): Pass true to it.
19209         (build_check_stmt): Pass false to it.
19210         (transform_statements): Flush hash table only on extended basic
19211         block boundaries, rather than at the beginning of every bb.
19212         Don't flush hash table on nonfreeing_call_p calls.
19213         * tree-flow.h (nonfreeing_call_p): New prototype.
19214         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19216 2013-02-13  David S. Miller  <davem@davemloft.net>
19218         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19220 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
19222         PR target/56184
19223         * ira.c (max_regno_before_ira): Move from ...
19224         (ira): ... here.
19225         (fix_reg_equiv_init): Use max_regno_before_ira instead of
19226         vec_safe_length.
19228 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19230         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
19232 2013-02-13  Richard Biener  <rguenther@suse.de>
19234         PR lto/56295
19235         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
19236         globals in MEM_REFs.
19238 2013-02-13  Richard Biener  <rguenther@suse.de>
19240         * loop-init.c (loop_optimizer_init): Clear loop state when
19241         re-initializing preserved loops.
19242         * loop-unswitch.c (unswitch_single_loop): Return whether
19243         we unswitched the loop.  Do not verify loop state here.
19244         (unswitch_loops): When we unswitched a loop discover new loops.
19246 2013-02-13  Kostya Serebryany  <kcc@google.com>
19248         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
19249         on x86_64 linux.
19250         * sanitizer.def: Rename __asan_init to __asan_init_v1.
19252 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
19254         Avoid instrumenting duplicated memory access in the same basic block
19255         * Makefile.in (asan.o): Add new dependency on hash-table.h
19256         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
19257         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
19258         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
19259         (free_mem_ref_resources, has_mem_ref_been_instrumented)
19260         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
19261         (get_mem_ref_of_assignment): New functions.
19262         (get_mem_refs_of_builtin_call): Extract from
19263         instrument_builtin_call and tweak a little bit to make it fit with
19264         the new signature.
19265         (instrument_builtin_call): Use the new
19266         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
19267         of is_gimple_builtin_call.
19268         (instrument_derefs, instrument_mem_region_access): Insert the
19269         instrumented memory reference into the hash table.
19270         (maybe_instrument_assignment): Renamed instrument_assignment into
19271         this, and change it to advance the iterator when instrumentation
19272         actually happened and return true in that case.  This makes it
19273         homogeneous with maybe_instrument_assignment, and thus give a
19274         chance to callers to be more 'regular'.
19275         (transform_statements): Clear the memory reference hash table
19276         whenever we enter a new BB, when we cross a function call, or when
19277         we are done transforming statements.  Use
19278         maybe_instrument_assignment instead of instrumentation.  No more
19279         need to special case maybe_instrument_assignment and advance the
19280         iterator after calling it; it's now handled just like
19281         maybe_instrument_call.  Update comment.
19283 2013-02-13  Richard Biener  <rguenther@suse.de>
19285         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
19286         Fix loop discovery code.
19288 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
19290         PR inline-asm/56148
19291         * lra-constraints.c (process_alt_operands): Match early clobber
19292         operand with itself.  Check conflicts with earlyclobber only if
19293         the operand is not reloaded.  Prefer to reload conflicting operand
19294         if earlyclobber and matching operands are the same.
19296 2013-02-12  Richard Biener  <rguenther@suse.de>
19298         PR lto/56297
19299         * lto-streamer-out.c (write_symbol): Do not output symbols
19300         for hard register variables.
19302 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
19304         PR target/54222
19305         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
19306         (umulsidi3_insn, mulsidi3_insn): New insns.
19308 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
19310         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
19311         (struct tune_params): Add vec_costs field.
19312         * config/arm/arm.c (arm_builtin_vectorization_cost)
19313         (arm_add_stmt_cost): New functions.
19314         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
19315         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
19316         (arm_default_vec_cost): New struct of type cpu_vec_costs.
19317         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
19318         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
19319         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
19320         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
19322 2013-02-12  Richard Biener  <rguenther@suse.de>
19324         PR lto/56295
19325         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
19326         decls again if possible.
19328 2013-02-12  Richard Biener  <rguenther@suse.de>
19330         PR middle-end/56288
19331         * tree-ssa.c (verify_ssa_name): Fix check, move
19332         SSA_NAME_IN_FREE_LIST check up.
19334 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
19335             Steven Bosscher   <steven@gcc.gnu.org>
19337         PR rtl-optimization/56151
19338         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
19339         equal to op0 or op1, and last_insn pattern is CODE operation
19340         with MEM dest and one of the operands matches that MEM.
19342 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
19344         * doc/extend.texi: Document Function Multiversioning and "default"
19345         parameter string to target attribute.
19346         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
19347         target attribute parameter is "default".
19348         (ix86_compare_version_priority): Remove checks for target attribute.
19349         (ix86_mangle_function_version_assembler_name): Change error to sorry.
19350         Remove check for target attribute equal to NULL. Add assert.
19351         (ix86_generate_version_dispatcher_body): Change error to sorry.
19353 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
19354             Jack Howarth  <howarth@bromo.med.uc.edu>
19355             Patrick Marlier  <patrick.marlier@gmail.com>
19357         PR libitm/55693
19358         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
19359         define ENDFILE_SPEC as TM_DESTRUCTOR.
19360         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
19362 2013-02-11  Alexander Potapenko  <glider@google.com>
19363             Jack Howarth  <howarth@bromo.med.uc.edu>
19364             Jakub Jelinek  <jakub@redhat.com>
19366         PR sanitizer/55617
19367         * config/darwin.c (cdtor_record): Rename ctor_record.
19368         (sort_cdtor_records): Rename sort_ctor_records.
19369         (finalize_dtors): New routine to sort destructors by
19370         priority before use in assemble_integer.
19371         (machopic_asm_out_destructor): Use finalize_dtors if needed.
19373 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
19375         PR rtl-optimization/56275
19376         * simplify-rtx.c (avoid_constant_pool_reference): Check that
19377         offset is non-negative and less than cmode size before
19378         calling simplify_subreg.
19380 2013-02-11  Richard Biener  <rguenther@suse.de>
19382         PR tree-optimization/56264
19383         * cfgloop.h (fix_loop_structure): Adjust prototype.
19384         * loop-init.c (fix_loop_structure): Return the number of
19385         newly discovered loops.
19386         * tree-cfgcleanup.c (repair_loop_structures): When new loops
19387         are discovered, do a full loop-closed SSA rewrite.
19389 2013-02-11  Richard Biener  <rguenther@suse.de>
19391         PR tree-optimization/56273
19392         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19393         first VRP run.
19394         (check_array_ref): Fix missing newline in dumps.
19395         (search_for_addr_array): Likewise.
19397 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
19399         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
19401 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
19403         PR target/56256
19404         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
19406 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
19408         PR rtl-optimization/56246
19409         * lra-constraints.c (simplify_operand_subreg): Try to reuse
19410         reload pseudo.
19411         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
19412         constraints are satisfied.
19414 2013-02-08  Jeff Law  <law@redhat.com>
19416         PR debug/53948
19417         * emit-rtl.c (reg_is_parm_p): New function.
19418         * regs.h (reg_is_parm_p): New prototype.
19419         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
19420         callee-clobbered registers.
19422 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
19424         PR target/56043
19425         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
19426         If there is no implicit builtin declaration, just return NULL.
19428 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19430         * config/i386/sse.md (FMAMODEM): New mode iterator.
19431         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
19432         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
19434 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19436         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
19437         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
19438         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
19440 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19442         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
19443         (microblaze*-*-elf): Likewise.
19444         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
19445         LINK_SPEC.
19446         * config/microblaze/microblaze-c.c: Add builtin defines for
19447         _LITTLE_ENDIAN and _BIG_ENDIAN.
19448         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
19449         add to TARGET_DEFAULT flags.
19450         Expand ASM_SPEC and LINK_SPEC.
19451         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
19452         * config/microblaze/microblaze.md: Update extendsidi2 and
19453         movdi_internal instructions to use low-order / high-order reg
19454         print_operands.
19455         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
19456         options and inversemask / mask of LITTLE_ENDIAN.
19457         * config/microblaze/t-microblaze: Expand multilib options to
19458         include mlittle-endian (le) and update exceptions patterns.
19460 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
19462         PR rtl-optimization/56195
19463         * lra-constraints.c (get_reload_reg): Don't reuse regs
19464         if they have smaller mode than requested, if they have
19465         wider mode than requested, try to return a SUBREG.
19467         PR tree-optimization/56250
19468         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
19469         if type is unsigned and code isn't MULT_EXPR.
19471 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19473         PR tree-optimization/56064
19474         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
19475         bits according to mode.
19476         * fixed-value.h (fixed_from_double_int)
19477         (const_fixed_from_double_int): Adjust comments.
19479 2013-02-08  Richard Biener  <rguenther@suse.de>
19481         PR lto/56231
19482         * lto-streamer.h (struct data_in): Remove current_file, current_line
19483         and current_col members.
19484         * lto-streamer-out.c (lto_output_location): Stream changed bits
19485         en-block for efficiency.
19486         * lto-streamer-in.c (clear_line_info): Remove.
19487         (lto_input_location): Cache current file, line and column
19488         globally via local statics.  Read changed bits en-block.
19489         (input_function): Do not call clear_line_info.
19490         (lto_read_body): Likewise.
19491         (lto_input_toplevel_asms): Likewise.
19493 2013-02-08  Michael Matz  <matz@suse.de>
19495         PR tree-optimization/52448
19496         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
19497         (nt_call_phase): New static.
19498         (add_or_mark_expr): Only mark accesses with newer phase than any
19499         call seen.
19500         (nonfreeing_call_p): New.
19501         (nt_init_block): Update nt_call_phase, mark blocks as visited.
19502         (nt_fini_block): Keep blocks marked as visited.
19503         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
19505 2013-02-08  Richard Biener  <rguenther@suse.de>
19507         * ira.c (ira): Free broken dominator information.
19509 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19511         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
19513 2013-02-08  Marek Polacek  <polacek@redhat.com>
19515         * cfgloop.c (verify_loop_structure): Add more checking of headers.
19517 2013-02-08  Richard Biener  <rguenther@suse.de>
19519         PR middle-end/56181
19520         * cfgloop.h (flow_loops_find): Adjust.
19521         (bb_loop_header_p): Declare.
19522         * cfgloop.c (bb_loop_header_p): New function split out from ...
19523         (flow_loops_find): ... here.  Adjust function signature,
19524         support incremental loop structure update.
19525         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
19526         * cfgloopmanip.c (fix_loop_structure): Move ...
19527         * loop-init.c (fix_loop_structure): ... here.
19528         (apply_loop_flags): Split out from ...
19529         (loop_optimizer_init): ... here.
19530         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
19531         in incremental mode, only remove dead loops here.
19533 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19535         PR target/54222
19536         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
19537         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
19538         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
19539         (*round<mode>3.libgcc): New insns for fixed-modes.
19540         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
19541         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
19542         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
19543         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
19544         implementations.  Define to __builtin_avr_absFX,
19545         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
19546         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
19547         __builtin_avr_countlsFX, respectively.
19548         * config/avr/avr-c.c (target.h): Include it.
19549         (enum avr_builtin_id): New enum.
19550         (avr_resolve_overloaded_builtin): New static function.
19551         (avr_register_target_pragmas): Use it to set
19552         targetm.resolve_overloaded_builtin.
19553         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
19554         tree nodes used by DEF_BUILTIN.
19555         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
19556         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
19557         <AVR_BUILTIN_xxBITS>: Same.
19559 2013-02-08  Richard Biener  <rguenther@suse.de>
19561         * cfgloop.c (verify_loop_structure): Properly handle
19562         a loop exiting to another loop header.
19563         * ira-int.h (ira_loops): Remove.
19564         * ira.c (ira_loops): Remove.
19565         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
19566         (do_reload): Use loop_optimizer_finalize.
19567         * ira-build.c (create_loop_tree_nodes): Use get_loops and
19568         number_of_loops to access the loop tree.
19569         (more_one_region_p): Likewise.
19570         (finish_loop_tree_nodes): Likewise.
19571         (rebuild_regno_allocno_maps): Likewise.
19572         (mark_loops_for_removal): Likewise.
19573         (mark_all_loops_for_removal): Likewise.
19574         (remove_unnecessary_regions): Likewise.
19575         (ira_build): Likewise.
19576         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19578 2013-02-08  Richard Biener  <rguenther@suse.de>
19580         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
19581         * ipa-pure-const.c (analyze_function): Avoid calling
19582         mark_irreducible_loops twice.
19583         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
19585 2013-02-07  David S. Miller  <davem@davemloft.net>
19587         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
19588         on 'reg'.
19589         * var-tracking.c (vt_add_function_parameter): Test the presence of
19590         HAVE_window_save properly and do not remap argument registers when
19591         we have a leaf function.
19593 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
19595         PR bootstrap/56227
19596         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
19597         instead of "ll".
19598         * config/i386/i386.c (ix86_print_operand): Ditto.
19600 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19602         * lra-constraints.c (process_alt_operands): Fix recently added comment.
19604 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
19606         PR rtl-optimization/56225
19607         * lra-constraints.c (process_alt_operands): Check that reload hard
19608         reg can hold value for strict_low_part.
19610 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19612         PR debug/56154
19613         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
19614         dwarf2out_end_function.
19615         (in_first_function_p, maybe_at_text_label_p,
19616         first_loclabel_num_not_at_text_label): New variables.
19617         (dwarf2out_var_location): In the first function find out
19618         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
19619         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
19620         functions.
19622 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
19624         PR rtl-optimization/56178
19625         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
19626         SUBREG of a register.  Tidy up related block of code.
19627         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
19628         note if the source is a register or a SUBREG of a register.
19630 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
19632         PR target/56228
19633         * config/rs6000/rs6000.md (ptrm): New mode attr.
19634         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
19635         call_value_indirect_aix<pttrsize>,
19636         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
19637         m in constraints.
19639 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
19641         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
19642         if -bnortl. Convert to strcmp and strncmp.
19644 2013-02-07  Alan Modra  <amodra@gmail.com>
19646         PR target/54009
19647         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
19648         addresses won't wrap when offsetting.
19649         (rs6000_secondary_reload): Provide secondary reloads needed for
19650         wrapping LO_SUM addresses.
19652 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
19654         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
19655         MACH, just __MACH__.
19657 2013-02-06  Richard Biener  <rguenther@suse.de>
19659         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
19660         instead of calling fix_loop_structure.
19662 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
19664         PR middle-end/56217
19665         * omp-low.c (use_pointer_for_field): Return false if
19666         lower_send_shared_vars doesn't generate any copy-out code.
19668 2013-02-06  Tom de Vries  <tom@codesourcery.com>
19670         PR rtl-optimization/56131
19671         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
19672         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
19673         of the label is NULL.  Add comment.
19675 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
19677         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
19679         PR sanitizer/55374
19680         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
19681         (STATIC_LIBTSAN_LIBS): Likewise.
19682         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
19683         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
19684         is defined, don't add anything else beyond that.
19685         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
19686         (LINK_COMMAND_SPEC): Use them.
19688         PR tree-optimization/56205
19689         * tree-stdarg.c (check_all_va_list_escapes): Return true if
19690         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
19691         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
19693 2013-02-05  Richard Biener  <rguenther@suse.de>
19695         PR tree-optimization/53342
19696         PR tree-optimization/53185
19697         * tree-vectorizer.h (vect_check_strided_load): Remove.
19698         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
19699         not disallow peeling for vectorized strided loads.
19700         (vect_check_strided_load): Make static and simplify.
19701         (vect_analyze_data_refs): Adjust.
19702         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
19703         correctly when vectorizing strided loads.
19705 2013-02-05  Richard Biener  <rguenther@suse.de>
19707         * doc/install.texi: Refer to ISL, not PPL.
19709 2013-02-05  Jan Hubicka  <jh@suse.cz>
19711         PR tree-optimization/55789
19712         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
19714 2013-02-05  Jan Hubicka  <jh@suse.cz>
19716         PR tree-optimization/55789
19717         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
19718         the dead call anyway.
19720 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
19722         PR sanitizer/55374
19723         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
19725 2013-02-04  Alexander Potapenko  <glider@google.com>
19726             Jack Howarth  <howarth@bromo.med.uc.edu>
19727             Jakub Jelinek  <jakub@redhat.com>
19729         PR sanitizer/55617
19730         * config/darwin.c (sort_ctor_records): Stabilized qsort
19731         on constructor priority by using original position.
19732         (finalize_ctors): New routine to sort constructors by
19733         priority before use in assemble_integer.
19734         (machopic_asm_out_constructor): Use finalize_ctors if needed.
19736 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
19738         PR libstdc++/54314
19739         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
19740         about visibility on artificial decls.
19741         * config/sol2.c (solaris_assemble_visibility): Likewise.
19743 2013-02-04  Kai Tietz  <ktietz@redhat.com>
19745         PR target/56186
19746         * config/i386/i386.c (function_value_ms_64): Add additional valtype
19747         argument and improve checking of return-argument types for 16-byte
19748         modes.
19749         (ix86_function_value_1): Add additional valtype argument on call
19750         of function_value_64.
19751         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
19752         handling infunction_value_64 function.
19754 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
19756         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
19758 2013-02-04  Richard Biener  <rguenther@suse.de>
19760         PR tree-optimization/56188
19761         * tree-ssa-structalias.c (label_visit): Consider case with
19762         initially non-empty points-to set.
19763         (perform_var_substitution): Dump node mapping and clean up.
19765 2013-02-04  Richard Guenther  <rguenther@suse.de>
19767         PR lto/56168
19768         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
19769         node prevail as last resort.
19770         (lto_symtab_merge_decls): Remove guard on LTRANS here.
19771         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
19773 2013-02-04  Richard Biener  <rguenther@suse.de>
19775         PR tree-optimization/56113
19776         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
19777         Merge into ...
19778         (equiv_class_lookup_or_add): ... this.
19779         (label_visit): Adjust and fix error in previous patch.
19780         (perform_var_substitution): Adjust.
19782 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
19784         * config/sh/divtab.c: Fix formatting and comments throughout the file.
19785         * config/sh/sh4-300.md: Likewise.
19786         * config/sh/sh4a.md: Likewise.
19787         * config/sh/constraints.md: Likewise.
19788         * config/sh/sh.md: Likewise.
19789         * config/sh/netbsd-elf.h: Likewise.
19790         * config/sh/predicates.md: Likewise.
19791         * config/sh/sh-protos.h: Likewise.
19792         * config/sh/ushmedia.h: Likewise.
19793         * config/sh/linux.h: Likewise.
19794         * config/sh/sh.c: Likewise.
19795         * config/sh/superh.h: Likewise.
19796         * config/sh/elf.h: Likewise.
19797         * config/sh/sh4.md: Likewise.
19798         * config/sh/sh.h: Likewise.
19800 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19802         * config/pa/constraints.md: Adjust unused letters.  Change "T"
19803         constraint to match_test floating_point_store_memory_operand().
19804         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
19805         (base14_operand): New.
19806         (floating_point_store_memory_operand): New.
19807         (integer_store_memory_operand): Revise to use base14_operand and
19808         reg_plus_base_memory_operand.
19809         (move_dest_operand): Allow symbolic_memory_operands.
19810         (symbolic_memory_operand): Check for LO_SOM.
19811         (symbolic_operand): Change default case to break.
19812         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
19813         CONST_DOUBLE values to be reloaded by putting them into memory when
19814         the destination is a floating point register.
19815         (movdf): Remove code to handle CONST_DOUBLE.
19816         (movsf): Likewise.
19817         (reload_indf_r1): New.
19818         (reload_insf_r1): New.
19819         Consistently use "Q" and "T" constraints with integer and floating
19820         point move instructions, respectively.
19821         (movdi): Remove FAIL.
19822         Change predicate for source operand unamed DImode move from
19823         general_operand to move_src_operand.
19824         (umulsidi3): Change predicate for destination operand to
19825         register_operand.
19826         Likewise for similar unamed patterns.
19827         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
19828         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
19829         (hppa_legitimize_address): Simplify mask calculation.
19830         (pa_emit_move_sequence): Revised handling of secondary reloads from
19831         REG+D addresses for floating point loads and stores.  Directly handle
19832         loading CONST0_RTX (mode) to a floating point register.
19833         (pa_secondary_reload): Handle reloading DF and SFmode constant values
19834         to floating point registers.  Don't restrict secondary reloads to
19835         floating point registers to integer modes.  Revise some comments and
19836         cleanup some code.
19837         (TARGET_LEGITIMATE_ADDRESS_P): Define.
19838         (pa_legitimate_address_p): New.
19839         (pa_legitimize_reload_address): New.
19840         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
19841         (STRICT_REG_OK_FOR_BASE_P): New.
19842         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
19843         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
19845 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
19846             Andrew Dixie  <andrewd@gentrack.com>
19848         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
19849         flag set.
19851 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
19853         * expmed.c (extract_bit_field_1): Pass the full width of the
19854         structure to get_best_reg_extraction_insn.
19856 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19858         PR target/54601
19859         * configure.ac (use_cxa_atexit): Add AIX.
19860         * configure: Regenerate.
19862         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
19864 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
19866         PR debug/54793
19867         * final.c (need_profile_function): New variable.
19868         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
19869         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
19870         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
19871         notes, targetm.asm_out.function_prologue doesn't emit anything,
19872         HAVE_prologue and profiler should be emitted before prologue,
19873         set need_profile_function instead of emitting it.
19874         (final_scan_insn): If need_profile_function, emit
19875         profile_function on the first NOTE_INSN_BASIC_BLOCK or
19876         NOTE_INSN_FUNCTION_BEG note.
19878 2013-02-01  Richard Henderson  <rth@redhat.com>
19880         * config/rs6000/rs6000.md (smulditi3): New.
19881         (umulditi3): New.
19883         * config/alpha/alpha.md (umulditi3): New.
19885 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
19887         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
19888         (ASM_OUTPUT_ALIGNED_LOCAL): New.
19890 2013-02-01  Richard Biener  <rguenther@suse.de>
19892         PR tree-optimization/56113
19893         * tree-ssa-structalias.c (label_visit): Reduce work for
19894         single-predecessor nodes.
19896 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
19898         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
19899         range isn't testing for zero.
19901 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
19903         PR middle-end/56113
19904         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
19906 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
19907             Nick Clifton  <nickc@redhat.com>
19909         * config/v850/constraints.md (Q): Define as a memory constraint.
19910         * config/v850/predicates.md (label_ref_operand): New predicate.
19911         (e3v5_shift_operand): New predicate.
19912         (ior_operator): New predicate.
19913         * config/v850/t-v850: Add e3v5 multilib.
19914         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
19915         (v850_gen_movdi): Prototype.
19916         * config/v850/v850.c: Add support for e3v5 architecture.
19917         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
19918         TARGET_V850E_UP.
19919         (construct_save_jarl): Add e3v5 long JARL support.
19920         (v850_adjust_insn_length): New function.  Adjust length of call
19921         insns when using e3v5 instructions.
19922         (v850_gen_movdi): New function: Generate instructions to move a
19923         DImode value.
19924         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
19925         (CPP_SPEC): Define __v850e3v5__ as appropriate.
19926         (TARGET_USE_FPU): Enable for e3v5.
19927         (CONST_OK_FOR_W): New macro.
19928         (ADJUST_INSN_LENGTH): Define.
19929         * config/v850/v850.md (UNSPEC_LOOP): Define.
19930         (attr cpu): Add v850e3v5.
19931         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
19932         (movdi): New pattern.
19933         (movdi_internal): New pattern.
19934         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
19935         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
19936         (cstoresf4): Likewise.
19937         (cstoredf4): Likewise.
19938         (insv): New pattern.
19939         (rotlso3_a): New pattern.
19940         (rotlsi3_b): New pattern
19941         (rotlsi3_v850e3v5): New pattern.
19942         (doloop_begin): New pattern.
19943         (fix_loop_counter): New pattern.
19944         (doloop_end): New pattern.
19945         (branch_normal): Add e3v5 long branch support.
19946         (branch_invert): Likewise.
19947         (branch_z_normal): Likewise.
19948         (branch_z_invert): Likewise.
19949         (branch_nz_normal): Likewise.
19950         (branch_nz_invert): Likewise.
19951         (call_internal_short): Add e3v5 register-indirect JARL support.
19952         (call_internal_long): Likewise.
19953         (call_value_internal_short): Likewise.
19954         (call_value_internal_long): Likewise.
19955         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
19956         (mloop): New option.
19957         * config.gcc: Add support for configuring v840e3v5 target.
19958         * doc/invoke.texi: Document new v850 specific command line options.
19960 2013-01-31  Paul Koning  <ni1d@arrl.net>
19962         PR debug/55059
19963         PR debug/54508
19964         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
19965         children if parent is a class.
19966         (prune_unused_types_prune): Don't add DW_AT_declaration.
19968 2013-01-31  Richard Biener  <rguenther@suse.de>
19970         PR tree-optimization/56157
19971         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
19972         match up operand with SLP child.
19974 2013-01-31  Jason Merrill  <jason@redhat.com>
19976         PR debug/54410
19977         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
19978         parameters the first time.
19979         (gen_scheduled_generic_parms_dies): Check completeness here.
19981 2013-01-31  Richard Biener  <rguenther@suse.de>
19983         PR middle-end/53073
19984         * common.opt (faggressive-loop-optimizations): New flag,
19985         enabled by default.
19986         * doc/invoke.texi (faggressive-loop-optimizations): Document.
19987         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
19988         infer_loop_bounds_from_undefined by it.
19990 2013-01-31  Richard Biener  <rguenther@suse.de>
19992         PR tree-optimization/56150
19993         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
19994         visit virtual operands.
19995         (find_uses_to_rename_bb): Likewise.
19997 2013-01-31  Richard Biener  <rguenther@suse.de>
19999         PR tree-optimization/56150
20000         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
20001         mixed store non-store stmts.
20003 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
20005         PR sanitizer/55374
20006         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
20007         LIBASAN_EARLY_SPEC is defined.
20008         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
20009         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
20010         before %o.
20011         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
20013         PR c++/55742
20014         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20015         invalid args instead of ICEing on it.
20016         (ix86_valid_target_attribute_tree): Return error_mark_node if
20017         ix86_valid_target_attribute_inner_p failed.
20018         (ix86_valid_target_attribute_p): Return false only if
20019         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
20020         target("default") attribute.
20021         (sorted_attr_string): Change argument from const char * to tree,
20022         merge in all target attribute arguments rather than just one.
20023         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
20024         instead of free.  Avoid using strcat.
20025         (ix86_mangle_function_version_assembler_name): Mangle
20026         target("default") as if no target attribute is present.  Adjust
20027         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
20028         instead of xmalloc and XDELETEVEC instead of free.
20029         (ix86_function_versions): Don't return true if one of the decls
20030         doesn't have target attribute.  If they don't and one of the decls
20031         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
20032         sorted_attr_string caller.  Use XDELETEVEC instead of free.
20033         (ix86_supports_function_versions): Remove.
20034         (make_name): Fix up formatting.
20035         (make_dispatcher_decl): Remove resolver_name and its initialization.
20036         Avoid leaking memory.
20037         (is_function_default_version): Return true if there is
20038         target("default") attribute rather than no target attribute at all.
20039         (make_resolver_func): Avoid leaking memory.
20040         (ix86_generate_version_dispatcher_body): Likewise.
20041         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20042         * target.def (supports_function_versions): Remove.
20043         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20044         * doc/tm.texi: Regenerated.
20046 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20048         PR rtl-optimization/56144
20049         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20050         for values with side effects.
20052 2013-01-30  Richard Biener  <rguenther@suse.de>
20054         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20055         (sparseset_pop): Likewise.
20056         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20057         to be able to use quick_push in the worker loop.
20059 2013-01-30  Marek Polacek  <polacek@redhat.com>
20061         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20063 2013-01-30  Richard Biener  <rguenther@suse.de>
20065         PR lto/56147
20066         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20068 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20070         PR tree-optimization/56064
20071         * fixed-value.c (fixed_from_double_int): New function.
20072         * fixed-value.h (fixed_from_double_int): New prototype.
20073         (const_fixed_from_double_int): New static inline function.
20074         * fold-const.c (native_interpret_fixed): New static function.
20075         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20076         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20077         (native_encode_fixed): New static function.
20078         (native_encode_expr) <FIXED_CST>: Use it.
20079         (native_interpret_int): Move double_int worker code to...
20080         * double-int.c (double_int::from_buffer): ...this new static method.
20081         * double-int.h (double_int::from_buffer): Prototype it.
20083 2013-01-30  Richard Biener  <rguenther@suse.de>
20085         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20086         New pointer-map and obstack.
20087         (init_alias_vars): Allocate pointer-map and obstack.
20088         (delete_points_to_sets): Free them.
20089         (find_what_var_points_to): Cache result.
20090         (find_what_p_points_to): Adjust for changed interface of
20091         find_what_var_points_to.
20092         (compute_points_to_sets): Likewise.
20093         (ipa_pta_execute): Likewise.
20095 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20097         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20098         * configure: Regenerate.
20099         * config.in: Regenerate.
20100         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20101         #nobits/#progbits if supported.
20103 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20105         PR target/56121
20106         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20107         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20108         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20110 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20112         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20113         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20115 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20117         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20118         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20120 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20122         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20123         declaration.
20124         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20125         * config/arm/cortex-a7.md: New bypasses using
20126         arm_mac_accumulator_is_result.
20128 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20130         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
20131         (cortex_a7_neon_mla): Likewise.
20132         (cortex_a7_fpfmad): New reservation.
20133         (cortex_a7_fpmacs): Use ffmas and update required units.
20134         (cortex_a7_fpmuld): Update required units and latency.
20135         (cortex_a7_fpmacd): Likewise.
20136         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20137         (cortex_a7_neon). Likewise.
20138         (bypass) Update participating units.
20140 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20142         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20143         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20144         from fmac to ffma.
20145         * config/arm/vfp11.md (vfp_farith): Use ffmas.
20146         (vfp_fmul): Use ffmad.
20147         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20148         (cortex_r4_fmacd): Use ffmad.
20149         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20150         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
20151         (cortex_a9_fmacd): Use ffmad.
20152         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20153         (cortex_a8_vfp_macd): Use ffmad.
20154         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20155         (cortex_a5_fpmacd): Use ffmad.
20156         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20157         (cortex_a15_vfp_macd): Use ffmad.
20158         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20160 2013-01-29  Jason Merrill  <jason@redhat.com>
20162         PR libstdc++/54314
20163         * varasm.c (default_assemble_visibility): Don't warn about
20164         visibility on artificial decls.
20166 2013-01-29  Richard Biener  <rguenther@suse.de>
20168         PR tree-optimization/56113
20169         * tree-ssa-structalias.c (equiv_class_lookup): Also return
20170         the bitmap leader.
20171         (label_visit): Free duplicate bitmaps and record the leader instead.
20172         (perform_var_substitution): Adjust.
20174 2013-01-29  Richard Biener  <rguenther@suse.de>
20176         PR tree-optimization/55270
20177         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20178         the CFG, schedule loops for fixup.
20180 2013-01-29  Nick Clifton  <nickc@redhat.com>
20182         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20183         SP_REG.
20185 2013-01-28  Leif Ekblad  <leif@rdos.net>
20187         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20188         * config/i386/i386.h (TARGET_RDOS): New macro.
20189         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20190         * config/i386/i386.c (ix86_option_override_internal): For 64bit
20191         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20192         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20193         DEFAULT_LARGE_SECTION_THRESHOLD.
20194         * config/i386/i386.md (R14_REG, R15_REG): New constants.
20195         * config/i386/rdos.h: New file.
20196         * config/i386/rdos64.h: New file.
20198 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
20200         PR other/54814
20201         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20202         TEST_HARD_REG_BIT.
20204 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20206         PR rtl-optimization/56117
20207         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20208         call cselib_lookup_from_insn on the MEM before calling
20209         add_insn_mem_dependence.
20211 2013-01-28  Richard Biener  <rguenther@suse.de>
20213         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20214         to a stmt that didn't have one.
20215         (copy_phis_for_bb): Likewise for PHI arguments.
20216         (copy_debug_stmt): Likewise for debug stmts.
20218 2013-01-28  Richard Biener  <rguenther@suse.de>
20220         PR tree-optimization/56034
20221         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20222         (partition_builtin_p): Adjust.
20223         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
20224         it is the last partition.
20225         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20226         up the vertex for the definition.
20227         (classify_partition): Classify whether a partition is a
20228         PKIND_REDUCTION, thus has uses outside of the loop.
20229         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20230         Merge all PKIND_REDUCTION partitions into the last partition.
20231         (tree_loop_distribution): Seed partitions from reductions as well.
20233 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20235         PR tree-optimization/56125
20236         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
20237         pow(x,c) into sqrt(x) * powi(x, n/2) or
20238         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
20239         optimizing for size.
20240         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
20241         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
20242         integer.
20244         PR tree-optimization/56094
20245         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
20246         to UNKNOWN_LOCATION while gimplifying expr.
20248 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
20250         PR target/56114
20251         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
20252         operand 0 in movabs insn template for -masm=intel asm alternative.
20253         (*movabs<mode>_2): Ditto for operand 1.
20255 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
20257         PR target/54663
20258         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
20259         of microblaze-c.o
20261 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20263         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
20264         tm_file.
20266 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20268         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
20269         Undef to avoid warning.
20271 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20273         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
20274         * configure: Regenerate.
20276 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
20278         PR tree-optimization/56098
20279         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
20280         for stmts with volatile ops.
20281         (cond_store_replacement): Don't optimize if assign has volatile ops.
20282         (cond_if_else_store_replacement_1): Don't optimize if either
20283         then_assign or else_assign have volatile ops.
20284         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
20285         volatile ops.
20287 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20289         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
20291 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20293         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
20294         missing ':' in asm example.
20296 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
20298         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
20299         entries into lane and laneq entries.
20300         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
20301         Remove AdvSIMD scalar modes.
20302         (aarch64_sq<r>dmulh_laneq<mode>): New.
20303         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
20304         modes.
20305         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
20306         builtin implementations to relfect changes in RTL in aarch64-simd.md.
20307         * config/aarch64/iterators.md (VCOND): New.
20308         (VCONQ): New.
20310 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20312         PR target/54222
20313         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
20314         Add NULL LIBNAME argument to existing definitions.
20315         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
20316         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
20317         * config/avr/avr.c (DEF_BUILTIN): Same.
20318         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
20319         (avr_expand_builtin): Expand to a vanilla call if a libgcc
20320         implementation is available (DECL_ASSEMBLER_NAME is set).
20321         (avr_fold_absfx): New static function.
20322         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
20323         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
20324         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
20325         AVR_BUILTIN_ABSLLK.
20326         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
20327         (abshk, absk, abslk, absllk): Provide as static inline functions.
20329 2013-01-25  Marek Polacek  <polacek@redhat.com>
20331         PR tree-optimization/56035
20332         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
20334 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
20336         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
20337         (*movtf_internal_rex64): Add (!o,C) alternative
20338         (*movxf_internal_rex64): Ditto.
20339         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
20341 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
20343         * doc/invoke.texi: fix typo.
20344         * doc/objc.texi: fix typo.
20346 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
20348         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
20349         for the first two alternatives.
20351 2013-01-24  Diego Novillo  <dnovillo@google.com>
20353         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
20354         (ggc-zone.o): Remove.
20355         * configure.ac: Remove option --with-gc.
20356         * configure: Re-generate.
20357         * doc/install.texi: Remove documentation for --with-gc.
20358         * gengtype.c (write_enum_defn): Remove.  Update all users.
20359         (write_Types_process_field): Remove generation of gt_e_* argument.
20360         (output_type_enum): Remove.  Update all users.
20361         (write_enum_defn): Remove.  Update all users.
20362         (enum alloc_zone): Remove.  Update all users.
20363         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
20364         * ggc-common.c (ggc_splay_alloc): Remove first argument.
20365         Update all callers.
20366         (struct ptr_data): Remove field TYPE.  Update all users.
20367         (gt_pch_note_object): Remove argument TYPE.  Update all users.
20368         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
20369         Update all users.
20370         * ggc-none.c (ggc_alloc_typed_stat): Remove.
20371         (struct alloc_zone): Remove.
20372         (ggc_internal_alloc_zone_stat): Remove.
20373         (ggc_internal_cleared_alloc_zone_stat): Remove.
20374         * ggc-page.c (ggc_alloc_typed_stat): Remove.
20375         (ggc_pch_count_object): Remove last argument.  Update all users.
20376         (ggc_pch_alloc_object): Remove last argument.  Update all users.
20377         (struct alloc_zone): Remove.
20378         * ggc-zone.c: Remove.
20379         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
20380         (struct alloc_zone): Remove.
20381         (ggc_alloc_typed_stat): Remove.
20382         (ggc_alloc_typed): Remove.
20383         (ggc_splay_alloc): Remove first argument.
20384         (rtl_zone): Remove.  Update all users.
20385         (tree_zone): Remove.  Update all users.
20386         (tree_id_zone): Remove.  Update all users.
20387         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
20388         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
20389         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
20390         * tree-ssanames.c: Remove references to zone allocator in comments.
20392 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
20394         * config/avr/avr.c (avr_out_fract): Make register numbers that
20395         might be outside of source operand signed.
20397 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
20399         * config/i386/constraints.md (Yf): New constraint.
20400         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
20401         of f constraint to conditionaly disable x87 register preferences.
20402         (*movdf_internal): Ditto.
20403         (*movsf_internal): Ditto.
20405 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
20407         PR inline-asm/55934
20408         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
20409         that have operands with impossible constraints.
20410         Add a FIXME for a speed-up opportunity.
20411         * lra-constraints.c (process_alt_operands): Verify that a class
20412         selected from constraints on asms is valid for the operand mode.
20413         (curr_insn_transform): Remove incorrect comment.
20415 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
20417         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
20418         TOC operand is a valid symbol ref in the constant pool.
20420 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20422         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
20424 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
20426         PR target/54222
20427         * config/avr/stdfix.h: New file.
20428         * t-avr (stdfix-gcc.h): New rule to build it.
20429         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
20431 2013-01-23  Kostya Serebryany  <kcc@google.com>
20433         * config/darwin.h: remove dependency on
20434         CoreFoundation (asan on Mac OS).
20436 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
20438         PR target/49069
20439         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
20440         instead of cmpdi_operand for first comparison operand.
20441         Don't assert that comparison operands aren't both constants.
20443 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20445         * doc/install.texi (Downloading the Source): Update references to
20446         downloading separate components.
20448 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20450         * doc/extend.texi (__int128): Improve grammar.
20452 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
20454         PR target/56028
20455         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
20456         alternative to (o,r).
20457         (*movdi_internal_rex64): Remove (!o,n) alternative.
20458         (DImode immediate->memory splitter): Remove.
20459         (DImode immediate->memory peephole2): Remove.
20460         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
20461         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
20462         alternative to (!o,*r).
20463         (*movtf_internal_sse): New pattern.
20464         (*movxf_internal_rex64): New pattern.
20465         (*movxf_internal): Disable for TARGET_64BIT.
20466         (*movdf_internal_rex64): Remove (!o,F) alternative.
20468 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
20470         PR middle-end/56074
20471         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
20472         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
20473         * tree-vect-loop-manip.c (find_loop_location): Also ignore
20474         stmt locations where LOCATION_LOCUS of the stmt location is
20475         UNKNOWN_LOCATION or BUILTINS_LOCATION.
20477         PR target/55686
20478         * config/i386/i386.md (UNSPEC_STOS): New.
20479         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
20480         *strsetqi_1): Add UNSPEC_STOS.
20482 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
20484         PR c++/56067
20485         * doc/invoke.texi: Remove left over -Wsynth example.
20487 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20489         PR tree-optimization/56051
20490         * fold-const.c (fold_binary_loc): Don't fold
20491         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
20492         a narrowing conversion, or widening conversion from signed
20493         to unsigned.
20495 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20497         PR rtl-optimization/56023
20498         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
20499         dependent on debug instruction.
20501 2013-01-21  Martin Jambor  <mjambor@suse.cz>
20503         PR middle-end/56022
20504         * function.c (allocate_struct_function): Call
20505         invoke_set_current_function_hook earlier.
20507 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20509         * reload1.c (init_reload): Only initialize reload_obstack
20510         during the first call.
20512 2013-01-21  Marek Polacek  <polacek@redhat.com>
20514         * cfgloop.c (verify_loop_structure): Fix up grammar.
20516 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
20518         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
20519         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
20521 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20523         PR target/56058
20524         * config/arm/marvell-pj4.md: Update copyright year.
20525         Fix up use of alu to alu_reg and simple_alu_imm.
20527 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20529         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
20531 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
20533         PR target/55433
20534         * lra-constraints.c (curr_insn_transform): Don't reuse original
20535         insn for secondary memory move when memory mode should be different.
20537 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20539         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
20540         atomic_storedi_1): New patterns.
20542 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
20544         btver2 pipeline descriptions.
20545         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
20546         descriptions.
20547         * config/i386/i386.md (btver2_decode): New type attributes.
20548         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
20549         type attributes.
20550         * config/i386/btver2.md: New file describing btver2 pipelines.
20552 2013-01-19  Andrew Pinski  <apinski@cavium.com>
20554         PR tree-optimization/52631
20555         * tree-ssa-sccvn (visit_use): Before looking up the original
20556         statement, try looking up the simplified expression.
20558 2013-01-19  Anthony Green  <green@moxielogic.com>
20560         * config/moxie/moxie.c (moxie_expand_prologue): Set
20561         current_function_static_stack_size.
20563 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20565         PR tree-optimization/56029
20566         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
20567         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
20569 2013-01-18  Sharad Singhai  <singhai@google.com>
20571         PR tree-optimization/55995
20572         * dumpfile.c (dump_loc): Print location only if available.
20573         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
20575 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
20577         PR target/55433
20578         * lra-constraints.c (curr_insn_transform): Reuse original insn for
20579         secondary memory move.
20580         (inherit_reload_reg): Use rclass instead of cl for
20581         check_secondary_memory_needed_p.
20583 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20585         PR middle-end/56015
20586         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
20587         the case where writing real complex part of target modifies op1.
20589 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
20591         * config/aarch64/aarch64-simd.md
20592         (aarch64_vcond_internal<mode>): Handle unordered cases.
20593         * config/aarch64/iterators.md (v_cmp_result): New.
20595 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
20596             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20598         * config/arm/marvell-pj4.md: New file.
20599         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
20600         * config/arm/arm.md (generic_sched): Add marvell_pj4.
20601         (generic_vfp): Likewise.
20602         * config/arm/arm-cores.def: Add marvell-pj4.
20603         * config/arm/arm-tune.md: Regenerate.
20604         * config/arm/arm-tables.opt: Regenerate.
20605         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
20606         * doc/invoke.texi: Document marvell-pj4.
20608 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
20610         * config/aarch64/arm_neon.h: Map scalar types to standard types.
20612 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20614         PR debug/54114
20615         PR debug/54402
20616         PR debug/49888
20617         * var-tracking.c (negative_power_of_two_p): New.
20618         (global_get_addr_cache, local_get_addr_cache): New.
20619         (get_addr_from_global_cache, get_addr_from_local_cache): New.
20620         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
20621         heading comment.
20622         (vt_stack_offset_p): Remove.
20623         (vt_canon_true_dep): Always canonicalize loc's address.
20624         (clobber_overlapping_mems): Make sure we have a MEM.
20625         (local_get_addr_clear_given_value): New.
20626         (val_reset): Clear local cached entries.
20627         (compute_bb_dataflow): Create and release the local cache.
20628         Disable duplicate MEMs clobbering.
20629         (emit_notes_in_bb): Clobber MEMs likewise.
20630         (vt_emit_notes): Create and release the local cache.
20631         (vt_initialize, vt_finalize): Create and release the global
20632         cache, respectively.
20633         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
20635 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20637         PR libmudflap/53359
20638         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
20639         not found in the symtab.
20641 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
20643         PR debug/56006
20644         PR rtl-optimization/55547
20645         PR rtl-optimization/53827
20646         PR debug/53671
20647         PR debug/49888
20648         * alias.c (offset_overlap_p): New, factored out of...
20649         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
20650         the conservative special case for symbolic constants.  Don't
20651         adjust zero sizes on alignment.
20653 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
20655         PR rtl-optimization/52573
20656         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
20657         REG_UNUSED for the same register.
20659 2013-01-17  Richard Biener  <rguenther@suse.de>
20660             Marek Polacek  <polacek@redhat.com>
20662         PR rtl-optimization/55833
20663         * loop-unswitch.c (unswitch_loops): Move loop verification...
20664         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
20665         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
20666         Set it to true when we're removing a loop from hierarchy tree in
20667         an irreducible region.
20668         (fix_bb_placements): Adjust caller.
20669         (fix_loop_placements): Likewise.
20671 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
20673         * config/avr/builtins.def (DEF_BUILTIN): Factor out
20674         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
20675         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
20676         Remove ID.  Adjust comments.
20677         * config/avr/avr-c.c (avr_builtin_name): Remove.
20678         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
20679         * config/avr/avr.c (avr_tolower): New static function.
20680         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
20681         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
20682         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
20683         default expansion.
20685 2013-01-17  Jan Hubicka  <jh@suse.cz>
20687         PR tree-optimization/55273
20688         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
20690 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
20692         PR target/55981
20693         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
20694         store through atomic_store<mode>_1.
20695         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
20697 2013-01-17  Martin Jambor  <mjambor@suse.cz>
20699         PR tree-optimizations/55264
20700         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
20701         for virtual methods.
20702         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
20703         virtual methods before inlining is over.
20704         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
20705         virtual functions.
20706         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
20707         non-virtual.
20709 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
20711         PR rtl-optimization/56005
20712         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
20713         pending reads for prefetch.
20715 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
20717         * config/aarch64/aarch64.md
20718         (*cstoresi_neg_uxtw): New pattern.
20719         (*cmovsi_insn_uxtw): New pattern.
20720         (*<optab>si3_uxtw): New pattern.
20721         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
20722         (*<optab>si3_insn_uxtw): New pattern.
20723         (*bswapsi2_uxtw): New pattern.
20725 2013-01-16  Richard Biener  <rguenther@suse.de>
20727         * tree-inline.c (tree_function_versioning): Remove set but
20728         never used variable.
20730 2013-01-16  Richard Biener  <rguenther@suse.de>
20732         PR tree-optimization/55964
20733         * tree-flow.h (rename_variables_in_loop): Remove.
20734         (rename_variables_in_bb): Likewise.
20735         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
20736         (copy_loop_before): Adjust and delete update-ssa status.
20737         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
20738         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
20739         (rename_variables_in_loop): Remove.
20740         (slpeel_update_phis_for_duplicate_loop): Likewise.
20741         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
20742         use available cfg machinery instead of duplicating it.
20743         Update PHI nodes and perform poor-mans SSA update here.
20744         (slpeel_tree_peel_loop_to_edge): Adjust.
20746 2013-01-16  Richard Biener  <rguenther@suse.de>
20748         PR tree-optimization/54767
20749         PR tree-optimization/53465
20750         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
20751         (vrp_visit_phi_node): For PHI arguments coming via backedges
20752         drop all symbolical range information.
20753         (execute_vrp): Compute backedges.
20755 2013-01-16  Richard Biener  <rguenther@suse.de>
20757         * doc/install.texi: Update CLooG and ISL requirements to
20758         0.18.0 and 0.11.1.
20760 2013-01-16  Christian Bruel  <christian.bruel@st.com>
20762         PR target/55301
20763         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
20764         (broken_move): Handle UNSPECV_SP_SWITCH_B.
20765         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
20767 2013-01-16  DJ Delorie  <dj@redhat.com>
20769         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
20770         (UNSPECV_SP_SWITCH_E): New.
20771         (sp_switch_1): Change to an unspec.
20772         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
20773         replace $r15.
20775 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
20777         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
20778         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
20779         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
20780         (expand_mem_thread_fence): Ditto.
20781         (expand_mem_signal_fence): Ditto.
20782         (expand_atomic_load): Ditto.
20783         (expand_atomic_store): Ditto.
20785 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
20787         PR rtl-optimization/55547
20788         PR rtl-optimization/53827
20789         PR debug/53671
20790         PR debug/49888
20791         * alias.c (memrefs_conflict_p): Set sizes to negative after
20792         AND adjustments.
20794 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20796         PR target/55940
20797         * function.c (thread_prologue_and_epilogue_insns): Always
20798         add crtl->drap_reg to set_up_by_prologue.set, even if
20799         stack_realign_drap is false.
20801 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
20803         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
20804         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
20805         *call): Fix indention.
20807 2013-01-15  Tom de Vries  <tom@codesourcery.com>
20809         PR target/55876
20810         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
20811         Update comment.
20813 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
20815         PR rtl-optimization/55153
20816         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
20818 2013-01-15  Martin Jambor  <mjambor@suse.cz>
20820         PR tree-optimization/55920
20821         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
20822         accesses as grp_to_be_debug_replaced.
20824 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20826         PR tree-optimization/55920
20827         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
20828         there is non-useless type conversion needed from debug rhs to lhs,
20829         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
20831 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
20832             Mikael Pettersson  <mikpe@it.uu.se>
20834         PR target/43961
20835         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
20836         Thumb.
20837         (ASM_OUTPUT_CASE_LABEL): Remove.
20838         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
20839         * final.c (shorten_branches): Update alignment of labels before
20840         jump tables if CASE_VECTOR_SHORTEN_MODE.
20842 2013-01-15  Richard Biener  <rguenther@suse.de>
20844         PR bootstrap/55961
20845         * system.h: Do not include gmp.h for building host tools.
20847 2013-01-15  Richard Biener  <rguenther@suse.de>
20849         PR middle-end/55882
20850         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
20851         account for bitpos when computing alignment.
20853 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
20855         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
20856         (ix86_target_macros_internal): Likewise.
20858         * config/i386/i386.c (m_CORE2I7): Removed.
20859         (m_CORE_HASWELL): New macro.
20860         (m_CORE_ALL): Likewise.
20861         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
20862         (initial_ix86_arch_features): Likewise.
20863         (processor_target_table): Initializations for Core avx2.
20864         (cpu_names): New names "core-avx2".
20865         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
20866         PROCESSOR_CORE_HASWELL.
20867         (ix86_issue_rate): New case.
20868         (ia32_multipass_dfa_lookahead): Likewise.
20869         (ix86_sched_init_global): Likewise.
20871         * config/i386/i386.h (TARGET_HASWELL): New macro.
20872         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
20873         (processor_type): New PROCESSOR_HASWELL.
20875 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
20877         PR tree-optimization/55955
20878         * tree-vect-loop.c (vectorizable_reduction): Give up early on
20879         *SHIFT_EXPR and *ROTATE_EXPR codes.
20881         PR tree-optimization/48766
20882         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
20883         -ftrapv disable -fwrapv.
20885 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20887         PR target/55974
20888         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
20889         etc. to 1 and not to __flash.
20890         Use LL suffix for __INT24_MAX__ with -mint8.
20891         Use ULL suffix for __UINT24_MAX__ with -mint8.
20893 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20895         * config/avr/avr-arch.h
20896         (struct base_arch_s): Use typedef avr_arch_t instead.
20897         (struct arch_info_s): Use typedef avr_arch_info_t instead.
20898         (struct mcu_type_s): Use typedef avr_mcu_t instead.
20899         * config/avr/avr.c: Same.
20900         * config/avr/avr-devices.c: Same.
20901         * config/avr/driver-avr.c: Same.
20902         * config/avr/gen-avr-mmcu-texi.c: Same.
20903         * config/avr/avr-mcus.def: Adjust comment.
20905 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
20907         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
20908         * config/aarch64/iterators.md (VALLDI): New.
20910 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
20911             Andi Kleen  <ak@linux.intel.com>
20913         PR target/55948
20914         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
20915         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
20916         memmodel flag.
20918 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20920         * config/avr/avr-stdint.h: Remove trailing blanks.
20921         * config/avr/avr-log.h: Same.
20922         * config/avr/avr-arch.h: Same.
20923         * config/avr/avr-devices.c: Same.
20924         * config/avr/avr-dimode.md: Same.
20925         * config/avr/predicates.md: Same.
20926         * config/avr/avr-c.c: Same.  And fix typo.
20928         * config/avr/avr-protos.h: Same.  And:
20929         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
20930         (init_cumulative_args): Rename to avr_init_cumulative_args.
20931         (expand_prologue): Rename to avr_expand_prologue.
20932         (expand_epilogue): Rename to avr_expand_epilogue.
20933         (adjust_insn_length): Rename to avr_adjust_insn_length.
20934         (notice_update_cc): Rename to avr_notice_update_cc.
20935         (final_prescan_insn): Rename to avr_final_prescan_insn.
20936         * config/avr/avr.c: Same.
20937         * config/avr/avr.h: Same.
20938         * config/avr/avr.md: Remove trailing blanks.
20939         (prologue): Use avr_expand_prologue.
20940         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
20942 2013-01-14  Richard Biener  <rguenther@suse.de>
20944         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
20945         verify_location, collect_subblocks): New functions.
20946         (verify_gimple_in_cfg): Verify that locations only reference
20947         BLOCKs in the functions BLOCK tree.
20949 2013-01-14  Richard Biener  <rguenther@suse.de>
20951         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
20952         PHI argument.
20953         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
20954         unshare reference.
20955         (insert_out_of_ssa_copy_on_edge): Likewise.
20956         (rewrite_close_phi_out_of_ssa): Likewise.
20957         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
20958         debug expressions.
20959         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
20960         propagated constants.
20961         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
20962         can not be shared.
20964 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
20966         * config/avr/avr-modes.def: Add GPL copyright notice.
20968 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
20970         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
20971         MEMMODEL_MASK to determine memory model.
20972         (atomic_store<mode>): Ditto from operands[2].
20973         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
20975 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
20977         PR fortran/55935
20978         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
20979         (fold_gimple_assign): Don't call unshare_expr here.
20980         (fold_ctor_reference): Call unshare_expr.
20982 2013-01-13  Terry Guo  <terry.guo@arm.com>
20984         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
20985         * doc/fragments.texi: Document MULTILIB_REUSE.
20986         * gcc.c (multilib_reuse): New internal spec.
20987         (set_multilib_dir): Also search multilib from multilib_reuse.
20988         * genmultilib (tmpmultilib3): Refactor code.
20989         (tmpmultilib4): Ditto.
20990         (multilib_reuse): New multilib argument.
20992 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
20994         * Makefile.in: Update copyright.
20996 2013-01-12  Tom de Vries  <tom@codesourcery.com>
20998         PR middle-end/55890
20999         * calls.c (expand_call): Check if arg_nr is valid.
21001 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
21003         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
21004         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21005         documentation.  Add missing '__' in front of
21006         __builtin_ia32_packssdw256.
21008 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21010         PR target/55719
21011         * config/s390/s390.c (s390_preferred_reload_class): Do not return
21012         NO_REGS for larl operands.
21013         (s390_reload_larl_operand): Use s390_load_address instead of
21014         emit_move_insn.
21016 2013-01-11  Richard Biener  <rguenther@suse.de>
21018         * tree-cfg.c (verify_node_sharing_1): Split out from ...
21019         (verify_node_sharing): ... here.
21020         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21022 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
21024         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21025         Substitute TREECHECKING.
21026         * configure: Regenerate.
21027         * Makefile.in (TREECHECKING): New.
21029 2013-01-11  Richard Guenther  <rguenther@suse.de>
21031         PR tree-optimization/44061
21032         * tree-vrp.c (extract_range_basic): Compute zero as
21033         value-range for __builtin_constant_p of function parameters.
21035 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
21037         Update copyright years.
21039 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21041         PR rtl-optimization/55672
21042         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21043         const to be eliminable.
21045 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21047         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21048         * configure: Regenerate.
21050 2013-01-10  Richard Biener  <rguenther@suse.de>
21052         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21054 2013-01-10  Richard Biener  <rguenther@suse.de>
21056         PR bootstrap/55792
21057         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21058         locations for virtual PHI arguments.
21059         (rewrite_update_phi_arguments): Likewise.
21061 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21063         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21064         on to assembler.
21066 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21068         PR tree-optimization/55921
21069         * tree-complex.c (expand_complex_asm): New function.
21070         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21072 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21074         PR target/55718
21075         * config/s390/s390.c (s390_symref_operand_p)
21076         (s390_loadrelative_operand_p): Merge the two functions.
21077         (s390_check_qrst_address, print_operand_address): Add parameters
21078         to s390_loadrelative_operand_p invokation.
21079         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21080         (s390_reload_larl_operand, s390_secondary_reload): Use
21081         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21082         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21084 2013-01-09  Mike Stump  <mikestump@comcast.net>
21086         * dse.c (record_store): Remove unnecessary assert.
21088 2013-01-09  Jan Hubicka  <jh@suse.cz>
21090         PR tree-optimization/55569
21091         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21092         * cfgloop.h (scale_loop_profile): Likewise.
21094 2013-01-09  Jan Hubicka  <jh@suse.cz>
21096         PR lto/45375
21097         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21098         functions.
21099         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21101 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21103         PR middle-end/55114
21104         * expr.h (maybe_emit_group_store): Declare.
21105         * expr.c (maybe_emit_group_store): New function.
21106         * builtins.c (expand_builtin_int_roundingfn): Call it.
21107         (expand_builtin_int_roundingfn_2): Likewise.
21109 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21111         PR rtl-optimization/55829
21112         * lra-constraints.c (match_reload): Add code for absent output.
21113         (curr_insn_transform): Add code for reloads of matched inputs
21114         without output.
21116 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21118         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21119         attribute of movddup insn to DF.
21120         (*vec_interleave_lowv2df): Ditto.
21121         (vec_dupv2df): Ditto.
21123 2013-01-09  Jan Hubicka  <jh@suse.cz>
21125         PR tree-optimiation/55875
21126         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21127         EVERY_ITERATION parameter.
21128         (number_of_iterations_exit): Check if exit is executed every iteration.
21129         (idx_infer_loop_bounds): Similarly here.
21130         (n_of_executions_at_most): Simplify
21131         to only test for cases where statement is dominated by the
21132         particular bound; handle correctly the "postdominance" test.
21133         (scev_probably_wraps_p): Use max loop iterations info
21134         as a global bound first.
21136 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
21137             Nick Clifton  <nickc@redhat.com>
21139         * config/v850/v850.md (cbranchsf4): New pattern.
21140         (cstoresf4): New pattern.
21141         (cbranchdf4): New pattern.
21142         (cstoredf4): New pattern.
21143         (movsicc): Disallow floating point comparisons.
21144         (cmpsf_le_insn): Fix order of operators.
21145         (cmpsf_lt_insn): Likewise.
21146         (cmpsf_eq_insn): Likewise.
21147         (cmpdf_le_insn): Likewise.
21148         (cmpdf_lt_insn): Likewise.
21149         (cmpdf_eq_insn): Likewise.
21150         (cmpsf_ge_insn): Use LE comparison.
21151         (cmpdf_ge_insn): Likewise.
21152         (cmpsf_gt_insn): Use LT comparison.
21153         (cmpdf_gt_insn): Likewise.
21154         (cmpsf_ne_insn): Delete pattern.
21155         (cmpdf_ne_insn): Delete pattern.
21156         * config/v850/v850.c (v850_gen_float_compare): Use
21157         gen_cmpdf_eq_insn for NE comparison.
21158         (v850_float_z_comparison_operator)
21159         (v850_float_nz_comparison_operator): Move from here ...
21160         * config/v850/predicates.md: ... to here.  Move GT and GE
21161         comparisons into v850_float_z_comparison_operator.
21162         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21163         Delete prototype.
21164         (v850_float_nz_comparison_operator): Likewise.
21166 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21168         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21169         with calls to gen_insvsi/gen_insvdi.
21171 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21173         * config/i386/i386.c (initial_ix86_tune_features): Set up
21174         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21176 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
21177             Jakub Jelinek  <jakub@redhat.com>
21179         PR tree-optimization/48189
21180         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21181         If nitercst is 0, don't predict the exit edge.
21183 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21185         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21186         in asm_fprintf with reg_names.
21187         (aarch64_print_operand_address): Likewise.
21188         (aarch64_return_addr): Likewise.
21189         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21191 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21193         * config/pa/pa.h (VAL_U6_BITS_P): Define.
21194         (INT_U6_BITS): Likewise.
21195         * config/pa/predicates.md (uint6_operand): New predicate.
21196         (shift5_operand, shift6_operand): Likewise.
21197         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21198         arith32_operand.
21199         (lshrdi3): Use shift6_operand.
21200         (shrpsi4, shrpdi4): New insn patterns.
21201         (extzv): Delete expander.
21202         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
21203         predicates in unamed zero extract patterns.  Tighten common constraint.
21204         (extv): Delete expander.
21205         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
21206         predicates in unamed sign extract patterns.  Tighten common constraint.
21207         (insv): Delete expander.
21208         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
21209         predicates in unamed insert patterns.  Tighten common constraint.
21210         Change uint32_operand predicate to uint6_operand predicate in unamed
21211         DImode pattern to insert constant values of type 1...1xxxx.
21213 2013-01-04  Jan Hubicka  <jh@suse.cz>
21215         PR tree-optimization/55823
21216         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21217         issue.
21219 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21220             Uros Bizjak  <ubizjak@gmail.com>
21222         PR rtl-optimization/55845
21223         * df-problems.c (can_move_insns_across): Stop scanning at
21224         volatile_insn_p source instruction or give up if
21225         across_from .. across_to range contains any volatile_insn_p
21226         instructions.
21228 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21230         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21231         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
21232         Declare.
21233         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
21234         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
21236 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21238         PR fortran/55341
21239         * asan.c (asan_clear_shadow): New function.
21240         (asan_emit_stack_protection): Use it.
21242 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21244         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
21245         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
21246         with tab instead of space.
21248 2013-01-08  Nick Clifton  <nickc@redhat.com>
21250         * config/rl78/rl78.c (rl78_expand_prologue): Always select
21251         register bank 0 at the start of an interrupt handler.
21252         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
21253         MDBH registers.
21255 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21257         * config/aarch64/aarch64-simd.md
21258         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
21259         (aarch64_simd_bsl): Likewise.
21260         (aarch64_vcond_internal<mode>): Likewise.
21261         (vcond<mode><mode>): Likewise.
21262         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
21263         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
21265 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21267         * config/aarch64/aarch64-builtins.c
21268         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
21270 2013-01-08  Martin Jambor  <mjambor@suse.cz>
21272         PR debug/55579
21273         * tree-sra.c (analyze_access_subtree): Return true also after
21274         potentially creating a debug-only replacement.
21276 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21278         PR middle-end/55890
21279         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
21281         PR tree-optimization/54120
21282         * tree-vrp.c (range_fits_type_p): Don't allow
21283         src_precision < precision from signed vr to unsigned_p
21284         if vr->min or vr->max is negative.
21285         (simplify_float_conversion_using_ranges): Test can_float_p
21286         against CODE_FOR_nothing.
21288 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21289             Richard Biener  <rguenther@suse.de>
21291         PR middle-end/55851
21292         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
21293         types instead of just INTEGER_TYPE types.
21295 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
21297         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
21298         TF_SIZE): Define.
21300 2013-01-07  Steve Ellcey  <sellcey@mips.com>
21302         PR target/42661
21303         * config/mips/mips.opt: Change mad to mmad to match documentation.
21305 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21307         PR target/55897
21308         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
21309         .progmemx.data now.
21311 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21313         PR target/55897
21314         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
21315         (avr_addrspace_t): Add .section_name field.
21316         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
21317         array size.
21318         (avr_addrspace): Same.  Initialize .section_name.  Remove last
21319         NULL entry.  Put __memx into .progmemx.data.
21320         (progmem_section_prefix): Remove.
21321         (avr_asm_init_sections): No need to initialize progmem_section.
21322         (avr_asm_named_section): Use avr_addrspace[].section_name to get
21323         section name prefix.
21324         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
21325         retrieve the progmem section.
21326         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
21327         boundary to run over avr_addrspace[].
21328         (avr_register_target_pragmas): Ditto.
21330 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
21332         * varasm.c (output_constant_def_contents): For asan_protect_global
21333         protected strings, adjust DECL_ALIGN if needed, before testing for
21334         anchored symbols.
21335         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
21336         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
21337         normal decls.
21338         (output_object_block): For asan protected decls, emit asan padding
21339         after their contents.
21340         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
21341         (asan_finish_file): Test it here instead.
21343 2013-01-07  Nick Clifton  <nickc@redhat.com>
21344             Matthias Klose  <doko@debian.org>
21345             Doug Kwan  <dougkwan@google.com>
21346             H.J. Lu  <hongjiu.lu@intel.com>
21348         PR driver/55470
21349         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
21351         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
21353         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
21355         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
21357         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
21359 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21361         PR target/54461
21362         * doc/install.texi (Cross-Compiler-Specific Options): Document
21363         --with-avrlibc.
21365 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
21367         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
21368         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
21369         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
21370         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
21371         vqmovun_high_s64): Fix source operand number and update copyright.
21373 2013-01-07  Richard Biener  <rguenther@suse.de>
21375         PR middle-end/55890
21376         * gimple.h (gimple_call_builtin_p): New overload.
21377         * gimple.c (validate_call): New function.
21378         (gimple_call_builtin_p): Likewise.
21379         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21380         Use gimple_call_builtin_p.
21381         (find_func_clobbers): Likewise.
21382         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
21383         (strlen_optimize_stmt): Likewise.
21385 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21387         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
21388         (vld1q_dup_*): Likewise.
21389         (vld1_*): Likewise.
21390         (vld1q_*): Likewise.
21391         (vld1_lane_*): Likewise.
21392         (vld1q_lane_*): Likewise.
21394 2013-01-07  Richard Biener  <rguenther@suse.de>
21396         * lto-streamer.h (LTO_minor_version): Bump to 2.
21398 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21400         * config/aarch64/aarch64-protos.h
21401         (aarch64_const_double_zero_rtx_p): Rename to...
21402         (aarch64_float_const_zero_rtx_p): ...this.
21403         (aarch64_float_const_representable_p): New.
21404         (aarch64_output_simd_mov_immediate): Likewise.
21405         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
21406         move immediate case.
21407         * config/aarch64/aarch64.c
21408         (aarch64_const_double_zero_rtx_p): Rename to...
21409         (aarch64_float_const_zero_rtx_p): ...this.
21410         (aarch64_print_operand): Allow printing of new constants.
21411         (aarch64_valid_floating_const): New.
21412         (aarch64_legitimate_constant_p): Check for valid floating-point
21413         constants.
21414         (aarch64_simd_valid_immediate): Likewise.
21415         (aarch64_vect_float_const_representable_p): New.
21416         (aarch64_float_const_representable_p): Likewise.
21417         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
21418         (aarch64_output_simd_mov_immediate): New.
21419         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
21420         (*movdf_aarch64): Likewise.
21421         * config/aarch64/constraints.md (Ufc): New.
21422         (Y): call aarch64_float_const_zero_rtx.
21423         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
21425 2013-01-07  Richard Biener  <rguenther@suse.de>
21427         PR tree-optimization/55888
21428         PR tree-optimization/55862
21429         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
21430         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
21431         not if it is contained therein.
21433 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21435         * config/avr/t-avr: Typo.
21437 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21439         PR55243
21440         * config/avr/t-avr: Don't automatically rebuild
21441         $(srcdir)/config/avr/t-multilib
21442         $(srcdir)/config/avr/avr-tables.opt
21443         $(srcdir)/doc/avr-mmcu.texi
21444         (avr-mcus): New phony target to build them on request.
21445         (s-avr-mlib, s-avr-mmcu-texi): Remove.
21446         * avr/avr-mcus.def: Adjust comments.
21448 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
21450         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
21452 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21454         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
21456 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21458         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
21460 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
21462         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
21463         to generate profiling.
21464         * config/rs6000/aix64.h (LIB_SPEC): Same.
21466 2013-01-04  Andrew Pinski  <apinski@cavium.com>
21468         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
21469         New function.
21470         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
21472 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
21474         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
21475         unconditionally.
21476         (ix86_expand_move): Ditto.
21477         (ix86_zero_extend_to_Pmode): Ditto.
21478         (ix86_expand_call): Ditto.
21479         (ix86_expand_special_args_builtin): Ditto.
21480         (ix86_expand_builtin): Ditto.
21482 2013-01-04  Richard Biener  <rguenther@suse.de>
21484         PR tree-optimization/55862
21485         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
21486         translating them through PHI nodes.
21488 2013-01-04  Martin Jambor  <mjambor@suse.cz>
21490         PR tree-optimization/55755
21491         * tree-sra.c (sra_modify_assign): Do not check that an access has no
21492         children when trying to avoid producing a VIEW_CONVERT_EXPR.
21494 2013-01-04  Marek Polacek  <polacek@redhat.com>
21496         PR middle-end/55859
21497         * opts.c (default_options_optimization): Clarify error message.
21499 2013-01-04  Richard Biener  <rguenther@suse.de>
21501         PR middle-end/55863
21502         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
21503         reassociation.
21505 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21507         PR target/53789
21508         * config/pa/pa.md (movsi): Revert previous change.
21509         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
21510         references.
21512 2013-01-03  Richard Henderson  <rth@redhat.com>
21514         * config/i386/i386.c (ix86_expand_move): Always assign to op1
21515         after eliminating TLS symbols.
21517 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
21519         PR bootstrap/50167
21520         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
21521         * graphite-poly.c (debug_gmp_value): Likewise.
21523 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
21525         PR target/55712
21526         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
21527         selected code model, define __code_mode_small__, __code_model_medium__,
21528         __code_model_large__, __code_model_32__ or __code_model_kernel__.
21529         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
21530         xchg temporary register with %k.  Declare temporary register as
21531         early clobbered.
21532         [__x86_64__]: For medium and large code models, preserve %rbx register.
21534 2013-01-03  Richard Biener  <rguenther@suse.de>
21536         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
21537         (dump_subscript): Adjust.
21538         (finalize_ddr_dependent): Do not dump redundant info.
21539         (analyze_siv_subscript): Adjust.
21540         (subscript_dependence_tester): Likewise.
21541         (compute_affine_dependence): Likewise.
21543 2013-01-03  Richard Biener  <rguenther@suse.de>
21545         Revert
21546         2013-01-03  Richard Biener  <rguenther@suse.de>
21548         PR tree-optimization/55857
21549         * tree-vect-stmts.c (vectorizable_load): Do not setup
21550         re-alignment for invariant loads.
21552         2013-01-02  Richard Biener  <rguenther@suse.de>
21554         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21555         invariant load do not generate a vector load from the scalar location.
21557 2013-01-03  Richard Biener  <rguenther@suse.de>
21559         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
21560         for not vectorizing.
21561         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
21562         not build INDIRECT_REFs, call get_name once only.
21563         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
21564         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
21566 2013-01-03  Richard Biener  <rguenther@suse.de>
21568         PR tree-optimization/55857
21569         * tree-vect-stmts.c (vectorizable_load): Do not setup
21570         re-alignment for invariant loads.
21572 2013-01-03  Richard Biener  <rguenther@suse.de>
21574         PR lto/55848
21575         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
21576         prefer a built-in decl.
21578 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21580         * gcc.c (process_command): Update copyright notice dates.
21581         * gcov.c (print_version): Likewise.
21582         * gcov-dump.c (print_version): Likewise.
21584         PR rtl-optimization/55838
21585         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
21586         iv0.step, iv1.step and step.
21588 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21589             Marc Glisse  <marc.glisse@inria.fr>
21591         PR tree-optimization/55832
21592         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
21593         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
21594         integer_{one,zero}_node.
21596 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
21598         PR debug/54402
21599         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
21600         * var-tracking.c (reverse_op): Don't add reverse ops to
21601         VALUEs that have already
21602         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
21604 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
21606         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
21608 2013-01-02  Teresa Johnson  <tejohnson@google.com>
21610         * dumpfile.c (dump_loc): Print filename with location.
21611         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
21612         new location_t parameter to emit complete unroll message with
21613         new dump framework.
21614         (canonicalize_loop_induction_variables): Compute loops location
21615         and pass to try_unroll_loop_completely.
21616         * loop-unroll.c (report_unroll_peel): New function.
21617         (peel_loops_completely): Use new dump format with location
21618         for main dumpfile message, and invoke report_unroll_peel on success.
21619         (decide_unrolling_and_peeling): Ditto.
21620         (decide_peel_once_rolling): Remove old dumpfile message subsumed
21621         by report_unroll_peel.
21622         (decide_peel_completely): Ditto.
21623         (decide_unroll_constant_iterations): Ditto.
21624         (decide_unroll_runtime_iterations): Ditto.
21625         (decide_peel_simple): Ditto.
21626         (decide_unroll_stupid): Ditto.
21627         * cfgloop.c (get_loop_location): New function.
21628         * cfgloop.h (get_loop_location): Declare.
21630 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21632         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
21633         NULL.
21635 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21637         PR middle-end/55198
21638         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
21639         BLKmode objects when EXPAND_MEMORY is specified.
21641 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
21643         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
21644         in loop predicate.
21645         (fold_builtin_cpu): Do not share cpu model decls across statements.
21647 2013-01-02  Jason Merrill  <jason@redhat.com>
21649         PR c++/55804
21650         * tree.c (build_array_type_1): Revert earlier change.
21652 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
21654         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
21655         "cortex-a57".
21656         * config/aarch64/aarch64-tune.md: Re-generate.
21658 2013-01-02  Richard Biener  <rguenther@suse.de>
21660         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21661         invariant load do not generate a vector load from the scalar location.
21663 2013-01-02  Richard Biener  <rguenther@suse.de>
21665         PR bootstrap/55784
21666         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
21667         * configure: Regenerate.
21669 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21671         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
21672         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
21673         (expand_builtin_int_roundingfn_2): Keep the original target around
21674         for the fallback case.
21676 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
21678         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
21679         to be clear for sign changes.
21681 2013-01-01  Jan Hubicka  <jh@suse.cz>
21683         * ipa-inline-analysis.c: Fix formatting.
21685 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
21687         PR tree-optimization/55831
21688         * tree-vect-loop.c (get_initial_def_for_induction): Use
21689         gsi_after_labels instead of gsi_start_bb.
21691 Copyright (C) 2013 Free Software Foundation, Inc.
21693 Copying and distribution of this file, with or without modification,
21694 are permitted in any medium without royalty provided the copyright
21695 notice and this notice are preserved.